Breast Cancer

Which Treatments Are Worth Doing, and for Whom?

This is a nonprofit site whose authors have no commercial interests.

• Introduction and overview; thinking critically about statistics
• Chemotherapy: How much difference does it make, and for whom?
• Taxol chemotherapy: Under what circumstances are the pluses greater than the minuses?
• An alternative to Taxol
• Radiation: For whom is it appropriate?
• Endocrine approaches: rationale and possible new paradigms
• Various updates and addenda
• Book recommendations for the newly-diagnosed
• Links

July 2010

This site provides an analysis of treatment decisions for postmenopausal, estrogen receptor-positive breast cancer. If you or someone you care about is postmenopausal and has an estrogen receptor-positive type of breast cancer, you may find it helpful. It is also meant to alert busy researchers to the thought-provoking research of colleagues. The author hopes it will inspire collaboration towards answering some of the questions raised, particularly in regards to complex hormone interactions in women.

This literature review is not written by a doctor—it is the result of a science writer putting aside a month of her life and finding out everything she can for a relative affected by breast cancer. So like anything else on the Web, it is not to be taken as gospel. However, it may save research time, point people towards interesting articles in the peer-reviewed scientific literature, and bring up questions patients may want to discuss with their doctor. Obviously, it is not meant as medical advice—it is just meant to prevent duplication of effort in this complicated field.

To help patients and family members decide whether this is likely to be applicable for them, the woman this is written for has the following characteristics:
• early sixties (postmenopausal)
• highly estrogen and progesterone receptor-positive tumor, 2.5 cm
• lymph node involvement: 3 of 11 lymph nodes
• Her2-neu negative
• no metastases
• pretty good health otherwise

Introduction and Overview; Thinking Critically about Statistics

Here is an interim report on the conclusions I have drawn regarding your cancer treatment. My main sources have been publications from peer-reviewed journals available through the PubMed system, and a report called the Moss Report which has helped me synthesize the information and pointed me towards some peer-reviewed publications to track down.

I will review your treatment plan chronologically, as it is in that order that you must make decisions. I have not yet done any reading about complementary or alternative therapies. So far I can make some comments on the following: prophylactic contralateral mastectomy (which you’ve already done), chemotherapy, taxol chemotherapy, and radiation. I will make preliminary comments on the aromatase inhibitor phase and potential alternatives. Most of this was written before you finished your last AC chemotherapy this week, so that will be included too.

First, from what I have read, I will say that your doctor is providing you with the "standard of care." Her recommendations fit with the various decision trees that one finds. The various elements used by the clinic also seem fairly up to date. For example, they are not using several older types of chemo which seem to do quite a lot of harm and not much good. (For example, if I remember right, the M part of the CMF regimen seemed to be singlehandedly responsible for a whole lot of "chemo-brain.") The AC regimen for chemo seems to be the current consensus on what has a more favorable benefit/harm ratio. Also, they are using 60 mg/m(2) of Adriamycin, in line with a large study (Henderson 2003, PubMed I.D. #12637460) which showed no benefit from higher doses.

From what I understand, the only part of your proposed treatment which is ambiguous under the "standard of care" is the radiation. Decision trees say that radiation can be considered in a case such as yours, but do not seem to argue one way or another. I believe your doctor is making her recommendation for it based on factors such as angiolymphatic invasion, perineural invasion, and a smaller clean margin than they wanted.

Now I would like to make comments on the "standard of care." The "standard of care" is very useful in making sure that patients in smaller regional hospitals or with doctors who are not keeping up to date get the same treatment as others. It takes the current consensus and makes it very clear, so that deviating from that consensus requires explanation. For example, a doctor who does not offer a woman tamoxifen or Arimidex because she is 70 and "too old" would be reminded that that is no longer standard practice. Or it can remind physicians that lumpectomy with radiation is currently considered as effective as mastectomy for very early cancers without node involvement. This can improve care, for example by slowing down surgeons who are excited about doing as much surgery as possible. It also reminds people of the latest research, for example the recent results showing Arimedex to be just as effective as Tamoxifen or even slightly more so.

That said, the "standard of care" also has some drawbacks. In some ways, it is the worst of both worlds on one measure: it simultaneously discourages use of newer treatments which have not yet been proven in randomized controlled trials, while encouraging continued use of treatments which are "grandfathered" because they were determined effective based on less than RCT's yet it is now considered unethical not to prescribe them, despite the sometimes rather tenuous evidence for their effect. Chemo is the prime therapy which falls into this category. For the past ten years or so it has been considered unethical to do an RCT comparing chemo to no chemo. It is considered a "given" that chemo works, but often based on sub-optimal trials. So we have primarily studies comparing one type of chemo to another, with no way to tell how much either type of chemo is improving survival versus placebo. The last large study to compare chemo to no chemo is the Trialists meta-analysis, which I will refer to later.

Here is a further blind spot of the "standard of care": treatments are often chosen based on whether the tumor shows a measurable response (shrinkage) and something called "disease-free survival." This seems rather logical, doesn't it? After all, one wants to shrink the tumor, and one doesn't want to get sick again. The idea is to get rid of the cancer and live a long, healthy life.

However, there is a major problem with this approach: the same treatment which shows great tumor shrinkage and a delay in recurrence often leads to NO benefit in overall survival. Radiation is particularly bad in this respect, if I remember correctly. For example, in some cases radiation can result in a 30% drop in the number of women who have recurrences at the five-year mark. But they don't then live any longer!

An understandable response at this point would be to say "I don't really care as much about living longer as about being healthy during the time I have. Even if I live the same length of time as that woman who didn't have treatment, I don't want to spend the last few years of it battling a cancer recurrence."

The problem with this reasoning is that it is not clear that when the overall survival length is the same, "less treatment plus recurrence" versus "more treatment without recurrence" leads to a worse quality of life. Let me here point out that the current tumor you just had removed had been growing in your breast for the past 8 to 10 years, without impacting your quality of life in the slightest. And simple statistics do not take into account the long-term impacts of some of these treatments, such as the permanent lung problems which can come with radiation, and the potentially permanent cognitive impairment which can come with chemo (although it seems that some of the older types of chemo were more notorious for this). Some of the negative effects of treatments are reflected in the survival rates (such as the increased heart attacks in women who have radiation, which in some groups more than negate the benefit of the radiation), but more chronic things such as lung problems are not reflected.

This issue is starting to be addressed by number-crunchers who count "survival-years free from side effects of either treatment or disease." These types of calculations almost always reduce the perceived benefit of treatments.

Finally, since there are now quite a few large statistical studies which show that disease-free survival does not correlate (yes, you read "does not correlate") with overall survival, I will base my estimation of benefit from here out on overall survival, not other markers such as "response rate" (number of women in whom a tumor shrinks) or women still disease-free at a certain point in time.

Now for perhaps the largest problem with the "standard of care": it is not fine-tuned according to some things that in the past few years have been shown to be quite important. Imagine the decisions that go into determining the "standard of care" as being determined by factors on an X axis and a Y axis. The X axis is stage and severity of the cancer, and the Y axis is all other factors which could have an impact on whether a treatment is likely to do more harm or good. (Okay, I know mathematically speaking you’re not supposed to have more than one factor per axis, but humor me here.) The problem is, there is currently practically no Y axis. Choice of therapies is determined by stage and severity, with little focus on other factors which have recently been shown to be predictive.

We do currently have a couple of things on the Y axis. One of them is that hormone-receptor status is paid attention to in one regard—if you have a hormone-receptor-positive tumor, you will be offered a hormone-based treatment after everything else is done (currently the favorite is Arimedex, which has supplanted tamoxifen in this role). Women with hormone-receptor-negative tumors will not be offered Arimedex. This is sensible and good. One other factor on the Y axis is Her2 status. Because your tumor was not Her2-positive, you were not offered Herceptin, because they knew it wouldn't do much good and it has a fair chance of doing harm. This is good and logical thinking. We just need more of it, regarding some other factors.

The most glaring omission in my mind is the intersection between age, premenopausal versus postmenopausal, and hormone-receptor status on the efficacy of chemotherapy and radiation. I will discuss this more specifically in those sections. But just to give the summary, post-menopausal women with hormone receptor-positive cancers and pre-menopausal women with hormone receptor-negative cancers react very differently. It is almost a different disease. Yet they are currently lumped together into the statistics which are used for decision-making.

Let me give you just a quick analogy. Suppose I tell you that 84% of women are happy with the results when they use bleaching creams to treat liver spots. Sounds like a pretty good reason to try a bleaching cream if you're troubled by liver spots, right? But supposed you're black/African American. Now I drop the bombshell on you by breaking that statistic down further: 98% of white women are happy with their results, and 100% of dark-skinned women (who only make up 10 or 20% of the population) are unhappy with the resulting discoloration of their skin. The overall statistic hid the fact that not only was the treatment not right for you, you would have been better off without it.

We’ve mentioned age, premenopausal versus postmenopausal status, and hormone receptor status as factors that need to be on the Y axis for chemotherapy (including taxol chemotherapy) and radiation. These factors are important individually and in combination. They can have surprising interactions. For example, it is thought that in premenopausal women who have estrogen receptor-positive cancers, part of the effect of the chemo is that their ovaries are taking such a hit that it in effect becomes an endocrine (hormone) treatment. Researchers who are really meticulous in their logic even suggest that the effect of chemo itself can only be determined by studying estrogen receptor-negative tumors in women (otherwise part of the effect is interference by the hormonal effect of shutting down the ovaries).

These factors come into play in the National Surgical Adjuvant Breast and Bowel Project B-27, which is quite relevant to you because they used the same chemo you're getting (AC) plus a taxane (docetaxel, not Taxol, but still, pretty close). In a brilliant commentary, Colleoni (Clinical Cancer Research Oct. 1, 2004) says, "The results across all of the age groups are not informative concerning chemosensitivity of the use of taxanes for estrogen receptor-positive disease." In other words, just because they found 26.1% remission when they added taxotere instead of 13.7% doesn't mean that has anything to do with you. "It is likely that more than two thirds of the patients were premenopausal, and differential endocrine effects of the chemotherapy regimens contributed to the finding in the estrogen receptor-positive cohort. Furthermore, because tamoxifen was given concurrently with chemotherapy, the tamoxifen duration was either 24 weeks or 12 weeks according to treatment assignment, making comparisons for the receptor-positive cohort quite problematic." In other words, it wasn't just a study of adding a taxane, it was also a study of twice as much tamoxifen!! So this study, though large, really can't be used as a justification for Taxol.

Another factor which I don't believe is adequately factored into treatment decisions is the impact of bilateral mastectomy. The consensus from various studies seems to be that though one cannot remove all breast tissue in a mastectomy, the chance of recurrence is roughly correlated with the amount of tissue remaining. So lets's say a 90% tissue removal reduces your chance of recurrence in that breast by 90%.

Now, we have many studies to look at which compare recurrence rates for this chemo versus that chemo or for lumpectomy with radiation versus without radiation, etc. And in these studies, a certain percentage of the recurrences are in the "ipsilateral" breast, a certain percentage in the "contralateral" breast, and a certain percentage are distant.

I would have to read back through the papers to refine these numbers, but let's say that a third are ipsilateral recurrences (especially in the lumpectomy rather than mastectomy studies), a third are contralateral, and a third are distant. So it seems reasonable to say that by removing both breasts, you are nearly wiping out two of those categories. Specifically, the chance of recurrence should be between one third and two thirds lower than in women getting lumpectomy or single-side mastectomy. Let's simplify by taking 50% as the number. (I will refine these numbers when I go back and re-read the relevant papers.)

This fits with studies—for example, we have a study of double-sided mastectomy (“contralateral prophylactic mastectomy”) versus single-sided mastectomy in which the women who had double-sided had 55% disease-free survival at 15 years and the single-sided women had only 28%. For actual overall survival, it was 64% still alive at 15 years instead of 48%. (Peralta 2000, PubMed I.D. 11182394)

The study was small, only 64 bilateral mastectomies and 182 matched controls, but it was big enough to get impressive statistically significant results (partly because of the long follow-up time). If we wanted to make the numbers look more remarkable—if this were a product we were trying to sell—we could play our “hazard ratio” game (which you will see later), and say that the hazard ratio was .625 for disease recurrence and .69 for death, or we could make that sound even better by saying the number of disease-free women at 15 years was almost double with bilateral mastectomy. We could say that there was a 38% drop in the number of women having recurrences, and a 31% drop in the number of women dying. (This is how numbers for substances such as Taxol are often reported in the press.)

So, I believe that when you read that a particular treatment post-mastectomy will result in a reduction in recurrence, you need to cut those recurrence numbers they’re using by 30-40% because of your bilateral mastectomy. You thus also need to cut the expected benefit from a treatment by 30-40%. So if anyone tells you that you should consider a treatment because it will reduce recurrence by a certain amount (for example, that radiation will reduce recurrence by 30%), you need to lower that benefit by a third or more. And for any expected survival benefit from a treatment, you should cut it by a little under a third.

Also, remember to think about the total percentages that are being discussed in the first place. What percentage do you require to put yourself through a procedure? Most doctors say they require upwards of a 10% difference before recommending something (though their prescribing actions actually aren't in line with their words). So if I tell you that without the bilateral mastectomy, you have a 50% chance of recurrence, and with it, you have a 30% chance of recurrence, a treatment that will drop that 30% by a third (to 20%) might be worth some hassle. But if you would have a 25% chance of recurrence with a single-sided mastectomy, and the bilateral mastectomy drops that to 15%, are you still willing to go through the same treatment if it is lowering your risk from 15% to 10%?

Then there's the "apples and oranges" issue. It is becoming very clear that (1) tumor response and (2) survival or lack of recurrence don't have much to do with each other. In fact, if you look just at the big groups, they can be inversely correlated. In Faneyte's study in the 2003 British Journal of Cancer, for example, the researchers unequivocably state, "Tumour response and marker expression did not predict disease-free or overall survival." In Colleoni's 2004 paper, he goes so far as to set up an inverse relationship between disease-free survival and responsiveness of the tumor. Basically, the nasty hormone receptor-negative tumors in young women responded great to the chemo and shrank like nobody's business—the faster replicating the tumor, the better. But despite the exciting shrinkage, the women got cancer again quite quickly. (Only 4 of 10 were free of cancer 5 years later.) Meanwhile, the women with estrogen receptor-positive tumors were ticking along, wondering why they went through all that treatment that didn't shrink their tumor—but only 2 or 3 out of 10 of them had recurrences by 5 years. Here it is in technical terms: "Despite the significantly higher incidence of pathological complete remission and node-negative status achieved by preoperative chemotherapy for patients with estrogen receptor and progesterone receptor absent disease, the disease-free survival was significantly worse for this cohort compared with the low/positive expression cohort (4-year disease-free survival 41% versus 74%)." So the moral of this story is, shrinking tumors does not translate directly into keeping you alive. In fact, a treatment that shrinks tumors may be just the thing for somebody else, not you.

Now, for some other factors which ought to be considered when deciding on treatment. These need to be considered on the Y axis along with all the stuff on the X axis (nodular involvement, extracapsular extension, size of clean margin, etc.).

One of these is which breast the cancer is in. This should be a factor when deciding whether to do radiation or not. Studies show that when women have radiation to the left breast, they have more resultant heart damage. The resulting heart problems can actually result in a decreased overall survival rate for women who have radiation to the left breast.

I imagine that the medical clinic is up to date on the various techniques of aiming radiation to try to minimize how much of it is aimed at the heart. But it seems to me that this sort of aim would be extremely difficult when irradiating a chest wall rather than a post-lumpectomy breast (which at least sticks out a little). You might want to ask your doctor about this.

Another factor to consider is which quadrant of the breast the tumor was in. Studies show a better prognosis for tumors on the outside (like yours) rather than the inside of the breast. This is another argument for a more gentle approach, remembering that it is yet another factor reducing the likelihood of recurrence.

Yet another factor: Ki-67 marker. In Colleoni's 2004 paper, women like you with Ki-67 under 25% (yours is 20%) had a 8.1% pathological complete remission in response to the chemo, rather than a higher 22.3% response for the women with Ki-67 greater than or equal to 25. So this is yet another factor saying that any tumor-shrinking that chemo is doing is likely to be in someone else, not you. (Again, just a reminder that tumor shrinkage doesn't seem to have much if anything to do with survival, so lack of tumor shrinkage doesn't necessarily mean much good or bad—but if the chemo isn't doing anything, why go through it?) Faneyte (2003) cites a another study (Wang et al 2002) in which the high-Ki-67 tumors also responded more to the chemo than the low-Ki-67 ones like yours did (just like in Colleoni's paper). That study was even using doxorubicin (Adriamycin), the same chemo you're taking.

So now let's turn to chronological decision points.

Back to Contents

AC Chemotherapy (Adriamycin and Cytoxan):
How Much Difference Does It Make, and for Whom?

I consider this marginal. I won't currently argue strongly for or against it.

Here are the stats from a meta-analysis (Lancet 1998 Trialists): for premenopausal women with node-negative disease (that's not you), chemotherapy conferred a 7% survival gain. For women 50-69 years old with node-positive disease, the gain was 3%. Yes, 3%. So at ten years, survival was 49% instead of 46% because of doing chemo rather than skipping it.

In a separate section of the discussion, when discussing tamoxifen, the Trialists have this cheery thing to say about their analysis of 30,000 women, which could make one wonder why exactly one is going through chemotherapy: "The reductions in recurrence with about 5 years of tamoxifen appeared to be about the same, and again highly significant, in the absence of adjuvant chemotherapy and in its presence." (Lancet p. 1460) This would make one wonder whether it might be better to skip the chemo and just get started on the endocrine therapy.

