VALIDATION OF A NEW PROGNOSTIC SCORE AIMED TO IDENTIFY THREE GROUPS WITH DIFFERENT PROBABILITIES OF DYING OF CHRONIC MYELOID LEUKAEMIA IN 1,120 PATIENTS WITH FIRST-LINE IMATINIB TREATMENT

Background
A new prognostic score for differentiating cumulative incidence probabilities (CIPs) of dying of chronic myeloid leukemia (CML) was developed in the IN-study section of the European Treatment and Outcome Study (EUTOS) registry. From 2002 to 2006, all patients were enrolled in prospective, controlled clinical trials. They had Philadelphia chromosome-positive chronic-phase (CP) CML and started first-line imatinib-based treatment within half a year after diagnosis. In the OUT-study section of the registry, patients were prospectively registered, but, as the only contrast to the IN-study, not part of clinical trials.

Aims
To compare the performance of the Sokal, the Euro, the EUTOS, and the new score in the OUT-study data

Methods
Survival was counted from start of imatinib treatment and censored at the time of allogeneic stem cell transplantation (SCT) in first CP. Only death after recorded disease progression was regarded as death due to CML. CIPs of dying of CML were estimated, treating death due to any other cause as a competing event. All prognostic scores were calculated at diagnosis of CML. Overall level of significance was 0.05.

Results
Applying the same inclusion criteria as for the IN-study section, 1,120 patients of the OUT-study section were evaluable with regard to all prognostic scores and survival. These patients came from registries in Russia, the Czech Republic, Poland, Spain, and Romania. Median observation time was 5.6 years. Allogeneic SCT in first CP was performed in 29 patients (3%, 7 died). Without prior SCT in first CP, 116 patients died, 6-year CIP was 11% [95%>confidence interval (CI): 9-14%]. In 56 cases (48% of 116), cause of death was CML. Six-year CIP of death due to CML was 5% [95%>CI: 4-7%] and 6% [95%>CI: 5-8%] for death due to other causes. The new prognostic score identified 681 low-risk (61% of 1,120), 302 intermediate- (27%), and 137 high-risk patients (12%). Their 6-year CIPs of dying of CML were 3% [95%>C.I.: 2-5%], 6% [95%>C.I.: 3-9%], and 17% [95%>C.I.: 11-25%]. In the corresponding risk groups of the Sokal score the 6-year CIPs resulted in 3% [n=450, 95%>C.I.: 2-5%], 5% [n=411, 95%>C.I.: 3-8%], and 11% [n=259, 95%>C.I.: 8-16%] and with the Euro score in 4% [n=466, 95%>C.I.: 2-6%], 5% [n=523, 95%>C.I.: 3-7%], and 15% [n=131, 95%>C.I.: 9-23%]. The EUTOS score found a 6-year CIP of 5% [n=971, 95%>C.I.: 3-6%] for its low-risk and of 12% [n=149, 95%>C.I.: 7-18%] for its high-risk group. Each of the 4 prognostic scores provided significantly higher probabilities of dying of CML for the high-risk group as compared with each of the non-high-risk group(s) (all p values < 0.005). The new prognostic score was also able to identify significantly different CIPs of dying of CML between the low- and the intermediate-risk group (p=0.04).

Summary
Judging from the CIPs and the distribution of the patients into the risk groups, the new score showed the best prognostic performance also in the out-study patients. Only the new model identified three risk groups with pairwise significantly different CIPs of dying of CML. With similar CIPs, the new score identified a much larger group of low-risk patients than the Sokal and the Euro score. Altogether, validation of the new score in the out-study data was successful. In the 681 patients of the low-risk group (61%), the new score showed an excellent long-term outcome when starting treatment with imatinib. Due to the worse results, for the high-risk patients (12%), an upfront comparison between different tyrosine kinase inhibitors would be highly desirable.

Keyword(s): Chronic myeloid leukemia, Imatinib, Long-term follow-up, Prognosis

www.eha.org