Introduction

The prediction of mediastinal and intra-abdominal lymphadenopathy poses a significant challenge, and definitive diagnosis is essential for treatment selection. Physicians performing endoscopic ultrasonography (EUS) can face difficulties in targetting suspicious lymph nodes, and lead unnecessary punctures when multiple lesions are present.

Aims & Methods

The aim of this study is to investigate the value of deep learning (DL) in aiding differentiating lymphadenopathy with EUS images. The retrospective single-center study included patients who underwent EUS for mediastinal or intra-abdominal lymphadenopathy from September 2016 to January 2023. Lymphadenopathy areas were outlined on EUS images for machine learning. UNet was selected for segmentation, and the classification was finished by k-Nearest-Neighbor approach and convolutional neural networks including ResNet 50, 101, 152 and DenseNet 121, 169, 201. Experts collected features of each lymph node for binary logistic regression analysis. The physician group, consisting of two experts and two beginners, reviewed EUS images from the validation set of the DL model and made respective diagnosis. Diagnostic efficacy was evaluated among these groups.

Results

A total of 200 suspicious lymph nodes with 630 EUS images from 93 patients were enrolled, and105 images were randomly selected as the validation set. ResNet 50 achieved a sensitivity of 95.38%, a specificity of 70.00%, an accuracy of 85.71%, and an area under the curve (AUC) of 0.827. The kappa statistic (kappa value=0.683) indicated substantial agreement compared to the final diagnosis. The logistic regression model achieved a sensitivity of 81.54%, a specificity of 60.00%, an accuracy of 73.33%, and an AUC of 0.708. The kappa statistic (kappa value=0.424) indicated moderate agreement compared to the final diagnosis. The diagnosis by experts achieved a sensitivity of 73.85%, a specificity of 65.00%, an accuracy of 70.47%, and an AUC of 0.694. The kappa statistic (kappa value=0.383) indicated fair agreement compared to the final diagnosis. The diagnosis by beginners achieved a sensitivity of 58.46%, a specificity of 55.00%, an accuracy of 57.14%, and an AUC of 0.575. The kappa statistic (kappa value=0.129) indicated slight agreement compared to the final diagnosis. Paired sample comparisons of their receiver operating characteristic curves (ROC) showed that ResNet 50 outperformed the other three groups (P<0.05).

Conclusion

DL has certain reliability and feasibility in assisting differentiating mediastinal and intra-abdominal lymphadenopathy with EUS images, which effectively supplements the diagnosis from clinical experience and yields benefits.

Introduction

Simple renal cyst is one of the most common diseases in the urinary system. Intervention is only considered under circumstances of large cysts (＞5cm), bleeding, recurrent infection or pain^[1]. Ethanol percutaneous sclerotherapy and laparoscopy are two main treatment modalities for simple renal cysts^[2]. However, in cases of cysts in superior renal pole, percutaneous sclerotherapy is often excluded due to the unsafety of percutaneous puncture. In recent years, endoscopic ultrasound (EUS) guided fine-needle aspiration has been proven as a safe and efficient treatment for biopsy of renal and renicapsule lesions^[3], inspired by which, we successfully performed an EUS-guided sclerotherapy on cyst in the superior renal pole^[4]. Since then, we performed nine more cases of EUS-guided sclerotherapy on cyst in the superior renal pole. This research is an retrospective analysis of the nine cases and is aimed to compare the efficacy, safety and cost-benefit analysis of endoscopic ultrasound guided sclerotherapy and laparoscopic surgery treating cyst in the superior renal pole.

Aims & Methods

To compare the efficacy, safety and cost-benefit analysis of endoscopic ultrasound guided sclerotherapy and laparoscopic surgery treating cyst in the superior renal pole.
Methods: A retrospective analysis was performed in 25 patients with cyst in the superior renal pole hospitalized in the Second Affiliated Hospital of Soochow University from January 2021 to August 2022.The efficacy, operation time, intraoperative blood loss volume, hospital stay, complications and treatment cost of EUS-guided sclerotherapy (9 cases) and laparoscopic surgery (16 cases) were compared.

Results

The EUS and laparoscopic groups had comparable efficacy (100% vs 87.5%, P=0.520). The EUS group exhibited shorter operation time (29.8±4.8 vs 70.1±11.1 min, P=0.000), less intraoperative blood loss (0 vs 26.1±5.9 ml, P=0.000), shorter length of postoperative hospital stay (3.5±0.7 vs 5.4±2.0 days, P=0.014) and lower total cost (10547.85±2388.19 vs15316.09±5352.45 yuan, P=0.019). There were no difference on total length of hospital stay(8.1±2.0 vs 9.3±3.1 days) and operation cost (3946.79±490.82 vs 3860.18.17±857.42 yuan) (both P＞0.05).

