Rajoriya, Neil

INTRODUCTION: Nosocomial infections (NIs) in cirrhosis are associated with high mortality but could be preventable. Logistic regression (LR) models have failed to identify high-risk patients. We aimed to develop machine learning (ML) models to predict NI. METHODS: The CLEARED consortium consists of prospectively enrolled cirrhosis inpatients from >120 centers. Using day-of-admission clinical data, 3 ML approaches (random forest [RF], extreme gradient boosting, and neural networks [NNs]) were used to predict NI. Data were split 80:20 for training and testing stratified by the outcome. Models were compared using area under the receiver operating characteristic curve (AUC). RESULTS: In total, 8,263 patients (55.90 ± 13.34 years; 64.1% men) from 127 centers in 37 countries were included. NI developed in 869 (10.5%), a median of 6 (4–11) days of postadmission. Major NIs were respiratory (29.6%) and urinary tract infection (15.7%), and spontaneous bacterial peritonitis (13.5%). NIs occurred more frequently in patients from low/low-middle income countries and those with severe liver disease, alcohol etiology, and admission infections. NIs were associated with inpatient mortality (31.9% vs 8.1%, P < 0.001) and liver transplantation (4.8% vs 1.9%, P < 0.001). Although the RF model (AUC 0.69) showed good calibration (Brier score 0.09), outperforming extreme gradient boosting, neural network, and LR models (AUC 0.66 for all; LR comparison P = 0.043), no model achieved AUC ≥0.80 for clinical utility. At 10% predicted probability threshold, the RF model demonstrated only 75.4% sensitivity, 52.9% specificity, and 15.9% positive predictive value (PPV). DISCUSSION: NIs cannot be accurately predicted from day-of-admission data using ML models, even in a large, prospective, global cirrhosis cohort. Every hospitalized patient with cirrhosis should receive protocolized infection control measures.

Abstract Philadelphia-negative Myeloproliferative Neoplasms (MPN), including Polycythemia Vera (PV), Essential Thrombocythemia (ET), Myelofibrosis (Primary [PMF] and secondary to PV and ET [PPV-, PET-MF] and unclassified MPN (U-MPN), are associated with an increased risk of venous thrombosis in unusual sites, such as splanchnic vessels (SVT). SVT can lead to complications such as portal hypertension, esophageal and gastric varices, ascites,hepatic failure and biliopathy. According to a meta-analysis MPN is the underlying cause of portal vein thrombosis (PVT) in 31.5% and Budd Chiari syndrome (BCS) in 40.9% of cases (Smalberg, 2012); a more in-depth analysis of clinical characteristics and evolution of MPN-associated SVT has been hampered by heterogeneity of cohorts comprising small number of cases. We conducted a retrospective multicenter study in patients (pts) with SVT associated with WHO2008-diagnosed MPN, with the aim to describe patient characteristics, disease course and prognostic factors with potential implications for clinical practice. Data were collected from 16 international hematologic centers in the framework of the Italian AGIMM and the IWG-MRT groups. We collected 519 cases of pts with PVT, splenic or mesenteric vein thrombosis (75.1%) and BCS (24.9%) associated with MPN. We used as comparator a