- Email: [email protected]
Effects of radiotherapy and surgery for early breast cancer – Authors' reply
Correspondence
3
4 5
Kunkler I, Williams L, Prescott R, King C. Re: Breast-conserving surgery with or without radiotherapy: pooled-analysis for risks of ipsilateral breast tumor recurrence and mortality. J Natl Cancer Inst 2004; 96: 1255–57. Kunkler I. PRIME II breast cancer trial. Clin Oncol (R Coll Radiol) 2004; 16: 447–48. Stephens R. The dangers of subgroup analysis. Lancet Oncol 2001; 2: 9.
Ratio breast cancer deaths
Every overview from the Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) has represented a milestone. The latest overview of post-surgery radiotherapy in breast cancer1 meets the high expectations to which we have been accustomed. It confirms the insight of our late colleague Jan Van de Steene who observed years ago that the survival benefit of radiotherapy became increasingly significant with longer follow-up and with improved maturity of the trials.2 This is important, because the overview’s confirmation challenges the common perception that radiation toxicity is delayed. Make no mistake: radiation is a harmful environmental hazard. But what Jan and the overview show is that the radiation hazard from controlled clinical trials is constant over time and hence reasonably predictable, and not an increasingly overwhelming hazard like that of accidental exposures or from nuclear devastations. The 2005 overview incorrectly cites our pooled analysis3 as a review of just the published results of radiotherapy trials after breast-conserving therapy. First, our pooled analysis identified more trials than any previously. Second, it ascertained whether or not the trials were still active. And third, contrary to 1·4
We declare that we have no conflict of interest.
1·2
Claire Verschraegen, *Vincent Vinh-Hung
1·0
[email protected]
0·8
University of New Mexico, Cancer Research and Treatment Center, Albuquerque, NM, USA (CV); and Oncology Center, AZ-VUB, 1090 Jette, Belgium (VV-H)
0·6 0·4 0·1
0·2
0·3
0·4
0·5
Ratio recurrence rates
Figure: Breast conserving surgery: lack of correlation between breast cancer mortality and isolated recurrences
1654
the overview’s affirmation, we also used unpublished data. Our pooled analysis concerned the overall survival benefit, or lack thereof, that could be expected from radiation treatment. We provided a conservative estimate of an 8·6% relative gain in reduction of mortality. A previous EBCTCG report indicated a 6% proportional reduction in any-cause mortality.4 The 2005 overview provides no updated overall mortality figure and hence is uninformative for treatment decisions. There is also no mention of the major difference between the overview and our pooled analysis. Whereas we considered that the reduction in recurrences did not explain the mortality reduction,3 the 2005 overview equates four local recurrences avoided with one breast cancer death avoided. This is not supported by the overview itself: the plot of the overview’s ratios of breast cancer deaths as a function of recurrences shows no correlation, and there is even a suggestion that the two trials with the most recurrence reductions are associated with an increased risk of breast cancer death (figure). In keeping with our modelling of population data,4 our pooled analysis cautions against any Halstedian interpretation of results and argues for conservative local management in early breast cancer. The overview’s interpretation forces aggressive local treatment, since radical mastectomy in node-negative patients is associated with the lowest risk of local recurrence.5 This is a critical issue that should be openly debated.
1
Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Effects of radiotherapy and of differences in the extent of surgery for early breast cancer on local recurrence and 15-year survival: an overview of the randomised trials. Lancet 2005; 366: 2087–106.
2
3
4
5
Van de Steene J, Vinh-Hung V, Cutuli B, Storme G. Adjuvant radiotherapy for breast cancer: effects of longer follow-up. Radiother Oncol 2004; 72: 35–43. Vinh-Hung V, Verschraegen C, The Breast Conserving Surgery Project. Breast-conserving surgery with or without radiotherapy: pooled-analysis for risks of ipsilateral breast tumor recurrence and mortality. J Natl Cancer Inst 2004; 96: 115–21. Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Favourable and unfavourable effects on long-term survival of radiotherapy for early breast cancer: an overview of the randomised trials. Lancet 2000; 355: 1757–70. Verschraegen C, Vinh-Hung V, Cserni G, et al. Modeling the effect of tumor size in early breast cancer. Ann Surg 2005; 241: 309–18.
