Inferring the dynamics of mutated hematopoietic stem and progenitor cells induced by IFNα in myeloproliferative neoplasms

Abstract

Classical BCR-ABL-negative myeloproliferative neoplasms (MPNs) are clonal disorders of hematopoietic stem cells (HSCs) caused mainly by recurrent mutations in genes encoding JAK2 (JAK2), calreticulin (CALR), or the thrombopoietin receptor (MPL). Interferon α (IFNα) has demonstrated some efficacy in inducing molecular remission in MPNs. To determine factors that influence molecular response rate, we evaluated the long-term molecular efficacy of IFNα in patients with MPN by monitoring the fate of cells carrying driver mutations in a prospective observational and longitudinal study of 48 patients over more than 5 years. We measured the clonal architecture of early and late hematopoietic progenitors (84 845 measurements) and the global variant allele frequency in mature cells (409 measurements) several times per year. Using mathematical modeling and hierarchical Bayesian inference, we further inferred the dynamics of IFNα-targeted mutated HSCs. Our data support the hypothesis that IFNα targets JAK2V617F HSCs by inducing their exit from quiescence and differentiation into progenitors. Our observations indicate that treatment efficacy is higher in homozygous than heterozygous JAK2V617F HSCs and increases with high IFNα dose in heterozygous JAK2V617F HSCs. We also found that the molecular responses of CALRm HSCs to IFNα were heterogeneous, varying between type 1 and type 2 CALRm, and a high dose of IFNα correlates with worse outcomes. Our work indicates that the long-term molecular efficacy of IFNα implies an HSC exhaustion mechanism and depends on both the driver mutation type and IFNα dose.

© 2021 by The American Society of Hematology.

Figures

Graphical abstract

Figure 1.

Inclusion criteria for various steps of analyses and IFNα dose. (A) Strategy of analysis and inclusion. Experimental observations were analyzed from progenitor and mature cells of the 48 patients of the cohort. We then excluded patients who had <2 data points (in progenitors and after the start of the therapy) for mathematical model calibration because there is no rationale to try to fit only 2 data points. To rigorously statistically analyze how IFNα doses differently impact molecular response according to the mutation type and zygosity in HSCs, we still had to exclude patients with no more than 5 progenitor type measurements from the start of the therapy for JAK2V617F patients. No CALRm patient was excluded because the model was less complex. The number in parentheses corresponds to the percentage of data used for the analyses. (B) To study the effect of IFNα on HSCs depending on the zygosity, it was necessary to exclude JAK2V617F patients whose clones exhibited too low CF over time. This analysis was not performed for CALRm MPN because only 2 of 12 patients had homozygous mutated cells. For our statistical analyses, a JAK2V617F patient was labeled as carrying homozygous (respectively heterozygous) subclones when CF >7% of homozygous (respectively heterozygous) progenitors were identified from ≥1 of the collected samples. Following this definition, some patients (8) could be considered carrying both heterozygous and homozygous subclones. Using this criterium, 17 patients carry heterozygous subclones and 10 patients carry homozygous subclones. (C) Averaged IFNα dose received over time by the 48 patients. Gray lines, individual dose; blue line, mean dose received by the 48 patients; shaded areas surrounding the curve, standard error of the mean.

Figure 2.

Characterization of the IFNα-treated MPN patient cohort. (A) Distribution of MPN diseases. (B) Distribution of MPN driver mutations. (C) Diseases and the molecular profile determined using an NGS myeloid panel of 77 genes of the first sample collected from the 48 patients in the studied cohort. P45 presented 2 diseases, ET/PV, based on its molecular profile (JAK2V617F and CALRm) and its high platelets count and hematocrit (75%) (D) Percent of hematologic response, nonresponse, or intolerance among patients with JAK2V617F or CALRm MPNs.

Figure 3.

Hematopoietic progenitors are targeted differently according to the driver mutation type or zygosity and the IFNα dose. (A) Effect of IFNα in different hematopoietic compartments during the clinical survey of the 48 patients. Graph lines indicate the VAF calculated in CD34+ progenitors (i-ii) and measured in granulocytes (iii-iv) for CALRm and MPLm patients (i-iii) and JAK2V617F patients (ii-iv). Thin lines, data from each patient harboring JAK2V617F (dotted blue), CALRm (dotted orange), or MPLm (green); thick curves, smoothed averages (floating averages ±100 days) from the 32 JAK2V617F (blue) or 14 CALRm (orange) patient data; surrounding shaded areas, standard error of the mean. P values were calculated between CALRm and JAK2V617F data. Within the first 300 days, there was no significant difference between CALRm and JAK2V617F patient VAF. Significant differences between CALRm and JAK2V617F cases were observed starting from 600 days of treatment in the progenitor compartment using a Mann-Whitney U test (P < .0005). Less difference between CALRm and JAK2V617F VAF was observed in mature cells (P < .025 after 650 days). (B) Effect of IFNα according to driver mutation type or zygosity and the IFNα dose during the clinical survey. The VAF were computed by pooling the data from each of the 3 progenitor compartments for (i) patients with CALRm MPN treated with IFNα at high doses (HD, >78 µg/wk on average) or low doses (LD, <78 µg/wk); (ii) patients with JAK2V617F MPN treated with IFNα at HD (>96.5 µg/wk) or LD (<96.5 µg/wk); and (iii) JAK2V617F heterozygous or homozygous progenitors independently of the doses. Thin line, data from a single patient; thick curves, smoothed averages (floating averages ±100 days) from the JAK2V617F (blue) or CALRm (orange) patient data; shaded areas surrounding the curve, standard error of the mean. Differences were calculated between heterozygous and homozygous JAK2V617F progenitors after 600 days of treatment (Mann-Whitney U test, P < .03) and were significant after 1000 days of treatment (Mann-Whitney U test, P < .003).

