Receiver Operating Characteristic Comparison Research Articles

Baseline Frailty Measured by the Risk Analysis Index and 30-Day Mortality After Surgery for Spinal Malignancy: Analysis of a Prospective Registry (2011-2020).

To evaluate the prognostic utility of baseline frailty, measured by the Risk Analysis Index (RAI), for prediction of postoperative mortality among patients with spinal malignancy (SM) undergoing resection. SM surgery cases were queried from the American College of Surgeons - National Surgical Quality Improvement Program database (2011-2020). The relationship between preoperative RAI frailty score and increasing rate of primary endpoint (mortality or discharge to hospice within 30 days, "mortality/hospice") were assessed. Discriminatory accuracy was assessed by computation of C-statistics (with 95% confidence interval [CI]) in receiver operating characteristic (ROC) curve analysis. A total of 2,235 cases were stratified by RAI score: 0-20, 22.7%; 21-30, 11.9%; 31-40, 54.7%; and ≥ 41, 10.7%. The rate of mortality/hospice was 6.5%, which increased linearly with increasing RAI score (p < 0.001). RAI was also associated with increasing rates of major complication, extended length of stay, and nonhome discharge (all p < 0.05). The RAI demonstrated acceptable discriminatory accuracy for prediction of primary endpoint (C-statistic, 0.717; 95% CI, 0.697-0.735). In pairwise ROC comparison, RAI demonstrated superiority versus modified frailty index-5 and chronological age (p < 0.001). Preoperative frailty, as measured by RAI, is a robust predictor of mortality/ hospice after SM surgery. The frailty score may be applied in clinical settings using a user-friendly calculator, deployed here: https://nsgyfrailtyoutcomeslab.shinyapps.io/spinalMalignancyRAI/.

Predictors and Outcomes of Immune Effector Cell Associated Neurotoxicity Syndrome in Patients Receiving Chimeric Antigen Receptor T-Cell Therapy for Aggressive B-Cell Non-Hodgkin Lymphoma

BACKGROUND Chimeric antigen receptor (CAR) T-cell therapy has changed the treatment landscape for aggressive large B-cell non-Hodgkin lymphoma (B-NHL). Immune effector cell-associated neurotoxicity syndrome (ICANS) is recognized as an adverse event of special interest associated with CAR T-cell therapy. However, determinants of ICANS presentation and outcome remain poorly understood. In this multicenter retrospective study, we aimed to determine risk factors for ICANS and clinical outcomes in patients who develop ICANS. METHODS Patients aged ≥18 years with aggressive large B-NHL who underwent apheresis from May 2018 to January 2021 for commercial CAR T-cell therapy at eight academic US medical centers were identified from the Cell Therapy Consortium (CTC) registry. All centers were approved to administer axicabtagene ciloleucel (axi-cel) and tisagenlecleucel (tisa-cel) per provider discretion. ICANS and CRS were graded using ASTCT criteria and classified as severe if grade 3-5. Cox regression analysis and the Kaplan Meier method were used for survival outcomes. Laboratory values collected at lymphodepletion were used for the previously reported Endothelial Activation and Stress Index (EASIX) and modified EASIX (m-EASIX) score. Log transformation was applied to reduce skew. Predictors of severe ICANS were identified using logistic regression analysis. Best subset selection and receiver operating characteristic (ROC) analyses were applied to determine optimal predictors. RESULTS Baseline characteristics are summarized in Table 1. A total of 352 patients received axi-cel (n=203) and tisa-cel (n=149) with a median follow-up of 21.6 months. A total of 125 (35.5%) developed ICANS while 227 (69.8%) did not. Of those with ICANS, severe ICANS occurred in 75 (60%) and mild ICANS occurred in 50 (40%). The median age at apheresis was similar at 64 (range 22-89) compared to 62 (range 18-86) among patients with and without ICANS. The majority of patients who developed ICANS (87.2%) received axi-cel. CRS was more common in those with ICANS (94.4% vs. 63.3%). IPI grade 3 or higher was more common in patients who developed ICANS (59.9% vs. 47%). There were similar numbers of patients with bulky, primary refractory, and active CNS disease among patients with and without ICANS. Most patients with ICANS (82%) received steroids compared to those without ICANS (9.4%). Results of cox regression analyses are summarized in Table 2. The development of ICANS did not significantly impact PFS or OS nor did steroid administration. Exploratory analysis showed that severity of ICANS did not influence PFS or OS. Bulky disease (≥10 cm), LDH greater than the upper limit of normal, and peak ferritin greater than 5000 in the first 28-days post-infusion were associated with worse OS. There was a significant interaction term between ICANS and product for OS. We conducted an exploratory analysis stratified by product which showed that the development of ICANS with tisa-cel was associated with inferior OS (HR 3.65, CI: 1.35-9.91, p=0.011) but no difference in PFS. The development of ICANS with axi-cel had no impact on OS or PFS. Risk factors for ICANS evaluated as continuous variables at lymphodepletion included included ferritin (p=0.021), LDH (p=0.004), CRP (p<0.001), EASIX (p=0.004), and m-EASIX (p<0.001). Thrombocytopenia, neutropenia, and creatinine were not associated with severe ICANS. Best subset selection showed that higher CRP was the best model to predict severe ICANS. ROC comparison analysis showed no difference between CRP (AUC 0.64), EASIX (0.64), and m-EASIX (0.66). Post infusion, CRS (OR 2.27, CI: 1.46-3.08, p<0.001) and earlier onset of CRS (OR 0.83, CI: 0.71-0.94, p=0.005) were predictive of severe ICANS while time to steroid administration was not. CONCLUSION In this retrospective analysis, we showed that outcomes were similar in patients with aggressive large B-NHL treated with CAR T-cell therapy who did or did not develop ICANS regardless of severity. In stratified analysis, ICANS was associated with inferior OS in those who received tisa-cel but was limited by small sample size. Pre lymphodepletion CRP alone had the highest discriminatory ability to predict severe ICANS. After infusion, developing CRS and earlier onset of CRS were predictive of ICANS. Future studies should identify novel predictive scores for ICANS which may lead to strategies to prevent neurologic toxicity after CAR T-cell therapy. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Application of Logistic Regression and Decision Tree Models in the Prediction of Activities of Daily Living in Patients with Stroke.