cohort of 1686 controls (Ctr) represented by MPN without (w/o) SVT: 741 ET (43.9%), 684 PV (39.7%), 261 PMF (15.5%). Frequency of MPN associated with SVT was 37.8% ET (n=196), 36.8% PV (n=191), 15.4% MF (n=80), 10% U-MPN (n=52). Median follow-up was 89.9 months (mo) (range 0.5-430). For SVT vs Ctr group females were 54.5% vs 44.4% in PV (P=0.001), 68.4 vs 63.5% (p=0.13) in ET, 63.7% vs 29.1% in PMF (p<0.0001); median age at MPN diagnosis (dg) was 43.5 yr (range 12-90) vs 60.6 yr (range 12-93) (p<0.0001). Age at SVT dg was 44 yr (range 15-85). In 240 cases (46.7%) MPN and SVT dg were coincident, in 121 (23.6%) SVT occurred before MPN dg (median 26 mo, range 4-307) and in 153 (29.8%) during MPN follow up (median 68 mo, range 4-362). JAK2V617F mutation was found in 94% PV vs 94% in Ctr, 84% vs 61% ET (p<0.0001), 88.1% vs 68% PMF (p=0.006) and in 93% U-MPN. Erythropoietin-independent colonies (EEC) were evaluated in 111 SVT pts and found in 80 (72%), accounting for 38/48 PV (79%), 31/44 ET (70.5%), 9/12 PMF (75%) and 2/7 U-MPN (28.6%). At dg, SVT PV pts had lower hemoglobin levels than Ctr: median was 17.4 g/dL vs 18.5 g/dL (p<0.0001) in male, 16.9 g/dL vs 17.7 g/dL (p=0.0006) in female. A co-existing thrombophilic status was found in 38.5% SVT vs 11.8% of Ctr (p<0.0001). Recurrent SVT occurred in 12.2% of pts with a rate of 1.6% person/year (CI 1.2-2.1); risk of venous thrombosis other than SVT was increased in SVT group vs Ctr (p=0.02), with no difference for arterial thrombosis. Hemorrhage was more frequent in SVT group (32%) vs Ctr (7.2%)(p<0.0001), mainly related to esophageal varices, which were present in 66.9% of SVT pts. There was no difference in evolution to MF and acute leukemia (AL) for PV and ET pts with and w/o SVT, while risk of AL was lower in MF with SVT (p<0.00001). Overall survival was shorter in ET pts with SVT vs Ctr (p<0.0001). In PMF survival was better in SVT group (p<0.00001) and was associated with a higher proportion of SVT pts in lowest risk categories: IPSS low 65%, intermediate-1 20%, intermediate-2 10% and high 5% compared with 15%, 34%, 25% and 26% in Ctr group. At last FU, 79/519 pts (15.2%) had died; causes of death were evolution to AL (15.4%), other cancers (13.8%), disease progression without AL (10.8%), SVT (10.8%), hepatic failure and venous thrombosis other than SVT (7.7% each), heart failure and arterial thrombosis (6.2% each), hemorrhage (5.5%), renal failure and infection (4.6% each). Therapy after SVT included anticoagulation in 77%, antiaggregant in 21.2% and combination in 1.8%; 70% received cytotoxic drugs; 12.4% were treated with transjugular porto-systemic shunt. Beta blocker therapy was used in 48.5% of pts and correlated with improved survival (p=0.041) MPN associated with SVT correlated with younger age and female sex and might antedate the clinical phenotype in a quarter of the patients. MPN-associated SVT equally affected PV and ET, was more likely to occur in the presence of JAK2V617F or underlying thrombophilia and predicted recurrent venous but not arterial thrombosis. The apparent association of SVT with better or worse prognosis in PMF and ET, respectively, requires further investigation. Disclosures No relevant conflicts of interest to declare.

