Faculty Bibliography

BACKGROUND:Clinical prediction models targeting patients for Barrett's esophagus (BE) screening include data obtained by interview, questionnaire, and body measurements. A tool based on electronic health records (EHR) data could reduce cost and enhance usability, particularly if combined with non-endoscopic BE screening methods. AIMS/OBJECTIVE:To determine whether EHR-based data can identify BE patients. METHODS:We performed a retrospective review of patients ages 50-75 who underwent a first-time esophagogastroduodenoscopy. Data extracted from the EHR included demographics and BE risk factors. Endoscopy and pathology reports were reviewed for histologically confirmed BE. Screening criteria modified from clinical guidelines were assessed for association with BE. Subsequently, a score based on multivariate logistic regression was developed and assessed for its ability to identify BE subjects. RESULTS:A total of 2931 patients were assessed, and BE was found in 1.9%. Subjects who met screening criteria were more likely to have BE (3.3% vs. 1.1%, p = 0.001), and the criteria predicted BE with an AUROC of 0.65 (95% CI 0.59-0.71). A score based on logistic regression modeling included gastroesophageal reflux disease, sex, body mass index, and ever-smoker status and identified BE subjects with an AUROC of 0.71 (95% CI 0.64-0.77). Both prediction tools produced higher AUROCs in women than in men. CONCLUSIONS:EHR-based BE risk prediction tools identify BE patients with fair accuracy. While these tools may improve the efficiency of patient targeting for BE screening in the primary care setting, challenges remain to identify high-risk patients for non-invasive BE screening in clinical practice.

INTRODUCTION: Gastrointestinal bleeding (GIB) in the setting of anti-coagulation is associated with considerable morbidity and mortality. Prediction models, such as the HAS-BLED score, are recommended by guidelines and used by clinicians to assess the risk of major bleeding among atrial fibrillation patients treated with warfarin. Whether gastric colonization by H. pylori (HP) confers bleeding risk that is unaccounted for by the HAS-BLED score is unknown. We hypothesized that HAS-BLED scores (ranging from 0-9, with higher scores indicating higher bleeding risk) would be lower among patients with upper GIB who have gastric colonization by HP than in those without HP.
METHOD(S): We examined all patients at a single medical center who had an upper endoscopy for suspected GIB between 2011-2018 with findings of a gastric or duodenal ulcer. Only patients who were tested for H. pylori by any modality and had taken warfarin within one week prior to endoscopy were included in the analysis. We calculated the HAS-BLED score for all patients, and compared the HAS-BLED scores of HP-positive and HP-negative patients using the Wilcoxon rank-sum test. In a secondary analysis, we classified patients as having a high risk for bleeding (score >=3) or low risk for bleeding (score <3), and used Fisher's exact test to compare the prevalence of a high risk score between the HP positive and negative groups.
RESULT(S): Of the 1,578 bleeding events reviewed, 62 patients were determined to be taking warfarin within one week of the GIB and had HP status checked. HP testing was positive in 12/62 (19%) patients and negative in 50/62 (81%) patients. Multiple individual components of the HASBLED score differed between HP positive and negative patients (Table 1). The median scores for HP positive and negative patients were 2 (IQR, 2-2.25) and 3 (IQR, 2-4), respectively (P < >01). In our secondary analysis, 3/12 (25%) of HP positive patients were classified as high risk for bleeding by the HAS-BLED score whereas 32/50 (64%) were classified as such among HP negative patients (P = 0.02).
CONCLUSION(S): Among patients with warfarin-associated upper GIB, those who were HP positive had significantly lower HAS-BLED scores as compared to those who were HP negative. The HASBLED prediction model may not account for GIB risk associated with HP colonization. Further studies should assess whether prediction models should include HP status, and whether screening for HP when starting anticoagulation is warranted. (Figure Presented)

BACKGROUND:Clostridioides difficile infection (CDI) is common after solid organ transplant (SOT) and is associated with high morbidity and mortality. METHODS:We assessed incidence, risk factors, and outcomes of CDI among SOT patients at a large multi-organ transplant center. Multivariable logistic regression was used to identify risk factors for initial and recurrent CDI. RESULTS:A total of 2622 SOT patients were included. 224 (8.5%) had CDI 1 year post-SOT. The highest incidence of CDI was among pancreas recipients (12.5%) followed by lung (11.7%), liver (11.0%), heart (10.8%), and kidney (5.8%). Median time to CDI was 56 days (range 2-354) post-SOT. About 64% of patients had severe CDI. About 56.3% were treated with metronidazole, 13.8% with oral vancomycin, and 28.6% with both. About 28.6% of patients had recurrent CDI. In multivariable modeling, lung transplant recipient status was the only significant predictor of recurrent CDI (OR 4.97, 95% CI 2.11-11.78, P < .001) controlling for age, severe CDI, and pre-SOT CDI. Post-SOT CDI nearly doubled the risk of mortality at one year, in particular among those with severe CDI. CONCLUSIONS:In summary, CDI is highly prevalent, occurs early in the post-transplant period, usually severe, with a high rate of recurrence, and associated with increased mortality within 1 year after transplant. The early post-transplant period may be a crucial window to reduce CDI rates.

