Collaborative workflow between pathologists and deep learning for evaluation of tumor cellularity in lung adenocarcinoma

Owing to the high demand for molecular testing, the reporting of tumor cellularity in cancer samples has become a mandatory task for pathologists. However, the pathological estimation of tumor cellularity is often inaccurate. We developed a collaborative workflow between pathologists and artificial intelligence (AI) models to evaluate tumor cellularity in lung cancer samples and prospectively applied it to routine practice. We also developed a quantitative model that we validated and tested on retrospectively analyzed cases and ran the model prospectively in a collaborative workflow where pathologists could access the AI results and apply adjustments (Adjusted-Score). The Adjusted-Scores were validated by comparing them with the ground truth established by manual annotation of hematoxylin-eosin slides with reference to immunostains with thyroid transcription factor-1 and napsin A. For training, validation, retrospective testing, and prospective application of the model, we used 40, 10, 50, and 151 whole slide images, respectively. The sensitivity and specificity of tumor segmentation were 97% and 87%, and the accuracy of nuclei recognition was 99%. Pathologists altered the initial scores in 87% of the cases after referring to the AI results and found that the scores became more precise after collaborating with AI. For validation of Adjusted-Score, we found the Adjusted-Score was significantly closer to the ground truth than non-AI-aided estimates (p<0.05). Thus, an AI-based model was successfully implemented into the routine practice of pathological investigations. The proposed model for tumor cell counting efficiently supported the pathologists to improve the prediction of tumor cellularity for genetic tests.

Associations of visual paired associative learning task with global cognition and its potential usefulness as a screening tool for Alzheimer’s Dementia

ABSTRACT Objective: Appropriate screening is integral to the early diagnosis and management of Alzheimer’s Dementia (AD). The Paired Associates Learning (PAL) task is a digital cognitive task that is free of cultural, language, and educational biases. This study examined the association between the PAL task performance and global cognition and the usefulness of the PAL task as a screening tool for AD. Design: Cross-sectional. Setting: Academic hospital. Methods: Twenty-five participants with AD and 22 healthy comparators (HC) were included. The Cambridge Neuropsychological Test Automated Battery PAL task and the Montreal Cognitive Assessment (MoCA) were used to assess cognition. We assessed the relationship between the PAL task and MoCA performance using Pearson correlation and linear regression. We also examined the PAL task’s ability to distinguish between AD and HC participants using Receiver Operating Characteristic curve (ROC) analysis. Measurements: MoCA Total Score had a strong positive correlation with PAL Stages Completed score (r = 0.8, p < 0.001), and a strong negative correlation with PAL Total Errors (adjusted) score (r = −0.9, p < 0.001). Further, PAL Total Errors (adjusted) score predicted the MoCA Total Score (F (4, 46) = 37.2, p < 0.001). On ROC analysis, PAL Total Errors (adjusted) score cut-off of 54 errors had 92% sensitivity and 86% specificity to detect AD. Conclusions: Performance on the PAL task is highly associated with global cognition. Further, the PAL task can differentiate patients with AD from HCs with high sensitivity and specificity. Thus, the PAL task may hold potential usage as an easy-to-administer screening tool for AD.

T56. RISKY DECISION-MAKING IMPAIRMENT IN EARLY-STAGE PSYCHOTIC BIPOLAR DISORDER

Background Previous research suggests that bipolar disorder may be associated with increased risk-taking / impulsivity. Risky decision-making paradigm is an objective, performance-based measure which has been increasingly applied in bipolar disorder research examining. Nonetheless, literature focused only on chronically ill samples, with illness chronicity, clinical heterogeneity and prolonged medication exposure being potential confounding factors of study results. The current study aimed to explore whether patients with early-stage psychotic bipolar disorder (BDP) exhibit impaired risky decision-making relative to healthy controls, using a well-validated, widely-applied experimental paradigm of Balloon Analogue Risk Task (BART). Methods Thirty-nine patients with early-stage BDP (defined by having received psychiatric treatment for first-episode BDP within 3 years since service entry) and 36 demographically matched healthy controls were recruited. BART was administered to examine risky decision-making performance. Deliberative risky behavior was operationalized as the willingness to inflate balloons as each pump was accompanied by an extra point gained in the temporary repository or balloon explosion. Three performance-based indices (adjusted score, explosion rate and cumulative score) were derived and analyzed. Results There were no significant differences between patients and controls in age, gender and educational levels. Independent samples t-tests illustrated that patients had significantly lower adjusted score (t = -3.45, p = .001, d = .791), explosion rate (t = -2.75, p = .007, d = .631) and cumulative score (t = -3.07, p = .003, d = .714) in BART compared to controls. Similar findings were obtained when comparison analyses were restricted to patients who were treated with antipsychotic medications at the time of study assessment (n = 30). No significant correlations between BART performance-based indices and measures of clinical and treatment variables were found in patient sample. Discussion Our results demonstrated that early-stage BDP patients displayed suboptimal risky decision-making compared with controls. Abnormal risky decision-making observed in the euthymic state of patients in early stage of bipolar disorder suggests that such impairment might represent a trait factor in the disorder. Further prospective research is warranted to clarify the longitudinal course of risky decision-making impairment in bipolar disorder.

Synoptic Operative Reporting: Documentation of Quality of Care Data for Rectal Cancer Surgery

Operative reports can be used to evaluate quality of care indicators in surgical patients. This study evaluated documentation of preoperative and intraoperative quality of care indicators for rectal cancer surgery in synoptic reports and traditional dictated reports. Two surgeons independently reviewed 40 prospectively collected synoptic operative reports from rectal cancer cases and a case-matched historical cohort of 40 dictated reports. Rectal cancer–specific quality measures were scored in both report groups using two separate, previously validated checklists. Synoptic reports had significantly higher overall scores on both checklists 1 (mean adjusted score ± SD 76 ± 4 vs 41 ± 19, P < 0.01) and 2 (54 ± 3 vs 24 ± 11, P < 0.01; maximum score of 100 for both checklists). Synoptic reports scored significantly higher in reporting preoperative and intraoperative care indicators. Data were extracted quickly from synoptic reports (mean 3:46 vs 6:21, minutes:seconds to complete checklists, P < 0.05). Synoptic reports are associated with accurate documentation of quality of care data for rectal cancer surgery. Refining the synoptic templates used will further enhance the collection of quality indicators and reporting in complex oncologic procedures.

OS9.7 Telomere length, TERTp mutation and ALT status in adult diffuse gliomas

BACKGROUND The current classification of adult diffuse gliomas integrates two alternative telomere maintenance mechanisms: reactivation of telomerase activity by TERT promoter (TERTp) mutations or ATRX mutations associated with alternative length telomere (ALT). We investigated here the relation between these two mechanisms, telomere length, and outcome in a large series of diffuse gliomas. MATERIAL AND METHODS We performed C-circle assay (CCA) to determine ALT status, determined telomere length in tumor (RTLt) and leukocyte (RTLl) in a cohort of 354 adult diffuse gliomas, and sequenced ATRX gene. We calculated an age-adjusted telomere score considering tumor and leukocyte (blood) telomere length and corrected by age. This score was used in univariate and multivariate survival analyses to evaluate the potential impact of telomere length on the prognosis of gliomas. We used the TCGA LGG-GBM dataset to validate our findings in an independent cohort. RESULTS RTLl and RTLt were associated with ATRX mutation and ALT phenotype, and negatively associated with age and TERTp mutations. ATRX mutations (found in 52% (64/123) of samples) were mostly transitions (C>T or T>C), and were associated with ALT phenotype. None of 1p/19q co-deleted oligodendrogliomas harbored an ALT phenotype. No patients with TERTp mutations had ALT phenotype except for a very small subgroup of patients (3/87, 3.4%) suggesting that multiple ways of telomere maintenance, may co-exist in a single tumor, probably expressed in different clones. Telomere age-adjusted score was independently associated with better outcome (HR= 0.73 [95% CI 0.56–0.97], p-value 0.03 adjusted for age, TERTp mutation, IDH mutation, 1p/19q co-deletion and WHO grade). These results were validated using the LGG-GBM TCGA dataset. CONCLUSION We unravel the relation between RTLl and RTLt, TERTp mutation and ALT phenotype and describe a novel telomere age-adjusted score independently associated with better prognosis in adult diffuse gliomas.

Adjustment of the multi-biomarker disease activity score to account for age, sex and adiposity in patients with rheumatoid arthritis

Objective To develop and evaluate an adjusted score for the multi-biomarker disease activity (MBDA) test to account for the effects of age, sex and adiposity in patients with RA. Methods Two models were developed to adjust MBDA score for age, sex and adiposity, using either serum leptin concentration or BMI as proxies for adiposity. Two cohorts were studied. A cohort of 325 781 RA patients who had undergone commercial MBDA testing and had data for age, sex and serum leptin concentration was used for both models. A cohort of 1411 patients from five studies/registries with BMI data was used only for the BMI-adjusted MBDA score. Univariate and multivariate linear regression analyses evaluated the adjusted MBDA scores and conventional clinical measures as predictors of radiographic progression, assessed in terms of modified total Sharp score (ΔmTSS). Results Two models were developed, based on findings that MBDA score was higher in females than males and increased with age, leptin concentration and BMI. In pairwise regression analyses, the leptin-adjusted (P = 0.00066) and BMI-adjusted (P = 0.0027) MBDA scores were significant independent predictors of ΔmTSS after adjusting for DAS28-CRP, whereas DAS28-CRP was not, after adjusting for leptin-adjusted (P = 0.74) or BMI-adjusted (P = 0.87) MBDA score. Moreover, the leptin-adjusted MBDA score was a significant predictor of ΔmTSS after adjusting for the BMI-adjusted MBDA score (P = 0.025) or the original MBDA score (0.027), whereas the opposite was not true. Conclusion Leptin-adjusted MBDA score significantly adds information to DAS28-CRP and the original MBDA score in predicting radiographic progression. It may offer improved clinical utility for personalized management of RA.