Authors’ reply The EBCTCG report presented metaanalyses of the randomised trials of local treatments in early breast cancer that began by 1995. Individual patients’ data were obtained on 7000 women in 10 trials of radiotherapy versus no radiotherapy after breast-conserving surgery (BCS), 10 000 women in 25 trials of radiotherapy versus no radiotherapy after mastectomy and axillary clearance, and 25 000 women in trials of various other local treatment comparisons. The main finding was that improvements in the initial treatment of local disease (by radiotherapy or by more extensive surgery) can eventually reduce breast cancer mortality. In aggregate, the treatment comparisons that involved a substantial reduction in the 5-year risk of local recurrence also involved a moderate reduction in the 15-year risk of death from breast cancer. The figure shows the combined evidence from all treatment comparisons that produced an absolute reduction of 10% or more in the 5-year local recurrence risk. (The other comparisons involved, on average, little difference in local recurrence or in breast cancer mortality.) These results suggest that, for every four local recurrences avoided by differences in local treatment, about one breast cancer death will eventually be avoided. This does not, as some correspondents suggest, assume a causal relation www.thelancet.com Vol 367 May 20, 2006
Correspondence
Isolated local recurrence (%)
60 50
5-year gain 18·7% (SE 0·5)
40
No radiotherapy 33·2%
30 20
Radiotherapy 11·5%
10 0
Breast cancer mortality (%)
60
15-year gain 5·0% (SE 0·8) Log-rank p <0·0001
50
49·5% 44·6%
40
No radiotherapy
Radiotherapy
30 20 10 0 0
5 10 Time (years)
15
Figure: Probabilities of 15-year local recurrence and breast cancer mortality for 12 randomised treatment comparisons that produced more than 10% absolute reduction in 5-year local recurrence risk Comparisons include 25 000 women, half with nodepositive disease, 90% in trials of radiotherapy versus no radiotherapy. A reduction of about 20% in isolated local recurrence, mostly during the first few years, is followed by a highly significant reduction of about 5% in breast cancer mortality, mostly in later years.
between local recurrence and mortality (indeed, some deaths avoidable by better local treatment might not be preceded by any diagnosis of local recurrence). Instead, it shows causal relations between local treatment and local recurrence, and between local treatment and mortality: additional local treatment that causes substantially fewer women to develop local recurrence also causes somewhat fewer to die of breast cancer within 15 years. For the particular case of radiotherapy, there were two important circumstances in which the reduction in breast cancer mortality carried through into a reduction in 15-year overall mortality: after BCS, the 15year mortality was 35·2% in the radiotherapy group versus 40·5% in controls (reduction 5·3%, SE 1·8; www.thelancet.com Vol 367 May 20, 2006
p=0·005), and after mastectomy with axillary clearance in node-positive disease the 15-year mortality was 59·8% in the radiotherapy group versus 64·2% in controls (reduction 4·4%, SE 1·2; p=0·0009). This is the first time that meta-analysis of individual patients’ data has shown that radiotherapy can improve longterm survival—although, as some correspondents note, relatively few older women were randomised. After BCS, radiotherapy substantially reduces local recurrence, chiefly in the conserved breast, both in nodenegative and in node-positive disease, so the mortality results for these were presented in combination as well as separately. After mastectomy and axillary clearance, however, the absolute effect of radiotherapy on local recurrence is small in node-negative disease and large in node-positive disease, so the mortality results for these were presented separately only. For similar reasons, we did not, as some correspondents request, present analyses that combine all radiotherapy trials, although the components of such an analysis are readily available on The Lancet’s website, linked to the EBCTCG report.
