Development and validation of deep learning classifiers to detect Epstein-Barr virus and microsatellite instability status in gastric cancer: a retrospective multicentre cohort study.

BACKGROUND: Response to immunotherapy in gastric cancer is associated with microsatellite instability (or mismatch repair deficiency) and Epstein-Barr virus (EBV) positivity. We therefore aimed to develop and validate deep learning-based classifiers to detect microsatellite instability and EBV status from routine histology slides. METHODS: In this retrospective, multicentre study, we collected tissue samples from ten cohorts of patients with gastric cancer from seven countries (South Korea, Switzerland, Japan, Italy, Germany, the UK and the USA). We trained a deep learning-based classifier to detect microsatellite instability and EBV positivity from digitised, haematoxylin and eosin stained resection slides without annotating tumour containing regions. The performance of the classifier was assessed by within-cohort cross-validation in all ten cohorts and by external validation, for which we split the cohorts into a five-cohort training dataset and a five-cohort test dataset. We measured the area under the receiver operating curve (AUROC) for detection of microsatellite instability and EBV status. Microsatellite instability and EBV status were determined to be detectable if the lower bound of the 95% CI for the AUROC was above 0·5. FINDINGS: Across the ten cohorts, our analysis included 2823 patients with known microsatellite instability status and 2685 patients with known EBV status. In the within-cohort cross-validation, the deep learning-based classifier could detect microsatellite instability status in nine of ten cohorts, with AUROCs ranging from 0·597 (95% CI 0·522-0·737) to 0·836 (0·795-0·880) and EBV status in five of eight cohorts, with AUROCs ranging from 0·819 (0·752-0·841) to 0·897 (0·513-0·966). Training a classifier on the pooled training dataset and testing it on the five remaining cohorts resulted in high classification performance with AUROCs ranging from 0·723 (95% CI 0·676-0·794) to 0·863 (0·747-0·969) for detection of microsatellite instability and from 0·672 (0·403-0·989) to 0·859 (0·823-0·919) for detection of EBV status. INTERPRETATION: Classifiers became increasingly robust when trained on pooled cohorts. After prospective validation, this deep learning-based tissue classification system could be used as an inexpensive predictive biomarker for immunotherapy in gastric cancer. FUNDING: German Cancer Aid and German Federal Ministry of Health.

Fatigue and its associated factors in microscopic colitis.

BACKGROUND: Fatigue is a well-recognized symptom in patients with inflammatory bowel disease and irritable bowel syndrome (IBS), and has been associated with psychological comorbidity and impaired quality of life in both. However, features associated with fatigue in patients with microscopic colitis (MC) are less clear. MATERIALS AND METHODS: We conducted a cross-sectional survey of patients with a new diagnosis of MC including levels of anxiety, depression, somatization, quality of life, and IBS-type symptoms. Levels and impact of fatigue were assessed using the Inflammatory Bowel Disease Fatigue self-assessment scale. Mean scores were compared against various patient characteristics, and were also correlated with anxiety, depression, somatization, and quality-of-life scores. RESULTS: In total, 129 patients with MC diagnosed between 2010 and 2015 returned completed postal questionnaires. Common histological subtypes were collagenous colitis (53.5%, n = 69) and lymphocytic colitis (38.8%, n = 50). Higher mean fatigue severity and impact scores were associated with the presence of irritable-bowel-syndrome-type symptoms, abnormal levels of anxiety and depression, and high levels of somatization (p < 0.0001 for all), but those reporting ongoing symptoms attributable to MC did not report significantly higher scores. There were significant positive correlations between total anxiety, depression, or somatization scores and fatigue severity and impact scores, and significant negative correlations with quality-of-life measures (p < 0.001 for all). CONCLUSIONS: Fatigue in MC appears to be associated with reporting IBS-type symptoms, psychological comorbidity and impaired quality of life. It may therefore represent an important target for treatment.

High prevalence of irritable bowel syndrome-type symptoms in microscopic colitis: implications for treatment.

