Underlying mechanisms in the genetic disposition of myeloproliferative neoplasms

Somatic driver mutations are known to exhibit a role as main promoters in myeloproliferative neoplasms (MPN) development. Previous studies have shown that MPN may also be associated with an increased hereditary risk in first-degree relatives of patients with MPN. However, there are limited analyses of the genetic basis of this predisposition.

Erik L. Bao et al. conducted a study in a large sample of patients with MPN and controls to investigate the basic elements of inherited risk, involved genes and cell types, as well as the underlying mechanisms contributing to this risk. Their results have recently been published in Nature.¹

Genome-wide association study (GWAS) revealed risk associations

This was a GWAS using three population-based cohorts (UK Biobank [UKBB], 23andMe, and FinnGen) resulting in a large study population of 2,949 patients with MPN and 835,554 controls. To discover hereditary traits of MPN, 7,343,617 variants were tested. The analysis of GWAS signals produced 15 linkage disequilibrium (LD)-independent loci with a significant association (p < 5 × 10⁻⁸) and ten loci with possible associations (p < 1 × 10⁻⁶). A total of 848 carriers of a somatic JAK2V617F mutation were used to simulate these links independently, as it was found to be associated with the risk of developing MPN. The genotypes of 24 out of 25 lead variants were assessed in these carriers (p < 1 × 10⁻⁶), and 17 variants produced significant results (binomial p < 2.2 × 10⁻¹⁶).

Combined analysis of 3,797 MPN cases and 115,299 controls revealed 17 genome-wide-significant independent risk associations, ten of which were previously found to be associated with MPN risk. These 17 loci were estimated to account for nearly 18.4% of the total familial relative risk.

A 104-variant polygenic risk score (PRS) was developed to further investigate the genome-wide architecture of inherited MPN risk, however, the impracticability of the use of PRS in a population-level analysis was well acknowledged. An analysis of one study cohort estimated a median PRS percentile of 67 in patients with MPN, while it was 50 in those without MPN.

• The risk of disease was similar between the 20th and 70th percentiles.
• Above the 90th percentile, the odds of MPN acquisition were greater:

• • 4.20-fold compared with the lowest decile (95% CI, 3.14–5.63).
  • 2.70-fold compared with the 50th percentile (95% CI, 2.11–3.46).

Along with PRS, individuals were also stratified by JAK2 46/1 haplotype—a known strong contributor to MPN risk—to investigate if the addition of a larger set of variants could change disease risk. Compared to those without JAK2 46/1, the risks were as follows:

• 0.87-fold for carriers of JAK2 46/1 with an intermediate PRS.
• 1.87-fold for those with JAK2 46/1 and a high PRS.

The association between cell types and MPN risk

Investigating 19 blood cell traits from 408,241 individuals of the UKBB revealed that there was a positive genetic correlation for MPN risk and counts of red blood cells, platelets, white blood cells, monocytes, neutrophils, and eosinophils, all of which derive from multipotent hematopoietic stem and progenitor cells (HSPCs). On Bayesian fine-mapping, posterior probability of causality (PP) was calculated for each variant. Risk variants with a PP estimate of > 0.10 (strongly fine-mapped) have shown pleiotropic blood trait associations (21.4%) compared with those with a PP < 0.10 (1.6%). This observation supported the assumption that MPN risk variants may affect several blood cell lineages by taking effect in hematopoietic progenitor cells.

Chromatin accessibility data have revealed that 11.6% of risk variants with a PP value > 0.01 and 4.82% of those with a PP < 0.01 were located within accessible chromatin of more than one hematopoietic community (p = 2.13 × 10^–6), indicating that these risk variants regulate multiple blood cell types. Notably, g-chromVAR, which is known to recognize variants from significant loci, revealed the greatest enrichments of MPN risk variants in multipotent progenitor cells and hematopoietic stem cells (HSCs). Genome-wide LD score regression (LDSC) also supported this finding by showing the strongest presence in progenitor cells (p = 7.10 × 10^–3) compared with other hematopoietic cells (p = 1.89 × 10^–2).

