* “Lumpectomy followed by breast irradiation, as compared with lumpectomy alone, was associated with a marginally significant decrease in deaths due to breast cancer (P = .04), as reported in the 20-year follow-up of the NSABP B-06 trial.”
At the time of our 2002 report of the B-06 study, as a consequence of radiation therapy administered after lumpectomy, there was a 7.2% decrease (43.6% v 36.4%) in breast cancer-related mortality. That advantage was mainly offset by a 6% increase in deaths from other causes in the irradiated group. Thus the 20-year decrease in all-cause mortality was only 1.2%. That small mortality decrease was associated with a substantial (24%) reduction in the incidence of IBTR as a result of radiation therapy. Thus those results support one of the tenets of our alternative hypothesis, that is, that variations in locoregional therapy are unlikely to substantially affect survival.
With regard to the effect of radiation therapy on survival, Rabinovitch and Kavanagh also assert the following:
* “There was a highly significant reduction in the annual breast cancer mortality rate for patients treated with radiotherapy after lumpectomy versus lumpectomy alone (breast cancer death rate ratio, 0.83; 95% CI, 0.75 to 0.91; 2P = .0002), as reported in a meta-analysis by Early Breast Cancer Trialists’ Collaborative Group [EBCTCG].”
Unfortunately, this statement fails to indicate the whole story with regard to the relationship between the use of postoperative radiation and survival outcome. In the 2000 EBCTCG overview, the favorable, as well as the unfavorable, effects of radiation therapy on long-term survival were considered. In their report, the authors of that meta-analysis presented a “forest-plot” that depicted the all-cause mortality in each of 40 randomized trials involving 20,175 women, half of whom received radiotherapy after surgery, and half of whom did not. Twenty-seven of the trials began in the 1960s and 1970s; 13 started in the 1980s. The trials were grouped according to the type of surgery performed, that is, mastectomy alone (five trials), mastectomy with axillary sampling (six trials), mastectomy with axillary clearance (23 trials), and breast conservation with axillary clearance (six trials). On examination of that plot, it is clearly evident that, among the 40 trials, with one exception, there is little, and mostly no significant difference in the annual death rates between the group that received radiation therapy and the control group. In all of the 40 trials combined, a nonsignificant 3.9% reduction in the mortality rate ratio was reported. In three of the four groups of trials, the reduction in the death rate resulting from postoperative radiation was not significant. Only in the group of trials in which patients were treated with mastectomy and axillary sampling was there a significant reduction in the ratio of the annual death rate due to the favorable results obtained from two large trials conducted by the Danish Breast Cancer Study Group. Thus, in keeping with the findings that we obtained in the B-06 trial, the evidence from the year 2000 EBCTCG meta-analysis demonstrated that, although postoperative radiation therapy resulted in a substantial benefit in local recurrence, there was only a small benefit (proportional reduction of 3.9%) in overall mortality.
The information in Table 1 is from several figures that appeared in the year 2000 EBCTCG overview and is related to the absolute effect of radiation therapy on cause-specific survival after 20 years of follow-up.
Because the previous statement by Rabinovitch and Kavanagh that there was a “highly significant reduction in annual breast cancer mortality rate from patients treated with radiotherapy…” was selected from the vast amount of data and commentary in a more recent EBCTCG overview (year 2005), we deemed it appropriate to provide balance to that statement by presenting additional information from that document. The authors of that overview noted that although the 5-year local recurrence was reduced in patients who received breast-conserving surgery, with and without radiation therapy (from 26% to 7%), they also indicated that radiation therapy produced a moderate absolute reduction not only in 15-year breast cancer mortality but also in 15-year overall mortality, ie, 5.4% and 5.3%, respectively.
Although almost all of the information presented in the 2005 EBCTCG overview was related to breast cancer mortality and only a small amount to all-cause mortality, it was noted by the authors that there was a “significant excess incidence of contralateral breast cancer… and a significant excess of non-breast-cancer mortality in irradiated women. ... ” The latter was mainly the result of heart disease and lung cancer. Thus, this significant excess in nonbreast cancer mortality reduced the impact of the reduction in breast cancer mortality putatively related to radiation therapy. The authors of the 2005 meta-analysis demonstrated that radiation therapy produced a moderate absolute reduction not only in 15-year breast cancer mortality but also in 15-year overall mortality, that is, 5.4% and 4.4%, respectively.
After scrutiny of the plethora of analyses conducted and the data obtained from the years 2000 and 2005 EBCTCG meta-analyses, several uncertainties exist with regard to both the process by which the findings were obtained and the interpretation of their meaning. In that regard, the question may be asked as to why, in the year 2000 overview, breast cancer–related mortality, nonbreast cancer–related mortality, and mortality from any cause were reported through 20 years of follow-up, whereas, 5 years later, in the report of the 2005 analyses, results were presented through 15 years. Also, why were data regarding nonbreast cancer deaths, and, consequently, all-cause mortality, less available than were data for estimating death due to cancer? Moreover, it seems that insufficient attention has been directed to the fact that, in the 2005 overview, women in 34 (74%) of the 46 trials received a variety of systemic therapy regimens in addition to radiation therapy. The composition of the regimens administered is often difficult to determine and to categorize by an examination of the data presented in a web table. Tamoxifen was given in some trials with—and in others without—single- or multiagent chemotherapy. In some instances, the use of tamoxifen was related to the estrogen content of tumors, and in others, it was not. Ovarian irradiation or ablation was performed in some of the trials with, and in others without, chemotherapy. Also, one cannot ignore the fact that the dosage of radiation therapy administered among trials varied relative to both site and dose. Finally, it remains a matter of concern regarding whether or not all of those variables might have confounded the findings with respect to radiation therapy.