Changes in neutrophil-to-lymphocyte ratio (NLR) in patients with polycythemia vera treated with ruxolitinib reflect changes of JAK2 variant allele frequency (VAF)

Abstract

Abstract Background. The neutrophil-to-lymphocyte ratio (NLR) has received considerable interest as a simple and reliable marker of disease associated-inflammation in different settings, including cardiovascular diseases and cancer. In polycythemia vera (PV), NLR was associated with venous thrombosis (Carobbio A, BCJ 2022), overall mortality (Barbui T, BCJ 2024), and achievement of primary endpoints in LOW-PV trial (Barbui T, BJH 2024). Other observations indicated that NLR changes in PV patients (pts) receiving ropeg-interferon alpha-2b (ropeg) were associated with JAK2V617F variant allele frequency (VAF) decrease (Barbui T; EHA2024, A1908). Aims. To evaluate levels and changes of NLR, total leukocytes (WBC), absolute neutrophil (ANC) and lymphocyte (ALC) counts, and correlate them with disease characteristics and JAK2 VAF modifications, in PV pts long-termtreated with ruxolitinib (ruxo). Results. Forty pts with PV who received ruxo, 10 mg BID for a median of 8.7y (4.7-9.8) as second line therapy because of intolerance (44.2%) or resistance (55.8%) to hydroxyurea, were included. Overall FU was 9.2y (1.3-14.5). 10 pts (25%) had history of major thrombosis. A complete clinical-hematologic response (CHR) and a molecular response (IWG-MRT criteria; Tefferi A, Blood 2013) was obtained in 32 (78%) and 41.5%, respectively. The median NLR value at baseline was 5.2 (1.6-25.0); we accordingly divided pts into a high (H-, above the median) and low (L-, below the median) NLR category, and compared them. At baseline, pts in the H-NLR had (median count x109/L; range) significantly higher WBC (12.6 vs 8.0; p=0.003), due to increased ANC (12.1; 5.1-26.3 vs 5.9; 4.0-15.8) compared to L-NLR (p<0.001); conversely, ALC was lower (1.4 (0.1-2.8) vs 2.2 (1.0-6.5) (p=0.001). Also, Hb level (15g/dL vs 14g/dL p=0.01) and PLT count (544 vs 323/L; p=0.004) were higher in H-NLR compared to L-NLR category. There was no baseline difference in disease duration, splenomegaly, ad history of cardiovascular events, between pts in the H- versus L-NLR category. In the entire cohort, NLR value at baseline was linearly correlated with JAK2 VAF (r=0.5, p=0.001). Accordingly, JAK2 VAF (%±SD) was significantly higher in H-NLR compared to L-NLR category: 81.8±13.1% vs 54.8±23.8% (P<0.001). After starting ruxo (10 mg BID), the changes in NLR observed in the 2 categories differed. In the H-NLR, the median baseline NLR value of 9.6 (5.3-25.0) decreased to 5.4 (2.7-11.0; p=0.002), 5.2 (1.9-8.7; p=0.001) and 4.2 (1.3-11.5; p=0.001) at 1y, 2y and 3y, remaining steadily stable afterwards. Conversely, in the L-NLR group, no significant changes of NLR were observed: 3.1 (1.6-5.0) at baseline, then 2.6 (1.1-8.2; p=0.5), 3.0 (1.0-5.0;p=0.16) and 3.2 (1.0-7.7; p=0.29) at 1y, 2y and 3y, respectively. Changes in H-NLR group were due to a decrease of WBC (-25% % (p=0.049), -32% (p=0.02) and -42% (p=0.002), at 1y, 2y and 3y), mostly accounted by decrease in ANC (-55% (p=0.001), -53% (p=0.002), -41%(p=0.001)); on the other hand, ALC remained unchanged (from a baseline 1.4 (0.1-2.8) to 1.35 (1.0-3.0; p=0.5), 1.24 (1.0-3.0,p=0.6) and 1.14 (1.0-5.0;p=0.4). During ruxo treatment, changes in JAK2 VAF proved different in the 2 NLR categories. In the H-NLR, the baseline VAF of 81.8±13.1% decreased to 72.0±17.6% (p=0.07), 62.1±26.0% (p=0.007) and 62.5±24.3% (p=0.006) at 1y, 2y and 3y, respectively. Conversely in L-NLR group, the baseline VAF of 54.8±23.8% decreased to 37.7±24.7% (p=0.11), 43.4±30.8% (p=0.16) and 41.9±34.8% (p=0.21) at 1y, 2y and 3y, respectively. No differences were observed in terms of CHR (n=p=0.7), thrombosis (n=2), progression to MF (n= 16; p=0.37), evolution to AML (p=1.0), or death (p=0.55). Conclusions. Among HU-resistant/intolerant PV pts, higher (above the median) NLR reflected signs of myeloproliferation, including higher JAK2 VAF. Upon ruxo treatment, however, H-NLR pts showed the most impressive reduction of NLR due to preferential ANC decrease, that was also associated with JAK2 VAF decline, when compared to L-NLR pts. We conclude that in H-NRL PV pts treated with ruxo, NLR proved to be a suitable biomarker mirroring changes in JAK2 VAF, clinically more friendly that PCR-based determination of VAF. Findings are reminiscent of observations in PV pts treated with ropeg-interferon in the LOW-PV and PROU-PV/CONTINUATION-PV study (Barbui T, EHA 2024). These findings deserve confirmation in other cohorts of ruxo-treated PV pts.