Screening for Signs of Portal Hypertension by Esophagogastroduodenoscopy in Patients with BCR-ABL Negative Myeloproliferative Neoplasms.

To the Editor: Non-cirrhotic portal hypertension (PH) is a well-recognized yet poorly characterized complication of BCR-ABL negative myeloproliferative neoplasms (MPN). Diagnosis is often delayed until decompensation occurs. Screening for gastroesophageal varices (GEVs), one of the most clinically significant manifestations of PH, is not routinely performed in MPN patients but might facilitate early diagnosis and intervention. Portal vein thrombosis (PVT) is a known cause of PH in MPN. PVT can remain asymptomatic and undetected until overt PH manifests. Indeed, GEVs are highly prevalent in patients with chronic PVT.1 PH can also occur in the absence of splanchnic circulation thrombosis, with portosinusoidal vascular disease (PSVD), sinusoidal obstruction due to extramedullary hematopoiesis (EMH), and increased portal blood flow due to splenomegaly as proposed mechanisms.2 How PH should be evaluated in MPN patients is unknown. Doppler ultrasonography (USG) can detect asymptomatic PVT, identifying patients who may benefit from anticoagulation. Analogously to liver cirrhosis, the early detection of GEV by esophagogastroduodenoscopy (EGD) may facilitate interventions to prevent variceal hemorrhage. However, there is a lack of epidemiological data on asymptomatic PVT and GEV in MPN patients to guide such screening. A study from Princess Margaret demonstrated increased hepatotoxicity in MF patients undergoing allogeneic hematopoietic stem cell transplantation (HCT) compared to patients with myelodysplastic syndrome.3 We hypothesized that this may be related to PH associated with the underlying MPN and therefore started EGD screening in MF patients planned for HCT, finding signs of PH in 15% (unpublished data, V. Gupta). As HCT is usually offered to patients with advanced disease, we wanted to determine the extent of the problem in the broader MPN patient population, which in turn led to design of this screening study (NCT01816256). We adopted a precision approach in calculating the sample size. Based on literature review and clinical judgment, we assumed that the prevalence of asymptomatic PH in MPN patients would be 5%–10%. Several scenarios were considered. The scenario adopted assumes a prevalence of 8% for these events, and a sample size of 100 estimates 90% confidence interval (CI) between 4% and 14%, meaning that 4% would be an acceptable lower limit for moving forward with the screening program (see methodology details in Supplementary Information). We enrolled adult patients with essential thrombocythemia (ET), polycythemia vera (PV), and primary myelofibrosis (PMF) or myelofibrosis (MF) evolving from ET or PV at Princess Margaret Cancer Centre and Sunnybrook Health Sciences Centre in Toronto between May 2013 and March 2019 (Figure S1). Patients with MF were required to have a spleen of at least 5 cm below the costal margin. Patients with prior history of GEV, PVT, Budd Chiari, or liver cirrhosis were excluded. EGD was used to screen for GEV and portal hypertensive gastropathy (PHG). The severity of EGD findings was classified according to the size of varices and/or severity of PHG, which correlate with bleeding risk. USG was performed to evaluate for PVT and spleen size. Of the 101 enrolled patients, 85 completed EGD and 84 completed USG (Figure S1). There was no difference in age, sex, or ECOG between patients who completed EGD versus those who did not (data not shown). Table S1 shows baseline characteristics of the study cohort. In keeping with the referral pattern at our centers, the majority of enrolled patients (93% [79/85]) had MF. EGD revealed GEV and/or PHG, in 33 patients (39%, [90% CI 30–48]); 13 patients (15% [90% CI 9–22]) had moderate-to-severe EGD findings (Table 1). The prevalence of GEV and/or PHG among MF and PV/ET patients was 37% (29/79) and 67% (4/6), respectively. Among MF patients, EGD findings were moderate-to-severe in 14% (11/79) (90% CI 8–20). USG revealed no cases of asymptomatic PVT. The median spleen size by USG was 20.6 cm (9.2, 31.5) for the entire cohort (N = 82). There was no difference in median baseline spleen size between patients with and without manifestations of PH on EGD (18.5 cm [9.2, 27.7] [n = 31] vs. 21.4 cm [10.8–31.5 cm] [n = 51], p = .11), or those with moderate-to-severe findings of PH compared to those without any findings of PH (22 cm [17–27.7 cm] [n = 12]) versus (21.4 cm [10.8–31.5 cm] [n = 51]), p = .99. There were no adverse events related to study screening investigations. Baseline variables were similar between patients with and without EGD findings of PH (Table S1). The frequency of JAK2, CALR, and MPL mutations among patients with GEV and/or PHG was 79% (26/33), 12% (4/33), and 6% (2/33), respectively; one patient had no canonical MPN driver mutations. There was no difference in distribution of MPN driver mutations between patients with and without EGD findings of PH (Table S1). Among MF patients with and without GEV and/or PHG, there was no difference in DIPSS risk or frequency of high molecular risk (HMR) mutations (Table S1). In univariate analysis, there were no factors predictive of EGD finding of PH (Table S2). Among the 13 patients with moderate-to-severe findings of PH on screening EGD, at least one follow-up EGD was done in 11 patients (Table S3); 10 patients received nonselective beta blocker (NSBB); 7 had variceal band ligation (VBL). Nine patients received JAK inhibitor therapy, and 3 patients underwent HCT. During follow-up, 2 patients developed variceal bleeding (VB) and 1 developed ascites. At last follow-up