Machine Learning for Decision-Support in Acute Abdominal Pain - Proof of Concept and Central Considerations.

Acute abdominal pain is a common presenting symptom in the emergency department and represents heterogeneous causes and diagnoses. There is often a decision to be made regarding emergency surgical care. Machine learning (ML) could be used here as a decision-support and relieve the time and personnel resource shortage.Patients with acute abdominal pain presenting to the Department of Surgery at Bonn University Hospital in 2020 and 2021 were retrospectively analyzed. Clinical parameters as well as laboratory values were used as predictors. After randomly splitting into a training and test data set (ratio 80 to 20), three ML algorithms were comparatively trained and validated. The entire procedure was repeated 20 times.A total of 1357 patients were identified and included in the analysis, with one in five (n = 276, 20.3%) requiring emergency abdominal surgery within 24 hours. Patients operated on were more likely to be male (p = 0.026), older (p = 0.006), had more gastrointestinal symptoms (nausea: p < 0.001, vomiting p < 0.001) as well as a more recent onset of pain (p < 0.001). Tenderness (p < 0.001) and guarding (p < 0.001) were more common in surgically treated patients and blood analyses showed increased inflammation levels (white blood cell count: p < 0.001, CRP: p < 0.001) and onset of organ dysfunction (creatinine: p < 0.014, quick p < 0.001). Of the three trained algorithms, the tree-based methods (h2o random forest and cforest) showed the best performance. The algorithms classified patients, i.e., predicted surgery, with a median AUC ROC of 0.81 and 0.79 and AUC PRC of 0.56 in test sets.A proof-of-concept was achieved with the development of an ML model for predicting timely surgical therapy for acute abdomen. The ML algorithm can be a valuable tool in decision-making. Especially in the context of heavily used medical resources, the algorithm can help to use these scarce resources more effectively. Technological progress, especially regarding artificial intelligence, increasingly enables evidence-based approaches in surgery but requires a strictly interdisciplinary approach. In the future, the use and handling of ML should be integrated into surgical training.

Surgical treatment for pancreatic cystic lesions-implications from the multi-center and prospective German StuDoQ|Pancreas registry.

PURPOSE: The detection of pancreatic cystic lesions (PCL) causes uncertainty for physicians and patients, and international guidelines are based on low evidence. The extent and perioperative risk of resections of PCL in Germany needs comparison with these guidelines to highlight controversies and derive recommendations. METHODS: Clinical data of 1137 patients who underwent surgery for PCL between 2014 and 2019 were retrieved from the German StuDoQ|Pancreas registry. Relevant features for preoperative evaluation and predictive factors for adverse outcomes were statistically identified. RESULTS: Patients with intraductal papillary mucinous neoplasms (IPMN) represented the largest PCL subgroup (N = 689; 60.6%) while other entities (mucinous cystic neoplasms (MCN), serous cystic neoplasms (SCN), neuroendocrine tumors, pseudocysts) were less frequently resected. Symptoms of pancreatitis were associated with IPMN (OR, 1.8; P = 0.012) and pseudocysts (OR, 4.78; P < 0.001), but likewise lowered the likelihood of MCN (OR, 0.49; P = 0.046) and SCN (OR, 0.15, P = 0.002). A total of 639 (57.2%) patients received endoscopic ultrasound before resection, as recommended by guidelines. Malignancy was histologically confirmed in 137 patients (12.0%), while jaundice (OR, 5.1; P < 0.001) and weight loss (OR, 2.0; P = 0.002) were independent predictors. Most resections were performed by open surgery (N = 847, 74.5%), while distal lesions were in majority treated using minimally invasive approaches (P < 0.001). Severe morbidity was 28.4% (N = 323) and 30d mortality was 2.6% (N = 29). Increased age (P = 0.004), higher BMI (P = 0.002), liver cirrhosis (P < 0.001), and esophageal varices (P = 0.002) were independent risk factors for 30d mortality. CONCLUSION: With respect to unclear findings frequently present in PCL, diagnostic means recommended in guidelines should always be considered in the preoperative phase. The therapy of PCL should be decided upon in the light of patient-specific factors, and the surgical strategy needs to be adapted accordingly.

Machine learning to guide clinical decision-making in abdominal surgery-a systematic literature review.

PURPOSE: An indication for surgical therapy includes balancing benefits against risk, which remains a key task in all surgical disciplines. Decisions are oftentimes based on clinical experience while guidelines lack evidence-based background. Various medical fields capitalized the application of machine learning (ML), and preliminary research suggests promising implications in surgeons' workflow. Hence, we evaluated ML's contemporary and possible future role in clinical decision-making (CDM) focusing on abdominal surgery. METHODS: Using the PICO framework, relevant keywords and research questions were identified. Following the PRISMA guidelines, a systemic search strategy in the PubMed database was conducted. Results were filtered by distinct criteria and selected articles were manually full text reviewed. RESULTS: Literature review revealed 4,396 articles, of which 47 matched the search criteria. The mean number of patients included was 55,843. A total of eight distinct ML techniques were evaluated whereas AUROC was applied by most authors for comparing ML predictions vs. conventional CDM routines. Most authors (N = 30/47, 63.8%) stated ML's superiority in the prediction of benefits and risks of surgery. The identification of highly relevant parameters to be integrated into algorithms allowing a more precise prognosis was emphasized as the main advantage of ML in CDM. CONCLUSIONS: A potential value of ML for surgical decision-making was demonstrated in several scientific articles. However, the low number of publications with only few collaborative studies between surgeons and computer scientists underpins the early phase of this highly promising field. Interdisciplinary research initiatives combining existing clinical datasets and emerging techniques of data processing may likely improve CDM in abdominal surgery in the future.