(Update May 2005: In the May 14 issue, The Lancet published numbers from an even bigger study—based on data from 145,000 women who had taken part in 194 trials around the world—in an article entitled “Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials.” The numbers appear to be getting better over time, and one must read that study carefully (in order to determine how much effect was due to chemotherapy vs. hormonal therapy in estrogen receptor-positive women) before coming to a conclusion. That data was not available when this paper was written.)

Not all the studies in this analysis used AC-Taxol chemo (in fact, it was maximum 19 out of 47 studies). Let's assume that the new types of chemo are better than the older ones in this study—say, a third better. So make that a 4% survival gain. Or, let's be more optimistic and figure the new chemo is twice as good-so that's 6%. Now remember that by cutting off both breasts, you eliminated many spots for recurrences, so you have to cut that expected benefit by a little under a third, resulting in being back down to a 4% gain versus no chemo.

What about other types of chemo? How about, for example, epirubicin? A study (Wils 1999) showed 28% less recurrence. Great, right? Not so fast. This should be beginning to sound familiar: "No evidence was found, however, for an improvement in overall survival."

Well, if you're not going to live longer or much longer, do you at least get a better quality of life? There is a meta-analysis of quality-adjusted survival (Gelber RD) based on 3920 patients 50 years or older with node-positive breast cancer (i.e. just like you). Ready? "Within 7 years of follow-up the modest benefit of relapse-free survival and overall survival for patients who received chemotherapy just balanced the costs in terms of acute toxic side-effects. Chemotherapy-treated patients gained an average of 5.4 months of relapse-free survival and 2 months of overall survival (neither statistically significant), but had to receive cytotoxic treatment for between 2 and 24 months to achieve these gains … Interpretation: Within 7 years of follow-up, adjuvant chemoendocrine therapy did not provide more quality-adjusted survival time than tamoxifen alone for women aged 50 or older with node-positive breast cancer."

What about getting down to specifics of estrogen-receptor status? Colleoni (Oct. 1, 2004) showed that the chemo had about a fourth of the chance of shrinking the tumor as in ER-negative women, for whatever that's worth.

There is other evidence pointing to that same thing, that chemo works better in women whose tumors are estrogen-receptor negative. In a report from the National Surgical Adjuvant Breast and Bowel Project Protocol B-27, which used AC chemo like yours, estrogen receptor-negative tumors also had a higher response rate (16.7%) than estrogen receptor-positive tumors (8.3%). This was another big study, with 2,411 women. Again, we can't confuse response rate with increased survival, but it is discouraging that it's only doing something in half as many of the women of your type (Bear 2003) . And as you know, the treatment is not benign: in this trial, for example, about 20% dropped out because of side effects. You get lovely stuff like this: "Of the 2,400 patients who received AC and who were assessable for toxicity, there were three deaths for which treatment could not be excluded as a contributing factor (one from a cerebral hemorrhage while on tamoxifen, one from liver failure thought to be secondary to breast cancer recurrence, and one from suicide), and 247 patients (10.3%) experienced at least one grade 4 adverse event episode." Things got even worse when they started giving docetaxel (the Taxol relative).

That was the B-27 trial. In the even larger B-20 trial, which used CMF chemo instead of AC chemo, they discuss the intersection between age and estrogen receptor status. They say, “Older women tend to have higher tumour estrogen-receptor concentrations and are more likely to benefit from tamoxifen than chemotherapy; in younger women, the converse is true.” (Fisher 2004 pmid 15351193) This is certainly relevant for you, since your tumor’s estrogen receptor positivity was 95%.

What else do we know about chemo? In a study by Faneyte (well, actually by van de Vijer, but Faneyte is probably his professor and gets all the credit), they did regular chemo or high-dose, and the high dose wasn’t any more effective. (However, this was FEC chemo, not AC, so it's not entirely parallel, and the women had to be under age 56, which changes things too). (Faneyte 2003 Clinical Cancer Research)

Henderson (2003, pmid 12637460) did this same study with AC chemo, the kind you’re taking, and found “no evidence of a doxorubicin dose effect” for the doses of Adriamycin of 60 (which you’re getting), 75, or 90 mg/m(2). So more was not better, at least not from what they could tell.

In Colleoni's letter to the editor (2004) in the Annals of Oncology, he discusses the International Breast Cancer Study Group (IBCSG), in which compared with a conventional dose regimen, a high-dose regimen was effective primarily in a subgroup of young women (under 40 years old).

In the Annals of Surgical Oncology (2003) , H. Bear discusses Petit et al.'s study comparing one blast of mitoxantrone chemo right after treatment to no chemo. Baer says "The addition of perioperative chemotherapy in this trial did not significantly improve overall or disease-free survival." He has several explanations for this, including that two thirds of the patients in the trial were post-menopausal and "post-menopausal patients have generally benefitted least from chemotherapy…"

In his editorial "Adjuvant Chemotherapy in Breast Cancer: Back to the Future," Colleoni (2003) discusses the 7-year Austrian Breast and Colorectal Cancer Study Group study, which studied chemo versus low-dose chemo, and described the authors as "almost being apologetic for not having shown a difference between the two treatment groups."

All this bad news! Aren’t there any studies where chemo makes more of a difference? Yes, there are some other types of cancers (in particular some rare childhood cancers) where chemo is quite miraculous. And for breast cancer, there’s one study that shows a pretty significant benefit, the Ludwig trial. Problem is, it’s pretty old (they started in 1978). Also, they used CMFp chemo, which isn’t what you’re getting (I’m not up on all the reasons they’ve switched to AC, but I do know that the “M” component carried a risk of long-term cognitive impairment— Schagen 2002 pmid 12196364 ).

The Ludwig Trial III only had 463 women, but still they got a significant survival difference: 32% survival (at 10 or 13 years, I’m not sure which) for surgery only, 36% survival with Tamoxifen, and 48% survival with Tamoxifen plus CMFp chemo.

I’m not sure what to make of that, since the numbers are a lot better than other studies. However, there is another study of CMF chemo that is available online (so I don’t have to go to the library and touch an actual book), so let’s look at it. It’s also an older CMF study, this time by Bonadonna (1995 pmid 7877646) . It too gets impressive results, and at an impressive 20-year follow-up. At 20 years, only about 77% as many women in the chemo group died (they got a 27% drop in deaths). That’s pretty substantial—the survival rate went from 24.5% to 33%, a tidy sum.

But as always, the devil is in the details. In summing up, Bonadonna had this to say: “With the exception of postmenopausal women, a benefit from adjuvant chemotherapy was evident in all subgroups of patients.”

Finally, I would like to include a piece from an editorial in the Journal of Clinical Oncology just two months ago (Schott 2004, pmid 15505274). It’s talking about women over 65, but otherwise similar to you:

" Taken together, do the results of the previously reported studies and from the current FASOG clinical trials provide conclusive evidence that all ER-positive, node-positive elderly women should receive adjuvant chemotherapy in addition to tamoxifen? If so, should the chemotherapy regimen include an anthracycline? We think not. First, the absence of a survival benefit in this study, and the dearth of data from other trials addressing this age group, speaks against widespread adoption of this approach. Although other studies have reported a modest survival benefit in all ER-positive, node-positive postmenopausal women, these studies have not been directly targeted towards the elderly, nor have retrospective subsets been presented from trials with women older than 65 years. One could argue that delay of recurrence is, by itself, an end point that justifies treatment, but this issue remains controversial. Second, these results must be placed in the context of ongoing progress in endocrine treatment of postmenopausal women in the modern era. Although still controversial, inhibition of aromatase activity has recently been shown in three published trials to significantly reduce the annual odds of recurrence by as much as 30% in ER-positive women. It seems intuitive that one can achieve greater benefit with less toxicity from the sequential use of hormonal treatments than by the addition of chemotherapy, although longer follow-up of these studies is needed. "

Summary for chemo: If you truly want to practice "evidence-based medicine," then the most benefit you should hope for from chemo would be a 3% increase in survival at 10 years. You can expect about 2 months of additional survival, but not a quality-adjusted increase in survival time.

It is up to you whether a potential 3% gain is worth it. You may want to see how it goes—if the chemo continues to be fairly manageable, continue it, but if the cumulative effect begins to significantly impact your current quality of life, stop it.

Based on the positive effects seen with endocrine treatments such as tamoxifen and Arimidex, there would be a theoretical argument that it would be just as good or better to get started on an endocrine treatment and skip the rest of the chemo. Again, this flies in the face of the current thinking that chemo is necessary to kill any distant micro-metastases. But if you just want to base your decisions on the numbers, not on theories, then it would be a highly logical course of action to go straight to the endocrine treatment.

Back to Contents

Taxol Chemotherapy:
Under What Circumstances Are the Pluses Greater Than the Minuses?

First, let’s go over the reasons Taxol is being proposed. Your doctor is suggesting Taxol because two large studies showed that it makes a difference. In one of them, only 82% as many bad things happened (i.e. at five years, or to be more exact, 69 months, there was an 18% reduction in the number of women who had kicked the bucket). That study had 3121 women, so the result was actually statistically significant ((i.e. there’s hardly any probability that the difference was due to chance—the probability of it being due to chance is right up there with pigs flying (p=0.001)). That study is called CALGB 9344 (Henderson 2003, pmid 12637460) . The primary investigator is at UCSF.

At the time Taxol got approved, the numbers were looking even more encouraging. The package insert says of women in the Taxol arm of the study, “They also had a 26% reduction in the risk of death.” (That was the analysis at 30.1 months. You’ll see that number a lot in older articles about Taxol.) That’s the sort of thing that makes the news and sticks in people’s minds.

In the other big study, in 3060 women, also a nice big number, the hazard ratio was 0.94 (i.e. with the taxol, 6% fewer women had died at the 5-year point). That’s good news for Taxol, although in this study, unlike in the other one, the results could be due to chance without any pigs taking flight (0.78-1.12, p=0.46, which means that they can’t be sure the real result is exactly .94, it could be lower or higher). This study is called the NSABP 28 trial (Mamounas EP 2003, described in Nowak 2004 pmid 15172358) .

There are a bunch of other studies, but these are the most important, for two reasons: 1) They use the exact same treatment you’re getting (AC chemo with or without Taxol), rather than some other older chemo, and 2) they’re big enough that you can really draw some conclusions (well, at least the first one). Two other fairly big studies used docetaxel instead of paclitaxel (Taxol), and one only had 524 women, so the two mentioned above give us the most information.

So that’s why Taxol is being recommended: because one study showed 18% fewer deaths, and one study showed 6% fewer deaths. If that were the end of the story, the decision would be pretty simple. But as often happens in life, that is not the end of the story.

First of all, before we go any further, we need to be sure we’re getting our statistics straight. There are so many ways to describe things with statistics. Let’s look at the first big Taxol study, for example, the one that showed the 18% risk reduction. So, does that mean that, say, instead of a 60% chance of being alive at 5 years you have a 78% chance? NO.

That number that came out of the study, 0.82, is called a hazard ratio. Hazard ratios are a great way to make your results look better than they are when the event you’re talking about is pretty rare. There’s nothing inherently wrong with hazard ratios, it’s just that they encourage people to get confused about how exciting the results actually are.

For example, suppose that this year two people in town die of some rare disease. Next year I give everybody in town daily vitamin C, and only one person dies. Wow! My hazard ratio is now 0.5! I’ve cut the deaths in half! But not so fast. You look back in the town records, and you find that the year before nobody died of that disease, and the year before that four people died, and the year before that one, and so on… The moral of the story is that when an event is very rare, you have to be careful not to make too much of hazard ratios.

So if a disease is so rare, how do you tell whether your treatment is making any difference? You need a huge group to study so that one or two people don’t tip the balance. You can’t get away with a small study.

Let’s use an analogy: Let’s say you’re Honda, and you’re trying to decide whether to change the seat slide-back mechanism in the new 2006 models. You want to know whether your new design is too complicated for people to figure out. Luckily, getting into the car is an everyday event. So you take 100 tall people, slide the seat all the way forward, have 50 of them get in the old car and 50 get in the new car, and time how long it takes them to figure out how to slide the seat back. By the end of the day, you have a pretty good answer to your question.

Now, let’s say you’re Honda, and you’re trying to decide whether putting in side airbags really saves lives. This time, you can’t take your group of 100 drivers and put them on the road for a day, because none of them is going to crash. Even a year isn’t good enough: Let’s say one in each group of 50 crashes during the year. First of all, they might be frontal crashes, which won’t tell you anything. Second of all, for whatever result you get you won’t be able to rule out the play of chance. What you really need is thousands and thousands in each group (say 5,000 in the pre-side airbag model and 5,000 in the airbag model). Then maybe you’ll get 50 crashes per group, and 20 in each group will be side, not frontal, crashes. Now if 16 of the 20 people in the cars without airbags die, and only 2 out of the 20 people in the cars with airbags die, you know maybe you’ve got something. In technical terms, your experiment was “powered” to give you a statistically significant result.

Now, let’s get away from the statistics for a moment and touch on politics. Suppose you’re Honda and you did that experiment, but you got different results: 16 people died without airbags, and 13 died with airbags. What are you going to do? It’s not really clear whether the airbags, or chance, made the difference. Could it be that in a side collision, you either die or you don’t die—if a car crushes your car from the side fast enough, no airbag can make the difference?

On the one hand, you hate to make everybody pay a thousand dollars more for their car if airbags don’t really make a difference, money the people could be spending on food for their children. On the other hand, every life is precious. Should you make everybody pay more for their car just in case you really can save those three lives?

Unfortunately, there’s another factor that you didn’t think of. While you’ve been doing this study, you’ve gotten a lot of good press for being an innovator with side airbags, and the public is desperate to feel safer on the road. It no longer matters whether the airbags make a difference or not. The public is now demanding side airbags, and you will be seen as negligent (and possibly even get sued in case of a death) if you don’t install them.

I think the parallels to Taxol are pretty obvious. But I will now get away from the political side and go back to the raw data, so that you can make a decision based only on facts.

So, let’s go back to the CALGB trial and that 82% hazard ratio (the 18% reduction in the number of women dead at 5 years). I think it is clear now that that’s not an 18% improvement in overall survival. So our first step is to figure out exactly how much of an increase in survival it is. To do that, we take the total number of women in each arm of the study and do some division and multiplication and stuff. What we come up with is the no-Taxol group had 74.2% chance of still being alive at 5 years, and the Taxol group had a 78.2% chance of being alive at five years, for a gain of 4%. There were enough women in that trial that the gain is statistically significant, not chance. (The data I’m working with is actually data for 69-month follow-up, not exactly 5 years, from the Nowak review, because it was even more recent.)

The next thing we have to do is read the rest of the paper. We’ve seen this “apples and oranges” phenomenon with estrogen receptor-negative versus estrogen receptor-positive women, so we want to be sure that the stuff is working in estrogen receptor-positive women like you. Since Taxol is more toxic to cells which are dividing more rapidly, it would not be surprising for it to be less toxic to slower-growing estrogen receptor-positive cancers than to those nasty estrogen receptor-negative cancers. At least in the test tube, this has been shown to hold true (Dougherty 2004 pmid 15020841) . So we want to know about humans.

Unfortunately they don’t analyze the survival data for separate groups, so we’re going to have to do something less than ideal and use the disease-free survival numbers. We find the following: “…in an unplanned subset analysis, the hazard ratio… was 0.72 … for those with estrogen receptor-negative tumors and only 0.91 (95% confidence interval 0.78 to 1.07) for patients with estrogen receptor-positive tumors…” In other words, it worked better and was clearly doing something in the ER-negative women, and it’s not clear whether it was doing anything in the ER-positive women.

Once you start divvying up the data it gets harder to reach statistically significant conclusions. But to be on the safe side, we can say that this study shows that Taxol clearly increases recurrence-free survival in estrogen receptor-negative women, and it’s not clear whether it has a disease-free survival or overall survival benefit for estrogen receptor-positive women, but if it does, the most it would be would be 4%. If you make the (probably erroneous) leap of faith that the overall survival would be like the disease-free survival, then the Taxol advantage for estrogen receptor-positive women would be slightly under 2% (78.2% instead of 76.3% survival). But based on the more robust data, a 4% increase in 5-year survival rates would be the upper limit of what one should hope for from Taxol.

Now let’s move on to the other big study, the NSABP B28. In this study, they didn’t find any difference in how well it worked in ER-positive versus ER-negative women, so we look at the group as a whole. Again, the first step is to translate the “hazard ratio” of 0.94 (the 6% drop in number of women dying) into actual survival rates. What we get is an 83% survival rate without Taxol, and an 84% survival rate with Taxol. It’s not even enough difference to rule out chance (it’s not statistically significant).

I’m not sure why, but the Crown review ((Crown J, 2004, The Oncologist 9(suppl 2):24-32)) reports these numbers differently than the Nowak review. It says “There was no difference in overall survival between treatment groups (85% both groups).” The actual NSABP paper isn’t online, so I’d have to go to the library to find out.

So where do we stand with Taxol? We have one big study showing up to a 4% survival difference, yet less or possibly no difference for ER-positive women, and one big study showing a 1% survival increase that’s not even statistically significant (or no survival increase, depending on whom you ask). Yet many papers that discuss these results (e.g. Nowak 2004 pmid 15172358) conclude by saying things such as “The results of this systematic review support the use of taxanes as adjuvant chemotherapy for women with early breast cancer and involved lymph nodes.” Henderson, the lead researcher on study #1, says, “The addition of four cycles of paclitaxel after the completion of a standard course of CA improves the disease-free and overall survival of patients with early breast cancer.” Well, yeah, if 0-3% improvement in survival that could be due to chance is your idea of improvement! (Or if you have estrogen receptor-negative cancer, where it seems more likely that it may be doing something.)

