One thing I’ve learned over the years is that there is a palpable hostility in the “alternative” medicine world towards chemotherapy. Many are the times I’ve posted examples, including rants by Mike Adams, cartoons, and a post about what I like to call the “2% gambit” that claims that chemotherapy only contributes 2% to survival in cancer. Basically, that last gambit uses and abuses a rather mediocre study whose design almost seemed intended to minimize any detected benefit from chemotherapy. On second thought, strike the word “seem.” It was pretty much designed to minimize any apparent benefit from chemotherapy, as it left out chemotherapy-responsive tumors without a good justification and didn’t look at ten year survivals, where chemotherapy effects are often more apparent.
Sadly, it’s not just the alt-med cranks who harbor a hatred of chemotherapy. There are otherwise rational people, some of them doctors even, who don’t like chemotherapy at all either, for example, Reynold Spector, who blighted Skeptical Inquirer with a depressingly nihilistic view of medicine, which earned him a heapin’ helpin’ of some rough and ready not-so-Respectful Insolence. I get it. I do, believe it or not. I understand why chemotherapy is one of the easiest aspects of science-based medicine to demonize. There’s no doubt about it; chemotherapy is toxic. Depending upon the specific drug, it can make your hair fall out, induce vomiting, result in immune suppression, and occasionally result in death. However, when weighed against the prospect of dying from cancer, the side effects, while extensive, can be considered in many cases to be an acceptable risk considering the alternative. Still, the toxicity of current cancer treatments clearly fuels the burgeoning alternative medicine cancer cure industry. People are afraid of chemotherapy and are thus susceptible to promises that they can be cured of cancer without chemotherapy.
That’s one reason why I’m always interested in studies that try to look objectively at the efficacy of chemotherapy in decreasing mortality from cancer. A doozy of just such a study was published yesterday. You’re not likely to see it on Mike Adams’ site, Joe Mercola’s site, or any other alt-med site, except in order to attack it or lie about it, which is why I can’t resist this little pre-emptive strike in which I discuss the study.
Basically, what I’m referring to is a large meta-analysis that appeared in The Lancet yesterday as an E-pub ahead of print. Funded by Cancer Research UK; British Heart Foundation; UK Medical Research Council, this study was carried out by the Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) at the Clinical Trial Service Unit at the University of Oxford, United Kingdom and entitled Comparisons between different polychemotherapy regimens for early breast cancer: meta-analyses of long-term outcome among 100 000 women in 123 randomised trials. Before I discuss the meta-analysis itself, I think it’s worth briefly noting exactly what the EBCTCG is:
400,000 women in 400 randomised trials
Every five years, CTSU brings together updated data on each woman randomised into all trials of the treatment of operable breast cancer. The EBCTCG process was initiated in 1983 and the first cycle collected data for hormonal and cytotoxic therapy in 1985 [6a, 6b]. The collaboration was extended in the 1990s to all aspects of early breast cancer management [7a – 7f] and its results informed the year 2000 NIH consensus development conference on the treatment of early breast cancer . The 2005 report on chemotherapy and endocrine therapy [9a] shows the substantial effects on 15-year survival of the chemotherapy regimens (such as about 6 months of FAC or FEC in women aged <70) and hormonal regimens (such as at least 5 years of tamoxifen in women with ER+ disease) that were being tested in the 1980s. The 2005 report on surgery and radiotherapy [9b] shows that treatments that substantially improve local control have little effect on breast cancer mortality during the first few years, but definite effects by 15 years.
Results from the fifth cycle are emerging (ER-poor [10a], DCIS [10b], aromatase inhibitors [10c], endocrine therapy [10d], radiotherapy after breast-conserving surgery [10e], chemotherapy [in preparation]) while the sixth (2010-2012) cycle of data collection is in progress.
It’s really quite an incredible effort, looking as it does at patient-level data for so many women in so many clinical trials. I sometimes say about meta-analyses the prototypical complaint about meta-analyses, namely that the quality of the output is critically dependent on the quality of the input. In other words, “garbage in, garbage out.” However, the inclusion criteria for the EBCTCG are actually pretty stringent. More importantly, the EBCTCG has access to unpublished data and patient-level information. As is explained here, this is very important as a means of avoiding bias as much as possible:
Where there are several trials that address similar, although not necessarily identical, questions, it is possible to obtain estimates of the differences between treatments by combining the data from them. This approach is much more precise than the estimates given by any individual trial. Inevitably, trials with extreme results tend to receive more attention than those with more moderate results. This produces a natural tendency for unduly selective emphasis on those trials or subcategories of patients where, by chance alone, the results are misleadingly positive or misleadingly negative. Most such biases can be avoided by appropriate combination of the results of all trials that address similar questions. This combination cannot be done satisfactorily from published data alone (Stewart and Parmar 1993), and the inclusion of unpublished as well as published data is necessary to avoid bias. Furthermore, the information available from the published trials is not sufficient to allow a uniform analysis of all the available data with appropriate stratification for factors that will affect survival such as age, time since diagnosis, or nodal status. Thus, analysis based on individual patient data is necessary.
The EBCTCG also goes to great lengths to try to include data from every randomized trial ever published, or an unbiased subset of them, in order to try to minimize selection bias that all too often results from too rigid selection criteria used for meta-analyses. All in all, it’s an enormous effort.
