Radiation: Oxford Overview

Date: Sun, 08 Dec 1996 22:12:28 -0800 (PST)
From: John Bonine
To: Breast Cancer Discussion List
Subject: Radiation: Oxford Overview

Dear B-C List friends,

I would appreciate any comments on what I've written here. I am nearly finished with a three-part survey of the literature, some of which I have previously posted. It is relevant to treatment decisions my friend is making, and conflicting advice from different doctors.

A doctor has suggested that in thinking about post-mastectomy radiotherapy, I review the study by the Early Breast Cancer Trialists' Collaborative Group, "Effects of Radiotherapy and Surgery in Early Breast Cancer," N. Engl. J. Medicine 333:1444-1455 (Nov. 30, 1995).

In fact, I have reviewed exactly that study over the past month, along with others. (I will refer to it as the "Oxford Overview," since some doctors have used that term.) Most of what I have read makes me favor post-mastectomy (and post-chemotherapy) radiation treatment in my friend's particular situation.

John ================================================================


First, eight short paragraphs of background. It helps me if I state my understanding of the science, in my own words. Please forgive me for trying to do this, without possessing medical credentials.

It is common knowledge that post-mastectomy radiotherapy substantially reduces the likelihood of "local recurrences" (e.g., recurrent tumors on the chest wall, in the mastectomy scar, in the supraclavicular nodes, etc.) by about 2/3. Even after mastectomies, some portion -- around 30%, from different studies -- of women still have some breast cancer cells in the region of the breast (on the chest wall, in the supraclavicular nodes, or elsewhere). Radiation eliminates those in about 2/3 of the women who have them (about 20% of all patients), thus reducing the number with such additional breast cancer cells in the region of the breast to about 10%.

An important question is whether such radiation treatment also increases the chances of survival or not. The answer must be based on at least two additional questions: (1) whether those regional cells would have spread later to dangerous sites in the body, so that for some percentage of women destroying them promotes survival by preventing future distant metastases; and (2) whether the radiation treatment itself causes other forms of damage (primarily heart damage) so that some percentage of women die of a disease caused by the radiation treatment itself.

If the radiation treatment prevents additional, future distant metastases, then this should show up as some benefit in "disease-free survival," meaning additional survival free of distant recurrences -- so that for some percentage who would have died of breast cancer, they do not die of breast cancer.

If radiation treatment causes other sorts of harm, such as to the heart, then this should show up as an increase in non-breast cancer causes of death. If this occurs and is not outweighed by the first, then it should also result in decreased percentages of overall survival. If it does not occur -- or is outweighed by a reduction in breast cancer recurrences -- this should result in increased percentages of overall survival from post-mastectomy radiation treatments.

To put this last sentence another way, if radiation treatment is designed in such a way that it does NOT cause these other effects (e.g., heart damage) and DOES prevent some deaths that would have occurred from distant metastases spreading from the chest region, then we should see an increase in overall survival in the population of women who receive radiotherapy after mastectomy.

To express this in yet a third way, we need to know if there is a type of modern radiation treatment that stops cancer spread, and that spares the heart from damage. If that is done, then any survival benefits of radiotherapy can emerge in the studies.

To apply this directly to the "Oxford Overview" meta-analysis, which I am about to discuss, it seems to me that it makes little sense to include, in a meta-analysis of post-mastectomy radiotherapy, data from heart-damaging techniques that are no longer in use. Neither does it make any sense to include data from radiotherapy techniques that are ineffective against breast cancer recurrences. Yet that is what appears to have happened in the largest attempt to pull data together on post-mastectomy radiotherapy. This aspect has been roundly criticized by some researchers. Furthermore, another recent overview, which avoids this inclusion of data from obsolete techniques, comes to the conclusion that radiotherapy DOES, indeed, enhance survival overall. And some recent studies point in the same direction. Everyone agrees that still more studies are needed. But some people must make decisions on the basis of their best attempts to weigh what we have now.



Early Breast Cancer Trialists' Collaborative Group, "Effects of Radiotherapy and Surgery in Early Breast Cancer," N. Engl. J. Medicine 333:1444-1455 (Nov. 30, 1995).


The Early Breast Cancer Trialists' Collaborative Group (the "Oxford Group") found in their meta-analysis (the "Oxford Overview") no "statistically significant" survival benefit from post-mastectomy radiotherapy, for either node-positive or node-negative women. (p. 1445.)

The authors speculated on the reasons, in terms of one of the reasons that I mentioned in the preceding section of this paper. They wrote, "Hormonal and cytotoxic therapies definitely improved 10-year survival, but radiotherapy did not, perhaps because moderate protection was counterbalanced by a moderate increase in risk." (p.1444.) That risk mostly involved heart disease.

