Correspondence
Long-Term Survival after Bone Marrow Transplantation
N Engl J Med 1999; 341:1394-1395October 28, 1999
- Article
To the Editor:
Socié et al. (July 1 issue)1 reported on the long-term survival of patients who have undergone allogeneic bone marrow transplantation and the rate of death more than two years after transplantation. They provided information on the characteristics and outcomes of long-term survivors after bone marrow transplantation, but the study was a retrospective survey in which data were collected by sending questionnaires to many institutions. We agree that a retrospective survey is a powerful method for collecting a large amount of information, but the results may be easily biased because it is difficult to collect complete data on the study population. If patients die suddenly or move to other hospitals, their records may be lost and thus excluded from the analysis.
We were surprised that Socié et al. did not mention suicide as a cause of death among long-term survivors after bone marrow transplantation. The incidence of suicide is higher among such survivors than among healthy subjects.2,3 Molassiotis and Morris reported that the incidence of suicide among patients who had undergone bone marrow transplantation was 2.5 to 3.0 percent three years after transplantation.2 We would like to know whether the incidence of suicide among the patients in the study by Socié et al. was negligible, as we inferred from the article.
3 ReferencesYasuhiro Oki, M.D.
Masahiro Kami, M.D.
Yoshitomo Muto, M.D.
Toranomon Hospital, Tokyo 105-8470, Japan1
Socie G ,Stone JV ,Wingard JR , et al. Long-term survival and late deaths after allogeneic bone marrow transplantation. N Engl J Med 1999;341:14-21
Full Text | Web of Science | Medline2
Molassiotis A ,Morris PJ . Suicide and suicidal ideation after marrow transplantation. Bone Marrow Transplant 1997;19:87-90
CrossRef | Web of Science | Medline3
Shaffer D . Preventing suicide in young people. Innov Res 1993;2:3-9
To the Editor:
Socié et al. report that patients with acute lymphoblastic leukemia (ALL) who received a single dose of total-body irradiation of 10 Gy or more had a higher risk of death that was not related to relapse than did the other groups. We think that this claim is incorrect and note that several stronger and more significant associations, reported in Table 4 of their article, are not discussed.
One of the difficulties of interpreting relative risk is that the strength of the association should be judged on the basis of the numerical interval between the relative risk and the reference value of 1. A relative risk of 1 would be observed if the risk of death unrelated to relapse was the same regardless of whether the patients received high-dose total-body irradiation or no total-body irradiation. Because relative risks are ratios, the values should be interpreted with caution, especially when comparing relative risks above and below 1.1
A way to illustrate the difficulty is to choose the reference category so that all the relative risks are above 1. In Table 1Table 1
Relative Risk of Late Death among Patients with Acute Lymphoblastic Leukemia in Continuous Complete Remission Two Years after Transplantation, According to Type of Conditioning Total-Body Irradiation., we show the results of Socié et al., but with the patients with ALL who received a single dose of total-body irradiation of less than 10 Gy as the reference group. The patients who received a single dose of total-body irradiation of 10 Gy or more now appear to have the highest risk of death not related to relapse. However, the relative risks for the patients who received fractionated total-body irradiation are much smaller than those for the patients who received no total-body irradiation, both for causes related to relapse and causes unrelated to relapse. These relative risks correspond to the strong associations that were originally reported, but not discussed, as values below 1.Furthermore, Socié et al. based their statement of statistical significance on overall P values but focused on a relative risk of 1.54, for which the confidence interval included the null value of 1. Because we can assume statistical significance when the confidence interval lies entirely above or below 1,3 only one association was obviously statistically significant — that between a single dose of total-body irradiation of <10 Gy and death from any cause (relative risk, 0.35) — and four were of borderline significance. These associations indicate that survival was improved among patients who received low-dose or fractionated total-body irradiation.
The results of recent randomized trials suggested that total-body irradiation can improve survival.4,5 We regret that the remarkable work of Socié et al. failed to convey the important message that pretransplantation regimens that include low-dose or fractionated total-body irradiation actually do improve survival.
