Purpose Development of a supervised machine-learning model capable of predicting clinically relevant molecular subtypes of pancreatic ductal adenocarcinoma (PDAC) from diffusion-weighted-imaging-derived radiomic features. Methods The retrospective observational study assessed 55 surgical PDAC patients. Molecular subtypes were defined by immunohistochemical staining of KRT81. Tumors were manually segmented and 1606 radiomic features were extracted with PyRadiomics. A gradient-boosted-tree algorithm was trained on 70% of the patients (N = 28) and tested on 30% (N = 17) to predict KRT81+ vs. KRT81- tumor subtypes. A gradient-boosted survival regression model was fit to the disease-free and overall survival data. Chemotherapy response and survival were assessed stratified by subtype and radiomic signature. Radiomic feature importance was ranked. Results The mean±STDEV sensitivity, specificity and ROC-AUC were 0.90±0.07, 0.92±0.11, and 0.93±0.07, respectively. The mean±STDEV concordance indices between the disease-free and overall survival predicted by the model based on the radiomic parameters and actual patient survival were 0.76±0.05 and 0.71±0.06, respectively. Patients with a KRT81+ subtype experienced significantly diminished median overall survival compared to KRT81- patients (7.0 vs. 22.6 months, HR 4.03, log-rank-test P = <0.001) and a significantly improved response to gemcitabine-based chemotherapy over FOLFIRINOX (10.14 vs. 3.8 months median overall survival, HR 2.33, P = 0.037) compared to KRT81- patients, who responded significantly better to FOLFIRINOX over gemcitabine-based treatment (30.8 vs. 13.4 months median overall survival, HR 2.41, P = 0.027). Entropy was ranked as the most important radiomic feature. Conclusions The machine-learning based analysis of radiomic features enables the prediction of subtypes of PDAC, which are highly relevant for disease-free and overall patient survival and response to chemotherapy.

/pdf/a-machine-learning-algorithm-predicts-molecular-subtypes-in-4clur3cpqm.pdf

A machine learning algorithm predicts molecular subtypes in pancreatic ductal adenocarcinoma with differential response to gemcitabine-based versus FOLFIRINOX chemotherapy

To develop a supervised machine learning (ML) algorithm predicting above- versus below-median overall survival (OS) from diffusion-weighted imaging-derived radiomic features in patients with pancreatic ductal adenocarcinoma (PDAC). One hundred two patients with histopathologically proven PDAC were retrospectively assessed as training cohort, and 30 prospectively accrued and retrospectively enrolled patients served as independent validation cohort (IVC). Tumors were segmented on preoperative apparent diffusion coefficient (ADC) maps, and radiomic features were extracted. A random forest ML algorithm was fit to the training cohort and tested in the IVC. Histopathological subtype of tumor samples was assessed by immunohistochemistry in 21 IVC patients. Individual radiomic feature importance was evaluated by assessment of tree node Gini impurity decrease and recursive feature elimination. Fisher’s exact test, 95% confidence intervals (CI), and receiver operating characteristic area under the curve (ROC-AUC) were used. The ML algorithm achieved 87% sensitivity (95% IC 67.3–92.7), 80% specificity (95% CI 74.0–86.7), and ROC-AUC 90% for the prediction of above- versus below-median OS in the IVC. Heterogeneity-related features were highly ranked by the model. Of the 21 patients with determined histopathological subtype, 8/9 patients predicted to experience below-median OS exhibited the quasi-mesenchymal subtype, whilst 11/12 patients predicted to experience above-median OS exhibited a non-quasi-mesenchymal subtype (p < 0.001). ML application to ADC radiomics allowed OS prediction with a high diagnostic accuracy in an IVC. The high overlap of clinically relevant histopathological subtypes with model predictions underlines the potential of quantitative imaging in PDAC pre-operative subtyping and prognosis.