The relevance of other traits

The association between leukocyte telomere length and MPN was also explored. The basis for this analysis was the correlations of leukocyte telomere length with the telomere length of earlier hematopoietic progenitors and the self-renewal capability of HSCs. The overlap between MPN risk and leukocyte telomere length-associated variants was evaluated on the telomere reverse transcriptase (TERT) gene in 78,592 subjects. Two independent variants, rs7705526 and rs2853677, were found to increase telomere length and to be associated with MPN risk. Data analyzing TERT and genome-wide LDSC could confirm a positive genetic correlation between MPN risk and leukocyte telomere length (p < 0.001 and p = 0.037, respectively). Two-sample Mendelian randomization also showed a consistent association between increased leukocyte telomere length and higher MPN risk, which was also related to other cancer types.

In addition, analyses revealed a similar correlation between inherited MPN risk and clonal hematopoiesis of indeterminate potential (p = 0.04). Notably, five of 17 main MPN variants were found to be relevant for acute myeloid leukemia, suggesting MPN risk loci may account for a predisposition to other clonal disorders.

Genes related to inherited risk and possible mechanisms

Three genes with coding variants (rs1800057, rs3184504, and rs17879961) were selected to investigate germline MPN predisposition. These genes were part of risk variants at PP > 0.10 with known consequences: rs1800057 leads to a variant gene in ATM; while rs3184504 and rs17879961 generate variant genes for SH2B3 and CHEK2, respectively.

These variants showed stronger than expected functional interactions as mapped by hematopoietic promotor capture, especially with genes that are known to play a modulator role in self-renewal and the activity of HSCs, namely ZNF521, GATA2, MECOM, RUNX1, HMGA1, ATM, FOXO1, TET2, JAK2, SH2B3, and TERT. Other target genes have been implicated in replicative senescence, cell cycle, and hematopoietic and lymphoid development, and were maximally enriched in HSCs, multipotent progenitor cells, and progenitors related to myeloid differentiation.

The mechanisms contributing to MPN predisposition were investigated through the I157T missense variant of the CHEK2 protein, which has previously been found to be associated with increased risk for other cancers. The structural analysis indicated that the I157T variant decreases the hydrophobic interfaces and destabilizes the CHEK2 dimer. The inhibition of CHEK2 by RNA interference led to a decrease in cell death following irradiation in human primitive lineage (Lin)⁻CD34⁺CD38⁻ cells, while increasing the expansion of human cord blood Lin⁻CD34⁺ cells in long-term cultures. These observations led the authors to conclude that CHEK2 is involved in the inhibition of HSPC growth and self-renewal, which could be alleviated by the I157T variant, thereby increasing MPN risk.

Other research was carried out analyzing a risk locus situated in an enhancer region of chromatin accessible in hematopoietic cells, which is located downstream of the GFI1B transcription factor gene, and involved in HSC quiescence and megakaryocytic erythroid differentiation. A rs524137 variant at this locus was found to be associated with a 40-fold increase in transcriptional activity and a 1.7-fold reduction in enhancer activity in HSPCs. CRISPR-guided deletion of a region encompassing the enhancer led to a decrease in the expression of GFI1B by 36.3%, suggesting GFI1B was a causal gene. This deletion was also associated with a 2.7-fold increase in long-term HSCs relative to the total CD34⁺ progenitor cells and with an increase in secondary colonies in replating assays with no effect on secondary erythroid colonies. These findings conclude that rs524137 decreases GFI1B expression in HSPCs and increases their self-renewal function.

Conclusion

This study exemplified the germline genetic risk of developing MPN related to changes in gene expression in HSCs, increasing the baseline pool of HSCs and augmenting the risk of acquiring MPN somatic driver mutations. This study identified a number of germline variants preceding the acquisition of somatic driver mutations and further elucidated the inherited risk associated with MPN development, which supports previous findings. Previous experience with other cancer types emphasizes the role of predisposing germline variants to better understand the mechanisms of MPN development and to inform future drug development.