As for the amount of potential breast cancer benefit from radiotherapy, the Oxford group authors noted that "radiotherapy may well produce a moderate reduction in deaths due to breast cancer," but they stated that the findings could also reveal only a "negligibly small effect from radiotherapy." (p. 1447.) They observed that radiotherapy did "reduce[] the rate of local recurrence by about two-thirds," but there was no equivalent reduction in deaths. The reason was that when local recurrences were eliminated, distant recurrences often showed up first in their place, and many women still died of breast cancer. They concluded that it was "impossible to assess, on the basis of the data currently available, whether radiotherapy had any protective effect against distant recurrence." (p. 1448.)

This alone would be disappointing if the data were based on modern techniques. In fact, however, the authors acknowledged that "the radiotherapy techniques differed substantially among the studies" that they amassed together. They asserted that despite this mixing of old and new studies, "the overall result still provides a valid measure of the value of such treatment." (p. 1450.) But they did not, however, explain why the value of modern radiotherapy could be assessed properly if the overall survival rates from older, heart-harmful radiotherapy techniques were mixed together with (and therefore canceled out) the overall survival rates from modern, heart-sparing radiotherapy.


Despite mixing old and new techniques together for the purpose of calculating single-number survival figures from radiotherapy, the Oxford group did seem to grasp the issues clearly. They posed the relationship between prevention of local recurrences and long-term survival clearly: "The central question about local therapy for early breast cancer is whether more-extensive treatment significantly reduces long-term morality from breast cancer." In other words, putting concerns over potential heart damage to the side for one moment, the first question to address is whether post-mastectomy radiotherapy prevents some distant metastases and therefore prevents some breast cancer deaths.

In the Oxford meta-analysis of numerous trials, the risk of death FROM BREAST CANCER was in fact found to be reduced by 6% through the use of radiotherapy. This was statistically significant at P = 0.03. They commented that six percent is not a large advantage. "But even a small difference could be important, especially if any hazards of the treatment could be limited. With radiotherapy, there is a small, marginally significant (P = 0.03) reduction in mortality due to breast cancer but not in overall mortality. However, because the 'breast-cancer' deaths do include some deaths from other causes [every death after a recurrence was counted as being 'from breast cancer' even if it was not, for example], the effect could be somewhat larger than the 6 percent reduction seen in this overview." (p. 1451.)


One question is this: What competing causes of death took the place of breast cancer in the trials that the Oxford Group analyzed, such that there was little overall survival benefit?

The answer, of course, is primarily heart damage. However, that damage showed up mostly with the older radiotherapy techniques, according to critics of the Oxford Overview (see below). These older techniques have now been replaced by newer, more heart-sparing techniques, at least in the West. The older techniques included orthovoltage instead of megavoltage therapy; large (and more heart-harmful) daily radiation fractions rather than smaller daily amounts given for more days; radiation of internal mammary nodes rather than avoiding that area, which is close to the heart; and radiation of left breasts, which exposes the heart to more damaging radiation. The overall survival figures in the Oxford meta-analysis, in other words, counted data from techniques that did raise incidences of death from heart damage, but which are not representative of the overall population receiving post-mastectomy radiotherapy today.

Dr. Abram Recht of Harvard Medical School wrote in the December 1995 issue of the Journal of Clinical Oncology (at the same time that the New England Journal of Medicine was publishing the Oxford Overview meta-analysis), "The long-term morbidity of modern radiotherapy is small. Long-term mortality rates from cardiac and other causes from currently used radiotherapy techniques are likely to be less than 1%."


The Oxford authors point out that choices regarding radiotherapy can be made for reasons of local control alone, but "if [survival] benefits differences could be reliably demonstrated they would be important. Any differences between these local therapies do not involve large effects on 10-year survival, but they could still involve worthwhile effects on longer-term survival."

In short, the Oxford authors do not recommend against the use of radiation treatment after mastectomy. Neither do they recommend it. They just report their results and speculate about potential additional benefits.


Where does this leave us? Do we take the minimal or almost non-existent survival benefits compiled in the Oxford Overview to be the final word -- or the most authoritative word?

The answer, I think, is "No." We must weigh carefully what others have said about the Oxford Overview and how the authors responded, and then look at both other research and a variety of opinions. For reasons that I will outline below, I think that we should take the Oxford Overview to be only part of the data that we must examine. My conclusion from all my reading is that modern radiotherapy can now be credited with giving a marginal survival advantage to some women.



When I examine opinions of other experts regarding whether all the Oxford Overview data should be weighed equally, or whether a few of the 35-36 studies involving radiotherapy that they examined are more indicative of what we can expect than the majority of these 35-36 studies, I find the criticisms by these experts to be cogent ones. I believe that a survival benefit from post-mastectomy radiotherapy emerges from parts of the Oxford data (when modern techniques are used).