Figure 4.

Mathematical model and inferred dynamics of JAK2V617F, CALRm, and MPLm cells. (A) Design of the mathematical model. Mature and fully differentiated cells no longer divide and die at a rate δm. We modeled progenitor cells as originating from active HSCs that divide and encounter several divisions (modeled by the parameter κi). Progenitors exit their compartment at the differentiation rate δi and proliferate (modeled by the parameter κm) before entering the mature compartment. We also introduced 2 stem cell compartments depending on whether the HSC is considered active or inactive (quiescent), parameters ɣ and β model the exchanges between these 2 compartments. We assumed that the active HSCs might be recruited to differentiate at a rate α to contribute to hematopoiesis. Parameter Δ models the propensity of the stem cell pool to be depleted (if Δ < 0) or to expand (if Δ > 0). (B) Examples of dynamics of inferred mutated progenitor, HSC (CF), and mutated mature cells (VAF) are presented. Dynamics focusing on (i) homozygous JAK2V617F cells for 3 patients, (ii) heterozygous JAK2V617F cells from 2 patients, (iii) heterozygous CALRm cells from 2 patients, and (iv) heterozygous MPLm cells from a patient. Dots, square, and triangles, experimental data values; curves, median values determined from the mathematical model; red line, inferred dynamics of mutated HSCs (overlaid with the heterozygous progenitor CF for CALRm cases); shaded areas surrounding each curve, 95% confidence intervals. When comparing mature cells dynamics to heterozygous progenitor dynamics, we must keep in mind that the VAF in progenitor cells would be half the CF.

Figure 5.

HSC are targeted differently according to driver mutation type and zygosity and the IFNα doses. Molecular stem cell response factor (R factor) was predicted at the end of the survey (3000 days) for (i) heterozygous CALRm HSCs in patients treated with high vs low IFNα doses, (ii) heterozygous type 1 vs heterozygous type 2 CALRm HSCs, (iii) heterozygous vs homozygous JAK2V617F HSCs, (iv) heterozygous JAK2V617F HSCs in patients treated with high vs low IFNα doses, (v) homozygous JAK2V617F HSCs in patients treated with high vs low IFNα doses, and (vi) global JAK2V617F HSCs in patients treated with high vs low IFNα doses. The R factor is defined as the ratio (median value) between the inferred mutated HSC proportion after a given time of treatment (t = 3000 days) over the proportion of mutated HSCs at the initial time. Depending on the context, it refers to heterozygous or homozygous CF or VAF. Dash lines, R = 1 for no response, R > 1 for a negative response, and R < 1 for a positive response. R < 0.5 corresponds to a PMR, and R ∼ 0 corresponds to a complete molecular response. Solid lines, R median. R significantly differs between heterozygous CALRm HSC with low vs high IFNα doses (Mann-Whitney U test, P = .0087) and between type 1 CALRm and type 2 CALRm (Mann-Whitney U test, P = .0162). R significantly differs between heterozygous and homozygous JAK2V617F HSCs (Mann-Whitney U test, P = .0047). R tends to differ between heterozygous JAK2V617F HSCs treated with high vs low IFNα doses (Mann-Whitney U test, P = .0745). R significantly differs in the global JAK2V617F VAF in HSC between treatment with high and low doses of IFNα (Mann-Whitney U test, P = .0288). For each patient, we computed an average of received IFNα doses over the first 450 days of treatment. HD vs LD are defined according to the median dose of the groups of considered patients. The threshold is automatically computed to compare 2 subgroups of patients of the same size. The dose thresholds of IFNα are 78 µg/wk for heterozygous CALRm HSCs, 96.5 µg/wk for JAK2V617F HSCs, 96 µg/wk for heterozygous JAK2VF617F HSCs, and 108 µg/wk for homozygous JAK2V617F HSCs. Statistical differences were calculated using a Mann-Whitney U test: *P < .05, **P < .01.

Figure 6.

IFNα differentially impacts on JAK2V617F and CALRm HSC homeostasis. (A) Graphs indicate the means values (solid lines) of the estimated parameters calculated using the mathematic model. (i) Δ* parameters were calculated in heterozygous and homozygous JAK2V617F HSCs and heterozygous CALRm HSCs. Δ*het significantly differs in patients with CALRm MPN vs those having JAK2V617F MPN (Mann-Whitney U test, P = .0031). Δ*het in patients with CALRm MPN is significantly different from Δ*hom of those having JAK2V617F MPN (Mann-Whitney U test, P < .0001). The dotted lines indicate Δ* = 0. Δ* > 0 corresponds to an expansion, and Δ* < 0 corresponds to a depletion of the stem compartment. (ii) Δ*het parameters were calculated in type 1 and type 2 CALRm HSCs. Δ*het significantly differs in patients with type 1 CALRm MPN vs those with type 2 CALRm MPN (Mann-Whitney U test, P = .004). (iii) Inverse ratios of the ɣ* parameter was calculated in heterozygous vs homozygous JAK2V617F HSCs. The 1/ɣ* value can be seen as a relative time spent by cells in the inactive compartment of our model. The 1/ɣ* significantly differs in heterozygous vs homozygous JAK2V617F HSCs (Mann-Whitney U test, P = .027). (B) Proposed mechanism of IFNα in JAK2V617F and CALRm HSCs.