Artificial intelligence for diagnosis and prognosis prediction of natural killer/T cell lymphoma using magnetic resonance imaging.

Accurate diagnosis and prognosis prediction are conducive to early intervention and improvement of medical care for natural killer/T cell lymphoma (NKTCL). Artificial intelligence (AI)-based systems are developed based on nasopharynx magnetic resonance imaging. The diagnostic systems achieve areas under the curve of 0.905-0.960 in detecting malignant nasopharyngeal lesions and distinguishing NKTCL from nasopharyngeal carcinoma in independent validation datasets. In comparison to human radiologists, the diagnostic systems show higher accuracies than resident radiologists and comparable ones to senior radiologists. The prognostic system shows promising performance in predicting survival outcomes of NKTCL and outperforms several clinical models. For patients with early-stage NKTCL, only the high-risk group benefits from early radiotherapy (hazard ratio = 0.414 vs. late radiotherapy; 95% confidence interval, 0.190-0.900, p = 0.022), while progression-free survival does not differ in the low-risk group. In conclusion, AI-based systems show potential in assisting accurate diagnosis and prognosis prediction and may contribute to therapeutic optimization for NKTCL.

Radiomic signatures associated with tumor immune heterogeneity predict survival in locally recurrent nasopharyngeal carcinoma.

BACKGROUND: The prognostic value of traditional clinical indicators for locally recurrent nasopharyngeal carcinoma (lrNPC) is limited due to their inability to reflect intratumor heterogeneity. We aimed to develop a radiomic signature to reveal tumor immune heterogeneity and predict survival in lrNPC. METHODS: This multicenter, retrospective study included 921 patients with lrNPC. A machine learning signature and nomogram based on pretreatment MRI features were developed for predicting overall survival (OS) in a training cohort and validated in two independent cohorts. A clinical nomogram and an integrated nomogram were constructed for comparison. Nomogram performance was evaluated by concordance index (C-index) and receiver operating characteristic curve analysis. Accordingly, patients were classified into risk groups. The biological characteristics and immune infiltration of the signature were explored by RNA sequencing (RNA-seq) analysis. RESULTS: The machine learning signature and nomogram demonstrated comparable prognostic ability to a clinical nomogram, achieving C-indexes of 0.729, 0.718, and 0.731 in the training, internal, and external validation cohorts, respectively. Integration of the signature and clinical variables significantly improved the predictive performance. The proposed signature effectively distinguished patients between risk groups with significantly distinct OS rates. Subgroup analysis indicated the recommendation of local salvage treatments for low-risk patients. Exploratory RNA-seq analysis revealed differences in interferon response and lymphocyte infiltration between risk groups. CONCLUSIONS: An MRI-based radiomic signature predicted OS more accurately. The proposed signature associated with tumor immune heterogeneity may serve as a valuable tool to facilitate prognostic stratification and guide individualized management for lrNPC patients.

Machine learning-based identification of contrast-enhancement phase of computed tomography scans.

Contrast-enhanced computed tomography scans (CECT) are routinely used in the evaluation of different clinical scenarios, including the detection and characterization of hepatocellular carcinoma (HCC). Quantitative medical image analysis has been an exponentially growing scientific field. A number of studies reported on the effects of variations in the contrast enhancement phase on the reproducibility of quantitative imaging features extracted from CT scans. The identification and labeling of phase enhancement is a time-consuming task, with a current need for an accurate automated labeling algorithm to identify the enhancement phase of CT scans. In this study, we investigated the ability of machine learning algorithms to label the phases in a dataset of 59 HCC patients scanned with a dynamic contrast-enhanced CT protocol. The ground truth labels were provided by expert radiologists. Regions of interest were defined within the aorta, the portal vein, and the liver. Mean density values were extracted from those regions of interest and used for machine learning modeling. Models were evaluated using accuracy, the area under the curve (AUC), and Matthew's correlation coefficient (MCC). We tested the algorithms on an external dataset (76 patients). Our results indicate that several supervised learning algorithms (logistic regression, random forest, etc.) performed similarly, and our developed algorithms can accurately classify the phase of contrast enhancement.

Evolution of radiation-induced temporal lobe injury after intensity-modulated radiation therapy in nasopharyngeal carcinoma: a large cohort retrospective study.