I’ve gone on at some length now, but I want to include a quotation that shows that I am not the only one coming to these conclusions from the data. This quotation is from a recent review, “Emerging role of taxanes in adjuvant and neoadjuvant therapy for breast cancer: the potential and the questions.” (Goble 2003, pmid 12875604) The second author on this review is Dr. H.D. Bear of the huge National Surgical Adjuvant Breast and Bowel Project. Here’s what they say:

"Adjuvant chemotherapy has gained increasing prominence in the treatment of nonmetastatic breast cancer, producing gradual improvement in the survival of these patients. The taxanes offer great hope for adding to the progress in adjuvant treatment, but data have been conflicting. Early results of multi-center trials testing the sequential addition of paclitaxel to anthracycline-based adjuvant chemotherapy have perhaps been prematurely reported, but have already made a major impact on patterns of care for node-positive and even some node-negative patients. The early dramatic improvements in CALG 9344 are fading with time, however, and have not been confirmed by a second similar trial, NSABP B-28. Moreover, it cannot be stated with certainty whether the modest improvements observed by sequential addition of paclitaxel reflect the ability of this drug to kill anthracycline-resistant cancer cells or the increased total duration and amount of treatment." [emphasis added]

I haven’t found any studies that factor in quality of life. But remember, the results are always less impressive if you subtract the sick-on-the-sofa days from any survival benefit. And remember, too, that you had a bilateral mastectomy, which is also likely to decrease any benefit Taxol might be able to give you, since you’ve already knocked out some of the common recurrence sites. Finally, don’t forget that Taxol is a neurotoxin, and there’s also the possibility of permanently increased joint pain. (If you don’t want to scare yourself, don’t read the thread at, or Actually, do read them. I think you know to take newsgroups with a large grain of salt and make up your own mind.)

There’s consensus in the medical literature as well that Taxol can cause myalgia, arthralgia, and peripheral neuropathy (i.e. joint pain and tingling, numbess, or pain in the hands and feet), but I couldn’t find any long-term follow-up. I wrote to the author of a big taxanes review paper asking if there is any news about side effects since her review in 2003, and she weighed in with the following:

"There have been no more substantial updates of long term toxicities since this paper was published. I don’t think we will get more information on the very long term toxicities for a few years.

My clinical impression of the neurotoxicity of paclitaxel is that it becomes noticeable during treatment in most patients who have 6 or more cycles, but is less of a problem in people having only 4 cycles. It is almost always reversible, but this may be slow. There are very, very few women who have noticeable long term problems with neurotoxicity if caution is used and symptoms of neurotoxicity are assessed at each visit. It does not seem to progress when the drug is stopped, so the most important lesson is to stop it in time.

Internet message boards are always full of those few people who have real problems. People who are not affected by these side effects tend not to write in. Don’t be too discouraged by what you read there." (Nowak 2/9/05)

One further factor to consider is the interaction with post-polio syndrome. Post-polio syndrome and peripheral neuropathy are often mentioned in the same breath. There would be a logical case for avoiding neurotoxins knowing that you probably have latent damage to the substantia nigra already from childhood polio—and especially since you’re already experiencing peripheral neuropathy from the AC chemo. There is also anecdotal evidence of chemotherapy and/or surgery hastening post-polio syndrome or bringing it on in people who were previously asymptomatic. No systematic research likely to come out in this area, because the study group is so small. However, I’ve written to the head of the UCLA post-polio clinic to ask what her clinical impression is (see short paper “Post-Polio Facts”).

(Update 3/2005: Dr. Perlman referred us to an excellent post-polio specialist in San Francisco; see end of Taxol section for results.)

We’ve established that according to the evidence, one should expect anywhere from nothing, to a maximum increased 69-month survival rate of 3-4% (or probably more like 2% in your case), with some substantial unknowns regarding side effects. Is it worth it? Only you can decide that.

Breaking news: Wait! There’s more! We may have our answer! There may be a rational, high-tech, numbers-based way to know whether Taxol is likely to do any good! I missed the significance of a study, and wouldn’t you know it, it was the one that may give us an answer. The study is called “Estrogen receptor expression and sensitivity to paclitaxel in breast cancer.” It comes out of Northwestern, May 2004 (Dougherty, pmid 15020841) . I mentioned it briefly above, but it turns out it’s really important, much more important than I realized, so listen up.

The researchers, who included V. Craig Jordan (a name you’ll see again in the endocrine section), wanted more information about the CALGB 9344 results reported above. The CALGB 9344 study, you’ll remember, is the one that found a 4% survival benefit for women taking Taxol, but most of that benefit was going to women with estrogen receptor-negative cancers. There’s been a lot of argument about how much weight to give that, especially since the subset analysis was unplanned:

"The results of the ER a subset analysis of CALGB 9344 is a frequently debated issue at national meetings and was an issue raised during the 2001 NIH consensus conference on breast cancer. The unplanned and retrospective nature of this analysis does not establish a firm basis on which to base treatment decisions. However the finding that patients with ER a - disease experienced a relatively large benefit from adjuvant paclitaxel, and patients with ER a + disease possibly no benefit at all, has [led] some oncologists to recommend using ER a status as a way to choose between a taxane and nontaxane based adjuvant chemotherapy regimens. This development underscores the importance of investigating the relationship between ER a expression and paclitaxel further."

In other words, some doctors have been starting to do what I’ve been proposing: use estrogen receptor status as a way to decide whether to prescribe Taxol or not. But if you’re going to do that, you’d better be sure that estrogen receptor status is the right thing to be using as a predictor!

So they thought that maybe in the lab, they could get some more clarity. They’re looking for something they can measure that will really tell them whether Taxol is likely to do any good in a particular estrogen receptor-positive woman: “In view of the narrow risk to benefit ratio for chemotherapy for some patients, the ability to confidently identify patients with ER a + tumors who benefit from chemotherapy would be a significant advance in patient management.”

They figured that since the cell-killing effects of Taxol are cell-cycle dependent, maybe the reason the estrogen receptor-positive cancers are less responsive to Taxol is that they’re growing more slowly, not something about them being estrogen receptor-positive per se. Seems like a pretty obvious conclusion, but apparently nobody had ever bothered to tease apart the two factors. So in the lab, they do this. They put the slow-growing estrogen receptor-positive cells in a culture that helps them grow faster, and presto, they become more responsive to Taxol! The researchers do several other manipulations with various cell lines, and they all point the same direction: it’s the growth rate, rather than the ER status, that determines whether Taxol does much of anything.

One of the researchers’ main points is that doctors should stop giving Taxol and tamoxifen at the same time, because it’s counterproductive. But for our purposes, the most important implication of this study is that we now have an objective way to decide whether you should do Taxol, free from all the complications of clinical study results. If your tumor was made up of slow-growing cells, then Taxol isn’t worth the “low but significant incidence of chemotherapy-induced leukemia, heart failure and sepsis.” If your tumor is fast-growing, then it may well be worth it to do Taxol. The decision becomes very objective. The trouble is, I don’t know how to convert the growth measures on your pathology report into the growth measures that they use in the study.

On your pathology report, you get 2 out of 3 points on “mitotic activity” on the Scharff-Bloom-Richardson scale, for a mitotic activity reading of “moderate.” You have 20% Ki-67 staining, which seems to be borderline. From brief research, I get the impression that less than 20 is considered lower growth, and over 25 is considered high growth. (Cortesi and Turchetti abstract,,1003,_12-002636-00_18-0017-00_19-0016670,00.asp). Another study also used 20 (pmid 10955790) . In fact, there seems to be a fair amount of clustering around the 20% number, with the Baylor College of Medicine also using it:

"The cut-off to define a high Ki-67 proliferation index is not well established or universally agreed upon. In our laboratory, we use 20% positive cells to define ‘positive’ (i.e. high risk) based on the results of studies we feel are currently the best available in the medical literature (1, 2)."
( )


In a National Cancer Institute clinical trial, they want “high risk” women to test their response to Taxol, “high risk” being defined as over 30% Ki-67 (or an S-phase fraction over 10%). (

They used different growth measures from Ki-67 in the Dougherty study: doubling time (which is pretty self-explanatory) and percent S-phase fractions. I’ve asked my biochemist friend to see if he can convert these figures (although I don’t know whether they can be directly correlated). We could also ask the pathology department at the clinic whether there is any more detail behind your Scharff-Bloom-Richardson number which would be useful. Also, my biochemist friend says that there is a company in Southern California called Oncotech that can do other growth rate tests, but I don’t know whether they could still do the test on your slides. It probably depends on whether the slides are frozen or in paraffin, and whether they were frozen in a particular way. I think my biochemist friend is going to find out about this for us, and your second opinion doctor would probably know too. Anyway, this has the potential to be a rational, numbers-based way to decide on the utility (or lack of utility) of Taxol.

One further predictive marker I haven’t explored much is Her2. One study showed women who are (unlike you) Her2-positive are three times as likely to respond to Taxol. ( Baselga J, Seidman AD, Rosen PP, et al. HER-2 overexpression and paclitaxel sensitivity in breast cancer: therapeutic implications. Oncology 1997; 11 (Suppl. 2):43-8.)

Finally, one team has come up with a very accurate assay to determine sensitivity to Taxol. It would be worth reading their paper and doing any additional tests needed. I have written for more information. ( Prediction of the therapeutic response to paclitaxel by gene expression profiling in neoadjuvant chemotherapy for breast cancer. Journal of Clinical Oncology , 2004 ASCO Annual Meeting Proceedings (Post-Meeting Edition). Vol 22, No 14S (July 15 Supplement), 2004: 500. Author(s): M. Yoshimoto, M. Makita, S. Nishimura, K. Tada, F. Kasumi, F. Akiyama, Y. Hoshikawa, Y. Miki, M. Matsuura, T. Noda; Cancer Institute Hospital, Tokyo, Japan. For reprints and all correspondence: Masataka Yoshimoto, Department of Breast Surgery, Cancer Institute Hospital, Kami-Ikebukuro 1-37-1, Toshima-ku, Tokyo 170-8455, Japan. E-mail: )

Taxol update Feb. 28: Never heard back from Dr. Yoshimoto. However, got a call back from William Goodson at California Pacific Medical Center in San Francisco, who has published on the prognostic value of Ki-67 and other proliferation indices.

My main question was what they use as a cutoff between high and low-growth tumors (and thus ones more or less responsive to chemo and Taxol chemo). He said they used the median, which was 23.2% in their case. So numbers in the low 20’s would be considered on the lower end, though still fairly average. (Your number was 20.)

He wasn’t familiar with the Dougherty paper (at least not referred to by that name), and we talked a bit about the issue of finding an equivalence between the Ki-67 numbers and proliferation index reported in your pathology report and the S-phase and doubling time used in the Dougherty/Jordan paper. He said to be careful with S-phase, though—that if they did it by flow cytometry, it’s not as reproducible (if I remember correctly). So it seems that we have the information we need about your tumor’s growth rate from the Ki-67 percentage and the Scharff-Bloom-Richardson mitosis index, we don’t need to translate those numbers into S-phase and doubling time to have useful information.

He didn’t seem to be strongly “pro” or “anti” Taxol—he mentioned several factors that would put you close to the average—but the balance shifted when we started to talk about the post-polio and peripheral neuropathy issues. I mentioned that Dr. Nowak had said the key is to stop treatment in time if peripheral neuropathy comes up, so I was a bit concerned about you going into the Taxol already having peripheral neuropathy from the AC chemo. He said, “Wait a minute, you just told me that the key is to stop it in time, so how do you square that with starting Taxol?” I explained that I hadn’t realized when I wrote to Dr. Nowak that one could get peripheral neuropathy even before starting the Taxol!

Anyway, after this discussion, he said he’d be “real unexcited” about you as a candidate for Taxol. He would encourage a second opinion. I asked whether he (Goodson) does second opinions, and yes he does, but he’s a surgeon, so he’s really not the person to talk to about this part of the process. He said your second opinion doctor, Dr. C., has his finger on the pulse of everything, that Dr. C would be where he would send someone. If Dr. C says you should do Taxol, he would defer to him; but if Dr. C says you shouldn’t do Taxol, he wouldn’t be surprised. He encouraged me to kidnap you if necessary and take you to the second opinion, and he said to tell you that delaying the Taxol a week is not going to make or break anything.

Another Taxol update 2/28: I got another kind reply from the author of the taxane review for the Lancet in regard to my two additional questions. She would be somewhat concerned by the peripheral neuropathy already from the AC chemo before having even started Taxol, although all contraindications are relative and it would depend on the amount of benefit expected from adding the taxane. It would require frank discussion with the patient. Docetaxel could also be considered as a less neurotoxic alternative. (I don’t know much of anything about docetaxel, not having studied it.) Since taxanes are a fairly new addition to the breast cancer treatment regimen, there isn’t any research on them in polio survivors that she’s aware of.

And another Taxol update: I heard back from my biochemist friend about proliferation measures! I told him about Dr. Goodson’s thoughts and his recommendation of Dr. C, and here’s his response:

"I'm so relieved! I think that sounds much more definitive than anything I could come up with. I did talk with the 2 cell biology folks here (unfortunately we have no pathologists on staff - we should), they were both of the opinion that this was a tough call since it was right on the borderline - but interestingly they both thought it was closer to the high growth side than the low growth side. But they also said that their measurements are done with FACS and the numbers are usually lower than with the IHC staining used on these.

I also spoke with Oncotech and they said that in their breast cancer samples, the %Ki67 varies from 3- 35%, rarely any higher, with a mean of about 17%. They also said that breast was one of the slower growing tumors and some of the faster growing ones like lung would have a mean %ki67 of 40-45%. So among all tumors, this would put 20% at maybe lower than the mean, but among breast tumors, this would be a bit higher than the mean.

Oncotech also said they would need a fresh biopsy to actually measure the doubling time directly in culture and can't do that with samples already in paraffin. They did confirm that for tumors they'd looked at, patients tended to respond less to taxol if they were slow growing tumors.

So, what to do? Definitely a second opinion is called for, and do present them with these numbers and the research you've done."

Taxol/polio update 4/2005: We saw a wonderful neurologist who specializes in polio, Dr. Miller, and he was of the opinion that Taxol is not likely to set off post-polio syndrome, because polio affects mostly muscle neurons whereas Taxol affects sensory neurons. So Taxol could cause its own problems, but it is not likely to bring about post-polio syndrome by having an additive effect on muscle neurons.

I asked him how this squares with the dose-determination studies of Taxol, in which invariably a few of the women at the highest doses end up in a wheelchair. How is that not damage to muscle neurons? He said that if it’s bad enough, sensory damage could put someone in a wheelchair too: if you can’t feel your feet, you are going to have a hard time walking. But this is a risk for anyone taking high doses of Taxol, not just polio survivors. One should go into it with one’s eyes open to the price that may be paid at the particular dose one is taking.

Back to Contents

An Alternative to Taxol

If Taxol gets down to the point where it’s maybe a 1% survival difference, or maybe half a percent, or maybe none, but you just don’t want to give up that sliver of a chance, how about an alternative that might be just as good, without the toxicity?

The alternative treatment hasn’t been studied as thoroughly as it should be, but there is quite a bit of evidence accumulated about it anyway. Let me quote (with modifications) from a meta-analysis (Chlebowski 2002) :

"Lack of this treatment as a negative influence on prognosis of breast cancer patients has been reported in the vast majority of studies over the past 25 years, with many analyses controlled for recognized prognostic variables. Currently, 34 studies have related this treatment to recurrence risk or survival in early-stage resected breast cancers. A statistically significant association between this treatment and recurrence or survival was seen in 26 reports incorporating 29,460 women (Table 2), whereas eight reports incorporating 3,727 women did not see such associations (Table 3). The positive effects of this treatment on breast cancer recurrence and survival are seen in both pre- and postmenopausal women. …

In a meta-analysis, the hazard ratio (HR) for effect of lack of this treatment on recurrence at 5 years was 1.78 (95% confidence interval [CI], 1.50 to 2.11) and for death at 10 years it was 1.36 (95% CI, 1.19 to 1.55). By another measure, recurrence risk at 5 years without this treatment was 1.91 (95% CI, 1.52 to 2.40) and for death at 10 years it was 1.6 (95% CI, 1.38 to 1.76), suggesting that women without this treatment were significantly more likely to develop recurrence and less likely to survive. Most recently, this treatment was again strongly associated (P _ .005) with disease-free survival and overall survival in a cohort of 535 women (median age, 50 years) with newly diagnosed breast cancer. … These associations of this treatment and positive breast cancer outcome are substantial, with differences comparable in magnitude to those associated with adjuvant hormonal and chemotherapy use and of potentially great clinical importance." [emphasis added]

Neglecting to get this treatment after diagnosis has been frequently reported for breast cancer patients, especially among women receiving systemic adjuvant chemotherapy. In a prospective cohort of 535 newly diagnosed breast cancer patients, use of adjuvant chemotherapy and onset of menopause were the strongest predictors of lack of this treatment. …

So what is this magic treatment that makes more difference than Taxol? I call it the “Tahoe Treatment.” You’ve probably guessed its other names: weight loss, weight control, waist-to-hip ratio...