BACKGROUND:Cystic fibrosis (CF) patients have increased risks of gastrointestinal cancers, including esophageal adenocarcinoma. Gastroesophageal reflux disease (GERD) is highly prevalent in CF and manifests at early ages. CF patients may be at increased risk for long-term sequelae of chronic GERD, including Barrett's esophagus (BE). We aimed to assess whether patients with CF have an increased risk of BE or related neoplasia. METHODS:A matched cohort study was performed of adults with and without CF who had undergone upper endoscopy. Non-CF patients were matched in a 4:1 ratio by age, sex, year of exam, and endoscopist. Odds ratios were calculated for the association between CF and BE or related neoplasia, and multivariable logistic regression modeling was performed to adjust for matching variables and additional potential confounders. RESULTS:122 CF patients underwent endoscopy, and 488 matched controls were identified. Seven (5.7%) CF patients had BE or related neoplasia, including one GE junction adenocarcinoma. Mean age of affected CF patients was 36.0, and 85.7% had a prior solid organ transplant. The odds of BE was significantly increased in CF patients (OR 2.91, 95% CI 1.08-7.81). The risk remained significantly increased in a multivariable model including matching variables (OR 3.32, 95% CI 1.19-9.22) and in a parsimonious model (OR 2.99, 95% CI 1.06-8.42). CONCLUSIONS:Adults with CF have a 3-fold increased risk of BE or related neoplasia and appears to develop at younger ages. Consideration should be given to screening for BE in select CF patients, especially those who have undergone solid organ transplantation.

OBJECTIVE:To evaluate the effect of pretransplant bridging locoregional therapy (LRT) on hepatocellular carcinoma (HCC) recurrence and survival after liver transplantation (LT) in patients meeting Milan criteria (MC). SUMMARY BACKGROUND DATA:Pre-LT LRT mitigates tumor progression and waitlist dropout in HCC patients within MC, but data on its impact on post-LT recurrence and survival remain limited. METHODS:Recurrence-free survival and post-LT recurrence were compared among 3601 MC patients with and without bridging LRT utilizing competing risk Cox regression in consecutive patients from 20 US centers (2002-2013). RESULTS:Compared with 747 LT recipients not receiving LRT, 2854 receiving LRT had similar 1, 3, and 5-year recurrence-free survival (89%, 77%, 68% vs 85%, 75%, 68%; P = 0.490) and 5-year post-LT recurrence (11.2% vs 10.1%; P = 0.474). Increasing LRT number [3 LRTs: hazard ratio (HR) 2.1, P < 0.001; 4+ LRTs: HR 2.5, P < 0.001), and unfavorable waitlist alphafetoprotein trend significantly predicted post-LT recurrence, whereas LRT modality did not. Treated patients achieving complete pathologic response (cPR) had superior 5-year RFS (72%) and lower post-LT recurrence (HR 0.52, P < 0.001) compared with both untreated patients (69%; P = 0.010; HR 1.0) and treated patients not achieving cPR (67%; P = 0.010; HR 1.31, P = 0.039), who demonstrated increased recurrence compared with untreated patients in multivariate analysis controlling for pretransplant and pathologic factors (HR 1.32, P = 0.044). CONCLUSIONS:Bridging LRT in HCC patients within MC does not improve post-LT survival or HCC recurrence in the majority of patients who fail to achieve cPR. The need for increasing LRT treatments and lack of alphafetoprotein response to LRT independently predict post-LT recurrence, serving as a surrogate for underlying tumor biology which can be utilized for prioritization of HCC LT candidates.

OBJECTIVE:We sought to develop a "Model Of Recurrence After Liver transplant" (MORAL) for hepatocellular carcinoma (HCC). BACKGROUND:The Milan criteria are used to allocate livers to patients with HCC requiring liver transplantation (LT) but do not include objective measures of tumor biology. Biological markers including the neutrophil-lymphocyte ratio (NLR) and alpha-fetoprotein (AFP) have been associated with recurrence risk. METHODS:Prospective cohort study of adults undergoing LT for HCC between January 2001 and December 2012. RESULTS:A total of 339 patients were included. On multivariable Cox regression analysis, 3 preoperatively available factors were independent predictors of worse recurrence-free survival (RFS), namely, an NLR ≥ 5 (P < 0.0001, hazard ratio, HR: 6.2), AFP > 200 (P < 0.0001, HR: 3.8), and Size >3 cm (P < 0.001, HR: 3.2). The Pre-MORAL score was constructed from the hazard ratios and assigning patients points in an additive fashion, with a minimum of 0 points (no factors) and a maximum of 13 points (all 3 factors). The highest risk patients in the Pre-MORAL had a 5-year RFS of 17.9% compared with 98.6% for the low risk group (P < 0.0001). The post-MORAL was constructed similarly using the 4 postoperatively available independent predictors of worse RFS, grade 4 HCC's (P < 0.0001, HR: 5.6), vascular invasion (P = 0.019, HR: 2.0), size >3 cm (P < 0.0001, HR: 3.2) and number >3 (P = 0.048, HR: 1.8). The pre- and post-MORAL were superior to Milan at predicting recurrence with c-statistics of 0.82 and 0.87, compared with 0.63, respectively. We then combined the scores to produce a combo-MORAL, with a c-statistic of 0.91 for predicting recurrence. CONCLUSIONS:The MORAL score provides a simple, highly accurate tool for predicting recurrence and risk-stratification pre- and postoperatively.

OPINION STATEMENT/UNASSIGNED:Performing endoscopic procedures in the elderly carries known enhanced risk compared to the general population. Weighing the benefits against the risks is easy when a patient is in immediate danger, but a gray area arises in screening protocols in an elderly patient of average risk. In this review, we compare national and international guidelines in average risk screening procedures (colonoscopic colorectal screening, Barrett's surveillance) to find consensus for screening practice in the elderly. With minor differences between societal guidelines, it is widely agreed that 75 years is the appropriate age to begin to weigh risks and benefits according to a patient's state of health and comorbidities. For colorectal screening, most guidelines advocate complete cessation of screening after the age of 85 years. Such consensus must take into account an aging population where patients are living healthier for longer and thus may be appropriate candidates for screening procedures even if beyond designated ages of screening cessation.