BACKGROUND: Previous studies have demonstrated conflicting findings regarding the initial MRI patterns of radiotherapy-induced temporal lobe injury (RTLI) and the evolution of different RTLI patterns. The aim of this study was to evaluate the initial MRI pattern and evolution of RTLI in patients with nasopharyngeal carcinoma (NPC) by means of a large cohort study. METHODS: Data of patients with RTLI were retrospectively collected from two hospitals between January 2011 and December 2021. The injured lobes were categorized into three patterns based on initial MRI patterns: isolated white matter lesions (WMLs), isolated contrast-enhanced lesions (CELs), and combined WMLs and CELs. The latency period, MRI appearances, and temporal changes in WMLs and CELs were evaluated. RESULTS: A total of 913 RTLI patients with 1092 injured lobes were included in this study. The numbers of isolated WMLs, isolated CELs, and combined WMLs and CELs identified at the first MRI detection were 7 (0.6%), 172 (15.8%), and 913 (83.6%), respectively. The evolution of bilateral RTLI was different in the same patient, and that of unilateral RTLI combined with WMLs and CELs also may occur asynchronously. The time intervals from the initial MRI detection of isolated WMLs, isolated CELs, combined WMLs and CELs to the last negative MRI scan were 8.6, 8.9 and 11.0 months, respectively. A significant difference was observed in the time intervals between the three patterns (H = 14.287, P = 0.001). And the time interval was identified as an independent factor influencing the initial MRI pattern of RTLI after Poisson regression (P = 0.002). CONCLUSION: Both WMLs and CELs could be the initial and only MRI abnormalities in patients with RTLI. This study is of great significance in accurately diagnosing RTLI early and providing timely treatment options. Additionally, it provides clinical evidence for guidelines on NPC, emphasizing the importance of regular follow-up of NPC patients.

N6-methyladenosine-modified circPLPP4 sustains cisplatin resistance in ovarian cancer cells via PIK3R1 upregulation.

BACKGROUND: Cisplatin (CDDP) is the first-line chemotherapeutic strategy to treat patients with ovarian cancer (OC). The development of CDDP resistance remains an unsurmountable obstacle in OC treatment and frequently induces tumor recurrence. Circular RNAs (circRNAs) are noncoding RNAs with important functions in cancer progression. Whether circRNAs function in CDDP resistance of OC is unclear. METHODS: Platinum-resistant circRNAs were screened via circRNA deep sequencing and examined using in situ hybridization (ISH) in OC. The role of circPLPP4 in CDDP resistance was assessed by clone formation and Annexin V assays in vitro, and by OC patient-derived xenografts and intraperitoneal tumor models in vivo. The mechanism underlying circPLPP4-mediated activation of miR-136/PIK3R1 signaling was examined by luciferase reporter assay, RNA pull-down, RIP, MeRIP and ISH. RESULTS: circPLPP4 was remarkably upregulated in platinum resistant OC. circPLPP4 overexpression significantly enhanced, whereas circPLPP4 silencing reduced, OC cell chemoresistance. Mechanistically, circPLPP4 acts as a microRNA sponge to sequester miR-136, thus competitively upregulating PIK3R1 expression and conferring CDDP resistance. The increased circPLPP4 level in CDDP-resistant cells was caused by increased RNA stability, mediated by increased N6-methyladenosine (m6A) modification of circPLPP4. In vivo delivery of an antisense oligonucleotide targeting circPLPP4 significantly enhanced CDDP efficacy in a tumor model. CONCLUSIONS: Our study reveals a plausible mechanism by which the m6A -induced circPLPP4/ miR-136/ PIK3R1 axis mediated CDDP resistance in OC, suggesting that circPLPP4 may serve as a promising therapeutic target against CDDP resistant OC. A circPLPP4-targeted drug in combination with CDDP might represent a rational regimen in OC.

A deep learning-based semiautomated workflow for triaging follow-up MR scans in treated nasopharyngeal carcinoma.

It is imperative to optimally utilize virtues and obviate defects of fully automated analysis and expert knowledge in new paradigms of healthcare. We present a deep learning-based semiautomated workflow (RAINMAN) with 12,809 follow-up scans among 2,172 patients with treated nasopharyngeal carcinoma from three centers (ChiCTR.org.cn, Chi-CTR2200056595). A boost of diagnostic performance and reduced workload was observed in RAINMAN compared with the original manual interpretations (internal vs. external: sensitivity, 2.5% [p = 0.500] vs. 3.2% [p = 0.031]; specificity, 2.9% [p < 0.001] vs. 0.3% [p = 0.302]; workload reduction, 79.3% vs. 76.2%). The workflow also yielded a triaging performance of 83.6%, with increases of 1.5% in sensitivity (p = 1.000) and 0.6%-1.3% (all p < 0.05) in specificity compared to three radiologists in the reader study. The semiautomated workflow shows its unique superiority in reducing radiologist's workload by eliminating negative scans while retaining the diagnostic performance of radiologists.