Abstract Philadelphia-negative Myeloproliferative Neoplasms (MPN) include Polycythemia Vera (PV), Essential Thrombocythemia (ET) and Myelofibrosis Primary (PMF) and secondary to PV and ET (PPV-, PET-MF); included are also some less characterized entities defined as unclassified MPN (U-MPN). Risk of arterial and venous thrombosis is increased in MPN patients, and thrombosis is one of most important causes of mortality and morbidity. The risk of venous thrombosis in unusual sites, such as splanchnic vessels (SVT), is particularly associated with MPN; SVT can lead to complications such as portal hypertension, esophageal and gastric varices, ascites and hepatic failure. A recent meta-analysis reported that a MPN is the underlying cause of portal vein thrombosis in 31.5% and of Budd Chiari syndrome in 40.9% of cases (Smalberg, 2012). A significant association of SVT with JAK2V617F mutated MPN was reported (Dentali, 2009) but study of other correlations has been hampered by heterogeneity of available patient cohorts comprising relatively small number of cases. We conducted a retrospective multicenter study collecting clinical and biological data of patients (pts) with SVT associated with MPN diagnosed according to WHO2008 criteria, aiming to describe patients’ characteristics, trends and prognostic factors, and their potential implications for clinical practice. Data were collected from 15 international hematology centers in the framework of IWG-MRT. We collected 475 cases of pts with portal, splenic or mesenteric vein thrombosis (75.2%) or Budd Chiari syndrome (24.8%) associated with MPN. In 32% of cases, simultaneous involvement of portal (69.1% of total thrombosis), splenic (30.5%) and mesenteric (25.3%) veins occurred, and in 1.7% they were associated with Budd Chiari syndrome. Frequency of MPN subtype: 38.1% ET (n=181), 34.9% PV (n=166), 16.2% MF (n=77), 10.8% U-MPN (n=51). Median follow-up 87.9 mo (range 0.5-430); female 61.3% (n=292; P<0.0001 vs male); median age at MPN diagnosis (dg) 44.4 y (range 12-90), at SVT dg 44.9 y (range 17-85). In 229 cases (48%) MPN and SVT dg were coincident, while in 104 (22%) SVT occurred before MPN dg (median 40 mo, range 5-335) and in 129 (27%) during MPN follow up (median 79 mo, range 5-394). JAK2V617F mutational status is available for 361 pts: 99% PV, 84.7% ET, 88.1% PMF and 92.9% U-MPN pts were JAK2V617F positive, with a mean allele burden of 56±27.4%, 33.1±25.5%, 39.3±19.4% and 23.8±11.9%, respectively. Erythropoietin-independent colonies (EEC) were present at diagnosis in 80/110 evaluated cases (72.7%), 38/47 PV (84.4%), 32/45 ET (71.1%), 8/11 PMF (72.7%) and 2/7 U-MPN (28.6%). A concurrent thrombophilic state was found in 38.9% of cases. A 12.3% of pts experienced a recurrence of SVT after a median of 29 mo (range 1-378.3) and 35.8% developed thrombosis in other sites (17.7% arterial, 19.3% venous). Esophageal varices were found in 70.6% from which 31.9% bled. MF transformation occurred in 32/166 PV (19%) and in 23/181 ET (13%) pts, with median time to progression of 122.3 mo (range 5.4-377.3) and 125.1 mo (range 39.3-255.3), respectively. Evolution to acute leukemia (AL) occurred in 12 pts (2.7%), of which 2 PMF, 6 PV and 4 ET. In 3 PV and 1 ET pts a PPV and PET-MF transformation occurred before AL. After SVT, 77% of pts received anticoagulation, 23.5% antiaggregant therapy and 1.5% both; 68.8% received cytotoxic drugs, 11.4% of pts were treated with trans jugular porto-systemic shunt. No differences in survival were noted with these approaches. Beta blocker therapy was used in 48.5% of pts and correlated with improved survival (p=0.041) At last follow up 70/473 pts (14.8%) died; causes of death are evolution to AL (16.4%), other cancers (14.5%), disease progression without AL (12.7%), SVT (10.9%), hepatic failure and venous thrombosis other than SVT (9.1% each), heart failure and arterial thrombosis (7.3% each), hemorrhage (5.5%), renal failure and infection (3.6% each). After 10 y follow up 8/166 PV (5%), 14/181 ET (8%), 14/77 PMF (18%) and 1/51 U-MPN (1.96%) pts died (p<0.01). Survival was significantly affected by occurrence of thrombosis other than SVT (p<0.0001) but not recurrence in splanchnic vessels (p=0.068). This large study confirms the strong association between JAK2V617F-mutated MPN and SVT and identifies the category of U-MPN as the prognostically more favorable; thrombosis at sites outside the splanchnic vasculature remains the leading cause of death. Disclosures: No relevant conflicts of interest to declare.