BACKGROUND: Patients with microscopic colitis (MC) often present with abdominal pain and diarrhoea, and previous data suggest that there may be overlap between MC and irritable bowel syndrome (IBS). We evaluated the prevalence of IBS-type symptoms in patients with MC, and assess the impact of these symptoms on psychological health and quality of life. METHODS: We conducted a cross-sectional survey of individuals with a histological diagnosis of MC, collecting demographic data, Rome III IBS-type symptoms, and mood, somatization, and quality of life data. RESULTS: In total, 151 (31.6%) of 478 individuals with a new diagnosis of MC completed questionnaires, 52 (34.4%) of whom reported IBS-type symptoms. The commonest histological subtype was collagenous colitis (51.7%, n = 78), followed by lymphocytic colitis (39.1%, n = 59). Individuals with IBS-type symptoms had significantly higher levels of anxiety [Hospital Anxiety and Depression Scale (HADS) anxiety score 8.6 versus 5.1, p < 0.001], depression (HADS depression score 6.2 versus 3.6, p = 0.001), and somatoform-type behaviour (Patient Health Questionnaire 15 score 12.7 versus 8.0, p < 0.001) compared with individuals who did not. Those with IBS-type symptoms scored significantly worse across all domains of the 36-item Short Form questionnaire, except for physical functioning. CONCLUSIONS: More than one third of individuals with MC reported IBS-type symptoms, although whether this is due to ongoing inflammation is unclear. These individuals had higher levels of anxiety, depression, and somatization, and impaired quality of life. Identifying concomitant IBS in individuals with MC may have important implications for management decisions.

Effect of psychological therapy on disease activity, psychological comorbidity, and quality of life in inflammatory bowel disease: a systematic review and meta-analysis.

BACKGROUND: Inflammatory bowel disease is associated with psychological comorbidity and impaired quality of life. Psychological comorbidity could affect the natural history of inflammatory bowel disease. Psychological therapies might therefore have beneficial effects on disease activity, mood, and quality of life in patients with inflammatory bowel disease. We did a systematic review and meta-analysis examining these issues. METHODS: In this systematic review and meta-analysis, we searched MEDLINE, Embase, Embase Classic, PsychINFO, and the Cochrane Central Register of Controlled Trials for articles published between 1947 and Sept 22, 2016. Randomised controlled trials (RCTs) recruiting patients with inflammatory bowel disease aged at least 16 years that compared psychological therapy with a control intervention or usual treatment were eligible. We pooled dichotomous data to obtain relative risks of induction of remission in active disease or prevention of relapse of quiescent disease, with 95% CIs. We pooled continuous data to estimate standardised mean differences in disease activity indices, anxiety, depression, perceived stress, and quality-of-life scores in patients dichotomised into those with clinically active or quiescent disease, with 95% CIs. We extracted data from published reports and contacted the original investigators of studies for which the required data were not available. We pooled all data using a random-effects model. FINDINGS: The search identified 1824 studies, with 14 RCTs of 1196 patients eligible for inclusion. The relative risk of relapse of quiescent inflammatory bowel disease with psychological therapy versus control was 0·98 (95% CI 0·77-1·24; p=0·87; I2=50%; six trials; 518 patients). We observed a significant difference in depression scores (standardised mean difference -0·17 [-0·33 to -0·01]; p=0·04; I2=0%; seven trials; 605 patients) and quality of life (0·30 [0·07-0·52]; p=0·01; I2=42%; nine trials; 578 patients) with psychological therapy versus control at the end of therapy for patients with quiescent disease. However, these beneficial effects were lost at final point of follow-up (depression scores -0·11 [-0·27 to 0·05], p=0·17, I2=0%, eight trials, 593 patients; quality of life 0·15 [-0·05 to 0·34], p=0·14, I2=22%, ten trials, 577 patients). When we assessed the effect of individual physiological therapies on quality of life, only cognitive behavioural therapy had any significant beneficial effect (0·37 [0·02-0·72]). We noted no effect on disease activity indices or other psychological wellbeing scores when compared with control in patients with quiescent disease. Dichotomous data for induction of remission and continuous data for change in clinical disease activity indices, depression, anxiety, and perceived stress scores were only reported in one RCT of patients with active disease. Quality of life was assessed in two RCTs of patients with active disease, but was not significantly different between intervention and control groups (0·27 [-0·05 to 0·59]). INTERPRETATION: Psychological therapies, and cognitive behavioural therapy in particular, might have small short-term beneficial effects on depression scores and quality of life in patients with inflammatory bowel disease. Further RCTs of these interventions in patients with coexistent psychological distress are required. FUNDING: None.

Minimal differences in prevalence and spectrum of organic disease at upper gastrointestinal endoscopy between selected secondary care patients with symptoms of gastro-oesophageal reflux or dyspepsia.