An improvement in the activities of daily living (ADLs) is significantly related to the quality of life and prognoses of patients with stroke. However, the factors predicting significant improvement in ADL (SI-ADL) have not yet been clarified. Therefore, we sought to identify the key factors affecting SI-ADL in patients with stroke after rehabilitation therapy using both logistic regression modeling and decision tree modeling. We retrospectively collected and analyzed the clinical data of 190 patients with stroke who underwent rehabilitation therapy at our hospital between January 2020 and July 2020. General and rehabilitation therapy data were extracted, and the Barthel index (BI) score was used for outcome assessment. We defined SI-ADL as an improvement in the BI score by 15 points or more during hospitalization. Logistic regression and decision tree models were established to explore the SI-ADL predictors. We then used receiver operating characteristic (ROC) curves to compare the logistic regression and decision tree models. Univariate analysis revealed that compared with the non-SI-ADL group, the SI-ADL group showed a significantly shorter course of stroke, longer hospital stay, and higher rate of receiving occupational and speech therapies (all P < 0.05). Binary logistic regression analysis revealed the course of stroke at admission (odds ratio (OR) = 0.986, 95%confidence interval (CI) = 0.979–0.993; P < 0.001) and the length of hospital stay (OR = 1.030, 95%CI = 1.013–1.047; P =0.001) as the independent predictors of SI-ADL. ROC comparisons revealed no significant differences in the areas under the curves for the logistic regression and decision tree models (0.808 vs. 0.831; z = 0.977, P = 0.329). Both models identified the course of disease at admission and the length of hospital stay as key factors affecting SI-ADL. Early initiation of rehabilitation therapy is of immense importance for improving the ADLs in patients with stroke.

Radiomics Predicts for Distant Metastasis in Locally Advanced Human Papillomavirus-Positive Oropharyngeal Squamous Cell Carcinoma.

