Abstract
Budd-Chiari Syndrome (BCS), characterized by thrombosis of the hepatic vein, is a classical complication of PNH. While BCS can be attributed to various causes, e.g., genetic and acquired thrombophilias and JAK2 mutations, many cases remain idiopathic. However, a portion of these cases may be due to PNH albeit missed because of lack of screening, especially if BCS is the initial presentation of PNH. The aim of this study was to determine the etiologic fraction of BCS, identify features of distinction between BCS and non-BCS PNH patients, as well as to identify simplified screening tools to increase the yield of PNH testing. For that purpose, we used our institutional cohort to identify unprovoked cases of BCS and leveraged large healthcare network and publicly available cohorts for comparative purposes.
In total, 302 cases of BCS were identified in our institution per stringent diagnostic criteria (2.6 per 100,000 patients). Among these cases, genetic causes of thrombophilia were responsible for 21% (n=62) of BCS cases, with Factor V Leiden (FVL) and antiphospholipid syndrome (APS) as the most common manifestations, at 33% (n=20) and 31% (n=19) of thrombophilic cases, respectively. The JAK2 V617F mutation was found in 15% (n=44) of patients. Finally, PNH was diagnosed in 23 (8%) of BCS patients. By comparison, in the NIH All of Us Database and TriNetX, the rates of BCS were 33 and 9 per 100,000 patients respectively, with only 0.6-1.5% of cases attributable to PNH.
To further explore the relationship between PNH and BCS, we next examined clinical and molecular data of PNH patients in our institutional cohort. We first investigated rates of BCS and other thromboses in PNH patients with different clone sizes. Remarkably, no patients with BCS had a clone size of < 40% granulocytes. The rate of BCS in patients with a clone between 40-59% was 24% and peaked at 26% for patients with clones between 60-100%. Controlling for PNH clone size, we then compared PNH patients with (BCS+, n=23) and without BCS (BCS-, n=71) to identify distinguishing characteristics; BCS+ patients had a higher ANC (4.4 vs. 2.4x10^9/L, p<0.01) but lower LDH (558 vs. 950 U/L, p=0.01) vs. BCS- patients. When we compared PNH patients with large (n=94) and small (n=30) clones, using the previously stated 40% as the cutoff, thrombotic events were more frequent in patients with larger clones. These patients also had significantly higher LDH and ANC levels. Interestingly, there were 4% of BCS+ PNH patients with thrombophila and 13% with co-occurrent JAK2 mutations.
To test the assumption that PNH is an underdiagnosed cause of BCS, the rates of screening and positivity of PNH in BCS was compared with JAK2, which is widely recognized for its associated with BCS. PNH showed a much lower rate of screening (49 vs 35%) albeit with comparable diagnostic yield (30 vs. 22%). Stratifying by decade, screening for PNH remained constantly lower than that of JAK2. To unravel potentially cryptic PNH cases, we created an algorithm to estimate the numbers of PNH diagnoses missed due to lack of testing. Parameters included in the diagnostic model were derived from comparison of BCS cases with/without PNH. Hemoglobin, platelets, LDH, and reticulocyte count were significantly different between the two groups. Using these parameters, a Bayesian approach was employed and trained using patients with known PNH status (n=106). The model showed a sensitivity and specificity for PNH diagnosis of 0.90 and 0.84, respectively. When we applied the algorithm to unscreened BCS patients (n=196), we estimate that 8 additional “missed” BCS cases were attributable to PNH. As a result, the sum of projected “missed cases and those formally diagnosed with PNH” resulted in a predicted PNH incidence of 11% in BCS, increasing the rate BCS in PNH by 35%.
Analysis of treatment patterns showed that, in 135 patient-years on C5 inhibitors, only 1 BCS case was recorded. The majority of PNH patients (53%) were given low molecular weight heparin as an acute anticoagulation strategy whereas 24% were acutely given DOACs. Coumadin was the most common chronic anticoagulant, given to 52% of patients, followed by DOACs (38%), and enoxaparin (10%).
In sum, our study provides a definitive etiologic landscape of BCS and further delineates the nature of the relationship between PNH and BCS. Critically, our results reveal a significant proportion of patients with BCS for whom causative therapy is available.