A Combined TLR7/TLR9/GATA3 Score Can Predict Prognosis in Biliary Tract Cancer.

Biliary tract cancer (BTC) refers to a heterogenous group of epithelial malignancies arising along the biliary tree. The highly aggressive nature combined with its silent presentation contribute to the dismal prognosis of this tumor. Tumor-infiltrating immune cells (TIICs) are frequently present in BTC and there is growing evidence regarding their role as therapeutic targets. In this study, we analyzed the immune cell infiltration in BTC and developed a promising immune signature score to predict prognosis in BTC. Immunohistochemistry (IHC) was carried out on tissue microarray sections from 45 patients with resectable cholangiocarcinoma for the detection of 6-sulfoLacNAc+ monocytes (slanMo), BDCA-2+ plasmacytoid dendritic cells (pDC), CD8+ or CD4+T-lymphocytes, CD103+ cells, GATA3+ cells, Toll-like receptor (TLR) 3, 7 and 9-expressing cells as well as programmed cell death protein 1 and programmed cell death ligand 1 positive cells. Data from the IHC staining were analyzed and correlated with clinicopathological and survival data. High expression of TLR7, TLR9, and GATA3 was associated with improved overall survival (OS, Log-rank p < 0.05). In addition, TLR9 was associated with better disease-free survival (Log-rank p < 0.05). In the multivariate Cox proportional-hazards model for OS, the TLR/TLR9/GATA3 score was found to be an independent prognostic factor for OS ("Score 2" vs. "Score 0": HR 11.17 95% CI 2.27-54.95, p < 0.01).

Tumor Infiltrating Neutrophils Are Frequently Found in Adenocarcinomas of the Biliary Tract and Their Precursor Lesions with Possible Impact on Prognosis.

Biliary tract cancer (BTC) is characterized by an intense stromal reaction and a complex landscape of infiltrating immune cells. Evidence is emerging that tumor-infiltrating neutrophils (TINs) have an impact on carcinogenesis and tumor progression. TINs have also been associated with outcomes in various solid malignant tumors but their possible clinical role in BTC is largely unknown. Tissue samples from patients with sporadic BTC ("spBTC" cohort, N = 53) and BTC in association with primary sclerosing cholangitis ("PSC-BTC" cohort, N = 7) were collected. Furthermore, tissue samples from 27 patients with PSC who underwent liver transplantation ("PSC-LTX" cohort) were investigated. All specimens were assessed for TIN density in invasive and precancerous lesions (biliary intraepithelial neoplasia, BilIN). Most spBTC showed low TIN density (LD, 61%). High TIN density (HD) was detected in 16% of the tumors, whereas 23% were classified as intermediate density (ID); the majority of both HD and ID groups were in T1-T2 tumors (83% and 100%, p = 0.012). TIN density in BilIN lesions did not significantly differ among the three groups. The HD group had a mean overall survival (OS) of 53.5 months, whereas the mean OS in the LD and ID groups was significantly shorter (LD 29.5 months vs. ID 24.6 months, log-rank p < 0.05). The results of this study underline the possible prognostic relevance of TINs in BTC and stress the complexity of the immune cell landscape in BTC. The prognostic relevance of TINs suggests a key regulator role in inflammation and immune landscape in BTC.

[Surgical treatment of primary sclerosing cholangitis : Experiences from 30 years in a single center cohort with 173 consecutive patients]. / Chirurgische Therapie der primär sklerosierenden Cholangitis : Erfahrungen aus 30 Jahren in einer monozentrischen Kohorte mit 173 konsekutiven Patienten.

BACKGROUND: In recent years substantial progress has been made in the treatment, surveillance and understanding of the pathogenesis of primary sclerosing cholangitis (PSC); however, in most cases liver transplantation (LTX) is still the only curative option for cancer or end-stage liver disease (ELD). In rare cases a partial liver resection is a possible curative treatment of a PSC-associated cholangiocellular carcinoma (CCC). These operations represent a significant additional burden for PSC patients. OBJECTIVE: Due to the rarity of PSC detailed studies regarding hepato-pancreato-biliary (HPB) surgery are lacking. The aim of this study was to analyze the surgical indications and prognosis of PSC patients. PATIENTS AND METHODS: A single center retrospective cohort study from 1990 to 2020 was carried out. In this study patients with PSC were included and investigated with respect to factors associated with surgery and the prognosis. RESULTS: As a consequence of PSC-associated conditions, in 62 patients (36%) a major HPB operation or explorative laparotomy was necessary. The prevalence of chronic inflammatory bowel disease was significantly higher in these patients (P < 0.019). An LTX was carried out in 46 patients (73%) because of ELD. A liver resection (LR) was performed in 8 patients (11%) and 9 patients only underwent an explorative laparotomy. The overall survival in the LTX subgroup was significantly longer than patients who underwent LR and explorative laparotomy (258 months; 95% confidence interval, CI 210-306 months vs. 88 months; 95% CI 16-161 months vs. 13 months; 95% CI 3-23 months; p < 0.05, respectively). CONCLUSION: The majority of patients with PSC have to be operated on because of the disease with substantial risks for morbidity and mortality. Curative treatment options are lacking, thus underlining the need for effective early detection and treatment strategies for PSC-CCC.