Simple SummaryThere is strong evidence that locally advanced human papillomavirus positive (HPV+) oropharyngeal squamous cell carcinoma (OPSCC) carries a significantly better prognosis than HPV negative OPSCC, suggesting the possibility of treatment de-escalation and, therefore, toxicity reduction in this patient population. The lack of success in clinical trials towards this end presses the need to risk stratify locally advanced HPV+ OPSCC patients who can safely have treatment de-escalated. The present study had recourse to radiomics for this purpose and showed that radiomics has the ability to discriminate patients with locally advanced HPV+ OPSCC who went on to develop distant metastasis after completion of definitive chemoradiation or radiation alone. The implications of this study aid in demonstrating the potential pivotal role of radiomics in predictive risk assessment and personalizing therapy for this patient population.(1) Background and purpose: clinical trials have unsuccessfully tried to de-escalate treatment in locally advanced human papillomavirus positive (HPV+) oropharyngeal squamous cell carcinoma (OPSCC) with the goal of reducing treatment toxicity. The aim of this study was to explore the role of radiomics for risk stratification in this patient population to guide treatment. (2) Methods: the study population consisted of 225 patients with locally advanced HPV+ OPSCC treated with curative-intent radiation or chemoradiation therapy. Appearance of distant metastasis was used as the endpoint event. Radiomics data were extracted from the gross tumor volumes (GTVs) identified on the planning CT, with gray level being discretized using three different bin widths (8, 16, and 32). The data extracted for the groups with and without distant metastasis were subsequently balanced using three different algorithms including synthetic minority over-sampling technique (SMOTE), adaptive synthetic sampling (ADASYN), and borderline SMOTE. From these different combinations, a total of nine radiomics datasets were derived. Top features that minimized redundancy while maximizing relevance to the endpoint were selected individually and collectively for the nine radiomics datasets to build support vector machine (SVM) based predictive classifiers. Performance of the developed classifiers was evaluated by receiver operating characteristic (ROC) curve analysis. (3) Results: of the 225 locally advanced HPV+ OPSCC patients being studied, 9.3% had developed distant metastases at last follow-up. SVM classifiers built for the nine radiomics dataset using either their own respective top features or the top consensus ones were all able to differentiate the two cohorts at a level of excellence or beyond, with ROC area under curve (AUC) ranging from 0.84 to 0.95 (median = 0.90). ROC comparisons further revealed that the majority of the built classifiers did not distinguish the two cohorts significantly better than each other. (4) Conclusions: radiomics demonstrated discriminative ability in distinguishing patients with locally advanced HPV+ OPSCC who went on to develop distant metastasis after completion of definitive chemoradiation or radiation alone and may serve to risk stratify this patient population with the purpose of guiding the appropriate therapy.

Attention Guided Lymph Node Malignancy Prediction in Head and Neck Cancer.

Accurate lymph node (LN) malignancy classification is essential for treatment target identification in head and neck cancer (HNC) radiation therapy. Given the constraints imposed by relatively small sample sizes in real-world medical applications, to classify LN malignancy status accurately, we proposed an attention-guided classification (AGC) scheme that (1) incorporates human knowledge (ie, LN contours) into model training to guide model's "learning" direction, alleviating the critical requirement of large training samples by deep learning approaches; and (2) does not require accurate delineation of LNs in the inference stage but can highlight the discriminative region nearby the LN, which is important for malignancy determination. In the proposed AGC scheme, there is an attention-guided convolutional neural network (agCNN) module, followed by a classification convolutional neural network (cCNN) module. The input of the proposed AGC scheme is a region of interest (ROI) containing the LN and its surrounding tissues. The agCNN is designed to find the discriminative region in the ROI, which outputs an activation map whose voxel values indicate the importance of the voxels in malignancy prediction. Through multiplying the activation map with the ROI, we obtain the input for the cCNN, which finally outputs the LN malignancy probability. To demonstrate the effectiveness of the proposed scheme, we performed experimental studies using positron emission tomography and contrast-enhanced computed tomography from 129 surgical HNC patients, including 791 LNs, with pathologic ground truth of malignancy status. To evaluate the performance, 5-folder cross validation was used. The sensitivity, specificity, accuracy, and area under the receiver operating characteristic (ROC) curve values obtained by the proposed AGC scheme were 0.91, 0.93, 0.92, and 0.98, respectively, significantly outperforming conventional convolutional neural network and radiomics approaches at a significance level of .05 under a paired ROC comparison statistical test. We developed an AGC scheme that can highlight the discriminative region in an image for LN malignancy prediction, outperforming a conventional radiomics method that requires accurate segmentation and a standard convolutional neural network model without involving segmentation.