Does dual-layer spectral detector CT provide added value in predicting spread through air spaces in lung adenocarcinoma? A preliminary study.

OBJECTIVES: To examine the predictive value of dual-layer spectral detector CT (DLCT) for spread through air spaces (STAS) in clinical lung adenocarcinoma. METHODS: A total of 225 lung adenocarcinoma cases were retrospectively reviewed for demographic, clinical, pathological, traditional CT, and spectral parameters. Multivariable logistic regression analysis was carried out based on three logistic models, including a model using traditional CT features (traditional model), a model using spectral parameters (spectral model), and an integrated model combining traditional CT and spectral parameters (integrated model). Receiver operating characteristic (ROC) curve analysis and decision curve analysis (DCA) were performed to assess these models. RESULTS: Univariable analysis showed significant differences between the STAS and non-STAS groups in traditional CT features, including nodule density (p < 0.001), pleural indentation types (p = 0.006), air-bronchogram sign (p = 0.031), the presence of spiculation (p < 0.001), long-axis diameter of the entire nodule (LD) (p < 0.001), and consolidation/tumor ratio (CTR) (p < 0.001). Multivariable analysis revealed that LD > 20 mm (odds ratio [OR] = 2.271, p = 0.025) and CTR (OR = 24.208, p < 0.001) were independent predictors in the traditional model, while electronic density (ED) in the venous phase was an independent predictor in the spectral (OR = 1.062, p < 0.001) and integrated (OR = 1.055, p < 0.001) models. The area under the curve (AUC) for the integrated model (0.84) was the highest (spectral model, 0.83; traditional model, 0.80), and the difference between the integrated and traditional models was statistically significant (p = 0.015). DCA showed that the integrated model had superior clinical value versus the traditional model. CONCLUSIONS: DLCT has added value for STAS prediction in lung adenocarcinoma. CLINICAL RELEVANCE STATEMENT: Spectral CT has added value for spread through air spaces prediction in lung adenocarcinoma so may impact treatment planning in the future. KEY POINTS: • Electronic density may be a potential spectral index for predicting spread through air spaces in lung adenocarcinoma. • A combination of spectral and traditional CT features enhances the performance of traditional CT for predicting spread through air spaces.

Radiomic signatures reveal multiscale intratumor heterogeneity associated with tissue tolerance and survival in re-irradiated nasopharyngeal carcinoma: a multicenter study.

BACKGROUND: Post-radiation nasopharyngeal necrosis (PRNN) is a severe adverse event following re-radiotherapy for patients with locally recurrent nasopharyngeal carcinoma (LRNPC) and associated with decreased survival. Biological heterogeneity in recurrent tumors contributes to the different risks of PRNN. Radiomics can be used to mine high-throughput non-invasive image features to predict clinical outcomes and capture underlying biological functions. We aimed to develop a radiogenomic signature for the pre-treatment prediction of PRNN to guide re-radiotherapy in patients with LRNPC. METHODS: This multicenter study included 761 re-irradiated patients with LRNPC at four centers in NPC endemic area and divided them into training, internal validation, and external validation cohorts. We built a machine learning (random forest) radiomic signature based on the pre-treatment multiparametric magnetic resonance images for predicting PRNN following re-radiotherapy. We comprehensively assessed the performance of the radiomic signature. Transcriptomic sequencing and gene set enrichment analyses were conducted to identify the associated biological processes. RESULTS: The radiomic signature showed discrimination of 1-year PRNN in the training, internal validation, and external validation cohorts (area under the curve (AUC) 0.713-0.756). Stratified by a cutoff score of 0.735, patients with high-risk signature had higher incidences of PRNN than patients with low-risk signature (1-year PRNN rates 42.2-62.5% vs. 16.3-18.8%, P < 0.001). The signature significantly outperformed the clinical model (P < 0.05) and was generalizable across different centers, imaging parameters, and patient subgroups. The radiomic signature had prognostic value concerning its correlation with PRNN-related deaths (hazard ratio (HR) 3.07-6.75, P < 0.001) and all causes of deaths (HR 1.53-2.30, P < 0.01). Radiogenomics analyses revealed associations between the radiomic signature and signaling pathways involved in tissue fibrosis and vascularity. CONCLUSIONS: We present a radiomic signature for the individualized risk assessment of PRNN following re-radiotherapy, which may serve as a noninvasive radio-biomarker of radiation injury-associated processes and a useful clinical tool to personalize treatment recommendations for patients with LANPC.