EGD, signs of PH severity improved in 6 patients. Among the 20 patients with mild baseline PH findings, 6 received NSBB prophylaxis; 11 patients underwent ongoing surveillance EGDs with 3 patients having an interval increase in size of varices that subsequently improved to baseline. There were no PH-related complications reported in this group. Taken together, our study shows that EGD manifestations of PH are common in MPN patients. Our findings are primarily generalizable to MF patients, who comprised the majority of the study population. Most patients were early in their disease course, with 83% having low or intermediate-1 risk disease by DIPSS. Further, we found that 25% of patients with PH findings on EGD, had varices deemed of sufficient risk by the treating hepatologist to warrant further intervention, including prophylaxis with NSBB, VBL, and/or ongoing endoscopic surveillance. Although our study was not powered to determine the influence of these interventions on outcomes in MPN patients, they are known to prevent VB in liver cirrhosis. Interestingly, 3 of 6 patients who experienced improvement in EGD findings during follow-up received a JAK inhibitor in addition to PH-directed treatment. The influence of JAK inhibitor therapy on PH warrants evaluation in larger studies. We did not identify any predictors of PH-related EGD findings in our study population, which were present across all age groups, ECOG categories, DIPSS risk strata, and driver or other myeloid malignancy mutations. Spleen size did not correlate with EGD findings of PH. The lack of readily identifiable risk factors for PH highlights the need for a heightened clinical suspicion for its presence in MPN patients. By comparison, one of the largest retrospective reviews comprising 228 MPN patients (53.5% PV/ET, 31.6% MF) revealed GEV on CT in only 4.8%4; this lower prevalence may be explained by the decreased sensitivity of CT for GEV compared to EGD. Further, all PH patients harbored JAK2 mutations, contrasting with our study, which found a similar distribution of MPN-driver mutations among patients with and without PH on EGD. In the only other prospective study involving 29 MPN patients, largely PV or ET, 14% were found to have PH as determined by USG features, and two patients were found to have GEV on EGD.5 Interestingly, among compensated liver cirrhosis patients, high-risk GEVs are found in 10%–20%. We did not find any cases of asymptomatic PVT. Patients with PVT are known to have high incidence of GEV.1 Patients with known liver disease and PVT were excluded from this study to avoid overestimation of PH related to non-MPN causes. The mechanism of PH in our study population may be related to PSVD, EMH in the liver, and increased splenic blood flow. Our findings must be interpreted within the context of our methodology. Firstly, our study population mainly consists of MF patients, limiting the application of our findings to this group. As not all patients with PH develop GEV or PHG, our study may have missed cases of PH without EGD manifestations. Further, our study was not designed to study the efficacy of PH-directed interventions in MPN patients, which were implemented at the discretion of the treating hepatologist. Moreover, the efficacy of these interventions is largely based on results in liver cirrhosis patients; further studies need to address their impact on MPN-related PH. This is the first study to systematically evaluate the feasibility and efficacy of EGD screening in MPN patients and is consistent with principles of effective screening programs.6 PH in MPN patients can be detected in early phase by acceptable low-risk screening procedures in a significant number of patients. Early detection of PH can potentially lead to interventions aimed at risk reduction of serious complications such as variceal bleeding. In summary, we found that manifestations of PH can be found in a significant number of MF patients early in the disease course. Our study suggests inclusion of EGD screening for signs of PH in the workup of MF patients. Vikas Gupta designed the study, obtained funding, and wrote the study protocol. Marta Davidson and Vikas Gupta verified the data and wrote the manuscript. Marta Davidson and Eshetu G. Atenafu performed data analysis. Florence Wong performed EGDs and decided on PH-directed therapy. Mostafa Atri reviewed USG findings. Marta Davidson, Vikas Gupta, Hassan Sibai, Dawn Maze, Jeannie Callum, and Verna Cheung, contributed to clinical care of patients. Marta Davidson and Vikas Gupta had full access to all the data in the study and had final responsibility for the decision to submit it for publication. All authors critically reviewed the manuscript and provided relevant input. All authors approved the final version of the manuscript. This investigator-initiated study was supported by grant support from Novartis (Vikas Gupta) and a program project grant from the Elizabeth and Tony Comper Foundation and the Princess Margaret Cancer Foundation (Vikas Gupta). Marta Davidson, Florence Wong, Mostafa Atri, Eshetu G. Atenafu, Dawn Maze, Hassan Sibai, and Verna Cheung declare no competing interests relevant to this study. Vikas Gupta declares consulting fees from Novartis, BMS Celegene, Sierra Oncology, AbbVie, Constellation Biopharma, and Pfizer; payment or honoraria from Novartis, BMS Celgene, and Constellation Biopharma; and participation on a data safety monitoring board of advisory board of BMS Celgene, Roche, Abb Vie, and Pfizer. All study participants provided written informed consent. We will share anonymized patient-level data on request. Data S1: Supporting Information. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.