Machine-learning algorithm that can improve the diagnostic accuracy of septic arthritis of the knee

PurposePrompt diagnosis and treatment of septic arthritis of the knee is crucial. Nevertheless, the quality of evidence for the diagnosis of septic arthritis is low. In this study, the authors developed a machine learning-based diagnostic algorithm for septic arthritis of the native knee using clinical data in an emergency department and validated its diagnostic accuracy.MethodsPatients (n = 326) who underwent synovial fluid analysis at the emergency department for suspected septic arthritis of the knee were enrolled. Septic arthritis was diagnosed in 164 of the patients (50.3%) using modified Newman criteria. Clinical characteristics of septic and inflammatory arthritis were compared. Area under the receiver-operating characteristic (ROC) curve (AUC) statistics was applied to evaluate the efficacy of each variable for the diagnosis of septic arthritis. The dataset was divided into independent training and test sets (comprising 80% and 20%, respectively, of the data). Supervised machine-learning techniques (random forest and eXtreme Gradient Boosting: XGBoost) were applied to develop a diagnostic model using the training dataset. The test dataset was subsequently used to validate the developed model. The ROC curves of the machine-learning model and each variable were compared.ResultsSynovial white blood cell (WBC) count was significantly higher in septic arthritis than in inflammatory arthritis in the multivariate analysis (P = 0.001). In the ROC comparison analysis, synovial WBC count yielded a significantly higher AUC than all other single variables (P = 0.002). The diagnostic model using the XGBoost algorithm yielded a higher AUC (0.831, 95% confidence interval 0.751–0.923) than synovial WBC count (0.740, 95% confidence interval 0.684–0.791; P = 0.033). The developed algorithm was deployed as a free access web-based application (www.septicknee.com).ConclusionThe diagnosis of septic arthritis of the knee might be improved using a machine learning-based prediction model.Level of evidenceDiagnostic study Level III (Case–control study).

Efficacy of high-resolution, 3-D diffusion-weighted imaging in the detection of breast cancer compared to dynamic contrast-enhanced magnetic resonance imaging.

PurposeTo evaluate the utility of high-resolution, 3-D diffusion-weighted imaging (DWI) in the detection of breast cancer and to compare the sensitivity, specificity, and area under the curves of DWI and dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI).Material and methodProspective IEC approved study included 131 breast lesions detected on mammography and breast ultrasound. Cases underwent MRI on a 3 Tesla scanner using a dedicated breast coil. T2WI, STIR, T1WI, and dynamic post contrast MR. DWI MRI with b value of 50, 800, and 1500 s/mm2. Post-processing data with apparent diffusion coefficient (ADC) calculations and kinetic curves were obtained. Characteristics for lesions were analysed as per ACR BI-RADS descriptors. Final histopathological diagnosis was considered as the standard of reference. c2 test, t-test, receiver operating characteristic (ROC) curve analysis, pairwise comparison of ROC curves, sensitivity, specificity, diagnostic accuracy, and area under the curve (AUC) were calculated.ResultsSixty-six (50.38%) malignant and 65 (59.62%) benign lesions were included in the study. The mean ADC of malignant lesions was 0.870 × 10–3 mm2/s and 1.637 × 10–3 mm2/s (p < 0.0001) for benign lesions. Sensitivity and specificity for DWI were 95.45% and 90.76%, respectively, and for DCE-MRI they were 96.97% and 87.69%, respectively. Positive predictive value (PPV) and negative predictive value (NPV) were obtained at 91.30% and 95.16%, respectively, in DWI while in DCE-MRI they were 88.88% and 96.61%, respectively. The AUC for ADC was 0.979. In ROC comparison of AUC for DWI 0.931 and for DCE-MRI 0.923, the difference between the areas was 0.00781 (p = 0.782).ConclusionsHigh-resolution DWI is a non-contrast MRI technique, which improves the lesion detection with diagnostic performance comparable to DCE-MRI and has potential as an adjunct with screening mammography.

Utility of T2 mapping in the staging of thyroid-associated ophthalmopathy: efficiency of region of interest selection methods.

Discriminating the stage of thyroid-associated ophthalmopathy (TAO) is crucial for the treatment strategy and prognosis prediction. Utility of conventional magnetic resonance imaging in the disease staging is limited. To investigate the performance of T2 mapping based on different region of interest (ROI) selection methods in the staging of TAO. Thirty-two patients with TAO were retrospectively enrolled. Two radiologists independently measured the T2 relaxation time (T2RT) of extraocular muscles using two different ROIs (hotspot [ROIHS]: T2RT-hot; single-slice [ROISS]: T2RT-mean, T2RT-max, T2RT-min). Independent-samples t test, Wilcoxon signed rank test, Spearman correlation analysis, receiver operating characteristic (ROC) curves analyses, multiple ROC comparisons, and intra-class correlation coefficient (ICC) were used for statistical analyses. No significant difference was found in the measuring time between ROIHS and ROISS methods (P = 0.066). T2RT-mean demonstrated the highest ICC for measurement, followed by T2RT-max and T2RT-min, and T2RT-hot showed the poorest reproducibility. Active TAOs showed significantly higher values for all the T2RTs than inactive mimics (all P < 0.001). Significant positive correlations were found between T2RTs and CAS (all P < 0.005). T2RT-hot and T2RT-max showed significantly higher areas under the curve than that of T2RT-mean (P = 0.013 and 0.024, respectively), while the difference between T2RT-hot and T2RT-max was not significant (P = 0.970). The T2RTs derived from both ROI selection methods could be useful for the staging of TAO. The results of measuring time, reproducibility, and diagnostic performance suggest that T2RT-max would be the optimal indicator for staging.