OBJECTIVES: Gastro-oesophageal reflux and dyspepsia are felt to be separate upper gastrointestinal (GI) conditions. We aimed to measure the degree of overlap between them, and assess whether endoscopic findings differed. MATERIAL AND METHODS: Demographic, symptom, upper GI endoscopy and histology data were collected from consecutive adults in secondary care. Patients were categorised according to whether they reported gastro-oesophageal reflux alone, dyspepsia alone or both, and patient demographics and endoscopic findings were compared. RESULTS: Of 1167 patients, 97 (8.3%) had gastro-oesophageal reflux alone, 571 (48.9%) dyspepsia alone, and 499 (42.8%) overlap. Patients with overlap symptoms were more likely to smoke, compared with those with gastro-oesophageal reflux alone, or dyspepsia alone (p = .009), but there were no other differences. Patients with gastro-oesophageal reflux alone or overlap had a higher prevalence of erosive oesophagitis (18.6% and 15.4% respectively, p < .001), but this was still the commonest diagnosis among those with dyspepsia alone (7.2%). No significant differences were seen in prevalence of other endoscopic findings. CONCLUSIONS: Gastro-oesophageal reflux and dyspepsia symptoms commonly overlap. There were minimal differences in demographics or spectrum of underlying organic disease between various symptom groups, suggesting that restrictive classifications according to predominant symptom may not be clinically useful.

Screening for Celiac Disease in Irritable Bowel Syndrome: An Updated Systematic Review and Meta-analysis.

OBJECTIVES: Celiac disease (CD) and irritable bowel syndrome (IBS) share similar symptoms, leading to confusion between the two and diagnostic delay. International guidelines recommend screening individuals with IBS for CD, via serological testing. However, studies published recently have cast doubt on the utility of this. We updated a previous meta-analysis examining this issue. METHODS: MEDLINE, EMBASE, and EMBASE Classic were searched through to May 2016. Eligible studies recruited adults with IBS according to symptom-based criteria, physician's opinion, or questionnaire data. Tests for CD included IgA-class antigliadin antibodies (AGA), endomysial antibodies (EMA), tissue transglutaminase antibodies (tTG), or duodenal biopsies following positive serology. The proportion of individuals meeting criteria for IBS testing positive for CD was combined to give a pooled prevalence for all studies, and compared between cases with IBS and, healthy controls without (where reported), using an odds ratio (OR) with a 95% confidence interval (CI). RESULTS: There were 36 eligible studies, recruiting 15,256 individuals, of whom 9,275 (60.8%) met criteria for IBS. Pooled ORs for positive IgA AGAs, EMA and/or tTG, and biopsy-proven CD in IBS subjects vs. controls were 3.21 (95% CI 1.55-6.65), 2.75 (95% CI 1.35-5.61), and 4.48 (95% CI 2.33-8.60), respectively. There was no increase in ORs for any test for CD among cases with IBS in North American studies, and results were inconsistent in population-based studies. The prevalence of biopsy-proven CD was significantly higher across all subtypes of IBS. Limitations included heterogeneity in some analyses, and few North American studies. CONCLUSIONS: Overall, prevalence of positive celiac serology and biopsy-proven CD was significantly higher in subjects with symptoms suggestive of IBS vs. healthy controls. However, the utility of screening for CD in individuals with suspected IBS in North America or in the community is less clear.

How does tolerability of double balloon enteroscopy compare to other forms of endoscopy?

BACKGROUND AND AIMS: Gastrointestinal endoscopy can be difficult for patients to tolerate. Studies on endoscopic tolerability mainly focus on gastroscopy or colonoscopy with a paucity of data on double balloon enteroscopy (DBE). We aimed to prospectively evaluate tolerability in patients undergoing several forms of endoscopy including DBE. METHODS: Consecutive patients undergoing colonoscopy, flexible sigmoidoscopy, gastroscopy, endoscopic retrograde pancreatography (ERCP), capsule endoscopy (CE) and DBE were prospectively recruited. A questionnaire recorded demographics, procedural data, patient tolerability (pain, discomfort and distress recorded on numerical rating scales) and the Hospital Anxiety and Depression Scale (HADS). RESULTS: 956 patients were recruited (512 women; median age 57 years). The median pain score for DBE was poor with a score of 5 compared with 1 and 0 for oesophagogastroduodenoscopy and ERCP, respectively (p<0.001). Colonoscopy and retrograde DBE scores were not dissimilar. CE was well tolerated with a median pain score of 0. Patients with DBE required significantly higher doses of sedation and analgesia than other patients. The HADS Anxiety Score was also associated with poorer tolerability. CONCLUSIONS: DBE is poorly tolerated when compared with other forms of endoscopy despite higher doses of sedation. Increasing demand to improve tolerability of DBE in the UK may be addressed with the use of propofol.

Point-of-care testing for celiac disease has a low sensitivity in endoscopy.

