Computer-aided classification of colorectal segments during colonoscopy: a deep learning approach based on images of a magnetic endoscopic positioning device

OBJECTIVE

Assessment of the anatomical colorectal segment of polyps during colonoscopy is important for treatment and follow-up strategies, but is largely operator dependent. This feasibility study aimed to assess whether, using images of a magnetic endoscope imaging (MEI) positioning device, a deep learning approach can be useful to objectively divide the colorectum into anatomical segments.

METHODS

Models based on the VGG-16 based convolutional neural network architecture were developed to classify the colorectum into anatomical segments. These models were pre-trained on ImageNet data and further trained using prospectively collected data of the POLAR study in which endoscopists were using MEI (3930 still images and 90,151 video frames). Five-fold cross validation with multiple runs was used to evaluate the overall diagnostic accuracies of the models for colorectal segment classification (divided into a 5-class and 2-class colorectal segment division). The colorectal segment assignment by endoscopists was used as the reference standard.

RESULTS

For the 5-class colorectal segment division, the best performing model correctly classified the colorectal segment in 753 of the 1196 polyps, corresponding to an overall accuracy of 63%, sensitivity of 63%, specificity of 89% and kappa of 0.47. For the 2-class colorectal segment division, 1112 of the 1196 polyps were correctly classified, corresponding to an accuracy of 93%, sensitivity of 93%, specificity of 90% and kappa of 0.82.

CONCLUSION

The diagnostic performance of a deep learning approach for colorectal segment classification based on images of a MEI device is yet suboptimal (clinicaltrials.gov: NCT03822390).

Endoscopic resection of high-risk T1 colorectal carcinoma prior to surgical resection has no adverse effect on long-term outcomes

OBJECTIVE

It is difficult to predict the presence of histological risk factors for lymph node metastasis (LNM) before endoscopic treatment of T1 colorectal cancer (CRC). Therefore, endoscopic therapy is propagated to obtain adequate histological staging. We examined whether secondary surgery following endoscopic resection of high-risk T1 CRC does not have a negative effect on patients' outcomes compared with primary surgery.

DESIGN

Patients with T1 CRC with one or more histological risk factors for LNM (high risk) and treated with primary or secondary surgery between 2000 and 2014 in 13 hospitals were identified in the Netherlands Cancer Registry. Additional data were collected from hospital records, endoscopy, radiology and pathology reports. A propensity score analysis was performed using inverse probability weighting (IPW) to correct for confounding by indication.

RESULTS

602 patients were eligible for analysis (263 primary; 339 secondary surgery). Overall, 34 recurrences were observed (5.6%). After adjusting with IPW, no differences were observed between primary and secondary surgery for the presence of LNM (OR 0.97; 95% CI 0.49 to 1.93; p=0.940) and recurrence during follow-up (HR 0.97; 95% CI 0.41 to 2.34; p=0.954). Further adjusting for lymphovascular invasion, depth of invasion and number of retrieved lymph nodes did not alter this outcome.

CONCLUSIONS

Our data do not support an increased risk of LNM or recurrence after secondary surgery compared with primary surgery. Therefore, an attempt for an en-bloc resection of a possible T1 CRC without evident signs of deep invasion seems justified in order to prevent surgery of low-risk T1 CRC in a significant proportion of patients.

Risk for Incomplete Resection after Macroscopic Radical Endoscopic Resection of T1 Colorectal Cancer: A Multicenter Cohort Study

OBJECTIVES

The decision to perform secondary surgery after endoscopic resection of T1 colorectal cancer (CRC) depends on the risk of lymph node metastasis and the risk of incomplete resection. We aimed to examine the incidence and risk factors for incomplete endoscopic resection of T1 CRC after a macroscopic radical endoscopic resection.

METHODS

Data from patients treated between 2000 and 2014 with macroscopic complete endoscopic resection of T1 CRC were collected from 13 hospitals. Incomplete resection was defined as local recurrence at the polypectomy site during follow-up or malignant tissue in the surgically resected specimen in case secondary surgery was performed. Multivariate regression analysis was performed to analyze factors associated with incomplete resection.