Quantitative computed tomography discriminates between postmenopausal women with low spine bone mineral density with vertebral fractures and those with low spine bone mineral density only: the SHATTER study.

We evaluated the ability of lumbar spine bone mineral apparent density (BMAD), trabecular bone score (TBS) and volumetric bone mineral density (vBMD) to discriminate between postmenopausal women with low areal bone mineral density (aBMD) by dual-energy X-ray absorptiometry (DXA) with and without vertebral fractures. The discriminatory ability of lumbar spine aBMD was compared with that of BMAD, TBS and vBMD. We studied three groups of postmenopausal women, i.e. group 1, aBMD T-score < - 1.0 and ≥ 1 vertebral fracture (n = 39); group 2, aBMD T-score < - 1.0 and no vertebral fracture, age- and aBMD-matched to group 1 (n = 34); group 3, aBMD score > - 1 and no vertebral fracture, age-matched to group 1 (n = 37). Lumbar spine aBMD was measured by DXA. BMAD was calculated using the DXA scan results. TBS was derived following DXA scan image reanalysis. Lumbar spine vBMD was assessed by quantitative computed tomography and Mindways Pro software. Differences in variables between groups 1, 2 and 3 were examined using general linear univariate modelling approaches. Area under the receiver operating characteristic (ROC) curve was calculated for BMAD, TBS and vBMD to determine the ability of lumbar spine measurement variables to discriminate between group 1 and group 2. A comparison of ROCs was performed. Lumbar spine BMAD and TBS measurement variables were similar for groups 1 and 2. However, vBMD was significantly lower in group 1 and could discriminate between those women with low aBMD with (group 1) and without vertebral fractures (group 2). We conclude that lumbar spine vBMD may discriminate well between postmenopausal women with low aBMD with and without vertebral fractures as it provides a 3D measure of bone mineral density, excludes the posterior elements of the vertebrae and takes into account bone size. A unique feature of the SHATTER study is that groups 1 and 2 were matched for aBMD, thus our study findings are independent of aBMD. Furthermore, we observed that neither BMAD nor TBS could distinguish between women with low aBMD with and without vertebral fractures. The knowledge gained from the SHATTER study will influence clinical and therapeutic decision-making, thereby optimising the care of patients with and without vertebral and other fragility fractures.

Probabilistic walking models using built environment and sociodemographic predictors

BackgroundIndividual sociodemographic and home neighborhood built environment (BE) factors influence the probability of engaging in health-enhancing levels of walking or moderate-to-vigorous physical activity (MVPA). Methods are needed to parsimoniously model the associations.MethodsParticipants included 2392 adults drawn from a community-based twin registry living in the Seattle region. Objective BE measures from four domains (regional context, neighborhood composition, destinations, transportation) were taken for neighborhood sizes of 833 and 1666 road network meters from home. Hosmer and Lemeshow’s methods served to fit logistic regression models of walking and MVPA outcomes using sociodemographic and BE predictors. Backward elimination identified variables included in final models, and comparison of receiver operating characteristic (ROC) curves determined model fit improvements.ResultsBuilt environment variables associated with physical activity were reduced from 86 to 5 or fewer. Sociodemographic and BE variables from all four BE domains were associated with activity outcomes but differed by activity type and neighborhood size. For the study population, ROC comparisons indicated that adding BE variables to a base model of sociodemographic factors did not improve the ability to predict walking or MVPA.ConclusionsUsing sociodemographic and built environment factors, the proposed approach can guide the estimation of activity prediction models for different activity types, neighborhood sizes, and discrete BE characteristics. Variables associated with walking and MVPA are population and neighborhood BE-specific.