/pdf/a-machine-learning-model-for-the-prediction-of-survival-and-361kzjrpcc.pdf

A machine learning model for the prediction of survival and tumor subtype in pancreatic ductal adenocarcinoma from preoperative diffusion-weighted imaging

Objective The aim of this study was to perform a meta-analysis on the accuracy of endoscopic biopsies, EUS, and 18F-FDG PET(-CT) for detecting residual disease after neoadjuvant chemoradiotherapy (nCRT) for esophageal cancer. Summary of background data After nCRT, one-third of patients have a pathologically complete response in the resection specimen. Before an active surveillance strategy could be offered to these patients, clinically complete responders should be accurately identified. Methods Embase, Medline, Cochrane, and Web-of-Science were searched until February 2018 for studies on accuracy of endoscopic biopsies, EUS, or PET(-CT) for detecting locoregional residual disease after nCRT for squamous cell- or adenocarcinoma. Pooled sensitivities and specificities were calculated using random-effect meta-analyses. Results Forty-four studies were included for meta-analyses. For detecting residual disease at the primary tumor site, 12 studies evaluated endoscopic biopsies, 11 qualitative EUS, 14 qualitative PET, 8 quantitative PET using maximum standardized uptake value (SUVmax), and 7 quantitative PET using percentage reduction of SUVmax (%ΔSUVmax). Pooled sensitivities and specificities were 33% and 95% for endoscopic biopsies, 96% and 8% for qualitative EUS, 74% and 52% for qualitative PET, 69% and 72% for PET-SUVmax, and 73% and 63% for PET-%ΔSUVmax. For detecting residual nodal disease, 11 studies evaluated qualitative EUS with a pooled sensitivity and specificity of 68% and 57%, respectively. In subgroup analyses, sensitivity of PET-%ΔSUVmax and EUS for nodal disease was higher in squamous cell carcinoma than adenocarcinoma. Conclusions Current literature suggests insufficient accuracy of endoscopic biopsies, EUS, and 18F-FDG PET(-CT) as single modalities for detecting residual disease after nCRT for esophageal cancer.

Accuracy of Detecting Residual Disease After Neoadjuvant Chemoradiotherapy for Esophageal Cancer: A Systematic Review and Meta-analysis

Background Fusobacterium nucleatum (F. nucleatum) is a gut microbe implicated in gastrointestinal tumorigenesis. Predicting the chemotherapeutic response is critical to developing personalised therapeutic strategies for oesophageal cancer patients. The present study investigated the relationship between F. nucleatum and chemotherapeutic resistance in oesophageal squamous cell carcinoma (ESCC). Methods We examined the relationship between F. nucleatum and chemotherapy response in 120 ESCC resected specimens and 30 pre-treatment biopsy specimens. In vitro studies using ESCC cell lines and co-culture assays further uncovered the mechanism underlying chemotherapeutic resistance. Results ESCC patients with F. nucleatum infection displayed lesser chemotherapeutic response. The infiltration and subsistence of F. nucleatum in the ESCC cells were observed by transmission electron microscopy and laser scanning confocal microscopy. We also observed that F. nucleatum modulates the endogenous LC3 and ATG7 expression, as well as autophagosome formation to induce chemoresistance against 5-FU, CDDP, and Docetaxel. ATG7 knockdown resulted in reversal of F. nucleatum-induced chemoresistance. In addition, immunohistochemical studies confirmed the correlation between F. nucleatum infection and ATG7 expression in 284 ESCC specimens. Conclusions F. nucleatum confers chemoresistance to ESCC cells by modulating autophagy. These findings suggest that targeting F. nucleatum, during chemotherapy, could result in variable therapeutic outcomes for ESCC patients.

Fusobacterium nucleatum confers chemoresistance by modulating autophagy in oesophageal squamous cell carcinoma.

