Counterpoising Immune Pressure and Privilege: Paroxysmal Nocturnal Hemoglobinuria Evolutionary Dynamics

Several lines of evidence posit that PIGA-mutant cells lacking GPI- anchored proteins may escape aplastic anemia (AA) immune pressure ensued by autoreactive cytotoxic T-cells. Whether conditional or immune-privileged, the subsequent expansion of GPI-deficient clones in AA bone marrow constitutes the pathobiological foundation for PNH development. Another hypothesis is that PNH clones have neither an absolute nor a conditional growth advantage: as PIGA mutations occur stochastically in hematopoietic stem cells (HSCs), the disease will manifest should PNH clones become sufficiently large. Around 40% of PNH cells designated as primary PNH (pPNH) show latent expansion without anamnestic evidence of AA; in these cases, time from diagnosis to progression is zero. Clinical evidence shows that 10% of AA cases presenting with PNH clones eventually expand and progress into secondary hemolytic PNH (sPNH), whereas others seem to remain smoldering. Besides PNH clone size, it is not clearly ascertainable the baseline risk/timing of sPNH evolution, whose identification early on may have important clinical and therapeutic implications. Here we took advantage of a large cohort of AA/PNH patients to study PNH evolutionary trajectories.

We accrued a total of 430 BMF patients, of whom 233 cases were noted as AA carrying small PNH clones and 85 patients with pPNH whose time to evolution is unknown but coincided with the initial diagnosis. Serial flow cytometry (FCM) results at follow up time points were recorded to monitor clonal size progression and sPNH was defined as a PNH granulocyte clone size expansion beyond 20%.

Overall, median age was 55 years (IQR 34-68) at diagnosis with a M:F ratio of 0.77. Baseline PNH clone size was <1% in 151 (65%), 1-5% in 42 (18%), 5-10% in 26 (11%), and 10-20% in 10 (4%) patients, with the majority (83%) classified as having type III dominant erythrocyte clones. Median and cumulative follow up times were 59.1 months (IQR 21.5-130.4) and 1533 patient-years, respectively. Evolution to sPNH occurred in 31 patients (13.3%) at a median time of 38.9 months (IQR 20-73.1) and associated with granulocyte clone size ≥5% (HR 30.6, 95% CI 7.3-128.7; P<.01), PIGA mosaicism (HR 5.6, 95% CI 1.2-25.7; P<.01), non-severe AA (HR 23.5, 95% CI 2.8-196.3; P<.01), no ATG therapy for AA (HR 15.8, 95% CI 1.6-153.9; P<.01), and failure to achieve CR (HR 4.4, 95% CI 1.5-13.4; P<.01). In contrast to sPNH, patients with pPNH without aplasia presented at younger age (median 36 years, IQR 25-57, P<.01), had a similar M:F ratio (0.74, P=.87) and percentage of type III dominant clones (75%, P=.3), while median clone size was 77 (IQR 51-93).

Next, we defined features driving PNH clonal dynamics over time. By setting an absolute Δ threshold of >5% increase of PNH clone size, we identified 33 AA/PNH patients and defined them slower/faster progressors based on a median time of 23.8 months (IQR 11.9-56.9) needed to meet the set criterion. Faster progressors were more likely to show non-severe AA phenotypes (OR 4, 95% CI 0.9-16.6; P=.05), not having received IST with ATG (OR 4.3, 95% CI 0.9-15.5; P=.05), and not achieving CR to IST (OR 3.7, 95% CI 0.8 to 15.1; P=.07). The latter was the only independent risk factor to rapid PNH clonal expansion (HR 6.8, 95% CI 1.1-43.5; P=.04). We further tested other progression thresholds and molecular parameters for pPNH vs fast and slow progressors, and the data for these will be presented at the meeting.

We then built a computational framework to simulate PNH cell dynamics which considered the features captured by our real life cohort analysis. By assuming an HSC division rate of about 1/year¹ and a PIGA mutational rate of 5×10⁻⁷ per replication² as well as the strong effect of IST and response, we developed an individual based well mixed birth-export process and simulated different settings of treatment with IST and no treatment with diverse PNH granulocyte clone size (between 5 and 20%). The simulations recapitulated both the stochastic nature of PNH clonal expansion and the association of treatment response with evolutionary dynamics.

In sum, our findings suggest that PNH clonal expansion is strongly related to AA disease course and outcomes. With both real life and evolutionary dynamics, we gave additional evidence to believe in the theory of selective survival of indolent PNH clones as an additional means of expansion - on top of stochastic expansion - in underpopulated BM in the setting of an ongoing immune pressure.