*S Darby, P McGale, R Peto, C Taylor, on behalf of the EBCTCG Secretariat [email protected] Clinical Trial Service Unit (CTSU), Richard Doll Building, Old Road Campus, University of Oxford, Oxford OX3 7LF, UK
Spinning stars: ASTEROID’s impact on atherosclerosis The publicity surrounding the ASTEROID trial1 does not reflect well on the pharmaceutical industry, the media, or many of our own profession. Virtually every newspaper and broadcast news bulletin carried statements to the effect that doctors had believed that atherosclerosis was irreversible and that drugs could do nothing about it.
This is of course nonsense. It has been known for many years that atherosclerosis is a dynamic process, which can regress as well as progress. Multifactorial intervention to reduce atherosclerosis progression and events is well established and forms part of the UK’s National Service Framework for Coronary Heart Disease from 2000.2 Many studies with statins and other drugs even using older imaging techniques and small numbers have shown that regression can actually happen.3 None of these studies followed the design of the ASTEROID study, which did have a relatively large number of participants (507) but had no control group and reported only evaluable lesions (349 of 507). The lipid changes in ASTEROID have been over-dramatised and the levels achieved simply follow the targets recommended in the Joint British Societies consensus guidelines of 2005.4 These lipid levels were achieved using a dose of rosuvastatin (40 mg) which most clinicians and the regulatory agencies would regard with caution.5 Thus we are left with the perspective of grand hyperbole when reviewing the attention lavished on this study as opposed to the science behind it. It is difficult to avoid the conclusion that the media quite uncritically used the press release from AstraZeneca even though at least some of the journalists concerned should have known that there was very substantial “spin” in the claims. Only one doctor (Peter Weissberg) properly stated that surrogate endpoint studies, however encouraging, require proof in outcome trials. Certainly some of the other doctors who commented must have known this and one can only wonder about their potential conflicts of interest given the close relationships between many clinicians (ourselves included) and pharmaceutical companies. What we have seen, therefore, in this depressing and probably prescient tale is a very successful avoidance of the UK’s ban on direct advertising to patients by disguising it as a sensational news 1655
3
4 5
Kunkler I, Williams L, Prescott R, King C. Re: Breast-conserving surgery with or without radiotherapy: pooled-analysis for risks of ipsilateral breast tumor recurrence and mortality. J Natl Cancer Inst 2004; 96: 1255–57. Kunkler I. PRIME II breast cancer trial. Clin Oncol (R Coll Radiol) 2004; 16: 447–48. Stephens R. The dangers of subgroup analysis. Lancet Oncol 2001; 2: 9.
Ratio breast cancer deaths
Every overview from the Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) has represented a milestone. The latest overview of post-surgery radiotherapy in breast cancer1 meets the high expectations to which we have been accustomed. It confirms the insight of our late colleague Jan Van de Steene who observed years ago that the survival benefit of radiotherapy became increasingly significant with longer follow-up and with improved maturity of the trials.2 This is important, because the overview’s confirmation challenges the common perception that radiation toxicity is delayed. Make no mistake: radiation is a harmful environmental hazard. But what Jan and the overview show is that the radiation hazard from controlled clinical trials is constant over time and hence reasonably predictable, and not an increasingly overwhelming hazard like that of accidental exposures or from nuclear devastations. The 2005 overview incorrectly cites our pooled analysis3 as a review of just the published results of radiotherapy trials after breast-conserving therapy. First, our pooled analysis identified more trials than any previously. Second, it ascertained whether or not the trials were still active. And third, contrary to 1·4
We declare that we have no conflict of interest.