Though as the authors point out, someone really needs to do a good randomized trial, the accumulated data strongly suggest that two months of bounding through the snow with your dog, and getting trim like you always do when you go to Tahoe, will do just as much good as Taxol—or more. And it’s sure to be a lot more pleasant! Plus, there’s no risk of permanent joint pain, unless you do something really stupid on your skis. (Though to be consistent, we should be clear that that the difference made by carrying or not carrying your 10 extra pounds, and the benefit to be gained from two months of exercise, is certainly smaller than the difference made for a person who carries or doesn’t carry 50-100 extra pounds for a lifetime. However, the benefits seen in these studies are so big that they could be reduced a lot and still be in the realm of any benefit expected from Taxol.)

Here is a summary from an analysis of post breast-cancer exercise to echo what the analysis of weight says: “ Most of the breast cancer survivors were not meeting the physical activity recommendations proposed for the general adult population. Efforts to encourage and facilitate physical activity among these women would be an important tool to decrease obesity, prevent postdiagnosis weight gain, and improve breast cancer prognosis.” (Irwin ML 2004) There’s also an exercise meta-analysis (Thune 2001, pmid 11427781) which says “An observed inverse association with a dose-response relationship between physical activity and breast cancer was also identified in the majority of the 41 studies including 108,031 breast cancer cases.” I also saw something about a steroid administered with chemo possibly being partially responsible for weight gain, but I haven’t pursued it.

If it were me looking at that data, I’d put my money on two months of intensive Tahoe Therapy.

Taxol update 4/2005: In January the FDA approved Abraxane (ABI-007), a new form of paclitaxel using nanotechnology to improve the delivery. Although I am not usually in favor of the latest whizbang variant which just happens to cost more and still be under patent protection, Abraxane seems to be a big improvement.

So far Abraxane is only approved for metastatic breast cancer, but at this point, I wouldn’t accept paclitaxel in any other form—if I couldn’t find a doctor who would give Abraxane off-label, I wouldn’t do taxane treatment. Why? Because Taxol uses the solvent polyoxyethylated castor oil (Cremophor) to get the paclitaxel in, a solvent which likely contributes to the hypersensitivity reactions that often occur with Taxol and are severe in about 3% of patients. Not only are solvents also likely culprits in peripheral neuropathy, but they may actually reduce the effectiveness of the paclitaxel (leading to the rather dismal results we’ve been reading about above). By contrast, Abraxane skips the solvents and uses nanoparticle technology to deliver the paclitaxel. In a study of 460 women, the women given Abraxane had less of a hit to their white blood cells and fewer cases of prolonged neuropathy than with Taxol. Oncologists can deliver a 50% higher dose of the active agent (paclitaxel) and still get fewer side effects (maybe that way one will actually get survival differences greater than 3%!). They also don’t have to give women steroids to prevent hypersensitivity reactions.

K. Garber’s January 2004 editorial in the Journal of the National Cancer Institute provides a helpful overview of the issues (pmid 14734692). Despite the expense, my position at this point would be “either Abraxane taxane or no taxane.”

Back to Contents

Radiation: For Whom Is It Appropriate?

Radiation is complicated, and the thinking has been changing over time.

It is clear that radiation makes a difference for some women. In the past, studies would show a decrease in local recurrence, but it would be more than made up for by the toxicities, so there would be no survival benefit to the radiation. But if you look at only the more recent studies with more modern equipment, there is actually a survival benefit. (Van de Steene, Soete, and Storme, 2000) According to Van de Steene, the benefit is probably only present when heart sparing techniques and optimal fractionation doses are used.

Radiation is clearly of benefit to somebody. Would it be likely to be of benefit to you? Does it depend on the number of involved lymph nodes? Does it depend on estrogen receptor status? Does it depend on age or menopausal status? Let’s look at the literature.

In 2001, the Journal of the National Cancer Institute published a set of recommendations about radiation called “Treatment guidelines and techniques in delivery of postmastectomy radiotherapy in management of operable breast cancer.” This is quite relevant, since it’s post-mastectomy (not post-lumpectomy) and it analyzes all the data to date. It starts out by saying “Recent trials have also demonstrated a significant survival benefit following radiotherapy in high-risk patients. Therefore, it is important to identify the patients who could potentially derive the survival benefit and to not offer treatment to those patients who are not at increased risk for failure.”

It talks about various recent studies that show a survival benefit of about 9% at 10 years (Pierce 2001, pmid 11773304) . It then says, “These data demonstrate a systemic benefit of radiation therapy in high-risk breast cancer patients and emphasize the importance of defining the patients who could potentially derive that survival benefit.” In other words, given the particulars of your case, should you expect to get that 9% survival benefit, or more, or less? Do you fit in the category of “high risk?” The 9% is an average, so we need to figure out whether the number is high enough to be worth it for women in situations similar to yours.

Unfortunately, as of 2001, they really didn’t know for sure how much difference estrogen receptor status makes in radiation’s effectiveness, or whether radiation is worth it for women with only one to three involved lymph nodes. They did come to one conclusion, about extracapsular extension: extracapsular extension is not really a big deal by itself, because it correlates highly with lymph node involvement. They say “Extracapsular extension does not appear to predict for a statistically significant increased risk of axillary recurrence, and correlation with increasing lymph node involvement may account for chest wall risk.” The guidelines end by saying “The role of locoregional therapy following mastectomy in patients with one to three axillary lymph nodes is currently undefined and is being evaluated in an ongoing randomized trial.”

Let’s see whether the American Society of Clinical Oncologists can tell us anything more. In 2001 they published an excellent piece called “Postmastectomy Radiotherapy: Guidelines of the American Society of Clinical Oncology.” The ASCO guidelines turn out to be incredibly useful. The committee members thoroughly and carefully summarize all the research through 2000, saying what they know and what they don’t know. If the evidence is high-quality and extensive, they’ll “recommend;” if it’s less clear, they’ll “suggest;” and if they can’t tell or they can’t agree, they’ll say “there is insufficient evidence on which to base a guideline.”

Regarding increased heart damage from left rather than right breast radiation, they say registry-based studies have clearly shown an effect (p. 1559) . ((The American Society for Therapeutic Radiology and Oncology, in their consensus statement in 1999, agrees with them. They say “Careful attention must be paid to the morbidity of treatment. In particular, [radiation] has the potential to cause late cardiac mortality in patients with left-sided cancers. Therefore, the amount of heart (and also lung) in the treatment field must be minimized and documented. Computerized tomography-assisted treatment planning is useful in accomplishing this goal.”)) The ASCO guidelines agree with ASTRA that modern state-of-the-art radiation techniques can make such heart damage less prevalent. Then they talk about one factor that may cause an additive detrimental effect on the heart: the modern tendency to use Adriamycin-based chemo, which is hard on the heart like radiation.

Regarding the bigger issue of radiation for women with one to three positive lymph nodes and whether it does any good, they have to say they have “insufficient evidence on which to base a guideline.” (Recht 2000, pmid 11230499)

They look at the issue of other tumor-related characteristics that might be predictive, and they can’t tell from the evidence at that point. They found the evidence to be fragmentary and sometimes contradictory for close or positive margins, and for extracapsular extension, they considered the effect uncertain. They couldn’t find enough data on estrogen receptor status to decide anything. They concluded by saying “Further investigation of the pathologic and molecular correlates of locoregional failure and treatment efficacy should be vigorously pursued.”

So the ASCO report concludes that women and their doctors must make the decision based on insufficient evidence. They urge researchers to do trials to answer the many questions. Researchers heeded their call. Let’s look at the results of those efforts—the research published since 2000.

We’ll jump to 2003, with a couple more years of data available. In that year, there was a study published on tumor location and prognosis (Gaffney, pmid 12560437) . Its main point is that women with tumors on the outer side of their breast (like you) have less chance of recurrence than women with tumors on the inner side (the researchers think maybe because the cancer isn’t as likely to have gotten into something called the internal mammary chain).

This information in itself is interesting, because it moves you slightly towards a lower-risk rather than higher-risk category. But to get this information, they had to do a huge, I mean huge, study. They took the records of 45,880 patients in a big government database, including death records and the whole bit. This meant that they had enough people to get meaningful results. And they found something interesting along the way: in women who had a mastectomy (like you) and “regional” disease (i.e. some involved lymph nodes), there was no significant difference in survival with or without radiation. This is important, because unlike many other studies, they had enough women to find a significant difference if there was one.

Now let’s move on to another 2003 paper, “Meeting Highlights: Updated International Expert Consensus on the Primary Therapy of Early Breast Cancer” (Goldhirsch, pmid 12847142). They have quite a bit to say about radiation, chemo, and the general approach to early breast cancer. First is a call for thinking about specific groups of women (“subset analysis”) instead of just the average:

"Although much useful information will come from new technologies, there is also a valuable resource of information already available from current studies. Subset analysis is inevitable in deriving information with which to tailor treatment to individual patients. Such analysis is statistically proper provided sufficient numbers of patients are available and provided hypotheses generated in one data set can be independently confirmed. This recently has been done to substantiate greater responsiveness to cytotoxic chemotherapy in tumors lacking estrogen receptors, (14,15) and the inadequacy of cytotoxic therapy alone in young patients with endocrine-responsive disease. (16,17) Indeed, concentration exclusively on overall average effects produced by pooling biologically different disease entities can result in dangerously misleading advice for individual patients."

Then they have a two-page chart giving their recommendations on everything. They don’t make a statement that everybody should or shouldn’t have radiation, they say" The balance between beneficial and harmful effects of postmastectomy radiation therapy depends on the risk of local recurrence and patient age (38,41)." In other words, one has to keep doing research and look at all these factors that determine one’s risk of recurrence! Also, the older one is, the less likely they are to recommend radiation.

They do finally have something to say about estrogen and progesterone receptors, in the context of treatment responsiveness in general, not just to radiation. They don’t go into detail, they just say that, as opposed to stuff like u-PA-PAI1 and other jumbled acronyms,

"Estrogen receptor and progesterone receptor expression of the primary tumor cells is the only tumor-related marker with clear predictive value for treatment response that has unequivocal clinical utility regarding adjuvant therapy (46)."

They say a little more about this:

"Gene expression profiling studies 4-7 support a clear separation of steroid hormone receptor–absent disease as an entity distinct from disease showing low or high levels of receptors, whereas some clinical studies already provide empirical data that receptor-absent disease is different from that with even low levels of receptor expression (51,75)."

In other words, the apples and oranges concept when it comes to estrogen receptor-positive and negative breast cancers.

Finally, we get to what they have to say about radiation. They don’t say anything about estrogen receptors, they just make life more difficult by creating even more confusion regarding making decisions about what’s best for women based on number of involved lymph nodes. It was starting to look like radiation was mostly of benefit to women with four or more involved lymph nodes. But then

"Recent analysis of the Danish and the European Organization for Research and Treatment of Cancer trials has cast doubt on the traditional perception that radiotherapy is relatively of most benefit to patients with higher risk of relapse (eg, four or more positive nodes). In these trials the highest survival benefit was seen predominantly among patients with one to three positive nodes, whereas the reduction in locoregional recurrences was largest in patients with more advanced cancer. 104 Tailoring postmastectomy radiation therapy treatment recommendations for individual patients remains a priority for additional research. "

Of course a large number of the women in the trials they’re talking about were premenopausal, which changes things, so we don’t know how much stock to put in the results. (There’s a lot of other argument back and forth about these results, with some people saying their recurrence rates were too high to start with because of lousy surgery, and other people saying that’s not true.) But basically, we’re back to the beginning.

Finally, they have something to say about patient preferences, and they encourage the kind of analysis and thinking you are doing:

"As in the previous meeting, 1 the Panel recognized the importance of ascertaining and allowing for patient preferences in defining the threshold of expected benefit at which treatment should be undertaken (73,86,87). The increasing importance of patient advocacy groups 88-90 and the availability of information on the Internet emphasize the need to incorporate this aspect into decision making. Individual patient’s decisions will involve their preferences with regard to the additional benefit that could reasonably be expected from additional effective therapy. … This concept particularly applies to decisions about the addition of chemotherapy to an endocrine regimen that might already offer most of the available improvement in survival, so that the residual potential gain must be evaluated against the morbidity of treatment.… Patients should be involved in the decision-making process to the extent they desire (91) and may be assisted by specific decisionmaking aids (73,92-94)."

So we’re at 2003 and we still don’t know whether radiation does any good for postmenopausal women with estrogen receptor-positive cancers and one to three involved lymph nodes (no wonder doctors’ recommendations vary!). Let’s see if in the remaining two years they were able to figure anything out.

Next up are two studies by Woodward (pmid 12957242 and 12957243) . They are quite relevant to your case: they use Adriamycin-based chemo after mastectomy, like you’re having. (However, the great majority of the women had one-side mastectomies, which would tend to increase the recurrence rate. Also, more than half were premenopausal, which could also have the same effect.)

They start by summing up previous research:

"Recently, several groups, including our own, have studied locoregional recurrences (LRR) after mastectomy and chemotherapy in an attempt to better define which patients may benefit from postmastectomy radiation (14, 15). The data from these series have shown that the long-term rate of locoregional recurrence after mastectomy and chemotherapy is greater than 20% in patients with four or more lymph nodes or T3 disease. In contrast, for the majority of patients with Stage II breast cancer and one to three positive lymph nodes, the long-term risk of LRR after mastectomy and chemotherapy is less than 15%. Within this subset, only patients with extracapsular extension (ECE) of disease measuring > or = to 2 mm or patients with fewer than 10 lymph nodes recovered at axillary dissection have been demonstrated to have substantially higher rates of LRR (15)."

(I don’t know how many millimeters your extracapsular extension was, but we do know that they took out more than 10 lymph nodes.)

Then they present their data, and in general, it’s encouraging for radiation. In the first study, they give radiation to 470 patients and find that not too many more of them have heart attacks compared to the controls, and the radiation didn’t seem to be causing extra cancers elsewhere in the body. In the second study of those women compared to the 1031 controls, they only look at recurrence (not survival), and they do indeed find less recurrence with the radiation.

After looking at their own and previous data, they conclude, “On the basis of these data, it is clear that postmastectomy radiation should be offered to all patients with a 20% to 30% locoregional recurrence risk.” But then they say:

“However, the locoregional recurrence (LRR) rates for patients with Stage II breastcancer with one to three positive lymph nodes are much lower. Specifically, in an analysis of 2016 patients treated with mastectomy and chemotherapy on Eastern Cooperative Group Trials, Recht et al. reported 10-year total LRR of 28.7% for patients with four or more positive lymph nodes, and 12.9% for patients with one to three positive lymph nodes (14). Similarly, our data suggested a 10-year total LRR rate of 25–34% for patients with four or more positive nodes, and 13% 10-year total LRR for patients with one to three positive nodes (15).

Based on the results of the randomized prospective clinical trials and the studies of LRR patterns after mastectomy and chemotherapy, both ASCO and ASTRO published consensus statements regarding postmastectomy radiation (12, 13). These statements both suggested that postmastectomy radiation should be considered as a standard component of therapy for patients with four or more positive lymph nodes. For patients with Stage II breast cancer and one to three positive lymph nodes, both statements concluded that there is insufficient evidence to make recommendations or suggestions for the routine use of postmastectomy radiation.

Perhaps the most interesting finding of this analysis is the demonstration that radiation reduced the risk of LRR for all categories of patients with lymph node–positive disease, including those with Stage II breast cancer and one to three positive lymph nodes. This is one of the first studies to evaluate the benefit of postmastectomy irradiation for patients with this stage of disease who were treated with a standard modified radical mastectomy and adjuvant doxorubicin-based chemotherapy.

In these patients, we found a 10% absolute improvement in the 10-year rate of LRR with the addition of radiation (3% vs. 13%, p _ 0.001). Furthermore, on multivariate analysis, no radiation was the strongest predictor for LRR of the 13 variables examined for this group. Whether this degree of improvement in LRR can affect overall survival is unknown.Because of selection biases that affected radiation use, we could not study whether radiation use improved overall or disease-free survival.

In conclusion, our data suggest that the addition of radiation therapy after mastectomy and doxorubicin-based chemotherapy improves LRR rates in all patients with lymph node-positive breast cancer. However, whether the magnitude of benefit we demonstrated for patients with Stage II breast cancer with one to three positive lymph nodes warrants postmastectomy radiation remains controversial."

So we’re still trying to figure out if radiation confers a survival benefit to women in your category. The next piece of evidence is a really recent article from the University of Texas M. D. Anderson Cancer Center (Garg 2004 pmid 15093909) . Finally, they divide it up by more factors! In plain English, the women having more recurrences had Stage T3 disease (which you don’t), four or more positive lymph nodes at surgery (which you didn’t), and were under 40 (which you’re not). For most women with T2 cancers (like yours was) and three or fewer positive lymph nodes (like you had), they think radiation may not be justified:

"The actuarial locoregional recurrence rate at both 5 and 10 years was 10%. Factors that correlated positively with locoregional recurrence included clinical Stage T3N0 (p = 0.0057), four or more positive lymph nodes at surgery (p = 0.0001), age < or =40 years at diagnosis (p = 0.0001), and no use of tamoxifen... The 5-year locoregional recurrence rate for the 42 patients with clinical Stage T1 or T2 disease and one to three positive lymph nodes at surgery was 5% (only two events). CONCLUSIONS: For patients with clinical Stage II breast cancer, T3 primary disease, four or more positive lymph nodes after chemotherapy, and age < or =40 years old predicted for locoregional recurrence. For most patients with clinical T1 or T2 disease and one to three positive lymph nodes, the 5-year risk for locoregional recurrence was low, and the routine inclusion of postmastectomy radiotherapy does not appear to be justified."

The authors do leave some room for judgement, though, within their overall conclusion. There are some women with one to three involved lymph nodes who they would not object to as candidates for radiation. They say

"However, within this subset, postmastectomy radiation may be appropriate for those aged <=40 years [no], those with <=10 recovered lymph nodes (or 20% positive lymph nodes)[no/yes], ECE of >=2 mm[unknown], lymphovascular space invasion [yes], or positive or close surgical margins[no]."