Large-scale pancreatic cancer detection via non-contrast CT and deep learning.

Pancreatic ductal adenocarcinoma (PDAC), the most deadly solid malignancy, is typically detected late and at an inoperable stage. Early or incidental detection is associated with prolonged survival, but screening asymptomatic individuals for PDAC using a single test remains unfeasible due to the low prevalence and potential harms of false positives. Non-contrast computed tomography (CT), routinely performed for clinical indications, offers the potential for large-scale screening, however, identification of PDAC using non-contrast CT has long been considered impossible. Here, we develop a deep learning approach, pancreatic cancer detection with artificial intelligence (PANDA), that can detect and classify pancreatic lesions with high accuracy via non-contrast CT. PANDA is trained on a dataset of 3,208 patients from a single center. PANDA achieves an area under the receiver operating characteristic curve (AUC) of 0.986-0.996 for lesion detection in a multicenter validation involving 6,239 patients across 10 centers, outperforms the mean radiologist performance by 34.1% in sensitivity and 6.3% in specificity for PDAC identification, and achieves a sensitivity of 92.9% and specificity of 99.9% for lesion detection in a real-world multi-scenario validation consisting of 20,530 consecutive patients. Notably, PANDA utilized with non-contrast CT shows non-inferiority to radiology reports (using contrast-enhanced CT) in the differentiation of common pancreatic lesion subtypes. PANDA could potentially serve as a new tool for large-scale pancreatic cancer screening.

Machine learning radiomics of magnetic resonance imaging predicts recurrence-free survival after surgery and correlation of LncRNAs in patients with breast cancer: a multicenter cohort study.

BACKGROUND: Several studies have indicated that magnetic resonance imaging radiomics can predict survival in patients with breast cancer, but the potential biological underpinning remains indistinct. Herein, we aim to develop an interpretable deep-learning-based network for classifying recurrence risk and revealing the potential biological mechanisms. METHODS: In this multicenter study, 1113 nonmetastatic invasive breast cancer patients were included, and were divided into the training cohort (n = 698), the validation cohort (n = 171), and the testing cohort (n = 244). The Radiomic DeepSurv Net (RDeepNet) model was constructed using the Cox proportional hazards deep neural network DeepSurv for predicting individual recurrence risk. RNA-sequencing was performed to explore the association between radiomics and tumor microenvironment. Correlation and variance analyses were conducted to examine changes of radiomics among patients with different therapeutic responses and after neoadjuvant chemotherapy. The association and quantitative relation of radiomics and epigenetic molecular characteristics were further analyzed to reveal the mechanisms of radiomics. RESULTS: The RDeepNet model showed a significant association with recurrence-free survival (RFS) (HR 0.03, 95% CI 0.02-0.06, P < 0.001) and achieved AUCs of 0.98, 0.94, and 0.92 for 1-, 2-, and 3-year RFS, respectively. In the validation and testing cohorts, the RDeepNet model could also clarify patients into high- and low-risk groups, and demonstrated AUCs of 0.91 and 0.94 for 3-year RFS, respectively. Radiomic features displayed differential expression between the two risk groups. Furthermore, the generalizability of RDeepNet model was confirmed across different molecular subtypes and patient populations with different therapy regimens (All P < 0.001). The study also identified variations in radiomic features among patients with diverse therapeutic responses and after neoadjuvant chemotherapy. Importantly, a significant correlation between radiomics and long non-coding RNAs (lncRNAs) was discovered. A key lncRNA was found to be noninvasively quantified by a deep learning-based radiomics prediction model with AUCs of 0.79 in the training cohort and 0.77 in the testing cohort. CONCLUSIONS: This study demonstrates that machine learning radiomics of MRI can effectively predict RFS after surgery in patients with breast cancer, and highlights the feasibility of non-invasive quantification of lncRNAs using radiomics, which indicates the potential of radiomics in guiding treatment decisions.

Magnetic resonance radiographic features which might lead to misdiagnosis of muscle-invasive bladder cancer based on vesical imaging reporting and data system: the application experience of a single center.