Conclusion

Compared with laparoscopic surgery, EUS-guided sclerotherapy was an effective but safer technique in treating cyst in the superior renal pole.

References

1. Richard PO, Violette PD, Jewett MA, et al. CUA guideline on the management of cystic renal lesions[J].Can Urol Assoc J. 2017, 11(3-4):E66-E73.DOI: 10.4253/wjge.v7.i3.253.
2. Eissa A, El Sherbiny A, Martorana E, et al. Non-conservative management of simple renal cysts inadults: a comprehensive review of literature[J]. Minerva Urol Nefrol. 2018, 70(2):179-192. DOI: 10.23736/S0393-2249.17.02985-X.
3. Lopes RI, Moura RN, Artifon E. Endoscopic ultrasound-guided fine-needle aspiration for the diagnosis of kidney lesions: A review[J]. World J Gastrointest Endosc. 2015, 7(3):253-7. DOI: 10.4253/wjge.v7.i3.253.
4. Wu W, Zhang Y, Hu D, Song K. EUS-guided sclerotherapy as a new therapy for renal cyst (with video)[J]. Gastrointest Endosc. 2022, 95(3):587-588. DOI:10.1016/j.gie.2021.10.029.

Introduction

Experimental data have shown increased expression of angiogenic factors in portal hypertension and disruption of endothelial integrity in small intestinal mucosal vessels that is possibly correlated with antigen translocation to the portal circulation. ZO-1 is a tight junction structural protein of the intestinal epithelium. Recent data revealed that ΖΟ-1 is also expressed by endothelial and mesenchymal cells and plays a dominant role in mucosal healing. Decreased ZO-1 expression may unsettle epithelial restorative capacity following mucosal damage and contribute to the impaired endothelial integrity and bacterial translocation.

Aims & Methods

We therefore aimed to assess changes in mucosal vascular density and evaluate intestinal vessel endothelial integrity in cirrhotic patients by assessing the immunohistochemical expression of ZO-1 in small intestinal biopsy samples. Sixty-three cirrhotic patients were enrolled (M/F:35/28, mean age ±SD: 62±13) during an 18-month period. Nine non-cirrhotic patients served as controls. Patients with current or recent alcohol consumption, porto-splenic vein thrombosis, non-cirrhotic portal hypertension, hepatocellular carcinoma, transjugular intrahepatic portosystemic shunt, chronic kidney disease, celiac disease, acute infection, or any other chronic intestinal disease were excluded. Small intestine biopsy samples were taken during endoscopy and endothelial ZO-1 expression was assessed by immunohistochemistry using the indirect streptavidin biotin peroxidase method. Mucosal vascular density was evaluated by counting the number of mucosal vessels per x20 optical field. Serum lipopolysaccharide-binding protein (LBP) levels were measured by ELISA.

Results

Thirty-four patients (54%) had compensated and 29(46%) had decompensated cirrhosis. Child-Turcotte-Pugh (CTP) classification was: A 46 (73%)/B 14 (22%)/ C 3 (5%). The mean MELD score was 10.5±3.7. Mean LBP levels were 7511.97±2868.90 μg/ml. The mean number of mucosal vessels per x20 optical field did not differ between cirrhotic patients and controls (3.7±1.1 vs 3.2±0.9, p=0.238) but patients with CTP-A class had increased number of mucosal vessels compared to controls (3.8±1.1 vs 3.2±0.9, p=0.048) and CTP-B class patients (3.8±1.1 vs 3.2±0.7, p=0.045). A negative correlation was observed between the number of mucosal vessels and CTP score (r: -0.428, p=0.016). In cirrhotic patients, ZO-1 immunostaining was detected in the whole perimeter of all counted vessels per optical field in only 10/63 (16.4%) samples. Loss of endothelial ZO-1 immunostaining in more than 50% of the vessel perimeter or total loss of ZO-1 in more than 50% of counted vessels was observed more frequently in cirrhotic patient samples (42/63, 66.6%) compared to controls (1/9, 11.1%, p=0.001). Mean (±SD) LBP levels were numerically, albeit not significantly, higher in patients with than without significant loss of ZO-1 endothelial expression (7801±3698 vs 6761±1032 μg/ml, p=0.342).

Conclusion

Cirrhotic patients show increased loss of small intestinal endothelial ZO-1 expression compared to control patients. Intestinal mucosal vessel number is increased in CTP-A patients compared to control and CTP-B patients. These findings are in agreement with novel theories proposing that mechanisms leading to bacterial translocation in cirrhosis are not confined to the small intestinal lumen and merit further investigation.

