Paroxysmal Nocturnal Hemoglobinuria Results from Initial Clonal PIG-a Mosaicism

The pathogenesis of PNH involves alteration of at least one HSC clone, characterized by a somatic mutation in the PIG-A gene. The resulting phenotype appears to offer a growth advantage over other unaffected HSCs, yet the failure of PIG-A mutant clones to expand in animal models suggests that additional factors are necessary for the development of PNH. The association of PNH with aplastic anemia (AA) is evidence of selection pressure by immune attack, but secondary hits in additional genes may also be clonally acquired to render PNH clones competitive irrespective of immune pressure. With the advent of NGS, clonal architecture analysis has been revolutionized. Yet, despite increased resolution, NGS is hampered by the presence of normal/non-clonal cells at varying frequencies in the sequenced sample, often decreasing the detection threshold and precluding recognition of subclonal mosaicism.

To study these issues in PNH, 133 patients with PNH (N=33), AA (N=33), and AA/PNH (N=67) were sequenced using deep targeted NGS covering all exons of the PIG-A gene. All patients had a detectable PNH clone by flow cytometry, with the mean WBC clone size of 0.22% in AA, 27.4% in AA/PNH, and 77.1% in PNH. 223 PIG-A mutations were identified in this cohort; 35% were frameshift mutations, 34% missense, 14% splice site, 13% nonsense, and 3% non-frameshift. In unpurified blood, PIG-A mutations were found in 27/33 (82%) of PNH patients, 36/67 (54%) of AA/PNH, and 2/33 (6%) of AA, with a clear association between PNH clone size and the ability to detect mutations by NGS (P<0.0001). In DNA extracted from cells flow sorted by FLAER negativity, detection rates were 26/29 (90%), with an average of nearly 3 mutations per patient, indicating that mechanisms such as microdeletions may affect as many as 10% of PNH or AA/PNH patients and that clonal mosaicism is common. Our results also show that indeed type III and II PNH cells are derived from 2 distinct clones, each with a distinct mutation.

To examine disease pathogenesis, we studied 47 patients longitudinally using both PNH flow and deep sequencing. Serial testing illustrates possible modes of evolution: mutations can fluctuate in the contribution to clonal burden over time with the variant allelic frequency of some mutations increasing while others decrease. Linear regression of the number of mutations vs . time from diagnosis suggests that in AA/PNH (N=29), the number of mutant clones detected increases over time (P=.012) while in classic PNH (N=18) the trend is reversed, as the number of mutations detected decreases (P<.0001) and a dominant clone emerges to supply a majority of blood cells.

Traditionally, PNH has been considered a monoclonal disease in which a PIG-A mutant HSC clone can outcompete normal hematopoiesis within the context of permissive conditions. Using deep NGS, our results clearly indicate that while a single clone ultimately may dominate, PNH begins as oligoclonal process with the eventual emergence of the most competitive HSC clone(s) populating hematopoietic space. Initially, an environment that "favors" a state of GPI-anchor deficiency thereby allows multiple clones to evolve. Over time, the most "fit" PNH HSC clones dominate, resulting in monoclonality. In addition, our results confirm the presence of distinct mutations with corresponding phenotype through sequencing of sorted type II and type III PNH cells. Clonal drift may also be observed, likely through a newly acquired advantage of one clone within the available pool of GPI-deficient PIGA mutant clones. Our previous work indicated that acquisition of additional enabling mutations results in a more complex clonal architecture comparable to that seen in MDS, an intrinsic mechanism of the selection process that involves mutations in genes altered in MDS. Additional intrinsic hits also explain patient-specific kinetics of PNH evolution and apparent independence of established PNH from the effect of immunosuppression. Thus, while immune selection may provide initial conditions favorable for oligoclonal PNH evolution, once pressure is relieved, intrinsic factors can lead to PNH monoclonality. Alternative mechanisms could include loss of dominance by exhaustion, senescence, or telomere shortening.

In sum, our study clearly demonstrates initial polyclonality early in the pathogenesis of PNH with selection of dominant clones during disease duration.