Background: The Vesical Imaging Reporting and Data System (VI-RADS) has been widely used for diagnosing muscle-invasive bladder cancer (MIBC), yet instances of misdiagnosis persist. However, limited research discusses the factors affecting its accuracy. This study aimed to evaluate the diagnostic efficacy of the VI-RADS in our center and to preliminarily identify possible magnetic resonance imaging (MRI) characteristics of misdiagnosis. Methods: From January 2018 to February 2023, a consecutive series of 211 participants pathologically diagnosed with bladder cancer (BC) who underwent an MRI exam were retrospectively enrolled. MRI was interpreted by 2 radiologists with different levels of experience, the diagnostic performance was validated using the receiver operating characteristic (ROC) curve, and VI-RADS ≥4 was considered to indicate MIBC-positive status. The clinical and radiographic characteristics of the true-positive (TP), true-negative (TN), false-positive (FP), and false-negative (FN) groups were analyzed using Kruskal-Wallis test or Fisher exact test. Results: With VI-RADS ≥4 as the cutoff value, the area under the ROC curves (AUCs) were 0.951 (0.912-0.976) and 0.847 (0.791-0.893) for the more-experienced reader and less-experienced reader, respectively, with good interobserver agreement (κ=0.74105). The median tumor size in the TP (more experienced: 57 cases; less experienced: 44 cases) and FP (more experienced: 8 cases; less experienced: 9 cases) groups was larger than that in the TN (more experienced: 141 cases; less experienced: 139 cases) group for the more-experienced reader (TP: 28 mm; FP: 31 mm; TN: 19 mm; P<0.001 and P=0.031, respectively) and the less-experienced reader (TP: 31 mm; FP: 28 mm; TN: 19 mm; P<0.001 and P=0.042, respectively). The tumor base in the TP and FP groups was larger than that in the TN group for the more-experienced reader (TP: 37 mm; FP: 48 mm; TN: 15 mm; both P<0.001) and for the less-experienced reader (FP: 42 mm; FP: 36 mm; TN: 15 mm; P<0.001 and P=0.022, respectively). The median tumor base in the TP group was larger than that in the FN group for the less-experienced reader (TP: 42 mm; FN: 17 mm; P=0.004). Conclusions: We observed good to excellent AUCs with good interobserver agreement among radiologists with different levels of expertise using VI-RADS. Large tumor size and wide tumor base affected the accuracy of VI-RADS in MIBC diagnosis.

Magnetic Resonance Deep Learning Radiomic Model Based on Distinct Metastatic Vascular Patterns for Evaluating Recurrence-Free Survival in Hepatocellular Carcinoma.

BACKGROUND: The metastatic vascular patterns of hepatocellular carcinoma (HCC) are mainly microvascular invasion (MVI) and vessels encapsulating tumor clusters (VETC). However, most existing VETC-related radiological studies still focus on the prediction of VETC status. PURPOSE: This study aimed to build and compare VETC-MVI related models (clinical, radiomics, and deep learning) associated with recurrence-free survival of HCC patients. STUDY TYPE: Retrospective. POPULATION: 398 HCC patients (349 male, 49 female; median age 51.7 years, and age range: 22-80 years) who underwent resection from five hospitals in China. The patients were randomly divided into training cohort (n = 358) and test cohort (n = 40). FIELD STRENGTH/SEQUENCE: 3-T, pre-contrast T1-weighted imaging spoiled gradient recalled echo (T1WI SPGR), T2-weighted imaging fast spin echo (T2WI FSE), and contrast enhanced arterial phase (AP), delay phase (DP). ASSESSMENT: Two radiologists performed the segmentation of HCC on T1WI, T2WI, AP, and DP images, from which radiomic features were extracted. The RFS related clinical characteristics (VETC, MVI, Barcelona stage, tumor maximum diameter, and alpha fetoprotein) and radiomic features were used to build the clinical model, clinical-radiomic (CR) nomogram, deep learning model. The follow-up process was done 1 month after resection, and every 3 months subsequently. The RFS was defined as the date of resection to the date of recurrence confirmed by radiology or the last follow-up. Patients were followed up until December 31, 2022. STATISTICAL TESTS: Univariate COX regression, least absolute shrinkage and selection operator (LASSO), Kaplan-Meier curves, log-rank test, C-index, and area under the curve (AUC). P < 0.05 was considered statistically significant. RESULTS: The C-index of deep learning model achieved 0.830 in test cohort compared with CR nomogram (0.731), radiomic signature (0.707), and clinical model (0.702). The average RFS of the overall patients was 26.77 months (range 1-80 months). DATA CONCLUSION: MR deep learning model based on VETC and MVI provides a potential tool for survival assessment. EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 3.

Deep learning-based recurrence detector on magnetic resonance scans in nasopharyngeal carcinoma: A multicenter study.

