Clinical and pathological differences between early- and late-onset colorectal cancer and determinants of one-year all-cause mortality among advanced-stage patients: a retrospective cohort study in Medellín, Colombia.

OBJECTIVE: To identify the differences between early- (EOCRC) and late-onset colorectal cancer (LOCRC), and to evaluate the determinants of one-year all-cause mortality among advanced-stage patients. METHODS: A retrospective cohort study was carried out. CRC patients ≥ 18 years old were included. Chi-Square test was applied to compare both groups. Uni- and multivariate regressions were performed to evaluate the determinants of one-year all-cause mortality in all advanced-stage patients regardless of age of onset. RESULTS: A total of 416 patients were enrolled; 53.1 % were female. Ninety cases (21.6 %) had EOCRC and 326 (78.4 %) had LOCRC. EOCRC cases were predominantly sporadic (88.9 %). Histology of carcinoma other than adenocarcinoma (p= 0.044) and rectum tumors (p= 0.039) were more prevalent in EOCRC. LOCRC patients were more likely to have smoking history (p < 0.001) and right colon tumors (p = 0.039). Alcohol consumption history (odds ratio [OR]: 3.375, 95 %CI: 1.022-11.150) and stage IV (OR: 12.632, 95 %CI: 3.506-45.513) were associated with higher one-year all-cause mortality among advanced-stage patients, the opposite was noted with left colon tumors (OR: 0.045, 95 %CI: 0.003-0.588). CONCLUSION: EOCRC was predominantly sporadic and had more cases of uncommon histological subtypes and rectal tumors. LOCRC was characterized by a higher prevalence of smoking history. Multivariate regression revealed an association between higher one-year all-cause mortality and alcohol consumption history and stage IV in advanced-stage patients. CRC exhibited differences based on age of onset. The evaluated factors associated with CRC mortality provide valuable insights for healthcare professionals, emphasizing the importance of adequate clinical assessment and early CRC diagnosis.

Human-recognizable CT image features of subsolid lung nodules associated with diagnosis and classification by convolutional neural networks.

OBJECTIVES: The interpretability of convolutional neural networks (CNNs) for classifying subsolid nodules (SSNs) is insufficient for clinicians. Our purpose was to develop CNN models to classify SSNs on CT images and to investigate image features associated with the CNN classification. METHODS: CT images containing SSNs with a diameter of ≤ 3 cm were retrospectively collected. We trained and validated CNNs by a 5-fold cross-validation method for classifying SSNs into three categories (benign and preinvasive lesions [PL], minimally invasive adenocarcinoma [MIA], and invasive adenocarcinoma [IA]) that were histologically confirmed or followed up for 6.4 years. The mechanism of CNNs on human-recognizable CT image features was investigated and visualized by gradient-weighted class activation map (Grad-CAM), separated activation channels and areas, and DeepDream algorithm. RESULTS: The accuracy was 93% for classifying 586 SSNs from 569 patients into three categories (346 benign and PL, 144 MIA, and 96 IA in 5-fold cross-validation). The Grad-CAM successfully located the entire region of image features that determined the final classification. Activated areas in the benign and PL group were primarily smooth margins (p < 0.001) and ground-glass components (p = 0.033), whereas in the IA group, the activated areas were mainly part-solid (p < 0.001) and solid components (p < 0.001), lobulated shapes (p < 0.001), and air bronchograms (p < 0.001). However, the activated areas for MIA were variable. The DeepDream algorithm showed the image features in a human-recognizable pattern that the CNN learned from a training dataset. CONCLUSION: This study provides medical evidence to interpret the mechanism of CNNs that helps support the clinical application of artificial intelligence. KEY POINTS: • CNN achieved high accuracy (93%) in classifying subsolid nodules on CT images into three categories: benign and preinvasive lesions, MIA, and IA. • The gradient-weighted class activation map (Grad-CAM) located the entire region of image features that determined the final classification, and the visualization of the separated activated areas was consistent with radiologists' expertise for diagnosing subsolid nodules. • DeepDream showed the image features that CNN learned from a training dataset in a human-recognizable pattern.