This study had some limitations, so it’s not the last word. For one thing, the women had chemo before surgery rather than after. For another, the study is quite small. For a third, it’s looking at disease-free survival, which we keep seeing isn’t as useful a measure as overall survival.

Next up we have a study in the New England Journal of Medicine (Fyles 2004:351;10, 963-70) . Finally someone has thought to study postmenopausal women separately—all 769 women were 50 or over—and it’s a recent study, so with fairly modern radiation equipment. Unfortunately for us they studied node-negative cancer (no lymph node involvement), so the levels of recurrence they would find would be expected to be lower than for with nodal involvement. On the other hand, these women only had lumpectomies, which would tend to increase the risk of recurrence versus your double-sided mastectomy. So maybe we can still learn something.

In the study, they got a lower rate of local recurrence with the radiation (91% instead of 84% were free of recurrence at 5 years), but (this should be sounding familiar) no significant differences in the rates of distant relapse or overall survival. Of course, if the study had been bigger or had gone on longer, they might have been able to find a difference by then, but they couldn’t within 5 years.

One thing that’s really nice about this study is that they gave statistics for disease-free survival for various categories. Again, this doesn’t seem to translate directly into survival, but it is better than just the rate of local recurrence. Of the hormone receptor-negative women, only 67% were disease-free at 5 years, while 89% of the women with estrogen receptor-positive tumors were disease-free. Of the women who had “second cancers” (as opposed to “recurrences” in the same breast), the most common site was the other breast. Thank goodness that this is a risk you have mostly eliminated.

Next up is another article in the same issue of the New England Journal of Medicine last September, this time on women 70 or older (Hughes 2004, pmid 15342805) . Again, it’s not a perfect match—they’re a little older than you, and the tumors were smaller with no node involvement—but we’ll have to make do. A counterbalancing factor is that they had lumpectomies, not even single-sided mastectomies, leaving more room for recurrence (but this probably doesn’t increase their risk as much as if they had lymph node involvement).

The authors begin by summarizing the background:

"Multiple trials of breast-conserving surgery for breast cancer (1-5) have shown that postoperative irradiation decreases the rate of ipsilateral recurrence but offers no survival benefit. However, the high rate of recurrence with surgery alone (10 to 40 percent) has suggested that the only two appropriate treatments are modified radical mastectomy and breast conserving surgery plus adjuvant radiation therapy.

Since tamoxifen, with (3) or without (4) radiation therapy, decreases the risk of recurrence, and given the cost and adverse effects of breast irradiation (5-12) and its negative effect on the quality of life (6,7), we designed a trial to determine whether women 70 years of age or older who have early, estrogen receptor-positive breast cancer can be safely treated with tamoxifen alone instead of irradiation plus tamoxifen. Such women were selected for the trial because they have a lower rate of recurrence (8-16) and a shorter time at risk for recurrence than younger women."

This is interesting, because the authors are implying that modified radical mastectomy, like you had, would give an equivalent benefit to lumpectomy plus radiation for these women (okay, actually, that’s not so interesting, since we’ve known that for a while from a large trial—otherwise people wouldn’t be doing lumpectomies). But now they want to know, since these women have a fairly low risk of recurrence anyway, could they do away with the radiation part as well? So we get some information about radiation, even though these women are probably at a lower risk than you (so we can’t generalize too much). Anyway, to cut to the chase, they did get a lower risk of recurrence like all the other studies (1% instead of 5%). But, as usual, they didn’t get any statistically significant change in distant recurrence or overall survival.

These two articles in the New England Journal of Medicine stirred up a flurry of press coverage and were accompanied by a very good editorial, which I will reproduce here because I think it’s quite useful. Once again, remember that even though these women only had lumpectomies, they are still probably at a lower risk than you because they didn’t have lymph node involvement. But the editorial brings up some good issues, including psychological issues, that one must consider in decision-making. Here it is:

"...medical resources. Paradoxically, women 70 years of age or older were specifically excluded from some of the earlier trials, including the NSABP B-06 trial. The CALGB trial also showed a significant increase in the five-year rate of local or regional recurrence among women treated with lumpectomy and tamoxifen alone, as compared with women who also received radiotherapy. In contrast to the results of other trials, however, the absolute difference was small (4 percent vs. 1 percent), and there were no significant differences in the subsequent need for mastectomy, the risk of distant metastases, or survival between the two groups.

Does this absolute difference in the rate of local recurrence of 3 percent matter clinically, when weighed against the cost of the use of additional resources and treatment-related adverse effects? An overview of 40 trials in this field has confirmed that local radiotherapy is associated with an increase in deaths from cardiac and other causes, which nullifies a long-term reduction in deaths from breast cancer. Modern radiotherapy techniques, including three-dimensional treatment planning and intensity-modulated radiotherapy, with cardiac shielding, can minimize the risk of cardiac irradiation in most patients. Some side effects persist, and the CALGB investigators reported an increased incidence of breast pain, fibrosis, breast edema, and poor cosmetic results in the group that received radiotherapy. In addition, breast radiotherapy is resource-intensive and time-consuming for the patient and may decrease the quality of life. There is, of course, the counter-argument that not having radiotherapy may increase a woman’s anxiety about the possibility of recurrence and could require more frequent follow-up. On balance, nevertheless, there are clear advantages in identifying subgroups in a large population of elderly women who do not require radiotherapy after lumpectomy and tamoxifen.

The long natural history of breast cancer is a further issue here, and the median follow-up of approximately five years in both these trials could certainly lead to an underestimate of the final difference between therapies. In the NSABP B-06 trial, 19 percent of local relapses after lumpectomy alone occurred 5 to 10 years after treatment, and a further 9 percent occurred after 10 years. In the Canadian trial, even in the subgroup with a good prognosis, the rate of local relapse rose from 5.9 percent after five years to 15.2 percent after eight years for women treated with tamoxifen alone. For women over 70 years of age, however, this is likely to be less of an issue, since breast cancer in this age group is biologically less aggressive and is associated with a lower risk of local recurrence after breast-conserving surgery and with less time at risk.

Adjuvant endocrine therapy is improving with the availability of the aromatase inhibitors. The ATAC (Arimidex, Tamoxifen, Alone or in Combination) trial showed that, as compared with tamoxifen, anastrozole resulted in a greater reduction in the risk of local recurrence as well as of systemic disease after a median follow-up of a mere 33 months, and similar benefits in local control have been reported with the use of exemestane and letrozole after 2 and 5 years of tamoxifen therapy, respectively. These findings raise the real possibility that treatment with aromatase inhibitors could further reduce the risk of local recurrence in the absence of radiotherapy.

Against this background, what course should we recommend to older women with small, hormone-receptor–positive breast tumors who have undergone lumpectomy? The Canadian trial indicates that women under the age of 70 years should still receive radiotherapy in addition to tamoxifen. This is frustrating advice, because although it is clear that the majority of women treated with lumpectomy and tamoxifen alone would not have a relapse, we still cannot confidently predict which women will be in the majority. We are entering an era in which the use of molecular markers, gene expression profiles, and other molecular prognostic indicators is being investigated as a means of individualizing adjuvant medical therapies, and there is no reason why the same approach should not be applied to radiotherapy. Future trials of radiotherapy should be required to include tissue biopsy for prospective molecular analyses and informed consent for this type of research.

In contrast to the Canadian study, the CALGB trial offers persuasive data that women 70 years of age or older who have estrogen-receptor–positive cancers up to 2 cm in diameter can now be offered lumpectomy and tamoxifen alone without additional radiotherapy. For many women, the mere 3 percent increase in the risk of local recurrence with no concomitant decrease in survival is likely to be an entirely acceptable tradeoff, even in view of the fact that the risk of recurrence may increase further with time or perhaps be reduced by treatment with an aromatase inhibitor rather than tamoxifen. Other women will undoubtedly still want radiotherapy. Women who have just received a diagnosis of early breast cancer rightly have their own strong views about treatment, and in this context, it is of interest that 57 percent of those eligible for the Canadian trial declined to enroll. The CALGB trial provides new data that will help women and their health care providers make informed choices about treatment."

I hope that was helpful. Even though it is talking about a slightly different group from you, it brings up many of the issues one must think about.

Finally, we have three more papers from the past two months to consider before coming to an end of the radiation discussion.

Just this past December, in the Journal of Clinical Oncology, another study came out from the University of Texas M. D. Anderson Cancer Center (Huang 2004, pmid 15570071) . They looked back at the records of 542 women who got radiation and 134 comparable controls. They found that radiation improved “cause-specific survival” in the following subsets: stage IIIB or worse disease (not you), clinical T4 tumors (not you), and four or more positive nodes (not you). In the group as a whole (as opposed to those high-risk women), they found no significant difference in overall survival or even “cause-specific survival.” The one thing you could say to criticize this study is that their subjects were taken over a very wide time period, so some of them got older radiation technology.

The authors conclude by saying “Whether radiation therapy benefits patients with stage II breast cancer with one to three positive lymph nodes remains an area of controversy. In this study, we did not find that radiation provided any benefit to these patients; however, the sample size was limited.”

Next we have a study from France published just last month (Ortholan 2005, pmid 15629606) about the long-term effects of radiation. It goes over survival too, but it’s too small (150 women) to mean anything about survival. We’ll just look at it for information about side effects.

Radiation is very hard in the short term, but the long-term side effects are probably less disturbing. Skin cancer at the irradiated site is very rare (only 58 cases in the literature) (2003 pmid 12963926) . The skin in the area often stays weird-looking or discolored for the first few years, and the tissue underneath can get fibrous-feeling in about half of women for the first two or three years (Ortholan 2005 pmid 15629606, Hughes) , but there are worse things in life. Sometimes there is long-term pain involved (Hughes) . (An interesting note: Women’s doctors report pain not being worse than in non-irradiated women after one to two years, but the women themselves still report it as worse at 4 years!) Arm or shoulder stiffness can be significant for the first few months. Rib breaks occur in only about 3% of cases, and they usually heal without intervention. (I would bet that the Clinic aims for a lower rate.) Limited lung fibrosis is seen in all or almost all patients, but it is usually asymptomatic (Ragaz 2005) .

Finally, we have the hot-off-the-press 20-year follow-up of the British Columbia randomized trial (Ragaz 2005, pmid 15657341) . Unfortunately, they don’t say anything about hormone receptor status, and since all the women were premenopausal (at least when the study started they were), many of them may have been estrogen-receptor negative (certainly more than in a menopausal group), which we have seen tends to make radiation more effective, at least in reducing recurrences. Also, the study was only 318 women, fewer than ideal, so not powered to find small differences.

Anyway, in contrast to the 15-year point, at which they were disappointed not to find any statistically significant improvement in survival from radiation, at the 20-year mark they did (47% instead of 37%). But when you look at women with 1-3 involved nodes versus women with 4 or more involved nodes, consistently the women with more involved nodes make up more of the effect. In the 183 women with 1-3 nodes, the survival difference was not statistically significant. (Either was the difference in breast cancer-specific survival.) Forty-nine of 92 women without radiation had died at twenty years and 41 of 91 women with radiation had died, which is within the realm of chance. We haven’t even accounted for estrogen receptor-status yet.

In their discussion, they say

"Current radiation planning techniques that use three-dimensional computer tomography planning are clearly superior in terms of their ability to shield the heart compared with the radiation therapy used in our trial, which was conducted in the era before such technology. However, cardiac morbidity remains an important issue that may become more evident, particularly for left-side lesions treated with radiation therapy and for radiation therapy used in conjunction with potentially more cardiotoxic schedules of chemotherapy, including anthracyclines, taxanes, and trastuzumab (i.e., Herceptin). Thus, careful attention to avoiding cardiac morbidity and mortality with more sophisticated radiation therapy planning must be a high priority in the new schedules of combined adjuvant chemotherapy and radiation therapy, even though the new radiation techniques provide substantially reduced radiation doses to the heart by use of three-dimensional computed tomography."

Summary: Radiation does seem to reduce the risk of locoregional recurrence, even in women in your category (but to a lesser extent than in women in other categories). However, you’ve already removed most of the local region! Whether it can make any difference in distant recurrence or survival in women in your category is still very much under debate. In older studies, radiation did not confer a survival benefit to the group overall. In newer studies that use more heart-sparing modern techniques, there does seem to be a difference—but the benefit is seen disproportionately in women outside your category. If there is a survival difference in women in your category, it is likely to be small.

Whether to do or not do radiation is a decision you’ll have to make taking into account all the data and the uncertainty. Since the next treatment phase, endocrine treatments, is universally acknowledged to make a substantial survival difference, there would be an argument for moving directly to that treatment. (No, doing them together is not recommended.)

Radiation update May 2005: The second opinion appointment changed the picture considerably. Dr. C’s basic stance was that with 1-3 nodes radiation would be borderline, but the extracapsular extension (and lymphovascular invasion) in his mind tipped the balance and made it like four nodes, which would call for radiation.

This inspired me to go back and read the ASCO recommendations more carefully. Upon doing so, it appears that the studies showing extracapsular extension not to be a prognostic factor separate from lymph node involvement were smaller than the trial showing it to be a separate factor, the British Columbia trial. The differences in this trial are pretty stark: for women with 1-3 positive nodes and no or limited extracapsular spread, radiation only gains them 5% survival benefit after many years (65% instead of 60% survival), versus for the women with extensive extracapsular spread, whom it gains a lot (65% instead of 39% survival, for a gain of 26 points).

There are some limitations to applying the results of this trial to your case: first of all, the women were pre-menopausal, and thus many of them presumably had estrogen receptor-negative cancers. But I just talked to Joseph Ragaz, who ran that trial, and he said their results were the same regardless of ER status. He said with the extranodal spread and lymphovascular invasion, you would be a “classic case” for radiation. Though it would be nice to have data for postmenopausal women, this information is hard to ignore.

Resources for extracapsular extension are as follows: ASCO recommendations; “Cancer” July 1, 2000 p. 223 Ragaz and reply; Proceedings of ASCO volume 18, 1999, abstract 274; abstract 411, source not written down: “Extracapsular spread (ES): Adverse prognostic factor in chemotherapy (CT)-treated stage I-II breast cancer (BrCa), predictive for survival impact of locoregional radiotherapy (RT). Observations from the British Columbia (BC) randomized trial.”

Back to Contents

Endocrine Approaches: Rationale and Possible New Paradigms

First, let me start by saying that endocrine approaches are the one treatment which clearly makes a difference in postmenopausal estrogen receptor-positive women. You will want to do some sort of endocrine treatment.

Your doctor is suggesting Arimidex. What makes Arimidex better than tamoxifen, the old standby? First of all, Arimidex has shown a lower recurrence rate. Second of all, women in studies find it more tolerable. You’ve seen online the articles about the results of the ATAC trial, announced at the big meeting in San Antonio last fall.

Unfortunately, the downside of Arimidex is something that is going to take a while to fully rear its head: Arimidex thins bones and results in more fractures. This would be expected to show up more at 10 years than at the current 5-year mark, since bones usually have to thin for a while before they break. Tamoxifen, unlike Arimidex, has a slight bone-strengthening effect.

Tamoxifen is not free of side effects either, though. It can increase the risk of endometrial cancer, and it is associated with more hot flashes. Both tamoxifen and Arimidex may be associated with memory and cognition effects: a study of 103 women from the ATAC trial showed detriments on a processing speed task and on immediate verbal memory. Immediate verbal recall in particular is thought to be an area especially sensitive to changes in estrogen levels. (Jenkins V, Psycho-Oncology 2004;13:61-66) Other areas, such as memory for numbers or faces, were not as affected. (It could be that some of these problems came from radiation—but they didn’t find any significant differences in level of impairment between the women who had and hadn’t gotten radiation.)

In terms of effectiveness, an analysis by B.E. Hillner (“Cancer”2004;101:1311-22) calculates that “even after 20 years, the overall survival rate in the anastrozole arm would be less than 2% greater than the corresponding rate in the tamoxifen arm. On average, this benefit would result in approximately 2 additional months of survival, with an incremental cost per life year of $40,600.” However, if the bone-thinning that is going on in Arimidex users leads to more hip fractures, then “the long-term benefit associated with this agent is reduced by approximately 25%” when considering quality of life.

The five-year results on the ATAC trial just came out (The Lancet, January 1, 2005, ATAC Trialists’ Group) . They say the overall survival was similar for anastrazole and tamoxifen. But they say it took at least 7 years for tamoxifen to show a survival difference versus placebo, so they think they’re on the right track, since this was a pretty low-risk group of women and it takes a while for there to be enough “events” to judge.

So tamoxifen or Arimidex? The Hillner analysis says that Arimidex is on the borderline on criteria usually used to decide whether something is “worth it” from a societal perspective, but that Arimidex is almost always seen as “worth it” from a patient perspective. Which is another way of saying that since your doctor is recommending Arimidex, and since it is better tolerated, and since someone else is paying for it, you might as well choose Arimidex over tamoxifen for now.

Endocrine update 5/5/2005 : Speaking with Dr. Joseph Ragaz on the phone today, he brought up an interesting issue: he said most of the benefit of Arimidex over tamoxifen were confined to the estrogen receptor-positive, progesterone receptor- negative group. Since you were highly progesterone receptor- positive, this would be another reason to feel free to choose Arimidex or tamoxifen based on whichever one is more comfortable for you. (Of course they think about such things up in Canada, where with a publicly-funded health system, they don’t want to be paying for the new drug that costs four times as much unless there is a clear benefit for that particular group of women.)