The name of the EBCTCG means exactly what it says, too. Basically, by “early” the EBCTCG means early stage breast cancer; i.e., breast cancer that can be cured by surgery alone. By concentrating on this subset of breast cancer the EBCTCG is able to concentrate on what factors impact survival in conjunction with surgery. The study that was published yesterday was thus designed to estimate the effect of adding various chemotherapy regimens to breast cancer treatment on survival. There are two common chemotherapy regimens for breast cancer. The first, commonly known as ACT, consists of Adriamycin (doxorubicin) and Cytoxan (cyclophosphamide), given together for several doses, followed by a taxane, such as paclitaxel. The other common regimen, more often used in Europe than here in the U.S., is known as CMF and consists of cyclophosphamide, methotrexate, and 5-fluorouracil.
Overall, this meta-analysis involved over 100,000 patients involved in 123 randomized trials over 40 years, and the authors made these comparisons: (1) taxane-based versus non-taxane-based regimens (data for 33 trials, begun in 1994-2003); (2) any anthracyclinebased regimen versus standard or near-standard CMF (20 trials, begun in 1978-97); (3) higher versus lower anthracycline dosage (six trials, begun in 1985-94); and (4) polychemotherapy versus no adjuvant chemotherapy (64 trials, begun in 1973-96, including 22 of various anthra cycline-based regimens and 12 of standard or near-standard CMF). Several meta-analyses were performed, which produced five main findings:
- Standard CMF and standard 4AC (ACT without the “T,” which is an older chemotherapy regimen used before taxanes were developed) were roughly equivalent in efficacy. Both of the roughly halved two-year recurrence rates and resulted in a proportional decrease in recurrence over the next eight years by approximately one-third. Overall, breast cancer mortality rates were reduced proportionally by 20-25%.
- Regimens were lower chemotherapy doses per cycle were less effective.
- Regimens with a lot more chemotherapy than the old standard 4AC (but not so nasty that they required stem-cell rescue) were somewhat more effective. They further decreased breast cancer mortality by 15-20%. the most prominent of these regimens is 4AC plus four cycles of “T” (a taxane), which became the standard of care for node-positive breast cancer after taxanses were developed.
- In all chemotherapy comparisons, the ten year overall mortality was reduced because there was not very much excess mortality due to causes other than breast cancer during the first year.
- In all meta-analyses looking at taxane-based regimens or anthracycline-based regimens (doxorubicin is an anthracycline), the proportional reductions in early recurrence, any recurrence, and breast cancer mortality were more or less independent of age, nodal status, tumor size, or even estrogen receptor status.
This latter finding is actually somewhat surprising, because more recent trials suggest that ER(+) tumors, although sensitive to antiestrogen therapy such as Tamoxifen, are less sensitive to chemotherapy than ER(-) tumors. Yet according to the findings of this meta-analysis, ER(+) and ER(-) tumors, the reduction of approximately one-third in mortality due to breast cancer due to modern chemotherapy regimens like ACT is the same. The autnors conclude:
While awaiting the results of these new trials, it appears that ER status, diff erentiation, and the other tumour characteristics available for the present meta-analyses had little eff ect on the proportional risk reductions with taxane-based or anthracycline-based regimens. The more effective of these regimens offer on average a one-third reduction in 10-year breast cancer mortality, roughly independently of the available characteristics. The absolute gain from a one-third breast cancer mortality reduction depends, however, on the absolute risks without chemotherapy (which, for ER-positive disease, are the risks remaining with appropriate endocrine therapy). Although nodal status and tumour diameter and differentiation are of little relevance to the proportional risk reductions produced by such chemotherapy (and by tamoxifen therapy), they can help in treatment decisions as they are strongly predictive of the absolute risk without chemotherapy, and hence of the absolute benefit that would be obtained by a one-third reduction in that risk.
One aspect of this trial that needs to be emphasized is that there were very few trials of patients with ER(+) tumors with favorable histology. These are the sorts of tumors that probably do not benefit much, if at all, from chemotherapy and can be effectively treated with estrogen-blocking drugs. There are a number of clinical trials right now looking at this very question, using signatures based on a number of genes to classify tumors as high, low, or intermediate risk. The Oncotype DX assay produces one such signature.
The bottom line is that, contrary to what you will hear from cranks and alt-med supporters who believe in “alternative” cancer cures, in the case of early stage breast cancer, chemotherapy saves lives. In women with breast cancer, it decreases the risk of their dying from breast cancer by approximately one-third. This is nothing to sneeze at, as it means thousands upon thousands of women who would have died but did not, thans to chemotherapy. This study simply represents yet another in a long line of studies, another strand in the web of evidence, that support the efficacy of chemotherapy in prolonging the lives of women with breast cancer. It’s not perfect, and it has a lot of potential complications, but it works and in many cases it’s better than the alternative.
While it’s true that chemotherapy decreases a woman’s risk of dying from her breast cancer, the vast majority of women do not individually benefit from chemotherapy. That decrease in risk is based on populations and probabilities. We can’t predict in advance whether chemotherapy will help in an individual patient with an acceptable degree of accuracy, only apply probabilities based on population data. In order to save that one-third, we have to treat most women, who segregate into three groups: those who would have done well without chemotherapy, who are treated unnecessarily; those who would do poorly regardless of chemotherapy, who are also treated unnecessarily; and those for whom chemotherapy is the difference between life and death. What would be far more effective (and far more desirable) would to be able to identify in advance which women would do poorly without chemotherapy but are likely to respond to chemotherapy. Fortunately, with the emerging era of genomic medicine, we are finally developing the tools necessary to identify these women. When that happens, we’ll finally be able to make sure that only the women who can be saved by chemotherapy are the women who receive chemotherapy. I’m hoping that day is not too far off.