1·2
Claire Verschraegen, *Vincent Vinh-Hung
1·0
[email protected]
0·8
University of New Mexico, Cancer Research and Treatment Center, Albuquerque, NM, USA (CV); and Oncology Center, AZ-VUB, 1090 Jette, Belgium (VV-H)
0·6 0·4 0·1
0·2
0·3
0·4
0·5
Ratio recurrence rates
Figure: Breast conserving surgery: lack of correlation between breast cancer mortality and isolated recurrences
1654
the overview’s affirmation, we also used unpublished data. Our pooled analysis concerned the overall survival benefit, or lack thereof, that could be expected from radiation treatment. We provided a conservative estimate of an 8·6% relative gain in reduction of mortality. A previous EBCTCG report indicated a 6% proportional reduction in any-cause mortality.4 The 2005 overview provides no updated overall mortality figure and hence is uninformative for treatment decisions. There is also no mention of the major difference between the overview and our pooled analysis. Whereas we considered that the reduction in recurrences did not explain the mortality reduction,3 the 2005 overview equates four local recurrences avoided with one breast cancer death avoided. This is not supported by the overview itself: the plot of the overview’s ratios of breast cancer deaths as a function of recurrences shows no correlation, and there is even a suggestion that the two trials with the most recurrence reductions are associated with an increased risk of breast cancer death (figure). In keeping with our modelling of population data,4 our pooled analysis cautions against any Halstedian interpretation of results and argues for conservative local management in early breast cancer. The overview’s interpretation forces aggressive local treatment, since radical mastectomy in node-negative patients is associated with the lowest risk of local recurrence.5 This is a critical issue that should be openly debated.
1
Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Effects of radiotherapy and of differences in the extent of surgery for early breast cancer on local recurrence and 15-year survival: an overview of the randomised trials. Lancet 2005; 366: 2087–106.
2
3
4
5
Van de Steene J, Vinh-Hung V, Cutuli B, Storme G. Adjuvant radiotherapy for breast cancer: effects of longer follow-up. Radiother Oncol 2004; 72: 35–43. Vinh-Hung V, Verschraegen C, The Breast Conserving Surgery Project. Breast-conserving surgery with or without radiotherapy: pooled-analysis for risks of ipsilateral breast tumor recurrence and mortality. J Natl Cancer Inst 2004; 96: 115–21. Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Favourable and unfavourable effects on long-term survival of radiotherapy for early breast cancer: an overview of the randomised trials. Lancet 2000; 355: 1757–70. Verschraegen C, Vinh-Hung V, Cserni G, et al. Modeling the effect of tumor size in early breast cancer. Ann Surg 2005; 241: 309–18.
Authors’ reply The EBCTCG report presented metaanalyses of the randomised trials of local treatments in early breast cancer that began by 1995. Individual patients’ data were obtained on 7000 women in 10 trials of radiotherapy versus no radiotherapy after breast-conserving surgery (BCS), 10 000 women in 25 trials of radiotherapy versus no radiotherapy after mastectomy and axillary clearance, and 25 000 women in trials of various other local treatment comparisons. The main finding was that improvements in the initial treatment of local disease (by radiotherapy or by more extensive surgery) can eventually reduce breast cancer mortality. In aggregate, the treatment comparisons that involved a substantial reduction in the 5-year risk of local recurrence also involved a moderate reduction in the 15-year risk of death from breast cancer. The figure shows the combined evidence from all treatment comparisons that produced an absolute reduction of 10% or more in the 5-year local recurrence risk. (The other comparisons involved, on average, little difference in local recurrence or in breast cancer mortality.) These results suggest that, for every four local recurrences avoided by differences in local treatment, about one breast cancer death will eventually be avoided. This does not, as some correspondents suggest, assume a causal relation www.thelancet.com Vol 367 May 20, 2006
Correspondence
Isolated local recurrence (%)
60 50
5-year gain 18·7% (SE 0·5)
40
No radiotherapy 33·2%
30 20
Radiotherapy 11·5%
10 0
Breast cancer mortality (%)
60
15-year gain 5·0% (SE 0·8) Log-rank p <0·0001
50
49·5% 44·6%
40
No radiotherapy
Radiotherapy
30 20 10 0 0
5 10 Time (years)
15
Figure: Probabilities of 15-year local recurrence and breast cancer mortality for 12 randomised treatment comparisons that produced more than 10% absolute reduction in 5-year local recurrence risk Comparisons include 25 000 women, half with nodepositive disease, 90% in trials of radiotherapy versus no radiotherapy. A reduction of about 20% in isolated local recurrence, mostly during the first few years, is followed by a highly significant reduction of about 5% in breast cancer mortality, mostly in later years.