Isn’t there any type of endocrine therapy that doesn’t cause endometrial cancer or speed osteoporosis? Do you really have to choose your poison? Let’s talk about endocrine approaches, how they work, the theory behind them, and whether there are any alternatives.

First, a quick overview of how these things work. Estrogen receptors on cells are like locks on a door, and estrogen, tamoxifen, and Arimidex react with the lock in different ways. Imagine you have a large office building with several tenants. The property manager has a master key which opens every door in the buiding, and individual tenants have keys which open the front door and the doors on their floor. Estradiol, the real-deal human estrogen, is like the master key. Put it in any lock, and it will open the door. Tamoxifen is like a tenant’s key. It opens the front door and the doors on that tenant’s floor, but if you try to force it to work on the other tenants’ floors, it will just jam the lock.

Now think of the body as that office building. The brain is the top floor; the breasts are one down; the heart is below that; below that is the endometrium; and the bones are another floor. Real human estrogen will make all those doors open: keep the brain functioning, stimulate the breasts in preparation for nursing a baby, help the endometrium get all nice and cushy to hold a pregnancy, and keep the heart and bones strong.

But what if you’re not planning to have a baby and you don’t need all those things? That’s where “selective estrogen receptor modulators” come in. They’re like a tenant’s key: they work some but not all of the locks. That’s what tamoxifen is. It works somewhat in the “bone” lock, which is why there’s not as much of a problem with osteoporosis on tamoxifen—for bones, tamoxifen is like a bit of hormone replacement therapy. But it doesn’t work on the “breast” lock—in fact, it just jams it—which we want in this case, because we don’t want the real estrogen to get in and stimulate those cells. (We’re talking about the little bit of estrogen that is still being produced by your adrenals and your fat cells.)

Unfortunately, tamoxifen also works in the “endometrium” lock, which is why there’s the problem of increased risk of endometrial cancer (because, just like with estrogen replacement therapy before they added progestins, the endometrium is all plumped up with nowhere to go).

People dream of developing a “designer SERM” that works in only the locks they want. It should stimulate bone growth but not increase hot flashes, help the memory but not stimulate the breast, lubricate the vagina but not stimulate the endometrium, etc…. It’s a nice dream, but don’t hold your breath. So far it’s been hard to have one’s cake and eat it too. There’s always been a trade-off. Which is why tamoxifen was used for so long: it jammed the “breast” lock without causing osteoporosis, and it seemed worth the price of having to watch like a hawk for endometrial cancer.

Where does Arimidex come in? Arimidex takes a different approach. It’s not a key that works or jams locks. Arimidex confiscates all the keys, including the master key. Then it runs the locksmith out of town, turns off the office building’s lights and furnace, and stops paying the building’s PG&E bill. All of which turns out to be a very effective way of keeping people out of that “breast” floor. Meanwhile the tenants on the bone floor are pretty upset that they can’t get in, but that’s the price you have to pay. So far the “bone” tenants aren’t causing too much problem—they’re just standing outside the building protesting, burning the occasional effigy and throwing the occasional paint bomb at the building. It’s if they sue (i.e. if fractures increase substantially) that there will really be a problem and the approach will have seemed problematic, but we won’t know that for a few years.

My main concern about Arimidex (aside from the bone effects) is, to continue the analogy, whether the rest of the locks will still be usable after 2 to 5 years of disuse. In other words, is there permanent damage to body-wide estrogen receptors from starving them for several years? Will the locks on the other floors be rusted beyond usability? Even if the locks can be worked again, will opening the other offices reveal water and rodent damage beyond repair? I don’t know the answer to these questions.

Endocrine update 3/2005: Heard back from VC Jordan. Regarding long-term effects of starving the body of estrogen, he says they just don’t know. The bones can be addressed with bisphosphonates, but the brain, which I also asked about, is another matter. He doesn’t believe there will be any permanent damage, but he says nobody has done the correct clinical evaluation to find out.

I do know, however, that a nine-week course of Arimidex in rodents actually super-sensitizes the estrogen receptors (the locks) to real estrogen (the master key). So if you use the master key on the “breast” floor after nine weeks of Arimidex, you can actually kill off cancer cells. But will you have that action inadvertently on healthy cells on the other floors too? I don’t know.

Now let’s take a step back and look at the theory behind all of this. The current way of understanding breast cancer is that all estrogen is evil. An accumulation of lifetime estrogen is responsible for breast cancer. This theory has been in favor since someone argued it convincingly in 1983.

Wouldn’t it be a bit simplistic, though, to declare something that occurs naturally in the body unequivocably “evil?” It might seem so, but there are several pieces of evidence which seem to fit with this theory. One is that women with earlier puberty have a higher risk of breast cancer, and another is that women with later menopause have a higher risk of breast cancer. (Cuzick 2001:93:10) The real clincher, though, was when the results of the WHI study came out and seemed to show that “hormone replacement therapy” caused breast cancer.

But is it possible for there to be another explanation for those facts? As you know, just because two things are correlated doesn’t mean there’s causation. A third factor can be at play. For example, could it be that early puberty is due to higher insulin levels, which are later a risk factor for increased weight, diabetes, and breast cancer? (Having been an athletic little kid and starting menstruation at 14, I look around at all the 8-year-olds with breasts and derive a certain amount of satisfaction from this theory.)

As for the late menopause, could it be that the problem is not additional cycles, but additional anovulatory (no ovulation) cycles? In the normal breast pre-menopause, estrogen stimulates cell growth and then progesterone keeps things in check each month. But in an anovulatory perimenopausal cycle, you have no estrogen peak, and thus no progesterone surge. So you’re getting a dribble of estrogen with no progesterone to keep it check.

In addition, there are some facts that don’t fit with the “all estrogen is evil” theory that people seem to conveniently ignore. If all estrogen is evil, then why aren’t most women getting breast cancer in their teens or 20’s, when the stuff is surging through them? Why don’t more women get breast cancer during pregnancy, when estrogen levels are off the chart? You could say, “Well, it’s the lifetime accumulation of estrogen exposure that matters.” But then why, oh why, are multiple pregnancies protective against breast cancer??

Could it be that we need a more sophisticated theory that can account for all of these things? I would propose that we do.

What if the problem is not estrogen itself, but a dribble of estrogen without progesterone to keep it in balance? Or worse yet, estrogen plus an androgenic progestin such as medroxyprogesterone (MPA)? Or, worse yet, a non-human estrogen that has a long half-life in the breasts (such as equine estrogens) plus an androgenic progestin such as MPA?

If this theory is true, then with tamoxifen and Arimidex, we’re only addressing half of the picture. Tamoxifen works by jamming up the breast estrogen receptors (like the wrong key in a lock). If a dribble of estrogen is a problem, then it makes sense that blocking the estrogen to the breast would make a difference. Arimidex works by shutting off all estrogen at the source, which would also make a difference. But maybe we’re not doing all we could, and we’re doing some collateral damage along the way. Maybe we should be looking at the overall balance.

To me, it seems a lot like the approach women have taken to dieting for years. All those years of counting calories or fat grams or carbs or whatever was in fashion, attempting to starve the problem at its source. Meanwhile, the other half of the equation, exercise, gets sorely neglected, no matter how good people’s intentions are. But the problem really isn’t that food is evil. With high-quality food and high-quality exercise, one can find a sustainable balance without starving oneself.

But what about the WHI trials? Don’t they show that estrogen is evil and should be avoided? Well, no.

First of all, the WHI wasn’t a trial of human estrogen, it was a trial of horse estrogen, which is less than 1% estradiol (Woodcock FDA PowerPoint “Bridging What Is Known…”) . Second of all, it was oral delivery through the liver first-pass mechanism, which changes it into entirely different metabolites. But let’s not split hairs. The real problem is that when the results of the Premarin-only arm came out a year or so after the results of the non-hysterectomized PremPro women, there was no longer a breast cancer increase! Yes, part of this is due to that arm of the study being smaller and statistically lower-powered—but still, the missing piece, the MPA progestin, should be getting some dirty looks at this point (and is). MPA (medroxyprogesterone) is known to be an estrogenic progestin in some tissues (Miyagawa, reported in Formby and Wiley 1998) . So, like women in the non-hysterectomized arm of the WHI study, you got years of continuous breast stimulation, instead of the growth-death-growth-death cell cycle that you would have had with transdermal progesterone. (How many years were you on the oral MPA before we got you switched to Prometrium? Ten or so?)

I just talked about MPA being an “estrogenic” progestin. What is meant by an “estrogenic” progestin? In order to make a patentable product, pharmaceutical companies have to make synthetic progestins out of other molecules, such as testosterone. In certain tissues, these synthetic progestins sometimes act more like their parent compound, or a relative of their parent compound, than like progesterone. I don’t know a lot about this, but that’s the basic idea.

Meanwhile, we have evidence that human estradiol and progesterone, when taken transdermally, should not be tarred with the same brush as their synthetic cousins such as MPA taken orally. Even publications aimed at general practitioners, such as the Association of Reproductive Health Professionals “Clinical Proceedings,” are starting to recognize that there is a difference, making statements such as “The risk of VTE is associated with oral but not with transdermal estrogen use” and “Unlike MPA, micronized progesterone does not attenuate the estrogen-associated protection against coronary vasospasm....” (June 2004) . Word is slowly getting out that transdermal estradiol has a different side effect profile than oral equine estrogens.

By no means are “estrogen” and “good cancer outcome” as mutually exclusive as they would appear. In fact, high doses of estrogen and testosterone were actually used as a cancer treatment in the 1950’s, and “ethinyl estradiol” is listed in decision trees regarding endocrine therapies along with tamoxifen and Arimidex (Comprehensive Cancer Care Center guidelines 1/2005) . In a 2002 review of endocrine therapy by respected scientists Goldhirsch, Colleoni, and Gelber (from Milan and Harvard), estrogens, androgens, and progestins are right there on the list of endocrine therapies, right next to SERMS and aromatase inhibitors. Yet nobody talks about it.

So here we have this nice theory. As Formby puts it, “Combined with our data on breast cancer, we speculate that progesterone occurs naturally as a homeostatic control of normal proliferation in healthy young women. Cancer then, together with other factors, could be a natural body-wide phenomenon when anovulation becomes prevalent.” Is there any data to support such a theory—that estradiol plus progesterone could keep cancer in check?

There is. Unfortunately, however, most of the studies in women use a mish-mash of different kinds of HRT, so we can’t put much stock in them and must rely more on animal and test tube data, where it’s easier to keep things pure. Which is just as well, since the human data is contradictory.

Pretty much all of the observational studies show increased survival (significant or not) in the HRT group, but in observational studies there’s lots of room for selection bias. But like with the WHI, at the end of all the observational studies comes the randomized trial bombshell. First I’ll go over a couple of the observational studies, and then I’ll describe the bombshell.

For an example of one of the more meaningful observational studies, in a fairly large study by O’Meara (2001, pmid 11353785) , HRT users had half as many recurrences, died from breast cancer a third as often, and died in general half as often. That’s a pretty big deal—more effect than tamoxifen. But the authors are very cautious about their conclusions: “Although residual confounding may exist, the results suggest that HRT after breast cancer has no adverse impact on recurrence and mortality.” In other words, though half as many women died, we’re not going to claim that HRT is better, but it doesn’t seem to be worse.

In a nice literature review from 2001, the authors go through all the published studies they can find and synthesize them (Col, pmid 11304788) . Unfortunately, they just look at recurrence. They found that the relative recurrence risk for women on unspecified HRT after breast cancer was 0.82 of controls, a lower risk for women with HRT than without, but not statistically significant (4.2% of women per year on HRT versus 5.4% of women per year not on HRT having recurrences). They too are modest in their conclusions, saying “Although our analyses suggest that HRT has no significant effect on breast cancer recurrence, these studies were based on observational studies subject to a variety of biases.”

Decker, Pettinga et al. (2003, Menopause) do a study with 277 post-breast cancer women on various regimens versus 277 matched controls. Overall survival favored the HRT group, and the result was statistically significant. Of interest, the women in the HRT group lived an average of 26.29 years, versus 21.83 years for the control group. (The women in the HRT group were on average a year or two younger than in the control group, though, so that should be subtracted out of the 4 1/2 -year difference). The researchers conclude by saying “In these selected patients, ERT relieved estrogen deficiency symptoms and did not increase the rate or time to an ipsilateral recurrence/new primary, contralateral new primary, local-regional recurrence, or systemic metastases.”

Another study looked at an average of 6 1/2 years of HRT after breast cancer (Peters 2001, pmid 11776498) . The largest number of women (24) used Premarin, 20 used PremPro, and 10 used estradiol. In the 56 women, there were no deaths from breast cancer. The average age at diagnosis was 61. This study has some major limitations, though, so I wouldn’t put too much stock in it.

They also did a table of seven other studies of hormone therapy after breast cancer, and the results are quite encouraging—only five deaths out of 600 women in an average follow-up of 2-3 years (so less than 1%). Again, this is not nearly as useful as data from a big randomized controlled trial, and it’s way too short to mean much—but at least it tells us that the women aren’t dropping like flies, like you might expect them to be doing given that they had estrogen-fueled cancer.

But then there’s the bombshell, the WHI-equivalent of post-cancer HRT studies: a Swedish study of post-cancer HRT that was stopped last year because of significantly more cancer recurrences in the HRT group (Holmberg, Lancet Feb. 7, 2004) .

Like the WHI for non-cancer HRT, the stopping of the HABITS trial was treated like the death knell for post-cancer HRT. This makes some sense, since it is the first randomized controlled trial, and observational trials are subject to so many types of biases. But despite the increased “recurrences” (26 in the HRT group instead of 7 in the control group), this trial is really not the last word, for a number of reasons:

So the studies in humans are ultimately not helpful in answering our question. They all mix together various types of hormone treatments, the observational studies have biases, and the HABITS study is too short and has this weird thing going on with concurrent tamoxifen in some of the women. So much to our dismay, we must turn to animal studies, where it is easier to keep the variables constant.

Here, the information is quite impressive. This may get a little boring, because there’s a lot of it, so you can skip this section if you get partway through and find it too detailed. But I want to get it all in, at least with a quick reference to each study. I’ll go chronologically, looking at studies “in vivo” (usually in rats) and overview studies. Later I’ll mention the studies with cancer cells in test tubes (“in vitro”).

Our rat studies start out with one by Swanson et al. (1997, pmid 9494582) . They use estradiol and progesterone (when you’re using rats, you can give them exactly what you want to!). They test toxicity and conclude that estradiol and progesterone treatments “are a non-toxic means of mimicking the protective effect of pregnancy against mammary cancer and that pregnancy or hormone treatments may achieve this prophylaxis through a differentiation mechanism.” (In other words, the breast cells go through a change that makes them more resistant to cancer.)

In 1998, Sivaraman et al. at Baylor (pmid 9771927) gave rats a high dose of estrogen and progesterone and then injected them with a carcinogen. Rats getting the hormones had an 82% reduction in number of mammary cancers compared to rats fed blanks. Decreasing doses of E/P (estrogen and progesterone) were partially protective, suggesting that complete differentiation of the gland was not required. (In other words, hormones, not just the high levels of pregnancy, can make a difference, although a smaller difference.)

In another study from the Baylor group (Lydon 2000, pmid 14973394) , they’ve gotten their hands on some genetically altered mice that don’t have working progesterone receptors, and they find that progesterone is necessary for the protective changes that take place in the breast during pregnancy. They have some ideas on why, and why it’s likely to be the same (“conserved”) in humans.

Later they write a paper going over their results about both estradiol and progesterone being necessary for mammary protection in rodents and talking about some about the theories of genes and proteins and how all of this works (Medina, Sivaraman et al., 2001 pmid 11795441) . They also write a paper in a different journal (pmid 12160088) in which they focus more on theories about reproductive history and its effects.

In 1999, Regine Sitruk-Ware (who’s now a big deal at the Population Council and a favorite of NIH funders in my field, and whom I met at a conference) wrote a review article on progestins and cancer (pmid 12227896) . This is back when she was still in France. She doesn’t completely distinguish between progestins and progesterone. But of relevance to our concerns about MPA, she says

"Progestins exert different effects according to the steroid from which they are derived, e.g. pregnanes derived from progesterone, estranes or gonanes derived from testosterone. Some estrane derivatives are able to stimulate breast cell multiplication in vitro through an estrogen receptor-mediated pathway. Most pregnanes do not exert such an effect. Also, some pregnane derivatives stimulate apoptosis, leading to cell death. However, it is well established that high doses of progestins have been successfully used in the treatment of advanced breast cancer as second-line endocrine therapy. Finally, striking differences have been observed in progestin use in Europe and in the USA. In France, where the rate of progestin use per head is higher than in the USA, the rate of breast cancer has not increased as sharply as observed in North America."

Of course we don’t want to make too much of that last statement, but the comment about different progestins having different action is confirmation. A 2003 paper by Druckmann (Maturitas, pmid 14670647) echoes this point, with specific attention to different progestins’ differential effects on the growth hormone axis (which is involved in cancer risk) and mammary gland differentiation.

Switching to just the estrogen side of the equation for a moment, Zhang (2001 pmid 11743748) at the University of Illinois points out that a metabolite of horse estrogens (Premarin) is a carcinogen. Remember, Premarin taken orally goes through the liver first pass and is turned into variants (metabolites). This particular one they study, 4-hydroxyequilinin, causes four different types of DNA damage when injected into rats’ mammary fat pads. By contrast, an “endogenous” (native to the body) estrogen, 4-hydroxyestrone, didn’t cause damage. They conclude that “4-hydroxyequilenin has the potential to be a potent carcinogen through the formation of a variety of DNA lesions in vivo.” This is just another reason to remember that studies of PremPro may have nothing to do with what happens when you take the real human stuff, estradiol and progesterone, without the liver first pass. (However, this DNA damage must not be as big a factor as some other factors, since they haven’t seen a breast-cancer increase in the Premarin-only arm of the WHI trial.)