Muscle quality (i.e., the expression of muscle function per unit of muscle mass) has been proposed as a clinically-relevant measure to detect individuals at risk of functional incapacity. Individuals with obesity might be at an increased risk of having poor muscle quality. Thus, we aimed to analyze the prevalence of poor muscle quality in obese individuals, to determine associated variables, and to provide normative values for this population. 203 individuals with obesity (103 women, age: 18–75 years, body mass index (BMI): 35–64 kg·m− 2) participated in this cross-sectional study. Their muscle strength (handgrip dynamometry), muscle power (sit-to-stand test) and muscle mass (bioelectrical impedance analysis) were measured, and muscle quality (strength/power to muscle mass ratio) was compared with reference values obtained in young healthy individuals. Muscle quality was individually categorized as normal, low or poor based on specific muscle strength and power (i.e., strength and power per unit of muscle mass, respectively). Sex and age-specific normative values of specific muscle strength and power were computed for the whole cohort. Age and being a woman were inversely associated with specific muscle strength, with age being also inversely associated with specific muscle power. A small proportion of participants (6%) presented with an impaired (i.e., low/poor) specific muscle power while most of them (96%) had impaired specific muscle strength. Eventually, 84% of the participants were deemed to have poor muscle quality. Being a woman (odds ratio [OR]: 18.09, 95% confidence intervals [CI]: 4.07–80.38), age (OR: 1.06, 95%CI: 1.03–1.10) and BMI (OR: 1.22, 95%CI: 1.07–1.38) were independently associated with a higher risk of poor muscle quality in adjusted analyses. These findings show a high prevalence of poor muscle quality among individuals with obesity, with age, sex and BMI being independent predictors.

https://link.springer.com/content/pdf/10.1186/s12891-020-03228-y.pdf

Obesity-associated poor muscle quality: prevalence and association with age, sex, and body mass index

Accurate T-staging is pivotal for predicting prognosis and selecting appropriate therapies for esophageal squamous cell carcinoma (ESCC). The diagnostic performance of fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) for its T-staging is uncertain. We investigated use of FDG PET/CT for preoperative T-staging of patients with ESCC. Patients with ESCC given preoperative FDG PET/CT scans, either with (CRT[+] group) or without (CRT[−] group) neoadjuvant chemoradiotherapy, were retrospectively reviewed. Maximal standardized uptake value (SUVmax) of the primary tumors on FDG PET/CT scans were measured, and histopathological results were used as the reference standard. The associations between pathological T-stage and potential factors of age, tumor location, tumor grade, tumor size, and tumor SUVmax were analyzed. The cut-off levels of SUVmax for predicting different T-stages and for residual viable tumors after neoadjuvant chemoradiotherapy were determined using receiver operating characteristic analyses. We enrolled 103 patients (45 in the CRT[−] group; 58 in the CRT[+] group). SUVmax, an independent predictive factor, positively correlated with the pathological T-stage in both groups (CRT[−] group: ρ = 0.736, p  13.0) and 67.2% in the CRT[+] group (SUVmax of T0: 0–3.4, T1: 3.5–3.9, T2: 4.0–5.5, T3: 5.6–6.2, T4: > 6.2). For CRT[−] group, the accuracy using an SUVmax cut-off of 4.4 to differentiate early (T0-1) from locally advanced disease (T2-4) was 82.2% (95% CI, 71.1–93.4%). For CRT[+] group, the accuracy using an SUVmax cut-off of 3.4 to predict residual viable tumors (non-T0) after completion of chemoradiotherapy was 82.8% (95% CI, 73.0–92.5%). The FDG avidity of a primary esophageal tumor significantly positively correlated with the pathological T-stage. PET/CT with thresholded SUVmax was useful for predicting T-stage and differentiating residual viable tumors.

/pdf/fdg-pet-using-suvmax-for-preoperative-t-staging-of-2eft2exgd0.pdf

Yi Chun Chiu

Papers

FDG PET using SUVmax for preoperative T-staging of esophageal squamous cell carcinoma with and without neoadjuvant chemoradiotherapy

Definitive chemoradiotherapy for clinical T4b esophageal cancer - Treatment outcomes, failure patterns, and prognostic factors.