Choi S, et al. J Korean Med Sci. 2021 Jul;36(40):e286. https://doi.org/10.3346/jkms.2021.36.e286

Microsatellite instability (MSI)/defective DNA mismatch repair (dMMR) is receiving more attention as a biomarker for eligibility for immune checkpoint inhibitors in advanced diseases. However, due to high costs and resource limitations, MSI/dMMR testing is not widely performed. Some attempts are in progress to predict MSI/dMMR status through histomorphological features on H&E slides using artificial intelligence (AI) technology. In this study, the potential predictive role of this new methodology was reviewed through a systematic review. Studies up to September 2021 were searched through PubMed and Embase database searches. The design and results of each study were summarized, and the risk of bias for each study was evaluated. For colorectal cancer, AI-based systems showed excellent performance with the highest standard of 0.972; for gastric and endometrial cancers they showed a relatively low but satisfactory performance, with the highest standard of 0.81 and 0.82, respectively. However, analyzing the risk of bias, most studies were evaluated at high-risk. AI-based systems showed a high potential in predicting the MSI/dMMR status of different cancer types, and particularly of colorectal cancers. Therefore, a confirmation test should be required only for the results that are positive in the AI test.

Before publication of the new classification system named the Milan System for Reporting Salivary Gland Cytopathology (MSRSGC) in 2018, there was no standard classification for salivary gland lesions obtained by fine-needle aspiration (FNA). We therefore aimed to evaluate the diagnostic utility of this system by retrospectively reviewing FNA samples using the MSRSGC and to determine their risk of developing into neoplasms and becoming malignant. Retrospective slide review and classification of salivary gland FNAs obtained over a 6-year period (2013-2018) at a single center were performed by two pathologists. The risks of neoplasm and malignancy for each category also were calculated. This study surveyed 374 FNAs (371 patients) performed over a six-year period and selected 148 cases that included documented surgical follow-up (39.6%). Among the surgically treated cases, the distributions of FNA categories were as follows: non-diagnostic (ND; 16.9%), non-neoplastic (NN; 2.7%), atypia of undetermined significance (AUS; 3.4%), benign (BN; 54.7%), salivary gland neoplasm of uncertain malignant potential (SUMP; 10.1%), suspicious for malignancy (SM; 6.8%), and malignant (M; 5.4%). The risk of malignancy (ROM) was 24.0% for ND, 0% for NN, 40.0% for AUS, 2.5% for BN, 46.7% for SUMP, 100% for SM, and 87.5% for M. The overall diagnostic accuracy was 95.9% (142/148 cases). The newly proposed MSRSGC appears to be a reliable system for classification of salivary gland lesions according to the associated ROM.

Jeon YH, et al. J Korean Med Sci. 2023 Mar;38(10):e78. https://doi.org/10.3346/jkms.2023.38.e78

Background: The deep learning-based nodule detection (DLD) system improves nodule detection performance of observers on chest radiographs (CXRs). However, its performance in different pulmonary nodule (PN) locations remains unknown. Methods: We divided the CXR intrathoracic region into non-danger zone (NDZ) and danger zone (DZ). The DZ included the lung apices, paramediastinal areas, and retrodiaphragmatic areas, where nodules could be missed. We used a dataset of 300 CXRs (100 normal and 200 abnormal images with 216 PNs [107 NDZ and 109 DZ nodules]). Eight observers (two thoracic radiologists [TRs], two non-thoracic radiologists [NTRs], and four radiology residents [RRs]) interpreted each radiograph with and without the DLD system. The metric of lesion localization fraction (LLF; the number of correctly localized lesions divided by the total number of true lesions) was used to evaluate the diagnostic performance according to the nodule location. Results: The DLD system demonstrated a lower LLF for the detection of DZ nodules (64.2) than that of NDZ nodules (83.2, p = 0.008). For DZ nodule detection, the LLF of the DLD system (64.2) was lower than that of TRs (81.7, p < 0.001), which was comparable to that of NTRs (56.4, p = 0.531) and RRs (56.7, p = 0.459). Nonetheless, the LLF of RRs significantly improved from 56.7 to 65.6 using the DLD system (p = 0.021) for DZ nodule detection. Conclusion: The performance of the DLD system was lower in the detection of DZ nodules compared to that of NDZ nodules. Nonetheless, RR performance in detecting DZ nodules improved upon using the DLD system. Critical relevance statement: Despite the deep learning-based nodule detection system’s limitations in detecting danger zone nodules, it proves beneficial for less-experienced observers by providing valuable assistance in identifying these nodules, thereby advancing nodule detection in clinical practice. Key points: • The deep learning-based nodule detection (DLD) system can improve the diagnostic performance of observers in nodule detection. • The DLD system shows poor diagnostic performance in detecting danger zone nodules. • For less-experienced observers, the DLD system is helpful in detecting danger zone nodules. Graphical

Aim The prognostic role of CD274 (programmed cell death ligand 1 (PD-L1)) overexpression has been examined in many studies. However, the results are controversial and conflicting. The present study aims to investigate the potential role of CD274 (PD-L1) immunohistochemical overexpression as a prognostic marker in malignant tumours. Methods We searched PubMed, Embase and Web of Science from inception to December 2021 to identify potentially eligible studies. The pooled HRs with 95% CIs were calculated to identify the association between CD274 (PD-L1) overexpression and overall survival (OS), cancer-specific survival, disease-free survival, recurrence-free survival and progression-free survival in 10 lethal malignant tumours. Heterogeneity and publication bias were also analysed. Results The study included 57 322 patients from 250 eligible studies (241 articles). The meta-analysis by tumour type using multivariate HR revealed worse OS in non-small cell lung cancer (HR 1.41, 95% CI 1.19 to 1.68), hepatocellular carcinoma (HR 1.75, 95% CI 1.11 to 2.74), pancreatic cancer (HR 1.84, 95% CI 1.12 to 3.02), renal cell carcinoma (HR 1.55, 95% CI 1.12 to 2.14) and colorectal cancer (HR 1.46, 95% CI 1.14 to 1.88). Estimated HRs showed associations between CD274 (PD-L1) overexpression and worse prognosis across different types of tumours in various survival endpoints, but no inverse correlation was identified. The heterogeneity for most of the pooled results was high. Conclusions This large meta-analysis suggests that CD274 (PD-L1) overexpression is a potential biomarker for multiple types of cancers. However, further studies are needed to reduce high heterogeneity. PROSPERO registration number CRD42022296801.