Next comes an Oncology paper from Japan (Kurebayashi 2000, pmid 10971176) . They want an answer to the question of why, in their previous study, breast cancer cells were stimulated by an anti-estrogen and inhibited by estradiol in animals, but not in test tubes. Why the difference? So for some reason they study estradiol, ovarian removal, and MPA. They got weird results: in the test tube, estradiol stimulated and MPA inhibited breast cancer cells; but in the mice, it was the opposite, with estradiol inhibiting and MPA stimulating. Unfortunately I don’t really understand their conclusions, which have something to do with VEGF expression and antiangiogenic agents, but maybe somebody else would. Anyway, we might want to keep in mind that substances can act differently in the test tube than in the body. That’s why I’m starting with the “in vivo” research.

From Belgium, we have a study called “Progesterone receptor activation: an alternative to SERMs in breast cancer.” They had women put either estradiol, progesterone, estradiol plus progesterone, or placebo gel on their breasts (Desreaux 2000, pmid 11056336) . They find that "in postmenopausal women, adding progesterone to percutaneously administrated oestradiol significantly reduces the proliferation induced by oestradiol… Progesterone seems to be important in normal breast homeostasis."

This is important in that it’s a test on women, not animals, and they did biopsies.

They also do a test tube study with a synthetic progestin called nomegestrol acetate, and they mention that “the peak of apoptosis after [nomegestrol acetate] withdrawal was absent in T47D cancerous breast cells, suggesting that these tumoral cells have lost their endocrine regulation of apoptosis.” I don’t know the significance of that, but it sounds of concern. Perhaps you can ask what that means when you get your second opinion, if you get down to that level of detail.

So far we have had just scattered estrogen and progesterone studies here and there. But next we see a sudden burst of activity in estrogen research. What could possibly cause renewed interest in a treatment that is more than half a century old? It comes from an interesting source: the need to do something about tamoxifen resistance.

In 2000, researchers at Northwestern entered this research area with a bang, with a study on tamoxifen-resistant mice (Yao, pmid 10815929) . This came out of V. Craig Jordan’s lab, a name you’ll see again. They find that in tumor cells that have grown resistant to tamoxifen after long tamoxifen treatment, stopping tamoxifen and starting estradiol can actually cause them to shrink. In fact, cycling the two may be positive. They conclude,

"On the basis of the new breast cancer models presented, we propose a cyclic sensitivity to [tamoxifen] that may have important clinical implications: (a) it is possible that a woman's own estrogen may produce an antitumor effect on the presensitized micrometastatic disease after 5 years of tamoxifen. Long-term antitumor action occurs because the drug is stopped, but resistance accumulates and tumors start to grow if adjuvant therapy is continued; and (b) although in the clinic tamoxifen-resistant tumors respond to second-line therapies that cause estrogen withdrawal, e.g., pure antiestrogens or aromatase inhibitors, estrogen therapy may also be effective and return the tumor to tamoxifen responsiveness. In this way, a hormone-responsive tumor may be controlled longer in the patient with advanced disease."

This is the first study which specifically mentions that it might be better to switch to estrogen therapy rather than continue a treatment such as tamoxifen past the point of diminishing returns. It also proposes estrogen therapy as an alternative to Arimidex post-tamoxifen.

Now we’re to 2001, with another study from Northwestern University (Chisamore, pmid 11595710) . V. Craig Jordan is involved again. It’s pretty technical, but the basic gist seems to be that in mice, estradiol can inhibit tamoxifen-resistant cells over-expressing something called PKC alpha. Think this may be important for treatment decisions after tamoxifen. Also, if a tumor over-expresses PKC a right from the beginning, it might be better to treat it with estradiol or an estrogen rather than tamoxifen. Their next step will be to examine biopsies from patients who took tamoxifen and had recurrences versus patients who took tamoxifen and remained disease-free. For our purposes here, the interesting thing is the envisioning of a possible role for estradiol in breast cancer treatment. (Hopefully they would use estradiol, not DES like cancer docs used to do.)

They also find, like previous researchers, that estradiol works differently in the test tube than in a living being, and they conclude that “these results suggest that some factor present in the tumor micro-environment is responsible for the differences in hormone responsiveness… observed in vitro and in vivo.”

Next up is an article in the Journal of the National Cancer Institute by Song et al. (2001 pmid 11717332) from the University of Virginia. They give the background that “High doses of estrogen can promote tumor regression in postmenopausal women with breast cancer, but the mechanism is unknown.” The point of their study is that they think they found the mechanism, or at least part of it: it may be because of estrogen activation of something called Fas-mediated apoptosis. They also find that it’s more effective the longer a woman has been post-menopausal or without much estrogen (which might be an argument for taking estradiol after a course of Arimidex).

But it’s the introduction of this paper that’ll really blow you away. They talk about the history of endocrine treatment of breast cancer:

"Once tamoxifen was developed, in the 1970s, the use of high-dose estrogen as hormone-additive treatment of breast cancer fell out of favor, and tamoxifen became standard first-line therapy. This change in practice resulted from a clinical trial with a 4-year follow-up that demonstrated that postmenopausal women with advanced breast cancer tolerated tamoxifen statistically significantly better than they tolerated DES (P = .02) while experiencing objective response rates that were not statistically significantly different (33% for tamoxifen and 42% for DES). A recent report with a 20-year follow-up observation of the same trial participants, however, revealed two surprising findings. First, DES-treated patients survived longer than patients receiving tamoxifen, with 35% of the DES-treated group and just 16% of the tamoxifen-treated group alive at 5 years (P=.03)."

In a letter to the editor about the Song paper, V. Craig Jordan (whose name we’ve seen before) praises the study and says we should look at what it means for the big picture:

"The recent excellent paper by Song et al. and the accompanying editorial deserve a broader perspective than that presented by the observations with estrogen-deprived MCF-7 cell lines in vitro. Song et al. state that their data on the apoptotic effects of high-dose estrogen could provide insights into the mechanism of pharmacologic estrogen used to treat breast cancer some 20–30 years ago. Estrogen-deprived cells actually become supersensitized to physiologic concentrations of estrogen (10 –10 – 10 –9M). This range (27–272 pg/mL) is equivalent to that observed in the circulation of perimenopausal and premenopausal women.

Although the goal of discovering a mechanism for estrogen-induced apoptosis when used as a therapy is valid, important additional applications of the new knowledge should be pursued. For example, breast cancer patients are currently being exposed to long-term estrogen withdrawal with aromatase inhibitors and selective estrogen receptor modulators (SERMs). The two SERMs, tamoxifen and raloxifene, are administered for at least 5 years for the treatment and prevention of breast cancer and osteoporosis, respectively. The existence of resistance to tamoxifen and raloxifene is evidenced by SERM-stimulated tumor growth in vivo.

In the case of tamoxifen resistance, MCF-7 tumors grow initially in response to both tamoxifen and estrogen, but after 5 years in a tamoxifen-treated environment, the tumors become supersensitive to the tumoricidal actions of physiologic estrogen and regress rapidly. Indeed, it has been suggested that a woman’s own estrogen may be responsible for the persistent survival benefit that patients enjoy following 5 years of adjuvant tamoxifen therapy. Interestingly, in the laboratory some tumors regrow in the estrogen environment, and these tumors again respond to tamoxifen as an antiestrogen.

We wish to echo the conclusion of Soto and Sonnenschein in their editorial — that estrogen-induced apoptosis may be caused by a complex interaction of estrogen with both the cancer cell and the stromal component of the tumor. We uniformly observe estrogen-stimulated growth of breast cancer cells in vitro, but when the cells are transplanted into animals, profound reductions in tumor size occur, with circulating estrogen concentrations in the 40–60 pg/mL range — that is, dramatically lower concentrations than are routinely found in premenopausal women. The fact that in vivo systems amplify the tumoricidal action of estrogen is good news for the clinical application of the knowledge, but the finding provides a complex paradigm to address precise mechanisms in the laboratory.

The laboratory demonstration of the tumoricidal actions of physiologic levels of estrogen should not persuade clinicians to rechallenge their patients with high-dose estrogen after safer endocrine therapies have been exhausted. Rather, we suggest that clinicians should think about ways to test how the emerging new knowledge about breast cancer sensitization to low-dose estrogen could ultimately aid patient survival. Although we appreciate that the wisdom of the time is that estrogen is bad and that it is fashionable to be “antiestrogenic,” we suggest that an integration of the concept of lower-dose estrogen after SERM therapy be evaluated, to improve both response and quality of life. Hormone replacement therapy after SERM treatment may have unexpected benefits." (Jordan VC 2002, pmid 12165645)

We’ve heard a lot from V. Craig Jordan. Who is he? Some radical? With a little help from Google, it turns out that he’s one of the most cited researchers in the field, and is considered the father of tamoxifen. This guy’s no slouch.

There follows an interesting exchange, with a response from the authors and then a weighing in from those who wrote the original editorial. They emphasize the plasticity of cancer cells: that under starvation of estrogen, a cell can become super-sensitized to it, and then swing back under other conditions. They suggest the idea of cyclic therapies, switching back and forth between aromatase inhibitors and estrogen. (What I find interesting here is that nobody has made the slightest mention of progesterone and its ability to keep things in check in premenopausal cycles—but at least we’re starting to learn something about estrogen.)

Finally, the authors of the original accompanying editorial weigh in (Soto and Sonnenschein 2002) , and they don’t pull any punches. Ever so politely, they deliver a scathing assessment of the state of critical thinking in the field:

"We all agree that more research into the mechanism of hormone action on the control of cell proliferation and cell death is needed. However, as we proposed more than 20 years ago, in a 1980 editorial in this Journal, researchers in the field should also reassess their hypotheses, spell them out in detail, and resolve any contradictory results and interpretations of the data. Furthermore, rather than generating new data now and thinking later, we should reassess our thinking now, postulate potentially new hypotheses, and then carry out the appropriate experiments to test them. Indeed, philosophy does matter in medicine, not only in regard to the ethics of exposing patients to the unintended deleterious effects of overzealous treatments but also in regard to sound epistemology." [emphasis added]

They also make a case against prolonged treatment with any one SERM or anti-estrogen, saying the Yao study points to the danger of developing resistant cancers. They say researchers should be studying the optimal length of time to take SERMS or aromatase inhibitors, and that after that, hormone replacement therapy as cancer treatment might be an option: “Hormone replacement therapy after selective estrogen receptor modulator therapy, as proposed by Jordan and his coworkers, may become an option after it is rigorously tested in a clinical setting.”

Where does this leave you? Here are respected scientists saying that hormone replacement therapy after a shortened course of tamoxifen or Arimidex might be a good idea and help women live longer. The only trouble is, according to all logic and lab data this seems like a good idea—but it won’t be proposed to you, a regular patient, until they’ve had a chance to check it out for five or ten years (just like if you had been going through this five or six years ago you wouldn’t have been given Arimidex). So following such a course would require a leap of faith that it will turn out to work the same way in humans as in animals. Unless you could find a doctor who was willing to be ahead of the curve, trying this would require enrolling in a clinical trial—and I am not aware of any.

Let’s see if there is additional knowledge gained since 2002. Next we have a study from Georgetown (Hilakivi-Clarke 2002, pmid 11897501) in which they point out that nobody has really been able to prove the theory that high lifetime estrogen exposure increases breast cancer risk. They propose that the timing of estrogen exposure might be more important. I haven’t read the full study, so I don’t know how convincing it is.

Going international, we come to a study in “Carcinogenesis” from researchers at the national academy of medicine in Argentina (Vanzulli 2002, pmid 12016147) . It’s a complicated study about tumor cell lines in mice. If I understand correctly, they get tumors to grow in mice taking MPA. They can then shrink the tumors by giving either estradiol or an anti-progestin. I’m not sure of the significance of this, other than that it’s another example of estradiol shrinking tumors, and unless I’m misunderstanding, it’s an example of MPA feeding tumors. It would take someone more knowledgeable to fully understand this study, though.

Now an interesting tidbit from Vienna (pmid 12429630) . Some people have said that if a woman has a contralateral breast cancer after tamoxifen treatment, it’s likely to be estrogen-receptor negative, since they think estrogen receptor-positive ones wouldn’t still be growing. The Vienna team’s research says that that’s not so. In their practice, 27 of the 35 patients with contralateral recurrences had estrogen receptor-positive ones. They think that the recurrences may be made of tamoxifen-resistant or tamoxifen-fed cells, and they have some other explanations too. It’s another example of cancer cells’ adaptability. If you wanted to draw a moral from it, you could say it’s yet another example of why not to stick with tamoxifen (or presumably Arimidex) for too long.

There are two more studies before another flurry of activity and press coverage. First, at the University of Illinois , SM Swanson and K Christov do a study on rats (2003 pmid 12931682) . In a previous study, they had found that a five-week course of estradiol and progesterone protected the rats from breast cancer, similar to the protection gained from a pregnancy and breastfeeding. They injected carcinogens, and all the rats not either taking estradiol and progesterone, or getting pregnant and nursing, got lots of mammary cancers. The pregnancy and the HRT rats got many fewer cancers. In this study, they showed that a three-week course of estradiol and progesterone (instead of a five-week course) works too, even when you’re injecting the carcinogen during the same period as the hormones. They also showed that the resulting cancers in the pregnancy and the HRT groups are not any more aggressive than the ones in untreated rats. I think this is pretty important stuff.

The next study, from Northwestern, took biopsies from a lot of premenopausal women and showed that you can tell what part of her menstrual cycle a woman is in by looking at the morphology of the breast tissue (Ramakrishnan 2002, pmid 12481017) . They could also correlate how the tissue looked with the progesterone levels at the time. This result had been gotten in the 1980’s with autopsy material, but this study confirms in living women that breast cells go through hormonal changes throughout the month.

Now comes another flurry of activity, once again inspired by the problem of cancer cells growing resistant to anti-estrogens (such as Arimidex), tamoxifen, and other SERMs. Again we hear from V. Craig Jordan, in an editorial in the Journal of the National Cancer Institute ( March 5, 2003 ).

He asks, “Is Tamoxifen the Rosetta Stone for Breast Cancer?” I’ll spare you the story about Napoleon and Egypt , and get straight to the point, which is that tamoxifen resistance may help decipher some of the mysteries about breast cancer. The editorial is quite technical, about cell signaling, etc. But it also makes the point that even before we know the details of exactly how drug resistance works, clinicians should ask themselves whether they can “already subvert the power of [treatment resistance] by the judicious use of new approaches to endocrine therapy in select patients.” (He’s referring to aromatase inhibitors and fulvestrant, not just estradiol.)

On the heels of Jordan’s editorial comes a flurry of articles about endocrine approaches, the first one from Stanford (Faridi, Clinical Cancer Research, 2003) in which they studied both in vitro and in vivo. They found that expression of Akt-3 indicates a tumor which is not fed by estradiol, it’s actually suppressed by estradiol and enhanced by tamoxifen. Such tumors are also resistant to fulvestrant. They conclude that “the amount of active Akt present in breast cancers may be important in the relative efficacy of different treatments.” I don’t know whether it’s too late or not to get your tumor slides tested for Akt. Some things you can still test for once it’s preserved, and some you can’t.

On November 5, 2003 two articles appear in the prestigious Journal of the National Cancer Institute, plus an accompanying editorial by Daniel Hayes at the University of Michigan . The first article is—yet again—from V. Craig Jordan’s lab (Liu, pmid 14600091) . They find that not only does estradiol kill cancer cells that are resistant to tamoxifen, it also kills cancer cells that are resistant to raloxifene, the SERM that is used to prevent osteoporosis! They do the study both in vitro and in mice. Once again, estradiol saves the day. They even figured out what phase of cell growth the estradiol arrests.

One particularly notable thing about this study is that they say “When a 9-week raloxifene or tamoxifen treatment was followed by a 5-week estradiol treatment, estradiol statistically significantly reduced the size of tumors stimulated by raloxifene or tamoxifen.” It would be interesting to ask the researchers how a 9-week course in mice would translate into humans, and what they think about its implications for an ideal length of Arimidex treatment.

The second study is also from V. Craig Jordan’s lab (lead author Clodia Osipo). They think they figured out how the estradiol kills tumors: “ER- a appears to mediate estradiol-induced apoptosis and tumor regression of MCF-7TAMLT tumors by suppressing the prosurvival/antiapoptotic factors HER2/neu and NF- k B and inducing the death receptor Fas.” In other words, the estradiol is turning things on and off through the estrogen receptor.

They propose three phases of cell life: in the first phase—initial treatment—they say estradiol makes the cells grow and tamoxifen makes them shrink. In the second phase, the cells acquire resistance such that they’ll grow for either estradiol or tamoxifen and will shrink for an aromatase inhibitor or an estrogen receptor-killer (e.g. fulvestrant). The third phase occurs with long-term treatment, 5 years of tamoxifen. This is the stage they study, and they find that at this point not only does tamoxifen not make them shrink, it makes them grow—and estradiol makes them rapidly shrink. This is the same type of drug resistance Liu et al. found after a year of raloxifene. They also find that fulvestrant is probably not a good idea in this phase, and that Arimidex in this phase will probably stabilize disease rather than cause actual shrinkage like estradiol does. They again bring up the concept of “extended adjuvant therapy, through progressive schedules of antihormonal and hormonal therapies.” So the impression one gets from this study is that estrogen is most appropriate after some amount of tamoxifen or Arimidex treatment.

