Allogeneic Hematopoietic Stem Cell Transplantation (HSCT) with a Related Donor in Chronic Myeloid Leukemia (CML): An Explanation for Fast Improvement of Survival in Two Consecutive German CML Studies.

In the two consecutive German CML studies III and IIIA (recruitment periods from 1995 to 2001 and 1997 to 2004), eligible patients were assigned to early HSCT by genetic randomization according to availability of a matched related donor. After randomization, 113 patients of study III (84% of 135) and 144 of study IIIA (87% of 166) were eventually transplanted in first chronic phase (CP) using a related donor. Despite comparable transplantation protocols and most centers participating in both studies, survival probabilities in study IIIA were significantly better, even when adjusted for the established EBMT risk score (Gratwohl et al., Lancet 1998 [1]), p + 0.0097. For further explanation, the German Registry for Stem Cell Transplantation (DRST) and the Swiss Transplant Working Group for Blood and Marrow Transplantation (STABMT) were asked for data support.

The main sample characteristics of the 257 transplanted CML study patients were also applied to the registry patients: diagnosis of CML between 1994 and 2004, first HSCT with a related donor performed in first CP between 1995 and 2004 at an age between 12 and 65 years, and blood or bone marrow as stem cell sources. Thus, additional data of 582 HSCT patients were retrieved from the two registries. Age, recipient sex, donor sex, time between diagnosis and HSCT, calendar year of HSCT, stem cell source, and HLA matching were investigated as potential predictive factors for survival. Then, a sample of patients with the same risk distribution as the 113 patients of study III was randomly drawn from the registry patients. By application of repeated resampling to this new patient group, bootstrap confidence intervals for survival probabilities at various times after HSCT were extractable. This provided the basis to judge whether the survival in study III could be seen as a typical random representation of a sample with an equivalent risk structure or not. The same method was applied to the 144 patients of study IIIA.

The 5-year survival probability of all 839 patients resulted in 73% (229 died). Median follow-up time of living patients was 6.7 years. Due to the characteristic plateau of post-transplant survival probabilities, the predictive influence was judged by the Kaplan-Meier method and the log rank statistic. Also consideration of age and time between diagnosis and HSCT as continuous variables seemed less appropriate than working with categorizations. Furthermore, the previously published cut-points “1 year” for time from diagnosis to HSCT ([1]) and “44 years” for age at HSCT (Maywald et al., Leukemia 2006) were independently confirmed to be the best. Cox model and logistic regression with survival status after 3 years both indicated that age at HSCT, HLA matching, time between diagnosis and HSCT, and calendar year of HSCT had independent statistically significant predictive influence on survival (p < 0.05). The first two factors had the strongest effects. Calendar year was only influential when distinction was made between HSCT until and after 1999. All possible combinations of the 4 factors could be summarized in 4 risk groups with significantly different survival probabilities (at 5 years: 87%, 76%, 63%, and 24%). Matched for the risk group distribution of study III [study IIIA], a maximum of 290 [428] registry patients could be drawn. For the 290 [428] patients, 5-year survival was 69% [77%] with a 95% bootstrap confidence interval from 63% to 74% [72% to 81%]. Thus, as for all yearly intervals within the first 5 years, the 5-year survival probabilities of studies III: 65% and IIIA: 79% lied within the corresponding confidence intervals.

Along with the registry patients, the study data enabled the identification of age at HSCT, HLA matching, time between diagnosis and HSCT, and calendar year of HSCT as factors with independent predictive impact on survival which led to 4 risk groups with statistically significantly different survival probabilities. More favorable-risk patients in study IIIA stood for a better transplantation strategy. In consideration of these different risks, the survival probabilities in both studies did not significantly vary from those of registry samples with matched risk structures. Accordingly, an improved transplantation strategy along with random variation could be considered as an explanation of the significantly different survival probabilities between the two studies.

Haferlach:MLL Munich Leukemia Laboratory: Equity Ownership. Hochhaus:Novartis, Bristol-Myers Squibb: Research Funding. Hasford:Novartis: Research Funding. Gratwohl:AMGEN, Roche, Bristol-Myers Squibb, Novartis, Pfizer: Research Funding; Novartis: Consultancy. German CML Study Group:Kompetenznetz Leukämie, European Leukemia Net, Roche, Essex, AMGEN: Research Funding.