RESULTS

In total, 877 patients with a median follow-up time of 36.5 months (interquartile range 16.0-68.3) were included, in whom secondary surgery was performed in 358 patients (40.8%). Incomplete resection was observed in 30 patients (3.4%; 95% confidence interval (CI) 2.3-4.6%). Incomplete resection rate was 0.7% (95% CI 0-2.1%) in low-risk T1 CRC vs. 4.4% (95% CI 2.7-6.5%) in high-risk T1 CRC (P=0.04). Overall adverse outcome rate (incomplete resection or metastasis) was 2.1% (95% CI 0-5.0%) in low-risk T1 CRC vs. 11.7% (95% CI 8.8-14.6%) in high-risk T1 CRC (P=0.001). Piecemeal resection (adjusted odds ratio 2.60; 95% CI 1.20-5.61, P=0.02) and non-pedunculated morphology (adjusted odds ratio 2.18; 95% CI 1.01-4.70, P=0.05) were independent risk factors for incomplete resection. Among patients in whom no additional surgery was performed, who developed recurrent cancer, 41.7% (95% CI 20.8-62.5%) died as a result of recurrent cancer.

CONCLUSIONS

In the absence of histological high-risk factors, a 'wait-and-see' policy with limited follow-up is justified. Piecemeal resection and non-pedunculated morphology are independent risk factors for incomplete endoscopic resection of T1 CRC.

Clinical risk factors of colorectal cancer in patients with serrated polyposis syndrome: a multicentre cohort analysis

OBJECTIVE

Serrated polyposis syndrome (SPS) is accompanied by an increased risk of colorectal cancer (CRC). Patients fulfilling the clinical criteria, as defined by the WHO, have a wide variation in CRC risk. We aimed to assess risk factors for CRC in a large cohort of patients with SPS and to evaluate the risk of CRC during surveillance.

DESIGN

In this retrospective cohort analysis, all patients with SPS from seven centres in the Netherlands and two in the UK were enrolled. WHO criteria were used to diagnose SPS. Patients who only fulfilled WHO criterion-2, with IBD and/or a known hereditary CRC syndrome were excluded.

RESULTS

In total, 434 patients with SPS were included for analysis; 127 (29.3%) were diagnosed with CRC. In a per-patient analysis ≥1 serrated polyp (SP) with dysplasia (OR 2.07; 95% CI 1.28 to 3.33), ≥1 advanced adenoma (OR 2.30; 95% CI 1.47 to 3.67) and the fulfilment of both WHO criteria 1 and 3 (OR 1.60; 95% CI 1.04 to 2.51) were associated with CRC, while a history of smoking was inversely associated with CRC (OR 0.36; 95% CI 0.23 to 0.56). Overall, 260 patients underwent surveillance after clearing of all relevant lesions, during which two patients were diagnosed with CRC, corresponding to 1.9 events/1000 person-years surveillance (95% CI 0.3 to 6.4).

CONCLUSION

The presence of SPs containing dysplasia, advanced adenomas and/or combined WHO criteria 1 and 3 phenotype is associated with CRC in patients with SPS. Patients with a history of smoking show a lower risk of CRC, possibly due to a different pathogenesis of disease. The risk of developing CRC during surveillance is lower than previously reported in literature, which may reflect a more mature multicentre cohort with less selection bias.

Prediction scores or gastroenterologists' Gut Feeling for triaging patients that present with acute upper gastrointestinal bleeding

INTRODUCTION

Several prediction scores for triaging patients with upper gastrointestinal (GI) bleeding have been developed, yet these scores have never been compared to the current gold standard, which is the clinical evaluation by a gastroenterologist. The aim of this study was to assess the added value of prediction scores to gastroenterologists' Gut Feeling in patients with a suspected upper GI bleeding.

METHODS

WE PROSPECTIVELY EVALUATED GUT FEELING OF SENIOR GASTROENTEROLOGISTS AND ASKED THEM TO ESTIMATE: (1) the risk that a clinical intervention is needed; (2) the risk of rebleeding; and (3) the risk of mortality in patients presenting with suspected upper GI bleeding, subdivided into low, medium, or high risk. The predictive value of the gastroenterologists' Gut Feeling was compared to the Blatchford and Rockall scores for various outcomes.

RESULTS

We included 974 patients, of which 667 patients (68.8%) underwent a clinical intervention. During the 30-day follow up, 140 patients (14.4%) developed recurrent bleeding and 44 patients (4.5%) died. Gut Feeling was independently associated with all studied outcomes, except for the predicted mortality after endoscopy. Predictive power, based on the AUC of the Blatchford and Rockall prediction scores, was higher than the Gut Feeling of the gastroenterologists. However, combining both the Blatchford and Rockall scores and the Gut Feeling yielded the highest predictive power for the need of an intervention (AUC 0.88), rebleeding (AUC 0.73), and mortality (AUC 0.71 predicted before and 0.77 predicted after endoscopy, respectively).

