Pb1899: development and validation of a predictive scoring system for the prognosis in patients with chronic myeloid leukemia (cml) in chronic phase receiving first-line imatinib

Background: Since the introduction of tyrosine kinase inhibitors (TKIs), the management of chronic myeloid leukemia (CML) has changed significantly, and in patients (pts) with optimal responses, survival is generally similar to that of age- and sex-matched normal population. World Health Organization stated that, the life expectancy at birth for women is 5 years longer than that of men. With knowing this fact, sex might have an impact on survival in patients with CML. There are 4 risk scores currently in use for the prediction of outcomes and survival in CML including Sokal, Euro-Hasford, European Treatment and Outcome Study (EUTOS), and EUTOS long-term survival (ELTS). But none of these scores contain sex as a parameter. Aims: In our study, we aimed to propose a new prognostic score for pts with CML in chronic phase (CML-CP), including sex as a parameter. Methods: Our training cohort (TC) included CML-CP pts diagnosed between 2003-2018 receiving imatinib as a first-line TKI therapy. While generating the new prognostic score – Cerrahpaşa CML Prognostic Score (CCPS), uni- and multivariate Cox regression analysis were performed in order to identify factors having an impact on progression and survival. Beta coefficients were used in creating this prognostic score. The area under the receiver operator characteristic curve (AUROC) was used to estimate accuracy of the predictive model. The CCPS had 3 parameters; age, blast %, and sex (Age (years) x 0.085 + Blast (%) x 0.489 + Sex (Male) x 1.449) and 3 risk groups [Low: <6.62, Intermediate: 6.62-7.85, and High: >7.85)]. Progression-free (PFS) and overall (OS) survivals were evaluated by Kaplan-Meier curves. For the validation of this new score, two independent validation cohorts (VCs) were used. Results: TC consisted of 185 pts, of which 103 (55.7%) were male, and the median age at diagnosis was 47 years (Table 1). First VC (VC1) had 86 pts, and median age was 44 years. All pts in VC1 received imatinib as upfront treatment. Second VC (VC2) consisted of 183 pts, and 62.8% of these cases were male. The median age of VC2 at diagnosis of was 59 years, which was higher than TC and VC1 (Table 1). While 148 (80.9%) had first-line imatinib, the remaining 35 pts (19.1%) received upfront second-generation TKI (2GTKI) therapy (10 pts with dasatinib and 25 pts with nilotinib). In TC, 60.5%, 25.9%, and 13.5% of the cases were in low-, intermediate-, and high-risk groups for the ELTS score, respectively. For the CCPS, the percentages of low-, intermediate-, and high-risk patients were 64.9%, 20.5%, and 14.6%, respectively (Table 1). Distribution of the ELTS and CCPS for VC1 and VC2 are displayed in Table 1. AUROC for the new prognostic score was the highest among others for TC, and for both VC1 and VC2 (Figure 1A-C). In the survival analysis of TC, ELTS score successfully dissected 3 risk groups both for PFS and OS (p=0.013 and p<0.001, respectively) (Figure 2A-B). Similarly, according to the CCPS, the low-risk group had significantly superior PFS and OS (p=0.025 and p<0.001, respectively) in TC (Figure 2C-D). For VC2, according to the CCPS, 3 risk groups significantly differed (p<0.001) for OS, however, although PFS was inferior for the high-risk group, the difference was not significant (p=0.063) (Figure 2E-F). Image:Summary/Conclusion: We developed and validated a score using sex as a parameter for the first time, which successfully predicts PFS and OS in pts with CML-CP receiving frontline imatinib. Although some pts in VC2 received upfront 2GTKIs, the CCPS should further be tested especially in cases receiving 2GTKIs in the first-line setting.