BACKGROUND: Celiac disease (CD) is a common but underdiagnosed condition. A rapid point-of-care test (POCT) could reduce lead times and missed diagnoses. OBJECTIVE: To assess the utility of an immunoglobulin (Ig) A tissue transglutaminase (TTG) antibody POCT in an endoscopic setting. DESIGN: Prospective observational study. SETTING: A single UK university hospital. PATIENTS: Patients presenting with suspected CD, known CD, and routine endoscopy for upper GI symptoms. INTERVENTIONS: All patients were tested with POCT, serum TTG, endomysial antibody (EMA), and upper GI endoscopy with duodenal biopsies at the same visit. MAIN OUTCOME MEASUREMENTS: Comparison was made with histology in all cases, with villous atrophy regarded as diagnostic of CD. RESULTS: A total of 576 patients (63.5% female, mean [± standard deviation] age 49.7 years [± 17.6 years]) were recruited. A total of 523 patients had no prior diagnosis of CD, and 53 patients had known CD coming for reassessment. A total of 117 patients were newly diagnosed with CD, and 82 were positively identified by the POCT. Sensitivity, specificity, positive predictive value, and negative predictive value were 70.1%, 96.6%, 85.4%, and 91.8%, respectively. In comparison, TTG and EMA both performed significantly better than the POCT. Sensitivity and specificity of TTG were 91.0% and 83.5%, respectively, and EMA were 83.8% and 97.5%, respectively. Of patients with known CD coming for reassessment, 26 had villous atrophy, and POCT results were positive in 16 (61.5%). There was poor agreement between POCT and standard serology. LIMITATIONS: High pre-test probability of CD. CONCLUSION: The performance of this POCT was disappointing compared with standard serology and cannot at present be recommended within the context of an endoscopy unit.

Rapid turnover of T cells in acute infectious mononucleosis.

During acute infectious mononucleosis (AIM), large clones of Epstein-Barr virus-specific T lymphocytes are produced. To investigate the dynamics of clonal expansion, we measured cell proliferation during AIM using deuterated glucose to label DNA of dividing cells in vivo, analyzing cells according to CD4, CD8 and CD45 phenotype. The proportion of labeled CD8(+)CD45R0(+) T lymphocytes was dramatically increased in AIM subjects compared to controls (mean 17.5 versus 2.8%/day; p<0.005), indicating very rapid proliferation. Labeling was also increased in CD4(+)CD45R0(+) cells (7.1 versus 2.1%/day; p<0.01), but less so in CD45RA(+) cells. Mathematical modeling, accounting for death of labeled cells and changing pool sizes, gave estimated proliferation rates in CD8(+)CD45R0(+) cells of 11-130% of cells proliferating per day (mean 47%/day), equivalent to a doubling time of 1.5 days and an appearance rate in blood of about 5 x 10(9) cells/day (versus 7 x 10(7) cells/day in controls). Very rapid death rates were also observed amongst labeled cells (range 28-124, mean 57%/day),indicating very short survival times in the circulation. Thus, we have shown direct evidence for massive proliferation of CD8(+)CD45R0(+) T lymphocytes in AIM and demonstrated that rapid cell division continues concurrently with greatly accelerated rates of cell disappearance.

Measurement and modeling of human T cell kinetics.

The ability to measure, describe and interpret T cell kinetics is pivotal in understanding normal lymphocyte homeostasis and diseases that affect T cell numbers. Following in vivo labeling of dividing cells with 6,6-D(2)-glucose in eight healthy volunteers, peripheral blood T cells were sorted by CD4, CD8 and CD45 phenotype. Enrichment of deuterium in DNA was measured by gas chromatography-mass spectrometry. A novel model of T cell kinetics, allowing for heterogeneity within T cell pools, was used to analyze data on acquisition and loss of label and calculate proliferation and disappearance rates for each subpopulation. Proliferation rates for CD45RO(+)CD8(+) cells and CD45RO(+)CD4(+) cells were 5.1% and 2.7% /day, respectively (equivalent doubling times: 14 and 26 days). CD45RA(+)CD8(+) lymphocytes and CD45RA(+)CD4(+) lymphocytes had slower proliferation rates, 0.5% and 0.6% / day, respectively (doubling time about 4 months). Disappearance rates of labeled cells were similar for all cell types (7%-12% / day) and exceeded corresponding proliferation rates. This disparity may be understood conceptually in terms of either phenotypic heterogeneity (rapid versus slow turnover pools), or history (recently divided cells are more likely to die). The new kinetic model fits the data closely and avoids the need to postulate a large external source of lymphocytes to maintain equilibrium.