Overall Survival and Prognosis In Patients With First-Line Imatinib Treatment Under Consideration Of Death Due To Any Cause and Death Due To Chronic Myeloid Leukemia Only

As a part of the European Treatment and Outcome Study (EUTOS), the IN-study registry collected data on patients with Philadelphia chromosome-positive chronic myeloid leukemia (CML) enrolled in prospective clinical trials. Patients had to have been in chronic phase (CP) and imatinib-based treatment been started within half a year after diagnosis. Meanwhile, median observation time seemed to allow the analysis of overall survival (OS) and a possible identification of prognostic factors.

For OS with death due to any cause and for cumulative incidences of mortality (CIM) due to CML only, we planned to describe probabilities as well as to analyze the prognostic influence of the candidate factors age, sex, spleen enlargement, hemoglobin, platelets, leukocytes, and percentages of blasts, eosinophils, and basophils in peripheral blood (PB) on the respective probabilities.

Survival of patients fulfilling the inclusion criteria was censored at the time of allogeneic stem cell transplantation (SCT) in first CP. Allowing any cause of death as an event, prognostic influence on OS was estimated with multiple Cox regression allowing fractional polynomials for continuous variables. These results were opposed to the findings when cause of death was restricted to CML only and then analyzing a) cause-specific hazards by standard Cox regression and b) subdistribution hazards of the CIM with the Fine and Gray model. All prognostic factors were measured at baseline. Level of significance was 0.05.

Patients joined study groups in Germany, France, Italy, Spain, the Netherlands, or the Nordic study group. Analyses were based on 2127 patients of five studies with similar follow-up patterns. Median observation time was 6.4 years. Eight-year survival probability was 89% [95%-confidence interval (CI): 87-90%]. Allogeneic SCT in first CP was performed in 94 patients (4%). The sample for regression analysis comprised 2067 patients with complete data for the candidate prognostic factors. Their median age was 51 years [range: 18-88], 60% were male.

Altogether, 187 patients died. With death due to any cause as event, higher age and bigger spleen size enlargement (cm below costal margin) significantly reduced OS probabilities. Besides, multiple modeling also suggested that the interaction between sex and hemoglobin had a significant influence on survival. Females with lower hemoglobin values had less favorable survival probabilities while hemoglobin showed no effect in males.

Judged by experienced physicians, in 89 cases (48% of 187), cause of death was CML-related (not related to CML: n=87 (47%), unknown cause: n=11 (6%)). Eight-year CIM due to CML was 5% [95%-C.I.: 4-6%] and 6% [95%-C.I.: 5-8%] due to other causes.

Cause-specific Cox regression with death due to CML as the only event resulted in the same prognostic factors as for the analysis with any cause of death as event – apart from one exception: now, higher percentages of blasts in PB had a significantly negative effect, too.

The analysis of the influence of prognostic factors on the CIM due to CML via the Fine and Gray model, which took the competing risk of dying due to other cause (87+11=98) into account, led to the same prognostic factors as for the cause-specific Cox model where death due to other cause was censored.

Median observation time and the number of events actually allowed a reasonable modeling of OS probabilities and CIM, even though the restriction on the cause of death due to CML reduced the number of events by about 50% in the second case. Also when considering only deaths due to CML, age remained significant and seems to be a true prognostic factor for long-term survival outcome when suffering from CML. However, age did not become part of the EUTOS score investigating the influence of various factors on the achievement of complete cytogenetic remission within 18 months (Hasford et al., Blood 2011) and this showed that imatinib-based treatment helps in any age. After all, significantly higher but still relatively low hazards of dying due to CML in older age result in small absolute numbers. It appears to be conclusive that blasts in PB influence OS probabilities and CIM due to CML, as blasts in PB became also part of the Sokal and the Euro Score, both developed at times when OS probabilities were much lower, with most CML patients actually dying of CML. All findings will be checked in independent patient data.

Pfirrmann:Novartis: Consultancy. Saussele:Pfizer: Honoraria; BMS: Honoraria, Research Funding, Travel, Travel Other; Novartis: Honoraria, Research Funding, Travel Other. Baccarani:Novartis: Research Funding. Lindörfer:Novartis: Research Funding. Hoffmann:Novartis Oncology: Research Funding. Castagnetti:Novartis: Consultancy, Honoraria; Bristol-Myers Squibb: Consultancy, Honoraria. Hehlmann:Novartis: Research Funding.