Clinical Case Summary

A 50 year old paraplegic gentleman (secondary to a road traffic accident) with a diversion colostomy and preserved small bowel length, and a history of idiopathic cirrhosis, was admitted with a one week history of high output stoma. Initially felt to be infectious, he was resuscitated with volume and electrolyte replacement. Despite a negative infection, autoimmune and extended inflammatory screen, the high stoma output persisted between 4-6L per 24 hours. He subsequently developed circulatory failure due to volume losses and required intensive care support for fluid management, parenteral nutrition (PN) and inotropes. Serial cross sectional imaging showed oedematous jejunal and ileal bowel without evidence of vascular insufficiency.

His stoma output failed to respond to maximum doses of loperamide, octreotide, proton pump inhibitors or opiates. He was commenced on a trial of high dose steroids without any benefit. A colonoscopy showed grossly oedematous bowel, with biopsies demonstrating preserved crypt architecture on a background of significant oedema. He was PN dependant with challenging management of his fluid balance. Due to a new acute renal injury with nephrotic range proteinuria, further stains on his colonic biopsies were requested. These stained positive with congo red, showing apple green birefringence which was suggestive of a diagnosis of Amyloidosis.
Following consultation with the National Amyloidosis Referral Centre, further investigations were requested which showed an amyloid load in the spleen (SAP scan) and chronic pelvic sepsis (PET CT) likely to be driving a chronic inflammatory state. Due to the advance nature of his disease, no active treatment could be offered and the patient unfortunately passed away 8 months into their hospital admission due to multi-organ failure.
Through reporting of this case, we want to raise awareness of Amyloidosis as a rare cause of Type 2 intestinal failure.

Introduction

Multiple treatment choices are available for patients with either Ulcerative Colitis (UC) or Crohn’s Disease (CD) requiring a first biologic therapy, but there is very limited data on comparative efficacy of different agents. Outcomes from robust, large-scale cohort studies can help to inform therapeutic decisions.

Aims & Methods

To compare the efficacy of first line biologic therapies [infliximab (IFX), adalimumab (ADA), golimumab (GLM), vedolizumab (VDZ)] in biologic-naïve patients with UC and CD and enrolled in the UK Inflammatory Bowel Disease (IBD) BioResource. Patients receiving first-line tofacitinib and ustekinumab were excluded from the analysis due to low numbers. Demographic, disease, treatment and outcome data were retrieved. Inverse probability of treatment weighting (IPTW) was used to balance groups using a propensity score-weighting approach accounting for baseline patient or disease related clinical characteristics. Balancing success was assessed with standardized mean differences (SMDs). Effectiveness of treatment was based on persistence free of treatment discontinuation or treatment failure. Failure was defined as occurring if any of the following occurred: (a) clinician-recorded treatment failure (b)start of treatment with a different biologic drug (c)occurrence of resectional or defunctioning bowel surgery. We compared Kaplan Meier curves for treatment failure-free survival after IPTW and used a log rank test for differences between the groups.Cox proportionate hazard ratios (HR)were used for comparisons between treatments.

Results

In total, 12747 patients were included in the analysis (UC=3967, CD=8780). Over a four-year follow up period, treatment survival free of treatment failure or treatment escalation was significantly different between different treatment groups both in UC and CD (log rank p=0.007 and p=0.011 respectively). For UC, we found that VDZ was superior to all anti-TNF agents [vs GLM HR 3.4 (95% CIs 2.2-5.2), vs ADA HR 3.1 (95% CIs 2.1-4.5), vs IFX HR 1.9 (95% CIs 1.3-2.8), all p values <0.001]. When compared between anti-TNF agents IFX was superior to both ADA and GLM [vs ADA HR 1.6 (95% CIs 1.4-1.8), vs GLM HR 1.7 (95% CIs 1.3-2.2)] while no significant difference was detected when ADA and GLM were compared. For CD, IPTW did not achieve optimal balancing for baseline differences in the presence of perianal disease. Restricting analysis to the subgroup with the absence of perianal disease, we did not observe a significant difference in treatment outcomes.

Conclusion

Our study showed superiority of VDZ over anti-TNF agents in biologic naïve patients with UC. This is in accordance with, and extends the findings of, a prior head-to-head trial in UC comparing VDZ and ADA¹. In patients with luminal CD, no difference was detected between VDZ and anti-TNF agents. Our data offer guidance to physicians and patients as to relative drug efficacy for first line therapies in IBD.

References

1. Sands BE, Peyrin-Biroulet L, Loftus EV Jr, et al. Vedolizumab versus Adalimumab for Moderate-to-Severe Ulcerative Colitis. N Engl J Med. 2019;381(13):1215-1226. doi:10.1056/NEJMoa1905725