between local recurrence and mortality (indeed, some deaths avoidable by better local treatment might not be preceded by any diagnosis of local recurrence). Instead, it shows causal relations between local treatment and local recurrence, and between local treatment and mortality: additional local treatment that causes substantially fewer women to develop local recurrence also causes somewhat fewer to die of breast cancer within 15 years. For the particular case of radiotherapy, there were two important circumstances in which the reduction in breast cancer mortality carried through into a reduction in 15-year overall mortality: after BCS, the 15year mortality was 35·2% in the radiotherapy group versus 40·5% in controls (reduction 5·3%, SE 1·8; www.thelancet.com Vol 367 May 20, 2006
p=0·005), and after mastectomy with axillary clearance in node-positive disease the 15-year mortality was 59·8% in the radiotherapy group versus 64·2% in controls (reduction 4·4%, SE 1·2; p=0·0009). This is the first time that meta-analysis of individual patients’ data has shown that radiotherapy can improve longterm survival—although, as some correspondents note, relatively few older women were randomised. After BCS, radiotherapy substantially reduces local recurrence, chiefly in the conserved breast, both in nodenegative and in node-positive disease, so the mortality results for these were presented in combination as well as separately. After mastectomy and axillary clearance, however, the absolute effect of radiotherapy on local recurrence is small in node-negative disease and large in node-positive disease, so the mortality results for these were presented separately only. For similar reasons, we did not, as some correspondents request, present analyses that combine all radiotherapy trials, although the components of such an analysis are readily available on The Lancet’s website, linked to the EBCTCG report.
*S Darby, P McGale, R Peto, C Taylor, on behalf of the EBCTCG Secretariat [email protected] Clinical Trial Service Unit (CTSU), Richard Doll Building, Old Road Campus, University of Oxford, Oxford OX3 7LF, UK
Spinning stars: ASTEROID’s impact on atherosclerosis The publicity surrounding the ASTEROID trial1 does not reflect well on the pharmaceutical industry, the media, or many of our own profession. Virtually every newspaper and broadcast news bulletin carried statements to the effect that doctors had believed that atherosclerosis was irreversible and that drugs could do nothing about it.
This is of course nonsense. It has been known for many years that atherosclerosis is a dynamic process, which can regress as well as progress. Multifactorial intervention to reduce atherosclerosis progression and events is well established and forms part of the UK’s National Service Framework for Coronary Heart Disease from 2000.2 Many studies with statins and other drugs even using older imaging techniques and small numbers have shown that regression can actually happen.3 None of these studies followed the design of the ASTEROID study, which did have a relatively large number of participants (507) but had no control group and reported only evaluable lesions (349 of 507). The lipid changes in ASTEROID have been over-dramatised and the levels achieved simply follow the targets recommended in the Joint British Societies consensus guidelines of 2005.4 These lipid levels were achieved using a dose of rosuvastatin (40 mg) which most clinicians and the regulatory agencies would regard with caution.5 Thus we are left with the perspective of grand hyperbole when reviewing the attention lavished on this study as opposed to the science behind it. It is difficult to avoid the conclusion that the media quite uncritically used the press release from AstraZeneca even though at least some of the journalists concerned should have known that there was very substantial “spin” in the claims. Only one doctor (Peter Weissberg) properly stated that surrogate endpoint studies, however encouraging, require proof in outcome trials. Certainly some of the other doctors who commented must have known this and one can only wonder about their potential conflicts of interest given the close relationships between many clinicians (ourselves included) and pharmaceutical companies. What we have seen, therefore, in this depressing and probably prescient tale is a very successful avoidance of the UK’s ban on direct advertising to patients by disguising it as a sensational news 1655