OBJECTIVES: Accuracy in the detection of recurrent nasopharyngeal carcinoma (NPC) on follow-up magnetic resonance (MR) scans needs to be improved. MATERIAL AND METHODS: A total of 5 035 follow-up MR scans from 5 035 survivors with treated NPC between April 2007 and July 2020 were retrospectively collected from three cancer centers for developing and evaluating the deep learning (DL) model MODERN (MR-based Deep learning model for dEtecting Recurrent Nasopharyngeal carcinoma). In a reader study with 220 scans, the accuracy of two radiologists in detecting recurrence on scans with vs without MODERN was evaluated. The performance was measured using the area under the receiver operating characteristic curve (ROC-AUC) and accuracy with a 95% confidence interval (CI). RESULTS: MODERN exhibited sound performance in the validation cohort (internal: ROC-AUC, 0.88, 95% CI, 0.86-0.90; external 1: ROC-AUC, 0.88, 95% CI, 0.86-0.90; external 2: ROC-AUC, 0.85, 95% CI, 0.82-0.88). In a reader study, MODERN alone achieved reliable accuracy compared to that of radiologists (MODERN: 84.1%, 95% CI, 79.3%-88.9%; competent: 80.9%, 95% CI, 75.7%-86.1%, P < 0.001; expert: 85.9%, 95% CI, 81.3%-90.5%, P < 0.001). The accuracy of radiologists was boosted by the MODERN score (competent with MODERN score: 84.6%, 95% CI, 79.8%-89.3%, P < 0.001; expert with MODERN score: 87.7%, 95% CI, 83.4%-92.1%, P < 0.001). CONCLUSION: We developed a DL model for recurrence detection with reliable performance. Computer-human collaboration has the potential to refine the workflow in interpreting surveillant MR scans among patients with treated NPC.

Artificial intelligence aided precise detection of local recurrence on MRI for nasopharyngeal carcinoma: a multicenter cohort study.

Background: MRI is the routine examination to surveil the recurrence of nasopharyngeal carcinoma, but it has relatively lower sensitivity than PET/CT. We aimed to find if artificial intelligence (AI) could be competent pre-inspector for MRI radiologists and whether AI-aided MRI could perform better or even equal to PET/CT. Methods: This multicenter study enrolled 6916 patients from five hospitals between September 2009 and October 2020. A 2.5D convolutional neural network diagnostic model and a nnU-Net contouring model were developed in the training and test cohorts and used to independently predict and visualize the recurrence of patients in the internal and external validation cohorts. We evaluated the area under the ROC curve (AUC) of AI and compared AI with MRI and PET/CT in sensitivity and specificity using the McNemar test. The prospective cohort was randomized into the AI and non-AI groups, and their sensitivity and specificity were compared using the Chi-square test. Findings: The AI model achieved AUCs of 0.92 and 0.88 in the internal and external validation cohorts, corresponding to the sensitivity of 79.5% and 74.3% and specificity of 91.0% and 92.8%. It had comparable sensitivity to MRI (e.g., 74.3% vs. 74.7%, P = 0.89) but lower sensitivity than PET/CT (77.9% vs. 92.0%, P < 0.0001) at the same individual-specificities. The AI model achieved moderate precision with a median dice similarity coefficient of 0.67. AI-aided MRI improved specificity (92.5% vs. 85.0%, P = 0.034), equaled PET/CT in the internal validation subcohort, and increased sensitivity (81.9% vs. 70.8%, P = 0.021) in the external validation subcohort. In the prospective cohort of 1248 patients, the AI group had higher sensitivity than the non-AI group (78.6% vs. 67.3%, P = 0.23), albeit nonsignificant. In future randomized controlled trials, a sample size of 3943 patients in each arm would be required to demonstrate the statistically significant difference. Interpretation: The AI model equaled MRI by expert radiologists, and AI-aided MRI by expert radiologists equaled PET/CT. A larger randomized controlled trial is warranted to demonstrate the AI's benefit sufficiently. Funding: The Sun Yat-sen University Clinical Research 5010 Program (2015020), Guangdong Basic and Applied Basic Research Foundation (2022A1515110356), and Guangzhou Science and Technology Program (2023A04J1788).

Efficacy and safety of sintilimab plus bevacizumab in metastatic nasopharyngeal carcinoma after failure of platinum-based chemotherapy: an open-label phase 2 study.