Open Abdomen Treatment in Acute Pancreatitis.

Background: Severe acute pancreatitis (SAP) is a heterogeneous and life-threatening disease. While recent guidelines recommend a stepwise approach starting with non-surgical techniques, emergency laparotomy remains inevitable in certain situations. Open abdomen treatment (OAT) may follow, potentially resulting in additional risks for severe morbidity. Causative factors and clinical impact of OAT in SAP are poorly understood and therefore issue of the present study. Materials and Methods: A retrospective analysis of patients admitted to the Department of General, Visceral, Thoracic and Vascular Surgery at University of Bonn suffering from acute pancreatitis (ICD K.85) between 2005 and 2020 was performed. Medical records were screened for demographic, clinical and outcome parameters. Patients who received primary fascial closure (PFC) were compared to those patients requiring OAT. SAP-specific scores were calculated, and data statistically analyzed (P = 0.05). Results: Among 430 patients included, 54 patients (13%) had to undergo emergency laparotomy for SAP. Patients were dominantly male (72%) with a median age of 51 years. Indications for surgery were infected necrosis (40%), suspected bowel perforation (7%), abdominal compartment syndrome (5%), and acute intra-abdominal hemorrhage (3%). While 22 patients (40%) had PFC within initial surgery, 33 patients (60%) required OAT including a median of 12 subsequent operations (SD: 6, range: 1-24). Compared to patients with PFC, patients in the OAT group had significantly fewer biliary SAP (P = 0.031), higher preoperative leukocyte counts (P = 0.017), higher rates of colon resections (P = 0.048), prolonged ICU stays (P = 0.0001), and higher morbidity according to Clavien-Dindo Classification (P = 0.002). Additionally, BISAP score correlated positively with the number of days spent at ICU and morbidity (P = 0.001 and P = 0.000002). Both groups had equal mortality rates. Discussion: Our data suggest that preoperative factors in surgically treated SAP may indicate the need for OAT. The procedure itself appears safe with equal hospitalization days and mortality rates compared to patients with PFC. However, OAT may significantly increase morbidity through longer ICU stays and more bowel resections. Thus, minimally invasive options should be promoted for an uncomplicated and rapid recovery in this severe disease. Emergency laparotomy will remain ultima ratio in SAP while patient selection seems to be crucial for improved clinical outcomes.

Diagnostic yield and clinical utility of a comprehensive gene panel for hereditary tumor syndromes.

BACKGROUND: In a considerable number of patients with a suspected hereditary tumor syndrome (HTS), no underlying germline mutation is detected in the most likely affected genes. The present study aimed to establish and validate a large gene panel for HTS, and determine its diagnostic yield and clinical utility. METHODS: The study cohort comprised 173 patients with suspected, but unexplained, HTS (group U) and 64 HTS patients with a broad spectrum of known germline mutations (group K). All patients in group U presented with early age at onset, multiple tumors, and/or a familial clustering of various tumor types; no germline mutation had been identified during routine diagnostics. Sequencing of leukocyte DNA was performed for the 94 HTS genes of the Illumina TruSight™Cancer Panel and 54 additional HTS genes. RESULTS: The sensitivity of the panel to identify known germline variants was 99.6%. In addition to known mutations, a total of 192 rare, potentially pathogenic germline variants in 86 genes were identified. Neither the proportion of rare variants per patient (group K: 0.9 variants; group U: 0.8 variants) nor the proportion of variants in the most frequently mutated, moderately penetrant genes CHEK2 and ATM showed significant inter-group difference. Four of the five patients from group U with a truncating CHEK2 mutation had thyroid cancer, pointing to a broader tumor spectrum in patients with pathogenic CHEK2 variants. In 22% of patients from group K, a further potential causative variant was identified. Here, the most interesting finding was an NF1 nonsense mutation in a child with a known TP53 frameshift mutation. In 17% of patients from group U, potential causative variants were identified. In three of these patients (2%), mutations in PMS2, PTEN, or POLD1 were considered to be causative. In both groups, incidental findings with presumptive predictive value were generated. CONCLUSIONS: The gene panel identified the genetic cause in some prescreened, unexplained HTS patients and generated incidental findings. Some patients harbored predicted pathogenic mutations in more than one established HTS gene, rendering interpretation of the respective alterations challenging. Established moderate risk genes showed an almost equal distribution among patients with known and unexplained disease.