5 ReferencesAli Lounici, M.D.
L. Rachid Salmi, M.D.
Université Victor Segalen Bordeaux 2, 33076 Bordeaux, France1
Hebert JR ,Miller DR . Plotting and discussion of rate ratios and relative risk estimates. J Clin Epidemiol 1989;42:289-290
CrossRef | Web of Science | Medline2
Socie G ,Stone JV ,Wingard JR , et al. Long-term survival and late deaths after allogeneic bone marrow transplantation. N Engl J Med 1999;341:14-21
Full Text | Web of Science | Medline3
Colton T. Statistics in medicine. Boston: Little, Brown, 1974.
4
Hartman AR ,Williams SF ,Dillon JJ . Survival, disease-free survival and adverse effects of conditioning for allogeneic bone marrow transplantation with busulfan/cyclophosphamide vs total body irradiation: a meta-analysis. Bone Marrow Transplant 1998;22:439-443
CrossRef | Web of Science | Medline5
Ringden O ,Remberger M ,Ruutu T , et al. Increased risk of chronic graft-versus-host disease, obstructive bronchiolitis, and alopecia with busulfan versus total body irradiation: long-term results of a randomized trial in allogeneic marrow recipients with leukemia. Blood 1999;93:2196-2201
Web of Science | Medline
Author/Editor Response
The authors and a colleague reply:
To the Editor: Oki et al. correctly note the difficulties involved in collecting complete data on many patients retrospectively. The International Bone Marrow Transplant Registry requires that patients be registered shortly after transplantation and after an annual follow-up. Consequently, this data base prospectively collects longitudinal information on a large cohort. The 221 centers that participated in our study provided follow-up data for more than 90 percent of their patients. Suicide was the reported cause of death in only 2 of 679 deaths. However, our study included only patients who had survived more than two years after bone marrow transplantation. It may be that most suicides occur earlier, as suggested by Molassiotis and Morris. We agree that suicide might be underreported because of the social stigma with which it is associated.
Lounici and Salmi take issue with a perceived emphasis on the association between late deaths unrelated to relapse and conditioning with a single dose of total-body irradiation of 10 Gy or more administered to patients with ALL. We cited all significant associations in the Results section and did not discuss this particular result further. “No irradiation” was presented as the reference variable because of the study design. As was noted, the reported P value reflects the association of the overall conditioning variable (with four categories) with the end point; P values for pairwise comparisons were not given. We agree that it may be easier to interpret values for which the category with the lowest risk serves as the reference so that all relative risks are >1.0. Their table, however, does not provide confidence intervals or P values, which are necessary for pairwise comparisons. In Table 1Table 1
Risk of Death Not Related to Relapse., we present the results for the relative risk of death unrelated to relapse, using the group with the lowest risk as the reference group. Patients receiving single-dose regimens of 10 Gy or more have a significantly higher risk of death not related to relapse than those receiving less than 10 Gy. The relative risk of death unrelated to relapse for the patients receiving low-dose radiation does not differ significantly from that for the patients receiving no radiation or fractionated radiation, and the last two groups do not differ significantly from each other (P=0.61).Although some P values are close to 0.05, the multiple comparisons in this study dictate that caution be used when interpreting associations with P values >0.01, as noted in the Methods section of our article. It is worth noting that single-dose irradiation is used infrequently in the 1990s; data from the International Bone Marrow Transplant Registry suggest that 90 percent of patients receiving total-body irradiation are treated according to a fractionation schedule. These and other changes may further improve the already excellent prognosis of long-term survivors.
Gérard Socié, M.D., Ph.D.
Hôpital Saint Louis, 75475 Paris, FranceJohn P. Klein, Ph.D.
Mary M. Horowitz, M.D.
Medical College of Wisconsin, Milwaukee, WI 53226for the Late Effects Working Committee of the International Bone Marrow Transplant Registry