Background: There are limited treatment options for patients with metastatic nasopharyngeal carcinoma (mNPC) after failure of platinum-based chemotherapy. In this trial, we assessed the efficacy and safety of sintilimab plus bevacizumab in patients with mNPC where platinum-based chemotherapy has been ineffective. Methods: This was a single-centre, open-label, single-arm, phase 2 trial in Guangzhou, China for patients with mNPC progressed after at least one line of systemic therapy. Eligible patients were between 18 and 75 years old, were histologically confirmed differentiated or undifferentiated non-keratinized NPC, were ineffective after platinum-based chemotherapy, and they had at least one measurable metastatic lesion assessed with Response Evaluation Criteria in Solid Tumors Version 1.1 (RECIST V.1.1) by investigators and unsuitable for local surgery or radiotherapy. Key exclusion criterion was previous treatment with anti-PD-1/PD-L1 antibodies plus anti-VEGF antibodies and high risk of hemorrhage or nasopharyngeal necrosis. Patients were enrolled and received sintilimab (200 mg) plus bevacizumab (7.5 mg/kg) intravenously every 3 weeks. Intention-to-treat population was included in primary endpoint analyses and safety analyses. The primary endpoint was objective response rate (ORR) assessed by investigators following the guidelines of RECIST V1.1. Key secondary endpoints were progression-free survival (PFS), overall survival (OS), duration of response (DOR), and safety. This trial is registered with ClinicalTrials.gov (NCT04872582). Findings: Between July 29, 2021 and August 16, 2022, 33 patients were enrolled. Median age was 46 years (range, 18-64 years), and 63.6% of patients had previously received two or more lines of chemotherapy for metastatic disease. Median follow-up was 7.6 months (range, 4.1-17.5 months). ORR was 54.5% (95% CI, 36.4-71.9%) with 3 complete responses (9.1%) and 15 partial responses (45.5%). Median PFS was 6.8 months (95% CI, 5.2 months to not estimable). Median DOR was 7.2 months (95% CI, 4.4 months to not estimable). Median OS was not reached. The most common potential immune-related adverse event (AE) was Grade 1-2 hypothyroidism (42.4%). Treatment-related grade 3 or 4 AEs occurred in 7 patients (21.2%), including nasal necrosis (3/33), hypertension (1/33), pruritus (1/33), total bilirubin increased (1/33) and anaphylactic shock (1/33). No treatment-related deaths and severe epistaxis occurred. Interpretation: This phase 2 trial showed that sintilimab plus bevacizumab demonstrated promising antitumour activity and manageable toxicities in patients with mNPC after failure of platinum-based chemotherapy. Further trials are warranted, and the detailed mechanisms need to be elucidated. Funding: The Guangdong Basic and Applied Basic Research Foundation, the National Natural Science Foundation of China, the Natural Science Foundation of Guangdong Province, and the Science and Technology Planning Project of International Cooperation of Guangdong Province.

An MRI-based scoring system for pretreatment risk stratification in locally advanced rectal cancer.

BACKGROUND: Accurately assessing the risk of recurrence in patients with locally advanced rectal cancer (LARC) before treatment is important for the development of treatment strategies. The purpose of this study is to develop an MRI-based scoring system to predict the risk of recurrence in patients with LARC. METHODS: This was a multicenter observational study that enrolled participants who underwent neoadjuvant chemoradiotherapy. To evaluate the risk of recurrence in these patients, we developed the mrDEC scoring system and assessed inter-reader agreement. Additionally, we plotted Kaplan-Meier curves to compare the 3-year disease-free survival (DFS) and 5-year overall survival (OS) rates among patients with different mrDEC scores. RESULTS: A total of 1287 patients with LARC were included in this study. We observed substantial inter-reader agreement for mrDEC. Based on the mrDEC scores ranging from 0 to 3, the patients were categorized into four groups. The 3-year DFS rates for the groups were 91.0%, 79.5%, 65.5%, and 44.0% (P < 0.0001), respectively, and the 5-year OS rates were 92.9%, 87.1%, 74.8%, and 44.5%, respectively (P < 0.0001). CONCLUSIONS: The mrDEC scoring system proved to be an effective tool for predicting the prognosis of patients with LARC and can assist clinicians in clinical decision-making.

Prognostic impact of deep learning-based quantification in clinical stage 0-I lung adenocarcinoma.

