Predicting the Long-Term Efficacy of Ifnα in JAK2^V617F and Calr-Mutated MPN Patients

Introduction

Classical BCR-ABL-negative myeloproliferative neoplasms (MPN) include Polycythemia Vera (PV), Essential Thrombocythemia (ET) and Primary Myelofibrosis (PMF). These are acquired clonal disorders of hematopoietic stem cells (HSC) leading to the hyperplasia of one or several myeloid lineages. MPN are caused by three main recurrent mutations: JAK2^V617F, mutations in the calreticulin (CALR) and thrombopoietin receptor (MPL) genes. Interferon alpha (IFNα) treatment induces not only a hematological response in around 70% of ET, PV and early myelofibrosis, but also a significant molecular response on both JAK2^V617F- and CALR-mutated cells. However, a complete molecular response is only achieved in around 20% of patients. Our aim is to predict the long-term efficacy of IFNα in JAK2^V617F- and CALR-mutated patients by monitoring the fate of the disease-initiating mutated HSC in order to better stratify the molecular responders.

Methods

A longitudinal observational study (3-5 years) was performed in 46 IFNα-treated patients. The MPN disease distribution was 42% ET, 47% PV and 11% PMF. We detected 33 patients with JAK2^V617F mutation, 11 with CALR mutations (7 type 1/type 1-like and 4 type 2/type 2-like), 1 with both JAK2^V617F and CALR mutation and 1 with JAK2^V617F, CALR mutation and MPL^S505N. At 4-month intervals, the JAK2^V617For CALR mutation variant allele frequency was measured in mature cells (granulocytes, platelets). Simultaneously, the clonal architecture was determined by studying the presence of the mutations in colonies derived from the different hematopoietic stem and progenitor cell (HSPC) populations (CD90⁺CD34⁺CD38^-HSC-enriched, CD90^-CD34⁺CD38^- immature and CD34⁺CD38⁺committed progenitors). We used a combination of mathematical modeling (Michor et al., Nature, 2005) and Bayesian analysis to infer the long-term behavior of mutated HSC.

Results

After a median follow-up of 40 months, IFNα targeted more efficiently and more rapidly the HSPC, particularly the HSC-enriched progenitors, than the mature blood cells in JAK2^V617Fpatients (p<.05). Moreover, kinetics of response of homozygous JAK2^V617FHSPC to IFNα were more rapid than that of heterozygous HSPC and mature cells. This IFNα-specificity towards homozygous HSPC slightly increased after a median follow-up of 51 months. In contrast, during a 40-month median follow-up of CALR-mutated patients, IFNα targeted similarly the HSPC and the mature cells. Moreover, IFNα was less efficient in targeting the CALR-mutated than the JAK2^V617FHSPC (p<.05).

Since it is very difficult to purify true HSC from patients, we used a combination of mathematical and statistical modeling to infer the behavior and the kinetics of IFNα-targeted mutated HSC. The model gave a good fit to the data and indicated that mutated HSC are exhausted slowly (> 1 year) with concomitant increase in mutated HSPC and granulocytes in well-responding patients. We calculated the rate of HSC decrease for each patient. Rates of decrease are very low for heterozygous JAK2^V617F and CALR-mutated HSC and greater for homozygous JAK2^V617FHSC, but all increase with high IFNα dose (>100 µg/week). Moreover, very low proportion of heterozygous mutated HSC compared to high proportion can be targeted more easily in patients. The associated mutations at diagnosis and at the last timepoint were also investigated using an NGS-targeted myeloid panel. Results indicate that IFNα does not induce any further mutations on additional genes and the mathematical approach predicts that associated mutations have no major impact on the ratio of HSC decrease.

Conclusion

Altogether, using a rigorous method of statistical inference, our results show that IFNα exhaust the human mutated HSC by differentiation in HSPC and mature cells. This is likely due to IFNα inducing a stronger proliferation of mutated compared to wild-type HSC, as previously shown in a mouse model (Mullally et al., Blood, 2013). Our study predicts that IFNα can slowly eradicate the mutated HSC, but this beneficial effect would be more efficient: i) in patients with homozygous JAK2^V617F versus those with heterozygous JAK2^V617F or CALR-mutated, ii) with high IFNα dose, iii) in patients with very low proportion of heterozygous JAK2^V617F and CALR-mutated HSC. Thus, this study will help to stratify patients for IFNα treatment.These results might also explain the different outcomes in current IFNα clinical trials.