Correspondence

Autologous Bone Marrow Transplantation versus MACOP-B in B-Cell Lymphoma

N Engl J Med 1997; 337:711-712September 4, 1997

Article

To the Editor:

The report by Gianni et al. (May 1 issue)1 concludes that high-dose sequential therapy is superior to a regimen of methotrexate, doxorubicin, cyclophosphamide, vincristine, prednisone, and bleomycin (MACOP-B) in a selected group of patients with aggressive non-Hodgkin's lymphoma. This conclusion could potentially change the standard treatment in a large number of patients with diffuse large-cell lymphoma of the B-cell type who are younger than 60 years. I should like to raise two questions.

The first question concerns the number of deaths. The authors report that four patients (8 percent) had fatal toxic reactions to the high-dose sequential therapy. However, according to an abstract the authors presented at the meeting of the American Society of Clinical Oncology in 1994, there were six deaths (four due to veno-occlusive diseases in patients who were positive for hepatitis B and two due to infection).2 As a result of these deaths, the authors replaced total-body irradiation with mitoxantrone (60 mg per square meter of body-surface area) as part of the myeloablative therapy and excluded from enrollment patients who were positive for hepatitis B. Were these six patients who died excluded from the final analysis in the current report? If the four patients who were positive for hepatitis B were excluded, can the study be justified as a prospective randomized trial? Or were all the patients who were positive for hepatitis B excluded retrospectively?

The second question concerns the efficacy of mitoxantrone in comparison with that of total-body irradiation. Thirty patients received high-dose melphalan and total-body irradiation as myeloablative therapy, and the other 18 patients received high-dose melphalan and mitoxantrone. The authors report that nine patients died during a median follow-up period of 55 months and that four of the nine deaths were related to toxic effects. The other five patients died from progressive disease. How many of these five patients received mitoxantrone?

Tony S. Mok, M.D.
Chinese University of Hong Kong, Hong Kong, China

2 References

1. 1

   Gianni AM, Bregni M, Siena S, et al. High-dose chemotherapy and autologous bone marrow transplantation compared with MACOP-B in aggressive B-cell lymphoma. N Engl J Med 1997;336:1290-1297
   Full Text | Web of Science | Medline

2. 2

   Gianni AM, Bregni M, Siena S, et al. 5-Year update of the Milan Cancer Institute randomized trial of high-dose sequential (HDS) vs MACOP-B therapy for diffuse large-cell lymphomas. Prog Proc Am Soc Clin Oncol 1994;13:373-373 abstract.
   

To the Editor:

The study by Gianni et al. reports significant differences (P<0.05) in rates of complete response, freedom from disease progression, and event-free survival, all favoring the group of patients who received high-dose sequential chemotherapy. The authors state that the difference in overall survival between the two treatment groups was marginally significant, with a P value of 0.09. Readers may interpret this statement as indicating that the improvement in overall survival with high-dose sequential therapy was marginally significant, whereas, in fact, the trial had no statistical power to detect clinically significant differences in survival.

With a median follow-up of 55 months, there were only 27 deaths: 9 in the group receiving high-dose sequential chemotherapy and 18 in the group receiving MACOP-B. On the basis of the reported number of deaths, we calculated that the study had less than 50 percent power to detect a difference in median survival of 100 percent between the two groups.1 Stated differently, if one of the treatments was effective enough to double the median survival, the study had less than a 50 percent chance of detecting this result. When authors make statements about survival in a trial, they should discuss negative results in the light of the trial's statistical power to detect a difference in survival.

The data on overall survival in the trial by Gianni et al. should not be interpreted as negative or even as marginally significant, but should instead be interpreted in the light of the limitations of the trial to make such comparisons. On the basis of the overall results of the trial, it is possible that a larger number of events would have resulted in the detection of a statistically significant difference in overall survival.

Renato G. Martins, M.D.
Michael V. Seiden, M.D., Ph.D.
Massachusetts General Hospital, Boston, MA 02114

1 References

1. 1

   Donner A. Approaches to sample size estimation in the design of clinical trials: a review. Stat Med 1984;3:199-214[Erratum, Stat Med 1990;9:1228.]
   CrossRef | Web of Science | Medline

Author/Editor Response

The authors reply:

To the Editor: The first question posed by Dr. Mok concerns an abstract in which we reported the results of an analysis that included all initially randomized patients. As we clearly mentioned in our final report,1 only 3 of 101 patients (3 percent) were excluded from the analysis because they presented with liver disease at the time of the diagnosis, which made them ineligible for the study. The results of the analysis of data from all 101 initially randomized patients were not substantially different from the reported results in the 98 patients. The rate of event-free survival in the 101 patients was 71 percent (95 percent confidence interval, 57 to 86 percent) for high-dose sequential therapy and 47 percent (95 percent confidence interval, 32 to 64 percent) for MACOP-B (two-sided P = 0.008). The rates of overall survival were 76 percent (95 percent confidence interval, 65 to 88 percent) and 53 percent (95 percent confidence interval, 35 to 72 percent), respectively (two-sided P = 0.11). As to Dr. Mok's second question, about the five patients receiving high-dose sequential chemotherapy who died from progressive disease, four had received total-body irradiation, and one had received mitoxantrone plus melphalan.

The issues raised by Drs. Martins and Seiden help clarify some statistical aspects of our study. In our report, we stated that the magnitude of the difference we were interested in detecting was at least a 25 percent improvement in the rate of freedom from disease progression. A one-sided test, at a significance level of 0.05, and a statistical power of 80 percent were selected to calculate the sample size required. Differences in survival were tested with the log-rank test, and we reported two-sided P values. The observed difference between the group receiving high-dose sequential chemotherapy and the group receiving MACOP-B, in terms of freedom from the progression of disease (crude difference, 35 percent; relative risk, 4.71; 95 percent confidence interval, 1.91 to 11.6), remained statistically significant with a two-sided test (P<0.001). When overall survival was examined, the observed difference between the two groups (crude difference, 26 percent; relative risk, 1.98; 95 percent confidence interval, 0.89 to 4.43) was not statistically significant with a two-sided test (P = 0.09). The constraints of space prevented a detailed specification of the various aspects of statistical interpretation, but we did not consider the results of the analysis of overall survival negative or marginally significant. We indeed stated, “By all measures used, we found that as initial treatment for large-cell lymphoma, high-dose sequential therapy was superior to MACOP-B.”

Alessandro M. Gianni, M.D.
Pinuccia Valagussa, B.S.
Gianni Bonadonna, M.D.
Istituto Nazionale Tumori, 20133 Milan, Italy

1 References

1. 1

   Gianni AM, Bregni M, Siena S, et al. High-dose chemotherapy and autologous bone marrow transplantation compared with MACOP-B in aggressive B-cell lymphoma. N Engl J Med 1997;336:1290-1297
   Full Text | Web of Science | Medline