OBJECTIVES: To evaluate the performance of automatic deep learning (DL) algorithm for size, mass, and volume measurements in predicting prognosis of lung adenocarcinoma (LUAD) and compared with manual measurements. METHODS: A total of 542 patients with clinical stage 0-I peripheral LUAD and with preoperative CT data of 1-mm slice thickness were included. Maximal solid size on axial image (MSSA) was evaluated by two chest radiologists. MSSA, volume of solid component (SV), and mass of solid component (SM) were evaluated by DL. Consolidation-to-tumor ratios (CTRs) were calculated. For ground glass nodules (GGNs), solid parts were extracted with different density level thresholds. The prognosis prediction efficacy of DL was compared with that of manual measurements. Multivariate Cox proportional hazards model was used to find independent risk factors. RESULTS: The prognosis prediction efficacy of T-staging (TS) measured by radiologists was inferior to that of DL. For GGNs, MSSA-based CTR measured by radiologists (RMSSA%) could not stratify RFS and OS risk, whereas measured by DL using 0HU (2D-AIMSSA0HU%) could by using different cutoffs. SM and SV measured by DL using 0 HU (AISM0HU% and AISV0HU%) could effectively stratify the survival risk regardless of different cutoffs and were superior to 2D-AIMSSA0HU%. AISM0HU% and AISV0HU% were independent risk factors. CONCLUSION: DL algorithm can replace human for more accurate T-staging of LUAD. For GGNs, 2D-AIMSSA0HU% could predict prognosis rather than RMSSA%. The prediction efficacy of AISM0HU% and AISV0HU% was more accurate than of 2D-AIMSSA0HU% and both were independent risk factors. CLINICAL RELEVANCE STATEMENT: Deep learning algorithm could replace human for size measurements and could better stratify prognosis than manual measurements in patients with lung adenocarcinoma. KEY POINTS: • Deep learning (DL) algorithm could replace human for size measurements and could better stratify prognosis than manual measurements in patients with lung adenocarcinoma (LUAD). • For GGNs, maximal solid size on axial image (MSSA)-based consolidation-to-tumor ratio (CTR) measured by DL using 0 HU could stratify survival risk than that measured by radiologists. • The prediction efficacy of mass- and volume-based CTRs measured by DL using 0 HU was more accurate than of MSSA-based CTR and both were independent risk factors.

Preoperative diagnosis of metastatic lymph nodes by CT-histopathologic matching analysis in gastric adenocarcinoma using dual-layer spectral detector CT.

OBJECTIVES: There still remain challenges to accurate diagnosis of lymph node (LN) involvement in gastric cancer (GC) on conventional CT. This study evaluated the quantitative data derived from dual-layer spectral detector CT (DLCT) for preoperative diagnosis of metastatic LNs compared to conventional CT images. METHODS: Patients with adenocarcinoma scheduled for gastrectomy were enrolled in this prospective study from July, 2021, to February, 2022. Regional LNs were labeled on preoperative DLCT. The LNs were located and matched using carbon nanoparticle solution during surgery according to their locations and anatomic landmarks on preoperative images. The matched LNs were randomly split into training and validation cohorts in a ratio of 2:1. The DLCT quantitative parameters in the training cohort were investigated using logistic regression models to identify independent predictors of metastatic LNs, and these predictors were subsequently applied to the validation cohort. Receiver operating characteristic curves were compared between the DLCT parameters and conventional CT images. RESULTS: Fifty-five patients were included in the study, with 267 successfully matched LNs (90 metastatic, 177 nonmetastatic). Independent predictors included arterial phase CT attenuation on 70-keV images, venous phase electron density, and clustered feature. These combination predictors had areas under the curve (AUC) of 0.855 and 0.907 in the training and validation cohorts, respectively. Compared to conventional CT criteria alone, the model had higher AUC and accuracy (0.741 vs. 0.907, 75.28% vs. 87.64%; p < 0.01) for LN diagnosis. CONCLUSION: Incorporating DLCT parameters improved preoperative diagnosis of LN metastasis in GC, increasing the accuracy of clinical N stage. CLINICAL RELEVANCE STATEMENT: Compared to conventional CT criteria, quantitative parameters from dual-layer spectral detector CT showed higher diagnostic efficacy for the preoperative diagnosis of lymph node metastases in gastric cancer, increasing the accuracy of clinical N stage. KEY POINTS: • Quantitative parameters from dual-layer spectral detector CT are useful for the preoperative diagnosis of lymph node metastases in gastric adenocarcinoma, increasing the accuracy of clinical N stage. • The values for metastatic lymph nodes are higher than those of nonmetastatic ones. The arterial phase of CT attenuation on 70-keV images, venous phase of electron density, and clustered feature independently predicted lymph node metastases. • Prediction model had area under the curve of 0.907, sensitivity of 81.82%, specificity of 91.07%, and accuracy of 87.64% for the preoperative diagnosis of lymph node metastasis.