Abram Recht, "Radiotherapy and Surgery in Early Breast Cancer," N. Engl. J. Med. 334:989 (April 11, 1996) (letter).

Four months after the Oxford Overview was published in the NEJM, Dr. Abram Recht of Harvard Medical School published a letter in the NEJM criticizing that Overview's failure to identify a survival benefit.

Dr. Recht pointed out, "Some treatments in the studies analyzed are obsolete (e.g., orthovoltage radiotherapy, radical mastectomy, and single-agent chemotherapy). It would be helpful to group together the trials that used only therapies currently in widespread use."

Dr. Recht next made an argument regarding four of the trials. Since he made that argument in more detail in a publication July 1996, I will delay discussing this argument until I reach that point. However, he found a survival advantage to post-mastectomy radiotherapy from four newer trials.


Richard Peto, et al., "Author's Reply," N. Engl. J. Med. 334:989 (April 11, 1996).

The authors of the Oxford Overview replied to Dr. Recht that they found "no clearly significant differences in survival, either when each each trial was considered separately or when a meta-analysis of all the results was performed."

Although Dr. Recht found in the four most "modern" trials a survival advantage, Peto, et al., doubted the reality of that advantage. They wrote that it is "entirely possible that unduly selective emphasis and the play of chance are chiefly responsible for the marginally significant survival advantage that Dr. Recht finds when he restricts his attention to just 4 of the studies (and ignores the other 32)."


Lori Pierce & Allen Lichter, "Defining the Role of Post-Mastectomy Radiotherapy: The New Evidence," Oncology, 10:991-1002 (July 1996).

In a review article mostly devoted to other studies (which I will discuss later), Drs. Lori Pierce and Allen Lichter of the University of Michigan Medical School made some comments on the information in the Oxford Overview. Like Dr. Abram Recht, they felt that looking at fewer than all of the studies gave more information, at least on the effectiveness of radiation therapy on breast cancer itself.

They stated that useful information on the causes of death was available from 28 out of 35 trials in the Oxford Overview. Excluding the other 7 trials, they said that by looking only at the 28, one finds that 36.9% of "controls" (women who did not receive radiotherapy) died from breast cancer, compared with 34.1% of the women who received radiotherapy. They repeated the fact that the overall survival benefit was just 1.1% and that that number did not achieve statistical significance.

"However, the reduction in breast cancer-related deaths among women treated with radiotherapy was ... suggestive of improved breast cancer-specific survival due to maximal local control." (p. 993.) In other words, they tend to believe that the radiotherapy did help prevent some deaths that would have been due to breast cancer. Of course, the Oxford authors agree on that narrow point; they were just not convinced of overall benefit, because of heart damage in some of the studies.


Abram Recht, "The Pierce-Lichter Article Reviewed," Oncology, 10:1006 (July 1996).

In the same issue of Oncology, seven months after the "Oxford Overview" was published in the Nov. 30, 1995, Dr. Abram Recht of Harvard Medical School broadened his criticisms of the Oxford Overview. He used a brief commentary on the Pierce & Lichter article as an opportunity to return again to the subject of the Oxford Overview. He felt that Pierce & Lichter should have been overtly critical of the Oxford Overview.

Dr. Recht criticized the fact that "the global results of the Oxford overview resulted from mixing together trials employing breast-conserving surgery, simple mastectomy without axillary dissection, modified radical mastectomy, or radical mastectomy and either orthovoltage or megavoltage radiotherapy, without regard to the relative effectiveness of these various treatments in preventing locoregional relapse."

Dr. Recht wrote that this massive mixing together of trials employing different therapies makes the small size of the 1.1% improvement in survival from radiotherapy (a number from the Oxford Overview) "irrelevant." He argued, "A more relevant grouping of trials ... would include only studies in which patients were treated by modified radical mastectomy and multiagent chemotherapy and were randomized to receive or not receive megavoltage radiotherapy."

Dr. Recht wrote that there were 4 such studies in the total of 35 in the Oxford Overview. These 4 studies (because they involved chemotherapy) consisted mostly of node-positive women. These 4 studies are the Danish BCG 82b, Vancouver (Canada), Helsinki (Finland), and BMFT 03 Germany trials.

Ignoring more than 30 studies that used older techniques or that involved node-negative women or other characteristics, and only looking at these 4, Dr. Recht calculated that 22.5% of a total of 1,072 control patients died during the periods of observation, but only 18.8% of the 1,064 irradiated patients died.

In short, Dr. Recht finds a survival benefit from modern radiotherapy for node-positive women who undergo mastectomies and multiagent chemotherapy.


I have been unable to find any reply to Dr. Recht by Dr. Peto, et al., of the Oxford Group after Recht's July 1996 criticism in the Journal of Clinical Oncology.


to be continued. John