Reading this, one would think that switching from tamoxifen to Arimidex to estradiol should be timed by Akt-3 and PKC alpha. I don’t know whether these can be measured in a blood test, though—if the measurement requires a tumor sample, that won’t work for women who have already had their mastectomy and don’t have any known tumors left.

Daniel Hayes at the University of Michigan Comprehensive Cancer Care Center is invited to write the editorial about these two studies. He puts the results in context:

"Importantly, what do these results mean to the investigative and clinical community? Clearly, they need confirmation in other preclinical models. Although the MCF7 cell line is arguably one of the most valuable models used in breast cancer, it is not entirely clear whether these results can be generalized to the vastly heterogeneous world of human breast malignancies. Nonetheless, they are provocative and raise yet another question: should we dust off the old standard, estrogen? If so, the results from Jordan and his colleagues suggest that rather than using pharmacologic [high] doses of estrogen as first-line therapy, as was done before tamoxifen's introduction, low doses should be used after SERM treatment. This theory is certainly too conjectural to apply in routine clinical practice. In fact, it would seem to run counter to the hypotheses that have led to completed and ongoing prospective clinical trials in which 5 years of adjuvant tamoxifen is followed by randomization to aromatase inhibitors or placebo, as well as to other trials in which tamoxifen and aromatase inhibitors are alternated within the first 5 years after diagnosis. To their credit, Jordan and his colleagues have proposed a clinical trial to test their theories. Such a trial would be fascinating but difficult to execute. Proper controls for the tamoxifen withdrawal effect and careful attention to the dose of estrogen are required to be certain that any results reflect the effects proposed by Jordan's group and do not just recapitulate findings from the past. Their results also suggest that the order in which serial endocrine therapies are administered might be critical, because the combination of fulvestrant and estrogen in their hands produced results in diametric opposition to those expected. Standard practice currently calls for fulvestrant therapy for patients after their disease has progressed on tamoxifen and aromatase inhibitor treatment. Given that physiologic levels of estrogen might reappear in such patients after discontinuation of aromatase inhibition, the data from Jordan and his colleagues would suggest that fulvestrant might be contraindicated in this setting!

Do these results mean that we should write a new arrangement for the old tune of estrogen therapy? Endocrine treatment is certainly a well-played piano, but perhaps even greater understanding of the biology of the estrogen-ER axis will permit us to write yet another song." (Hayes, pmid 14600082, comment and replies pmid 15069120)

You can find well-written press coverage of these results at the following link: Serena Gordon, “The Estrogen Paradox”

We are coming to the end of our animal studies on estradiol and endocrine approaches, with just four more papers. Berstein (2004 pmid 14977857) and team bring us a study confirming that “long-term tamoxifen exposure enhances sensitivity to the estrogenic effects of tamoxifen and also of estradiol itself.” V. Craig Jordan writes an article called the “Changing role of the oestrogen receptor in the life and death of breast cancer cells,” in which he suggests “that future clinical trials incorporate an ‘oestrogen purge’ to both enhance the actions of chemotherapy or completely reverse endocrine resistance and restore endocrine sensitivity. These new data build on the idea that breast cancer can be controlled as a chronic disease and will permit patients to live long and productive lives during targeted maintenance treatment.” Clodia Osipo and other members of V. Craig Jordan’s team also write a “mini-review” on the subject (2004, pmid 15337826) . If you want to plow through one thing on endocrine approaches, I would say it should be this, or at least the “Conclusion and Clinical Applications” section.

Finally, we come to a new article by Shou et al. (pmid 15199112) which gives further information about the mechanisms of tamoxifen resistance, plus accompanying editorial by Daniel Hayes. Shou and team also have a 2005 paper on the same subject.

Where does this leave us? Wanting to explore the idea of, at the very least, an “estrogen purge” after some length of Arimidex treatment. A more radical approach would be to see whether there is theoretical support for the idea of switching to cyclic estradiol/progesterone after a short course of Arimidex. [Note: Maybe not after a short course—see endocrine update 3/2005.] The most radical would be to do cyclic estradiol/progesterone from the beginning, but it would be hard to get anyone to support that course outside of a clinical trial, even though it has been effective in rodents (and even with the long-term results showing DES to have been MUCH more effective than tamoxifen).

I will not go into all the in vitro work, since we have seen that cells can respond differently in the petri dish than in a living being. I will just mention that I think one of the most interesting is a 1998 paper called “Progesterone Inhibits Growth and Induces Apoptosis in Breast Cancer Cells: Inverse Effects on Bcl-2 and p53” (Formby B and Wiley TS) . Unfortunately, it’s not available online, but it was available at the medical library. Beyond that, I will limit myself to just listing other papers of possible interest (with PubMed I.D. number) for future reference:

1: Uray IP, Liang Y, Hyder SM. Estradiol down-regulates CD36 expression in human breast cancer cells. Cancer Lett. 2004 Apr 15;207(1):101-7. PMID: 15050739
2: Lamote I, Meyer E, Massart-Leen AM, Burvenich C. Sex steroids and growth factors in the regulation of mammary gland proliferation, differentiation, and involution. Steroids. 2004 Mar;69(3):145-59. Review. PMID: 15072917
3: Franke HR, Vermes I. Differential effects of progestogens on breast cancer cell lines. Maturitas. 2003 Dec 10;46 Suppl 1:S55-8. PMID: 14670646
4: Frasor J, Danes JM, Komm B, Chang KC, Lyttle CR, Katzenellenbogen BS. Profiling of estrogen up- and down-regulated gene expression in human breast cancer cells: insights into gene networks and pathways underlying estrogenic control of proliferation and cell phenotype. Endocrinology. 2003 Oct;144(10):4562-74. Epub 2003 Jul 10. PMID: 12959972
5: Franke HR, Kole S, Ciftci Z, Haanen C, Vermes I. In vitro effects of estradiol, dydrogesterone, tamoxifen and cyclophosphamide on proliferation vs. death in human breast cancer cells. Cancer Lett. 2003 Feb 10;190(1):113-8. PMID: 12536084
6: Franke HR, Vermes I. The effect of continuous combined 17beta-oestradiol and dihydrodydrogesterone on apoptotic cell death and proliferation of human breast cancer cells in vitro. Eur J Cancer. 2002 Nov;38 Suppl 6:S69-70. No abstract available. PMID: 12409082
7: Brueggemeier RW, Gu X, Mobley JA, Joomprabutra S, Bhat AS, Whetstone JL. Effects of phytoestrogens and synthetic combinatorial libraries on aromatase, estrogen biosynthesis, and metabolism. Ann N Y Acad Sci. 2001 Dec;948:51-66. PMID: 11795395
8: Horita K, Inase N, Miyake S, Formby B, Toyoda H, Yoshizawa Y. Progesterone induces apoptosis in malignant mesothelioma cells. Anticancer Res. 2001 Nov-Dec;21(6A):3871-4. PMID: 11911261
9: Gompel A, Somai S, Chaouat M, Kazem A, Kloosterboer HJ, Beusman I, Forgez P, Mimoun M, Rostene W. Hormonal regulation of apoptosis in breast cells and tissues. Steroids. 2000 Oct-Nov;65(10-11):593-8. Review. PMID: 11108864
10: Detre S, Salter J, Barnes DM, Riddler S, Hills M, Johnston SR, Gillett C, A'Hern R, Dowsett M. Time-related effects of estrogen withdrawal on proliferation- and cell death-related events in MCF-7 xenografts. Int J Cancer. 1999 Apr 12;81(2):309-13. PMID: 10188736
11: Schneider HP, Jackisch C. Potential benefits of estrogens and progestogens on breast cancer. Int J Fertil Womens Med. 1998 Nov-Dec;43(6):278-85. Review. PMID: 9920536
12: This P. [Hormone replacement therapy in menopausal women previously treated for breast cancer] Presse Med. 1996 Feb 3;25(4):141-4. Review. French. PMID: 8728897
13: Rostagno P, Moll JL, Birtwisle-Peyrottes I, Ettore F, Caldani C. Cell cycle expression of estrogen receptors determined by image analysis on human breast cancer cells in vitro and in vivo. Breast Cancer Res Treat. 1996;39(2):147-54. PMID: 8872323
14: Villalobos M, Aranda M, Nunez MI, Becerra D, Olea N, Ruiz de Almodovar M, Pedraza V. Interaction between ionizing radiation, estrogens and antiestrogens in the modification of tumor microenvironment in estrogen dependent multicellular spheroids. Acta Oncol. 1995;34(3):413-7. PMID: 7779433
15: Bonnier P, Romain S, Giacalone PL , Laffargue F, Martin PM, Piana L. Clinical and biologic prognostic factors in breast cancer diagnosed during postmenopausal hormone replacement therapy. Obstet Gynecol. 1995 Jan;85(1):11-7. PMID: 7800305
16: Schneider PG, Jackisch C, Brandt B. Endocrine management of breast cancer. Int J Fertil Menopausal Stud. 1994;39 Suppl 2:115-27. Review. PMID: 7874188
17: Screpanti I, Felli MP, Toniato E, Meco D, Martinotti S, Frati L, Santoni A, Gulino A. Enhancement of natural-killer-cell susceptibility of human breast-cancer cells by estradiol and v-Ha-ras oncogene. Int J Cancer. 1991 Feb 1;47(3):445-9. PMID: 1993553

Where does this leave us from a theoretical perspective? First of all, we have some pretty good data that estradiol can actually kill cancer cells when they have become resistant to tamoxifen or raloxifene. It is thought that the same thing will happen after Arimidex, but I didn’t find any studies showing this for sure. It is also thought that if there hasn’t been a sensitization period, estradiol will grow, not shrink, tumors. This is based on a 1985 paper, I believe (and maybe a lot of other evidence—I’m not sure). However, there is some theoretical conflict between this and the positive survival benefits seen with high-dose DES in women.

So we have evidence that estradiol plus progesterone has a preventative effect—yet the researchers studying an “estrogen purge” and proposing a cycling back and forth between an estrogen and an anti-estrogen (or SERM) have not integrated the idea of progesterone into their work, and have not explored the idea of estradiol plus progesterone before SERM resistance develops. (Of course, that’s not surprising—one can’t do everything at once, and that team has a background in tamoxifen and tamoxifen resistance.) It’s a shame, though, because Swanson (who did the study with estradiol and progesterone) and VC Jordan (who did the studies on an “estrogen purge” after tamoxifen resistance) both worked in Chicago until recently.

Endocrine update 3/2005 : Heard back from V. Craig Jordan. Apparently the “estrogen purge” only works after multiple years of SERM (e.g. tamoxifen) or anti-estrogen (e.g. Arimidex) therapy. He didn’t say anything about the idea of estrogen plus progesterone, though. So this would somewhat call into question the idea of starting estradiol plus progesterone without first having done a lot of Arimidex, if in humans it takes more than the 9 weeks seen in rodents for estrogen to go from causing cell growth to cell death. It seems to me that it doesn’t completely take it off the table, though, if the cycling progesterone is enough to cause apoptosis every month and keep things in check.

Are there any missing pieces in what seems to data favoring the use of estradiol and especially progesterone? Yes, there are a couple of theoretical problems with the idea that progesterone is necessary to keep cell growth and death in balance. One is, if progesterone is so important, then why aren’t even more postmenopausal women getting breast cancer? (Heaven knows there are enough carcinogens around to get things started.) And speaking of which, why didn’t the women in the Premarin-only (hysterectomized) arm of the WHI trial get extra breast cancer, taking equine estrogens (which we have seen can cause DNA damage) and no progesterone to keep things in check? I don’t know the answers to these questions.

What can you do with this information? You can, in order from most conservative to least conservative,

1) Ignore it and take Arimidex
2) Ignore it and take Arimidex, but keep your eye on the results of studies that come out on the subject
3) Take tamoxifen for a year or two before Arimidex
4) Take Arimidex (either right away or after a couple years of tamoxifen) and do a one or two month “estrogen purge” after a year or two of it
5) Take Arimidex (either right away or after a couple years of tamoxifen) and do continuous or cyclical estradiol plus progesterone hormone replacement when you are finished
6) Take Arimidex (either right away or after a couple years of tamoxifen), and if the side effects become uncomfortable, switch to continuous estradiol plus progesterone
7) Take Arimidex (either right away or after a couple years of tamoxifen), and if the side effects become uncomfortable, switch to cyclical estradiol plus progesterone
8) Take continuous or cyclical estradiol plus progesterone instead of Arimidex from the beginning

If you take tamoxifen, it seems like there would be an argument for occasionally doing a cycle of progesterone and having a withdrawal bleed in order to minimize the risk of endometrial cancer. I’ve never seen this idea mentioned, and I’m not sure why.

Suppose you wanted to try one of those things? How would you do it? To do an “estrogen purge,” you would probably want to get the advice of V. Craig Jordan. He seems to favor one or at most two months of “estrogen purge.”

Back to Contents

Various Updates and Addenda

Endocrine update 4/2005: From talking with Dr. Steven Swanson (who did the rodent studies at the University of Illinois) and Dr. Daniel Medina (who has done a ton of endocrine studies at Baylor and actually did his post-doc in the same lab as Swanson at Berkeley), it becomes apparent that there are major additional factors, namely timing and parity. Has the rodent (or human) had a full-term pregnancy? At what age? Dr. Medina says 30 years is about the break-even point for humans for whether a pregnancy is protective or not (it’s protective before then but can actually be a negative after then).

Is the cancer already in place when the hormones are applied? (Okay, I can’t resist: Which came first—the cancer or the hormones?) Dr. Medina is occupied with trying to prevent cancer in the juvenile rodent.

I asked whether they’ve tried to simulate a post-menopausal situation, and he said no, because of the HRT studies in humans (presumably the WHI etc.). He saw my point about those studies not being the same, but that’s where they stand right now…

He did mention that Dr. Sandra (Sandy) Haslam at Michigan State has done work mimicking menopause in rodents. I looked her up, and indeed she has, by doing ovariectomy in mice to simulate menopause. Her work brings up a further distinction, between the early postmenopausal stage and late postmenopausal stage (time since menopause appears to be more important than actual age). She has also written a 2002 review article (pmid 12160089) on HRT effects on the breast in humans and mice.

Regine Sitruk-Ware also turns out to be extremely knowledgeable about these issues, especially progestins, and to have published extensively in addition to the references listed here. She also informed me about Dr. Russo in Philadelphia, who is a guru of breast differentiation and is experimenting with short courses of HCG (human chorionic gonadotropin), which, in rats, has been able to create the type of cancer resistance that an early pregnancy does (pmid 15701889). This is not so directly relevant to postmenopausal women who have already had children, but could be a big boon to younger women who do not plan to have children.

With so many factors—age, time since menopause, parity (and thus differentiation of the mammary gland), type of estrogen, type of progestin, delivery route, and cyclic vs. steady-state delivery—the only way to find out whether a theoretically optimal regimen would work would be for someone to do the study.

Update 5/2005—Complementary approaches: Once I finished going through the traditional treatments being proposed, I found some time to look at the data on complementary and alternative approaches. A few of them seem to have substantial data that they may help and little likelihood of doing any harm. Chief among these is glutamine, 18-30 grams a day, which may help reduce gastric tract effects and mouth sores from chemotherapy and possibly even make chemo (and radiation) more effective. It’s cheap, too, so there’s not a lot lost if it doesn’t do anything. Good review articles are Savarese 2003 (PMID#14585260) and D’Souza 2004 (PMID #15340021).

Also possibly helpful and unlikely to be harmful are green tea, and the selenium in two to three Brazil nuts (the kind still in their shell, though) per day. Rigorous critical analysis of complementary and alternative approaches can be found in the Moss Report, and there’s also a pretty decent book available from the library: “Breast Cancer: Beyond Convention” edited by Tagliaferri et al. (although since the book chapters are each written by a different author, they sometimes end up contradicting each other). The Moss Report is nice in that he keeps it updated, although that doesn’t come cheap.

Back to Contents

Book Recommendations for the Newly Diagnosed

Some of our favorite books in general, in recommended order of reading:

• “Just Get Me Through This: The Practical Guide to Breast Cancer” by Deborah Cohen.
A couple years older than Dr. Link’s books, but a very helpful overview of how to make it through conventional treatment. Also good for family members to read to increase understanding of the psychological issues (especially the "dos and don’ts” sections).

• “The Breast Cancer Survival Manual” by John Link

• “Breast Cancer Husband” by Marc Silver (if there is a male partner in the picture).
Though one would expect this book to be either condescending or obnoxious, it somehow manages to be neither.

• “Take Charge of Your Breast Cancer” by John Link.
Indispensible for helping one learn how to get the best treatment possible.

• The Moss Report on breast cancer
Available at, but it's expensive.

• Last but not least, “Dr. Susan Love’s Breast Book”

Back to Contents


One of the few breast cancer organizations that isn’t beholden to pharmaceutical companies:
Breast Cancer Action

the National Library of Medicine’s portal to medical literature:

For those unfamiliar with Medline: To search for any of the studies listed in this paper, either enter their “PubMed I.D. number” or PMID into the PubMed search box and hit return, or enter the researcher’s last name and initials in the search box (e.g. “Savarese DM”) and then sort the results by date. Abstracts (summaries) for most papers are usually quite helpful and are available for free. If there is a box at the top of the entry with a link for the full-text article, it will most likely not be free; to get free access, go to your local state university. Many public universities provide access to their library and electronic subscriptions as a public service. One can usually print out the PDFs for a per-page printing fee.

Back to Contents

This review was written during the spring of 2005.

Reciprocal links: