Specificity Threshold Research Articles

Optimizing Automated Hematoma Expansion Classification from Baseline and Follow-Up Head Computed Tomography

Hematoma expansion (HE) is an independent predictor of poor outcomes and a modifiable treatment target in intracerebral hemorrhage (ICH). Evaluating HE in large datasets requires segmentation of hematomas on admission and follow-up CT scans, a process that is time-consuming and labor-intensive in large-scale studies. Automated segmentation of hematomas can expedite this process; however, cumulative errors from segmentation on admission and follow-up scans can hamper accurate HE classification. In this study, we combined a tandem deep-learning classification model with automated segmentation to generate probability measures for false HE classifications. With this strategy, we can limit expert review of automated hematoma segmentations to a subset of the dataset, tailored to the research team’s preferred sensitivity or specificity thresholds and their tolerance for false-positive versus false-negative results. We utilized three separate multicentric cohorts for cross-validation/training, internal testing, and external validation (n = 2261) to develop and test a pipeline for automated hematoma segmentation and to generate ground truth binary HE annotations (≥3, ≥6, ≥9, and ≥12.5 mL). Applying a 95% sensitivity threshold for HE classification showed a practical and efficient strategy for HE annotation in large ICH datasets. This threshold excluded 47–88% of test-negative predictions from expert review of automated segmentations for different HE definitions, with less than 2% false-negative misclassification in both internal and external validation cohorts. Our pipeline offers a time-efficient and optimizable method for generating ground truth HE classifications in large ICH datasets, reducing the burden of expert review of automated hematoma segmentations while minimizing misclassification rate.

Abstract A056: Performance of multi-biomarker class reflex testing in a prospectively-collected cohort

Abstract Background: A multi-biomarker blood test has the potential to detect cancers in a broad range of organ types and stages. We previously trained and independently assessed the performance of two biomarker classes (methylation and protein; MP) for multi-cancer early detection (MCED) in a prospectively collected study (Cancer Res,2024;84(7Supp):LB100). Here, we assessed the addition of a somatic mutation reflex approach to samples with initially negative MP-results (MP-reflex; MP-r). Methods: This study included 3,160 samples (739 cancer and 2421 non- cancer) representing 20 organ types and all stages. Samples with negative MP results at a specificity threshold of ≥ 98.5% and detectable MP signals above pre-specified lower thresholds were analyzed for mutations using next generation sequencing. Classifier models and thresholds were developed using an independent training set. Sensitivity of the MP biomarker classifier was determined at a combined specificity of ≥98.5%. We also compared the performance of MP and MP-r configurations with prostate (n=52) and breast (n=88) organ types excluded. Results: At 98.5% specificity, the observed overall sensitivity for MP was 52.9%, with sensitivities of 15.3%, 39.8%, 71.2%, 87.2%, 26.7% and 37.5% for stages I, II, III, IV, I/II and unknown, respectively. At the same specificity of 98.5%, when an MP-r configuration was utilized, overall sensitivity was 55.8%, with sensitivities of 19.1%, 42.2%, 72.3%, 89.9%, 29.9% and 46.9% for stages I, II, III, IV, I/II and unknown, respectively. MP-r demonstrated absolute sensitivity increases of 3.8%, 2.5%, 1.1 %, 2.8%, 3.2%, 9.4%, and 2.8%, for stages I, II, III, IV, I/II, unknown, and overall, respectively. When breast and prostate cancers were excluded, test sensitivity for MP was 59.3%, with sensitivities of 17.2%, 51.9%, 76.4%, 88.3%, 31.9% and 41.4% for stages I, II, III, IV, I/II and unknown, respectively. When an MP-r configuration was utilized, overall sensitivity was 62.3% %, with sensitivities of 22.1%, 54.7%, 76.4%, 91.4%, 35.9% and 51.7% for stages I, II, III, IV, I/II and unknown, respectively. MP-r demonstrated absolute sensitivity increases of 4.8 %, 2.8%, 0.0%, 3.1%, 4.0%, 10.3%, and 3.0% for stages I, II, III, IV, I/II, unknown, and overall, respectively. Conclusions: Compared to MP alone (breast and prostate cancers excluded), an MP-r approach results in a relative sensitivity increase of 28% in Stage I cancers and 12.5% for Stage I/II cancers. MP-r is an efficient strategy to add a third biomarker class and enhance sensitivity of MP for the detection of early-stage cancer. Citation Format: Vladimir Gainullin, Jin Bae, Violeta Beleva Guthrie, Fanglei Zhuang, Chen Ji, Christopher Tyson, Mark Evans, Kevin Arvai, Melissa Gray, Madhav Kumar, Mael Manesse, Xi Chen, Philip Uren, Gustavo C. Cerqueira, Amin Mazloom, Jorge Garces, Tomasz M. Beer, Frank Diehl. Performance of multi-biomarker class reflex testing in a prospectively-collected cohort [abstract]. In: Proceedings of the AACR Special Conference: Liquid Biopsy: From Discovery to Clinical Implementation; 2024 Nov 13-16; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2024;30(21_Suppl):Abstract nr A056.

Deep Learning Algorithms for Breast Cancer Detection in a UK Screening Cohort: As Stand-alone Readers and Combined with Human Readers.

Background Deep learning (DL) algorithms have shown promising results in mammographic screening either compared to a single reader or, when deployed in conjunction with a human reader, compared with double reading. Purpose To externally validate the performance of three DL algorithms as mammographic screen readers in an independent UK data set. Materials and Methods Three commercial DL algorithms (DL-1, DL-2, and DL-3) were retrospectively investigated from January 2022 to June 2022 using consecutive full-field digital mammograms collected at two UK sites during 1 year (2017). Normal cases with 3-year follow-up and histopathologically proven cancer cases detected either at screening (that round or next) or within the 3-year interval were included. A preset specificity threshold equivalent to a single reader was applied. Performance was evaluated for stand-alone DL reading compared with single human reading, and for DL reading combined with human reading compared with double reading, using sensitivity and specificity as the primary metrics. P < .025 was considered to indicate statistical significance for noninferiority testing. Results A total of 26 722 cases (median patient age, 59.0 years [IQR, 54.0-63.0 years]) with mammograms acquired using machines from two vendors were included. Cases included 332 screen-detected, 174 interval, and 254 next-round cancers. Two of three stand-alone DL algorithms achieved noninferior sensitivity (DL-1: 64.8%, P < .001; DL-2: 56.7%, P = .03; DL-3: 58.9%, P < .001) compared with the single first reader (62.8%), and specificity was noninferior for DL-1 (92.8%; P < .001) and DL-2 (96.8%; P < .001) and superior for DL-3 (97.9%; P < .001) compared with the single first reader (96.5%). Combining the DL algorithms with human readers achieved noninferior sensitivity (67.0%, 65.6%, and 65.4% for DL-1, DL-2, and DL-3, respectively; P < .001 for all) compared with double reading (67.4%), and superior specificity (97.4%, 97.6%, and 97.6%; P < .001 for all) compared with double reading (97.1%). Conclusion Use of stand-alone DL algorithms in combination with a human reader could maintain screening accuracy while reducing workload. Published under a CC BY 4.0 license. Supplemental material is available for this article.

Energy spectrum computed tomography multi-parameter imaging in preoperative assessment of vascular and neuroinvasive status in gastric cancer.

Vascular and nerve infiltration are important indicators for the progression and prognosis of gastric cancer (GC), but traditional imaging methods have some limitations in preoperative evaluation. In recent years, energy spectrum computed tomography (CT) multiparameter imaging technology has been gradually applied in clinical practice because of its advantages in tissue contrast and lesion detail display. To explore and analyze the value of multiparameter energy spectrum CT imaging in the preoperative assessment of vascular invasion (LVI) and nerve invasion (PNI) in GC patients. Data from 62 patients with GC confirmed by pathology and accompanied by energy spectrum CT scanning at our hospital between September 2022 and September 2023, including 46 males and 16 females aged 36-71 (57.5 ± 9.1) years, were retrospectively collected. The patients were divided into a positive group (42 patients) and a negative group (20 patients) according to the presence of LVI/PNI. The CT values (CT40 keV, CT70 keV), iodine concentration (IC), and normalized IC (NIC) of lesions in the upper energy spectrum CT images of the arterial phase, venous phase, and delayed phase 40 and 70 keV were measured, and the slopes of the energy spectrum curves [K (40-70)] from 40 to 70 keV were calculated. Arterial phase combined parameter, venous phase combined parameters (VP-ALLs), and delayed phase association parameters were calculated for patients with late-stage disease. The differences in the energy spectrum parameters between the positive and negative groups were compared, receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC), sensitivity, specificity, and optimal threshold were calculated to measure the diagnostic efficiency of each parameter. In the delayed phase, the CT40 keV, CT70 keV, K (40-70), IC, NIC, and CT70 keV and the NIC in the upper arterial and venous phases of energy spectrum CT were greater in the LVI/PNI-positive group than in the LVI-negative group. The representative parameters for the arterial phase NIC were 0.14 ± 0.04 in the positive group and 0.12 ± 0.04 in the negative group. The venous phase NIC was 0.5 (0.5, 0.6) in the positive group and 0.4 (0.4, 0.5) in the negative group. Last, for the delayed phase NIC, it was 0.6 ± 0.1 in the positive group and 0.5 ± 0.1 in the negative group (all P values are less than 0.05). ROC curve analysis demonstrated that the diagnostic efficacy of each parameter during the venous stage was superior to that during the arterial and delayed stages. Furthermore, the diagnostic efficacy of the combined parameter throughout all three stages was superior to that of any single parameter. The AUC, sensitivity, and specificity of the optimal parameter, VP-ALL, were 0.931 (95% confidence interval: 0.872-0.990), 80.95%, and 95.00%, respectively. When assessing the condition of LVI and PNI (perineural invasion) in patients with GC prior to surgery, the ability to diagnose these conditions using venous stage parameters was superior to that using arterial stage and delayed stage parameters. Furthermore, the diagnostic accuracy of using a combination of parameters was better than that of using individual parameters alone.

Accuracy of an Artificial Intelligence System for Interval Breast Cancer Detection at Screening Mammography.

Background Artificial intelligence (AI) systems can be used to identify interval breast cancers, although the localizations are not always accurate. Purpose To evaluate AI localizations of interval cancers (ICs) on screening mammograms by IC category and histopathologic characteristics. Materials and Methods A screening mammography data set (median patient age, 57 years [IQR, 52-64 years]) that had been assessed by two human readers from January 2011 to December 2018 was retrospectively analyzed using a commercial AI system. The AI outputs were lesion locations (heatmaps) and the highest per-lesion risk score (range, 0-100) assigned to each case. AI heatmaps were considered false positive (FP) if they occurred on normal screening mammograms or on IC screening mammograms (ie, in patients subsequently diagnosed with IC) but outside the cancer boundary. A panel of consultant radiology experts classified ICs as normal or benign (true negative [TN]), uncertain (minimal signs of malignancy [MS]), or suspicious (false negative [FN]). Several specificity and sensitivity thresholds were applied. Mann-Whitney U tests, Kruskal-Wallis tests, and χ2 tests were used to compare groups. Results A total of 2052 screening mammograms (514 ICs and 1548 normal mammograms) were included. The median AI risk score was 50 (IQR, 32-82) for TN ICs, 76 (IQR, 41-90) for ICs with MS, and 89 (IQR, 81-95) for FN ICs (P = .005). Higher median AI scores were observed for invasive tumors (62 [IQR, 39-88]) than for noninvasive tumors (33 [IQR, 20-55]; P < .01) and for high-grade (grade 2-3) tumors (62 [IQR, 40-87]) than for low-grade (grade 0-1) tumors (45 [IQR, 26-81]; P = .02). At the 96% specificity threshold, the AI algorithm flagged 121 of 514 (23.5%) ICs and correctly localized the IC in 93 of 121 (76.9%) cases, with 48 FP heatmaps on the mammograms for ICs (rate, 0.093 per case) and 74 FP heatmaps on normal mammograms (rate, 0.048 per case). The AI algorithm correctly localized a lower proportion of TN ICs (54 of 427; 12.6%) than ICs with MS (35 of 76; 46%) and FN ICs (four of eight; 50% [95% CI: 13, 88]; P < .001). The AI algorithm localized a higher proportion of node-positive than node-negative cancers (P = .03). However, no evidence of a difference by cancer type (P = .09), grade (P = .27), or hormone receptor status (P = .12) was found. At 89.8% specificity and 79% sensitivity thresholds, AI detection increased to 181 (35.2%) and 256 (49.8%) of the 514 ICs, respectively, with FP heatmaps on 158 (10.2%) and 307 (19.8%) of the 1548 normal mammograms. Conclusion Use of a standalone AI system improved early cancer detection by correctly identifying some cancers missed by two human readers, with no differences based on histopathologic features except for node-positive cancers. © RSNA, 2024 Supplemental material is available for this article.

A prospective evaluation of the diagnostic accuracy of the point-of-care VISITECT CD4 Advanced Disease test in seven countries.

CD4 measurement is pivotal in the management of advanced HIV disease. VISITECT® CD4 Advanced Disease (AccuBio Limited, Alva, UK; VISITECT) is an instrument-free, point-of-care, semi-quantitative test allowing visual identification of a CD4 ≤200 cells/µl, or >200 cells/µl from finger-prick or venous blood. As part of a diagnostic accuracy study of FUJIFILM SILVAMP TB LAM (clinicaltrials.gov: NCT04089423), people living with HIV of ≥18 years old were prospectively recruited in seven countries from outpatient departments if a tuberculosis symptom was present, and from inpatient departments. Participants provided venous blood for CD4 measurement using flow cytometry (reference standard) and finger-prick blood for VISITECT (index text), performed at point-of-care. Sensitivity, specificity, and positive and negative predictive values of VISITECT to determine a CD4 ≤200 cells/µl were evaluated. Among 1604 participants, the median flow cytometry CD4 was 367 (IQR 128-626) cells/µl and 521 (32.5%) had a CD4 ≤200 cells/µl. VISITECT sensitivity was 92.7% (483/521, 95% CI 90.1-94.7%) and specificity was 61.4% (665/1083, 95% CI 58.4-64.3%). For participants with a CD4 between 0-100, 101-200, 201-300, 301-500, and >500 cells/µl, VISITECT misclassified 4.5% (95% CI 2.5-7.2%), 12.5 (95% CI 8.0-18.2%), 74.1% (95% CI 67.0-80.5%), 48.0% (95% CI 42.5-53.6%), and 22.6% (95% CI 19.3-26.3%), respectively. VISITECT's sensitivity, but not specificity, met the World Health Organization's minimal sensitivity and specificity threshold of 80% for point-of-care CD4 tests. VISITECT's quality needs to be assessed and its accuracy optimized. VISITECT´s utility as CD4 triage test should be investigated.

Comparative Performance of 4 Penicillin-Allergy Prediction Strategies in a Large Cohort

A safe and pragmatic guide for labelling and delabelling patients with suspected penicillin allergy is mandatory. To compare the performance of 4 penicillin-allergy prediction strategies in a large independent cohort. We conducted a retrospective study for subjects presenting between January 2014 and December 2021 at the University Hospital of Montpellier, with a history of hypersensitivity to penicillins. The outcome targeted by the study was a positive penicillin-allergy test. Of the 1,884 participants included, 382 (20.3%) had positive penicillin-allergy tests. The ENDA (European Network on Drug Allergy) and Blumenthal strategies yielded relatively high sensitivities and low specificities and, by design, did not misclassify any positive subjects with severe index reactions. The PEN-FAST <3 score had a negative predictive value of 90% (95% confidence interval [95% CI] 88%-91%), with a sensitivity of 66% (95% CI 62%-71%) and a specificity of 73% (95% CI 71%-75%), and incorrectly delabelled 18 subjects with anaphylaxis and 15 with other severe nonimmediate reactions. For the adapted Chiriac score, the specificity corresponding to 66% sensitivity was 73% (95% CI 70%-75%). Conversely, at a 73% specificity threshold, the sensitivity was 65% (95% CI, 61%-70%). Attempts to improve these prediction algorithms did not substantially enhance performance. The ENDA and Blumenthal strategies are safe for high-risk subjects, but their delabelling effectiveness is limited, leading to unnecessary avoidance. Conversely, the PEN-FAST and Chiriac scores are performant in delabelling, but more frequently misclassify high-risk subjects with positive penicillin-allergy tests. Selection of the most appropriate tool requires careful consideration of the target population and the desired goal.

A safe, cost-effective, and high-throughput SARS-CoV-2 antigen capture ELISA suitable for large-scale screening in low-resource settings

Diagnostics employing multiple modalities have been essential for controlling and managing COVID-19, caused by SARS-CoV-2. However, scaling up Reverse Transcription–Quantitative Polymerase Chain Reaction (RT-qPCR), the gold standard for SARS-CoV-2 detection, remains challenging in low and middle-income countries. Cost-effective and high-throughput alternatives like enzyme-linked immunosorbent assay (ELISA) could address this issue. We developed an in-house SARS-CoV-2 nucleocapsid capture ELISA, and validated on 271 nasopharyngeal swab samples from humans (n = 252), bovines (n = 10), and dogs (n = 9). This ELISA has a detection limit of 195 pg/100 µL of nucleocapsid protein and does not cross-react with related coronaviruses, ensuring high specificity to SARS-CoV-2. Diagnostic performance was evaluated using receiver operating characteristic curve analysis, showing a diagnostic sensitivity of 67.78 % and specificity of 100 %. Sensitivity improved to 74.32 % when excluding positive clinical samples with RT-qPCR Ct values > 25. Furthermore, inter-rater reliability analysis demonstrated substantial agreement (κ values = 0.73–0.80) with the VIRALDTECT II Multiplex RT-qPCR kit and perfect agreement with the CoVeasy™ COVID-19 rapid antigen self-test (κ values = 0.89–0.93). Our findings demonstrated that the in-house nucleocapsid capture ELISA is suitable for SARS-CoV-2 testing in humans and animals, meeting the necessary sensitivity and specificity thresholds for cost-effective, large-scale screening.

AI-based opportunistic quantitative image analysis of lung cancer screening CTs to reduce disparities in osteoporosis screening

Osteoporosis is underdiagnosed, especially in ethnic and racial minorities who are thought to be protected against bone loss, but often have worse outcomes after an osteoporotic fracture. We aimed to determine the prevalence of osteoporosis by opportunistic CT in patients who underwent lung cancer screening (LCS) using non-contrast CT in the Northeastern United States. Demographics including race and ethnicity were retrieved. We assessed trabecular bone and body composition using a fully-automated artificial intelligence algorithm. ROIs were placed at T12 vertebral body for attenuation measurements in Hounsfield Units (HU). Two validated thresholds were used to diagnose osteoporosis: high-sensitivity threshold (115–165 HU) and high specificity threshold (<115 HU). We performed descriptive statistics and ANOVA to compare differences across sex, race, ethnicity, and income class according to neighborhoods' mean household incomes. Forward stepwise regression modeling was used to determine body composition predictors of trabecular attenuation. We included 3708 patients (mean age 64 ± 7 years, 54 % males) who underwent LCS, had available demographic information and an evaluable CT for trabecular attenuation analysis. Using the high sensitivity threshold, osteoporosis was more prevalent in females (74 % vs. 65 % in males, p < 0.0001) and Whites (72 % vs 49 % non-Whites, p < 0.0001). However, osteoporosis was present across all races (38 % Black, 55 % Asian, 56 % Hispanic) and affected all income classes (69 %, 69 %, and 91 % in low, medium, and high-income class, respectively). High visceral/subcutaneous fat-ratio, aortic calcification, and hepatic steatosis were associated with low trabecular attenuation (p < 0.01), whereas muscle mass was positively associated with trabecular attenuation (p < 0.01). In conclusion, osteoporosis is prevalent across all races, income classes and both sexes in patients undergoing LCS. Opportunistic CT using a fully-automated algorithm and uniform imaging protocol is able to detect osteoporosis and body composition without additional testing or radiation. Early identification of patients traditionally thought to be at low risk for bone loss will allow for initiating appropriate treatment to prevent future fragility fractures. ClinicalTrials.gov identifierN/A

Prognostic significance of Pleural Fluid triglyceride levels based on a low-Fat Diet Management Strategy in patients with Chylothorax following pulmonary resection

BackgroundChylothorax is a postoperative complication in patients with lung cancer. Diet-control approaches have been the mainstay for managing this condition. However, a surgical intervention is needed for the patients if conservative treatment is ineffective. Because of the lack of accurate indicators to assess the prognosis of the postoperative complication at an early stage, the criteria of surgical treatment were not consistent.MethodsWe reviewed 2942 patients who underwent pulmonary resection and lymph node dissection for primary lung cancer at our hospital between March 2021 and December 2022. The prognostic implications of clinical indicators were assessed in patients with postoperative chylothorax who were managed with a low-fat diet. Binary logistic regression was used to explore the predictive value of these indicators for patient prognosis.ResultsPostoperative chylothorax occurred in 108 patients and 79 patients were treated with a low-fat diet management while 29 patients were managed with TPN. In contrast to drainage volume, the pleural effusion triglyceride level after 2 days of low-fat diet exhibited enhanced predictive efficacy in predicting patient prognosis. When the pleural fluid triglyceride level of 1.33 mmol/L was used as the diagnostic threshold for prognosis, the sensitivity and specificity reached 100% and 80.6%, respectively.ConclusionsThe pleural effusion triglyceride level after 2 days of low-fat diet can serve as a valuable prognostic indicator in patients undergoing lung surgery and experiencing chylothorax. This predictive approach will help thoracic surgeons to identify patients with poor prognosis in a timely manner and make decision to perform necessary surgical interventions.

Predicting Out-of-Hospital Cardiac Arrest in the General Population Using Electronic Health Records.

The majority of out-of-hospital cardiac arrests (OHCAs) occur among individuals in the general population, for whom there is no established strategy to identify risk. In this study, we assess the use of electronic health record (EHR) data to identify OHCA in the general population and define salient factors contributing to OHCA risk. The analytical cohort included 2366 individuals with OHCA and 23 660 age- and sex-matched controls receiving health care at the University of Washington. Comorbidities, electrocardiographic measures, vital signs, and medication prescription were abstracted from the EHR. The primary outcome was OHCA. Secondary outcomes included shockable and nonshockable OHCA. Model performance including area under the receiver operating characteristic curve and positive predictive value were assessed and adjusted for observed rate of OHCA across the health system. There were significant differences in demographic characteristics, vital signs, electrocardiographic measures, comorbidities, and medication distribution between individuals with OHCA and controls. In external validation, discrimination in machine learning models (area under the receiver operating characteristic curve 0.80-0.85) was superior to a baseline model with conventional cardiovascular risk factors (area under the receiver operating characteristic curve 0.66). At a specificity threshold of 99%, correcting for baseline OHCA incidence across the health system, positive predictive value was 2.5% to 3.1% in machine learning models compared with 0.8% for the baseline model. Longer corrected QT interval, substance abuse disorder, fluid and electrolyte disorder, alcohol abuse, and higher heart rate were identified as salient predictors of OHCA risk across all machine learning models. Established cardiovascular risk factors retained predictive importance for shockable OHCA, but demographic characteristics (minority race, single marital status) and noncardiovascular comorbidities (substance abuse disorder) also contributed to risk prediction. For nonshockable OHCA, a range of salient predictors, including comorbidities, habits, vital signs, demographic characteristics, and electrocardiographic measures, were identified. In a population-based case-control study, machine learning models incorporating readily available EHR data showed reasonable discrimination and risk enrichment for OHCA in the general population. Salient factors associated with OCHA risk were myriad across the cardiovascular and noncardiovascular spectrum. Public health and tailored strategies for OHCA prediction and prevention will require incorporation of this complexity.

Novel gene-specific Bayesian Gaussian mixture model to predict the missense variants pathogenicity of Sanfilippo syndrome.

MPS III is an autosomal recessive lysosomal storage disease caused mainly by missense variants in theNAGLU, GNS, HGSNAT, and SGSH genes. Thepathogenicity interpretation of missense variants is still challenging. We aimed to develop unsupervised clustering-based pathogenicity predictor scores using extracted features from eight in silico predictors to predict the impact of novel missense variants of Sanfilippo syndrome. The model was trained on a dataset consisting of 415 uncertain significant (VUS) missense NAGLU variants. Performance The SanfilippoPred tool was evaluated by validation and test datasets consisting of 197-labelled NAGLU missense variants, and its performance was compared versus individual pathogenicity predictors using receiver operating characteristic (ROC) analysis. Moreover, we tested theSanfilippoPred tool using extra-labelled 427 missense variants to assess its specificity and sensitivity threshold. Application of the trained machine learning (ML) model on the test dataset of labelled NAGLU missense variants showed that SanfilippoPred has an accuracy of 0.93 (0.86-0.97 at CI 95%), sensitivity of 0.93, and specificity of 0.92. The comparative performance of theSanfilippoPred showed better performance (AUC = 0.908) than the individual predictors SIFT (AUC = 0.756), Polyphen-2 (AUC = 0.788), CADD (AUC = 0.568), REVEL (AUC = 0.548), MetaLR (AUC = 0.751), and AlphMissense (AUC = 0.885). Using high-confidence labelled NAGLU variants, showed that SanfilippoPred has an85.7% sensitivity threshold. The poor correlation between the Sanfilippo syndrome phenotype and genotype represents a demand for a new tool to classify its missense variants. This study provides asignificant tool forpreventing themisinterpretation of missense variants of the Sanfilippo syndrome-relevant genes. Finally, it seems that ML-based pathogenicity predictors and Sanfilippo syndrome-specific prediction tools could be feasible andefficient pathogenicity predictors in the future.

Manometric esophagogastric junction barrier metrics as predictors of gastroesophageal reflux.

High-resolution manometry (HRM) tools, like esophagogastric junction contractile integral (EGJ-CI), assess EGJ barrier function. This study aimed to evaluate the relationships between manometric EGJ metrics with esophageal acid exposure. We conducted a retrospective review of 284 patients who underwent HRM and ambulatory reflux testing between 11/2017-1/2020. EGJ-CI and total-EGJ-CI were manually calculated. Pathologic acid exposure was defined as pH < 4 with esophageal acid exposure time (EAET) exceeding 6.0%. Pearson's correlation, univariable and multivariable regression models were utilized to assess the relationships between pathologic acid exposure and EGJ parameters. Sensitivity and specificity thresholds for EGJ-CI and total EGJ-CI were optimized with ROC analyses. On univariable analysis, patients with pathologic acid exposure had increased odds of having lower mean basal LES pressures, EGJ-CI, and total EGJ-CI than patients without pathologic acid exposure. On multivariable analysis, age, EGJ-CI and mean DCI were significant predictors of pathologic acid exposure. There were significant, though weak, correlations between EAET and EGJ-CI and total EGJ-CI (r = - 0.18, - 0.19, p < 0.01, respectively). An EGJ-CI cutoff of 44.16 as a predictor for pathologic acid exposure had a sensitivity of 46% and specificity of 42% (AUC 0.60). Total EGJ-CI cutoff of 11,461.3 for pathologic acid exposure had a sensitivity of 44% and a specificity of 43% (AUC 0.62). EGJ-CI can independently predict pathologic acid exposure. However, the poor correlation between EGJ-CI and acid exposure, as well as the low sensitivity and specificity of calculated thresholds, indicate that mechanisms other than EGJ barrier function may impact acid exposure.

Abstract LB100: Performance of a multi-analyte, multi-cancer early detection (MCED) blood test in a prospectively-collected cohort

Abstract Background: A multi-analyte blood test has potential for robust sensitivity in detecting a broad range of cancer types and stages. Previously, using retrospectively collected samples, we trained and independently assessed in a case-control study the performance of various biomarker classes for the detection of cancers (Douville, Annals of Oncology,2022:pS575). The aim of this study was to further assess two of the four original biomarker classes (methylation and protein) using samples from a large, multi-center, prospectively collected study: Ascertaining Serial Cancer patients to Enable New Diagnostic 2 (ASCEND 2). Methods: The study included 6,314 samples (1,426 cancer and 4,888 non-cancer) collected in LBgard® tubes. The training cohort included 3,026 samples (654 cancers and 2,372 non-cancer controls) and the test set included 3,288 samples (772 cancers and 2,516 non-cancer controls). By measuring biomarkers that capture shared cancer signals, the test is designed to detect a broad range of cancer types. In this analysis, cancer stages were well-represented across 21 solid and hematologic tumor organ sites that collectively represent &amp;gt;85% of incident cancers. We combined DNA methylation and protein biomarkers in a single overarching classifier. Machine learning models were calibrated using stratified cross-validation to reach optimal cancer detection at a combined specificity &amp;gt;99.0%. Results: The specificity threshold identified in training resulted in an overall test set sensitivity of 49.9% and 99.0% specificity, with sensitivities of 23.6% (n=364), 68.6% (n=191), 84.6% (n=182), and 40.0% (n=35) for stages I/II, III, IV, and unknown, respectively. We will also report the cross-validation results of the multi-analyte classifiers in this newly enrolled case-control study. Conclusions: In ASCEND2, a large prospectively collected case-control study that included &amp;gt;85% of cancer burden, new training data and updated models demonstrated robust performance of methylation and protein biomarkers. These results demonstrate reproducibility of the multi-biomarker class approach to cancer detection in the ASCEND 2 cohort and represent an evolution in cancer classifier design. Citation Format: Vladimir Gainullin, Hee Jung Hwang, Larson Hogstrom, Kevin Arvai, Amira Best, Melissa Gray, Madhav Kumar, Mael Manesse, Xi Chen, Philip Uren, Amin Mazloom, Gustavo Cerqueira, Jorge Garces, Tomasz M. Beer, Abigail Mcelhinny, Frank Diehl. Performance of a multi-analyte, multi-cancer early detection (MCED) blood test in a prospectively-collected cohort [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(7_Suppl):Abstract nr LB100.

Abstract 2427: The development of a tissue-agnostic genome-wide methylome enrichment MRD assay for applications across the cancer care continuum for head and neck malignancies

Abstract Background: Plasma-based tests to quantify circulating cell-free DNA cancer signal have emerged as viable applications across the cancer continuum, from early detection to optimal disease management. Here we demonstrate the feasibility of a tissue-agnostic, genome-wide methylome enrichment platform based on cell-free methylated DNA immunoprecipitation and high throughput sequencing (cfMEDIP-seq) for cancer detection, cancer signal quantification, and prognostication in head and neck cancer (HNC). Methods: Pre-treatment plasma samples from individuals with newly diagnosed stage I-IV HPV+ or HPV- HNC were analyzed with a bisulfite-free, non-degradative, genome-wide methylome enrichment platform using 5-10 ng of cell-free DNA. For cancer detection, a machine learning classifier used differentially methylated regions to distinguish cancers from non-cancer controls. The area under the receiver operating characteristic curve (AUC) and 95% confidence intervals were calculated. Cancer signals were quantified from average normalized counts across informative methylated regions and a 95% specificity threshold. For prognostication, events were defined as recurrence, progression, or death due to HNC, whichever occurred earliest. Time to event was compared for samples with cancer signal quantities above versus below the threshold. Post-treatment and longitudinal plasma samples from individuals with Stage I-IVB HNC (HPV+ and HPV- included) will be analyzed for recurrence prediction and detection of relapse. More than 100 patients and 300 samples will be analyzed. Results: For cancer detection, 92 pre-treatment plasma samples from HNC cases were distinguished from 674 controls with an AUC of 0.96 (0.94, 0.98). The AUC was 0.93 (0.86, 1.0) for Stage I, 0.93 (0.83, 1.0) for Stage II, 0.96 (0.94, 0.99) for Stage III, and 0.97 (0.96, 0.99) for Stage IV. For prognostication, 91 pre-treatment samples were included (7 stage I, 16 stage II, 23 stage III, 45 stage IV). Median follow-up time was 50.6 months with 27 events. Likelihood of recurrence or progression was significantly higher in samples with cancer signal above the threshold [hazard ratio 5.4 (95% CI 2.25, 12.95), log-rank P&amp;lt;0.001]. In the upcoming analysis, data will be reported on the ability to predict recurrence and relapse in post-treatment samples. Conclusions: The cfMeDIP-seq approach demonstrated robust detection of HNC, across all stages and subtypes, and the ability to predict recurrence and progression from pre-treatment samples. We will report training data with cross validation to predict recurrence and relapse using post-treatment and longitudinal sampling. Collectively, data from these studies indicate that genome wide methylome enrichment has multiple use cases across the care continuum for patients with HNC. Citation Format: Geoffrey Liu, Jun Min, Yarong Wang, Justin Burgener, Ben Brown, Karen Budhraja, Junjun Zhang, Owen Hall, Shu Yi Shen, Martha Pienkowski, Shao Hui Huang, Laurie Ailles, Katrina Rey-McIntyre, Jeremy B. Provance, Eduardo Sosa, Cynthia Frye, Scott Bratman, Brian Allen, Joshua T. Jones, Abel Licon, Jing Zhang, Anne-Renee Hartman, Daniel D. De Carvalho. The development of a tissue-agnostic genome-wide methylome enrichment MRD assay for applications across the cancer care continuum for head and neck malignancies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2427.

Abstract 4928: REFINE Method: Novel strategy for signal enhancement

Abstract Fragmentomics has emerged as a powerful tool in the detection of early cancer malignancies. One of the challenges for fragmentomics is the low signal to noise ratio in early stage cancer samples. Fragmentomic sequencing data is also inherently noisy and susceptible to batch effects. Common strategies to address these challenges include using normalization methods, such as frequency or Z-score normalization, to scale data. However, these existing methods struggle to differentiate true signals from noise and are susceptible to experimental bias. To address the need for a strategy that removes noise and batch effects, Genece Health is employing a novel normalization technique, the REFINE method, to improve the results of fragmentomic cancer detection. Whole genome sequencing (Illumina) fragmentomic data were generated at an average 2.5x coverage from 3,669 normal and 1,346 cancer samples. Six different cancer types were represented: colorectal, esophageal, liver, lung, ovarian, and pancreatic. Fragment end-motifs and fragment sizes (FEMS) data were then extracted from the sequence data. REFINE first establishes a baseline FEMS signal from a representative and random panel of normals (PoN) comprised of 100 healthy samples. This baseline, which represents the normal background FEMS noise in healthy donors, is subsequently decomposed using truncated singular value decomposition. Next, REFINE then eliminates this decomposed baseline FEMS signal from the raw FEMS data in all other samples using linear regression. A convolutional neural network (CNN) was trained (2,206 normal and 1,000 cancer samples) using frequency normalized data versus REFINE normalized data using 5-fold cross validation. The performances of both models were then assessed using the remaining 1,463 normal with 346 cancer samples. With a training dataset specificity threshold of 90%, REFINE significantly improves the sensitivity performance of the CNN model &amp;gt;30% (from 50.4% to 83.0%). REFINE also improves the auROC from 0.753 to 0.944. We also show that increasing the size of the PoN to 812 samples and including hyper-parameter optimizations further improves the sensitivity to 85.9% (with an auROC of 0.955). The REFINE method demonstrates a dramatic enhancement in the detection of ctDNA from fragmentomic data, overcoming the limitations of conventional normalization techniques. REFINE effectively enhances the signal to noise ratio, facilitating more accurate cancer detection. Future work will focus on validating the REFINE method across diverse datasets and exploring its application in other areas of oncology. Citation Format: Mengchi Wang, Jin Mo Ahn, Junnam Lee, Dasom Kim, Eun-Hae Cho, Byung In Lee, Andrew Carson. REFINE Method: Novel strategy for signal enhancement [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4928.

Effects of low-tube voltage coronary CT angiography on plaque and pericoronary fat assessment: intraindividual comparison.

To investigate the effects of low tube voltage on coronary plaques and pericoronary fat assessment, and to compare their extent among various levels of low voltage. Patients were recommended for high-pitch low-tube voltage coronary computed tomography angiography (CCTA), and they were included if they had poor image quality and were referred to a conventional CCTA. The patients were classified into a low-voltage group (with 70-kV, 80-kV, and 90-kV subgroups) and a conventional group (100/120 kV). Their total plaque and subcomponent volumes and pericoronary fat attenuation index (FAI) were measured. A total of 1002 image slices (from 65 patients and 74 plaques) were included, including 21, 31, 13, 4, and 61 patients in the 70-kV, 80-kV, 90-kV, 100-kV, and 120-kV groups respectively. The CT values of noncalcified plaques in the conventional and low-voltage groups were 54.6 ± 21.3 HU and 31.5 ± 22.6 HU, respectively (p < 0.05). Compared with the conventional group, the necrotic core and calcification volume were increased, and the fibrolipid volume, periplaque, and right coronary artery FAI were decreased in the low-voltage group and its subgroups (p < 0.001). The magnitude of changes in fibrous and calcification volumes increased in the 70-kV subgroup compared with that in the 90-kV subgroup (p < 0.05). Low tube voltages, particularly of 70 kV, have a significant effect on coronary plaque and FAI. The effect of low voltage on plaque composition is characterized by a polarization pattern, i.e., a decrease in fibrolipid (medium density) and an increase in necrotic core (low density) and calcification (high density). Our results highlight the comparability and repeatability of plaque and pericoronary fat assessments facilitated by the same or a similar tube voltage. It is necessary to carry out studies on the specificity threshold of low tube voltage at each level. • Low tube voltage had a significant effect on coronary plaque and pericoronary fat, particularly 70 kV. • The effect of low tube voltage on plaque composition shows the shift from medium-density mixed components to low- and high-density components. • It is necessary to correct the specificity threshold or attenuation difference for low tube voltage at each level.

Diagnostic utility of point-of-care ultrasound and optical coherence tomography for papilloedema in children: a prospective pilot study

Background/AimsPapilloedema is an important sign of serious neurological disease, but it can be difficult to detect on funduscopy. The purpose of this study was to determine the diagnostic accuracy of...

Methylation Biology of a Blood-Based MDS Risk Stratification Test

Background: Myelodysplastic syndromes (MDS) are a heterogeneous group of myeloid malignancies associated with a myriad of deleterious outcomes and a 5-year relative survival of 37%. MDS risk stratification is key for optimal treatment decisions. We previously demonstrated that methylation-based markers may help stratify MDS risk and achieved comparable performance relative to IPSS-R (the Revised International Prognostic Scoring System) by assessing the risk prediction ability of a classifier trained with features from either Bone Marrow (BM) whole-genome bisulfite sequencing (WGBS) or serum targeted methylation (TM) sequencing. Here, we elucidate the biological underpinnings of the methylation-based classifier and explore the survival stratification performance of our previously reported classifier. Methods: Utilizing datasets from BM WGBS and matched Serum TM sequencing of 127 patients with MDS (N=104) and secondary AML (N=23), we delineated the features used by the classifier by identifying differentially methylated regions (DMRs) that distinguished patients with overall survival of &amp;gt;3 years from those with &amp;lt;3 years. Our approach also included a functional enrichment and driver gene analysis associated with these DMRs, as well as cluster analysis across patients. We performed further evaluations of the survival outcomes of our BM WGBS and matched Serum TM prognostic classifier using concordance indices and log-rank tests. Results: Our analysis revealed 7,742 DMRs in BM WGBS and 14,093 DMRs in the Serum TM that significantly separated the long and short-term survival groups. Unsupervised hierarchical clustering of differential methylation patterns in both BM WGBS and Serum TM highlighted three distinctive subgroups of samples: short-term survivors, long-term survivors, and a group with intermediate survival outcomes. Notably, the short-term survivor group exhibited extensive hypermethylation across a majority of DMRs, which were linked to key biological pathways including calcium signaling, cAMP, MAPK, Rap1, and PI3K-Akt. Enrichment analysis of the BM WGBS DMRs between individuals with &amp;lt;3 and &amp;gt;3 years of survival underscored the differential hypermethylation of promoters, CpG islands, and CpG shores in association with worse survival outcomes. Our analysis of BM WGBS DMR driver genes via Hypergeometric Optimization of Motif EnRichment (HOMER) revealed potential contributors to differential survival outcomes, including known genes associated with MDS/AML, such as EWS/FLI-1, ERG, Pu.1, and RUNX1. Both pathway and driver gene analysis recapitulated previously known drivers of MDS to AML progression. The methylation classifier achieved comparable risk stratification performance for both BM WGBS comparing with IPSS-R (C-index: 0.69 [95% CI: 0.61-0.77] for methylation vs 0.71 [95% CI: 0.64-0.77] for IPSS-R) and Serum TM (C-index: 0.68 [95% CI: 0.60-0.75] for methylation vs. 0.70 [95% CI: 0.63-0.78] for IPSS-R). At a specificity threshold of 0.8, both BM WGBS and Serum TM prognostic classifiers were able to stratify between predicted short and predicted long survivors (BM WGBS HR: 2.83, p&amp;lt;0.0001; Serum TM HR: 3.01, p=0.0008). Conclusions: Our findings generated biological insights into the methylation-based risk stratification of MDS, shedding light on the specific genomic features and biological pathways that underlie these methylation patterns. We also demonstrated that methylation-based MDS risk stratification performs comparably with IPSS-R. These insights not only expand our understanding of the complexity of MDS, but also reinforce the potential of methylation-based sequencing as a viable, possibly less-invasive, alternative for MDS risk stratification compared to traditional IPSS-R methods.

Implementation, Evolution, and Laboratory Performance of Methods-Based Proficiency Testing for Next-Generation Sequencing Detection of Germline Sequence Variants.

Next-generation sequencing (NGS)-based assays are used for diagnosis of diverse inherited disorders. Limited data are available pertaining to interlaboratory analytical performance of these assays. To report on the College of American Pathologists (CAP) NGS Germline Program, which is methods based, and explore the evolution in laboratory testing practices. Results from the NGS Germline Program from 2016-2020 were analyzed for interlaboratory analytical performance. Self-reported laboratory testing practices were also evaluated. From 2016-2020, a total of 297 laboratories participated in at least 1 program mailing. Of the 289 laboratories that provided information on tests offered, 138 (47.8%) offered only panel testing throughout their enrollment, while 35 (12.1%) offered panels and exome testing, 30 (10.4%) offered only exomes, 9 (3.1%) offered only genomes, and 15 (5.2%) offered panels, exomes, and genomes. The remainder (62 laboratories, 21.4%) changed their test offerings during the 2016-2020 timeframe. Considering each genomic position/interval, the median detection percentage at variant positions across the 2016-2020 mailings ranged from 94.3% to 100%, while at reference positions (no variant detected), the median correct response percentage was 100% across all mailings. When considering performance of individual laboratories, 89.5% (136 of 152) to 98.0% (149 of 152) of laboratories successfully met the detection threshold (≥90% of the variants present), while 94.6% (87 of 92) to 100% (163 of 163) of laboratories met the 95% specificity threshold across mailings. Since the inception of this program, laboratories have consistently performed well. The median sensitivity and specificity of detection of sequence variants included in this program (eg, single nucleotide variants, insertions, and deletions) were 100.0%.