Polycythemia Vera Patients with >20% Reduction in Whole Blood JAK2^V617F Allele Frequency Have Improved Myelofibrosis-Free Survival but Not Overall Survival

Background: There is a great need for monitoring biomarkers to predict the success of treatment in preventing progression and improving survival of patients (pts) with myeloproliferative neoplasms (MPN). Owing to the high prevalence of the JAK2^V617F driver mutation in MPNs and the ready availability of whole blood (WB) specimens, it is common to monitor WB JAK2^V617F mutation allele frequency (MAF) in clinical practice and in clinical trials. Nonetheless, diverse criteria are used to define molecular response (MR) and none are validated as prognostic for treatment response or survival outcomes. WB is a composite mixture of different, mature cell types, and although WB MAF falls short in capturing core MPN stem cell biology¹, it is easily and frequently tested. Therefore, it is critical to determine whether a reduction in WB MAF correlates with an improved survival, until better biomarkers are routinely available.

Methods: Under an institutional board approved study at Weill Cornell Medicine (WCM), and using the Observational Medical Outcomes Partnership infrastructure, we performed a Structured Query Language extraction of all clinical and research genomic data on WB JAK2^V617F MAF in WCM polycythemia vera (PV) pts. We categorized the change in MAF from baseline (first MAF test) to the lowest MAF within 5 years (yr) as MR10, 20, 30 and 50 for a relative reduction of ≥ 10%, 20%, 30%, and 50%, respectively. MR was linked to clinical features and outcomes registered in our MPN Research Database Repository (RDR) that includes 470 PV pts. Our RDR survival data and cause of death was complete from the CDC national death index. Survival outcomes of pts who achieved MR were compared to pts not achieving a response using Kaplan-Meier methods and Cox proportional hazards multivariable analysis (MVA). Area under the curve (AUC) of Receiver-Operator Characteristic analysis was used to assess performance of MR as a biomarker for survival outcomes.

Results: WB MAF data was available for 260 PV pts between 10/2003-4/2021; and serial WB MAF within 5 yr from baseline, before progression, for 165. Median (range) of age at baseline, MAF at baseline, time between samples, and follow-up was 60 yr (24-91), 44% (1-100%), 2 yr (0.1-5), and 7 yr (1-16), respectively. MR10, 20, 30, and 50 was achieved in 74 (45%), 63 (38%), 57 (35%), 42 (25%). Pts achieving MR did not differ in age, sex, baseline MAF, sampling duration, or follow-up duration when compared to those who did not achieve equivalent MR. Pts achieving MR at any level, however, were more likely to be receiving interferon (IFN) (p<0.05). As previously reported¹, decrease in MAF was not prognostic for event-free survival (EFS) - the composite endpoint of thrombosis, progression to myelofibrosis (MF) or acute leukemia (AML), and death - or overall survival (OS). Nevertheless, MR20-50 was associated with improved myelofibrosis-free survival (MFS) (MR20, Figure 1), independent of age and interferon in MVA (MR20 HR = 0.20, p=0.03). The optimal AUC across MR levels and over time was modest (range 0.57-0.70) indicating that MR had only fair performance as a biomarker for MFS. Although MFS was higher in pts achieving MR, deaths occurred at a similar rate, but they were mostly due to other causes. In MR20, deaths (n=10, 16%) were due to AML without MF (1), infectious disease (2), valvular heart disease (2), ischemic heart disease (1), prostate cancer (1), malignant melanoma (1), and lymphoma (1). Deaths in those not achieving MR20 (n=12, 13%) were most commonly due to MPN progression (n=6).

Discussion: Given the widespread adoption of JAK2^V617F MAF monitoring in clinical trials, we conducted this study to determine the value of WB MAF reduction (molecular response) in predicting the most important outcomes in PV (EFS and overall survival) and found only a weak prognostic link to MFS-but no prognostic value for EFS or OS-in patients achieving MR20 or greater. Larger, long-term, prospective studies are necessary and could establish optimal definition of MR as a monitoring biomarker. But the preliminary signal is modest and we believe this effort would be better spent developing new biomarkers, such as MPN fitness¹, that capture disease biology.

References: [1] Abu-Zeinah et al. Blood Adv. 2022

Scandura:AbbVie: Consultancy, Membership on an entity's Board of Directors or advisory committees; MPN-RF: Research Funding; Sumitomo Pharma Oncology, Inc: Consultancy; European Leukemia net: Honoraria, Other: Travel fees; CR&T: Research Funding; Constellation Pharmaceuticals, Inc., a MorphoSys Company: Research Funding.