CONCLUSIONS

Gut Feeling is an independent predictor for the need of a clinical intervention, rebleeding, and mortality in patients presenting with upper GI bleeding; however, the Blatchford and Rockall scores are stronger predictors for these outcomes. Combining Gut Feeling with the Blatchford and Rockall scores resulted in the most optimal prediction.

Candidate genotypes associated with functional dyspepsia

There is accumulating evidence of a genetic predisposition for developing a functional gastrointestinal (GI) disorder. Identification of the genetic factors may improve understanding of underlying pathophysiological mechanisms. We aimed to test the association of functional polymorphisms in genes involved in serotonergic signalling and G-protein-mediated signal transduction, both affecting gastroduodenal sensory and motor function, with functional dyspepsia (FD). FD patients, send to our tertiary referral centre, were studied (n = 112). Healthy controls (n = 336) free of GI symptoms were matched 1 : 3 for age and gender. Polymorphisms in genes encoding the serotonin receptor type three A subunit (HTR3A), the serotonin transporter (SERT) and the G-protein beta3 subunit (GNB3) were analysed. The FD patients displayed a higher prevalence of the T allele of the GNB3 C825T polymorphism compared to healthy controls (OR = 1.60, 95% CI: 1.03-2.49, P = 0.038). No association between FD and the genotype of the insertion/deletion polymorphism in the promoter of SERT (SERT-P) or HTR3A C178T polymorphism was observed. Tertiary referral FD is associated with the 825T allele of the GNB3 gene. The increased signal transduction associated with this allele may contribute to the abnormalities in gastroduodenal sensory and motor function observed in FD.

Serotonergic signalling in the stomach and duodenum of patients with gastroparesis

Serotonin (5-HT) is involved in the regulation of motoric and sensory functions of the upper gastrointestinal tract. The aim of the current study was to determine whether serotonergic signalling is altered in patients with idiopathic gastroparesis. Mucosal biopsy specimens were collected from the duodenum, antrum and fundus of 11 patients with idiopathic gastroparesis and 11 healthy controls. Neuroendocrine cells, specifically 5-HT producing cells, were counted after immunohistochemistry, and non-neuronal mRNA expression levels of tryptophan hydroxylase (TPH)-1, 5-HT transport protein (SERT), 5-HT3 and 5-HT4 receptor were quantified by real time RT-PCR. The number of 5-HT producing cells was comparable between patients and controls. No difference in expression of TPH-1 (rate limiting enzyme in 5-HT biosynthetic pathway) and SERT (responsible for 5-HT uptake) was found between patients and controls (P > 0.05). In the duodenum, the expression of the 5-HT3 receptor subunits and the 5-HT4 receptor was comparable between both groups. However, the 5-HT4(c) splice variant was expressed more abundantly in healthy controls compared to patients (P = 0.015). This study suggests that the delayed gastric emptying and upper abdominal symptoms in idiopathic gastroparesis do not result from altered mucosal 5-HT biosynthetic and uptake capacity.

Regional differences in expression of TPH-1, SERT, 5-HT(3) and 5-HT(4) receptors in the human stomach and duodenum

The aim of this study was to increase the understanding of the role of serotonergic signalling in normal gastroduodenal function at a molecular level. Mucosal biopsy specimens were collected from the fundus, antrum and duodenum of 11 healthy subjects. Serotonin (5-HT)-positive cells were counted and the mRNA levels of tryptophan hydroxylase (TPH), serotonin transporter (SERT), 5-HT(4) receptor and 5-HT(3) receptor subunits were quantified by real-time reverse transcription polymerase chain reaction. The number of 5-HT-positive cells was larger in the duodenum compared with the stomach (P < 0.001). Serotonin transport protein expression was 19-fold higher in the duodenum compared with the antrum and 457-fold higher compared with the fundus (P < 0.001). Tryptophan hydroxylase-1 expression was lower in the duodenum compared with the antrum and fundus (regional differences -2.3 and -3.6, respectively). The 5-HT(4) receptor and the 5-HT(3C) and 5-HT(3E) receptor subunits were more abundantly expressed in duodenum compared with the stomach (P < 0.001). The larger number of 5-HT-positive cells, the higher expression of 5-HT(3) and 5-HT(4) receptors, and in particularly the higher uptake capacity of 5-HT in the duodenum, point out to a more prominent role of serotonergic signalling at the mucosal level in the duodenum compared with the stomach.