Cytological Immunostaining of HMGA2, LRP1B, and TP63 as Potential Biomarkers for Triaging Human Papillomavirus-Positive Women

Background: Since human papillomavirus (HPV) DNA testing has been promoted as primary screening strategy, the triage method has also evolved from morphological testing to a molecular biomarker detection to improve screening efficiency. In this study, we investigated the performance of three HPV integration hot-spots, HMGA2, LRP1B, and TP63, as potential triage markers in HPV screening tests. Materials and Methods: This cross-sectional study was conducted from November 2016 to December 2017 in the First Affiliated Hospital of Sun Yat-sen University. Immunocytochemistry was carried out using residual cervical cell samples from 121 HPV-positive cases (23 normal, 24 cervical intraepithelial neoplasia (CIN) 1, and 74 CIN2+). Results: Of the 121 cases, 77 showed completely paired for the three biomarkers. In these 77 cases, receiver operating characteristic (ROC) analysis of HMGA2 showed the best potential for detecting CIN2+ among HPV+ cases (sensitivity 70%; specificity 91.89%; AUC 0.839). TP63 was second most effective biomarker (AUC 0.838; sensitivity 80%; specificity 81.08%). In contrast, LRP1B had the smallest AUC (0.801) among the three biomarkers but had the highest sensitivity (90%) and specificity (56.76%). To test the triage value of combining the three biomarkers, logistic regression was conducted followed by ROC comparison analysis. Promisingly, the combination of the three biomarkers gave the largest AUC of 0.951 with 92.5% sensitivity and 89.1% specificity (P < .0001 compared to liquid-based cytology test by Z-test). Conclusions: A combination of HMGA2, LRP1B, and TP63 as potential biomarkers may be useful for screening during triage of HPV-positive patients, particularly for detecting CIN2 + .

SelectMDx versus Prostate Health Index in the identification of high-grade prostate cancer.

30 Background: A major clinical challenge is to identify patients at increased risk of high-grade prostate cancer (PCa) (Gleason scores ≥ 7) before biopsy. Thus, we evaluated the clinical utility of SelectMDx and Prostate Health Index (phi) tests for diagnosis of high-grade PCa as compared with transperineal mapping biopsy (TMB) results, where an average of 80 prostate needle biopsies are taken every 5 mm for a diagnostic accuracy of 98%. Methods: 70 patients were selected who had TMB. They were evaluated with both SelectMDx and phi tests from before or after TMB; all had serum and post-digital rectal examination urine samples collected prior to treatment, stored in our biorepository. phi was evaluated using proPSA, free PSA and total PSA in serum. SelectMDx test measured mRNA levels of the HOXC6 and DLX1 in post-DRE urine. Published data shows that phi &lt;27 and SelectMDx score = 0% correlate with patients free of HG PCa. Test results were compared against TMB histopathology data. Multivariate logistic regression (MLS) analyses and receiver operating characteristic (ROC) curves were used to determine diagnostic accuracy. DeLong test was used to determine statistical significance of ROC curves. All analyses were done for diagnosis of any PCa and high-grade PCa. Results: TMB histopathology showed 17/70 patients with no PCa and 22/53 with high-grade PCa. The sensitivity, specificity, positive (PPV) and negative (NPV) predictive values of each test are shown in the table below. Pairwise ROC comparison showed no statistically significant difference in the area under the curve of diagnosing PCa (0.75 vs 0.65) and high-grade PCa (0.71 vs 0.81) by phi and SelectMDx tests respectively. MLS analyses showed phi was significantly better than SelectMDx for diagnosing PCa (β = 0.054; p=0.005) and SelectMDx was significantly better than phi for diagnosing high-grade PCa (β = 8.45; p=0.0002). Conclusions: With high sensitivity and NPV, SelectMDx test is more useful than phi for screening patients at risk of high-grade PCa prior to biopsy. [Table: see text]

Fair lineups are better than biased lineups and showups, but not because they increase underlying discriminability.

Receiver Operating Characteristic (ROC) analysis has recently come in vogue for assessing the underlying discriminability and the applied utility of lineup procedures. Two primary assumptions underlie recommendations that ROC analysis be used to assess the applied utility of lineup procedures: (a) ROC analysis of lineups measures underlying discriminability, and (b) the procedure that produces superior underlying discriminability produces superior applied utility. These same assumptions underlie a recently derived diagnostic-feature detection theory, a theory of discriminability, intended to explain recent patterns observed in ROC comparisons of lineups. We demonstrate, however, that these assumptions are incorrect when ROC analysis is applied to lineups. We also demonstrate that a structural phenomenon of lineups, differential filler siphoning, and not the psychological phenomenon of diagnostic-feature detection, explains why lineups are superior to showups and why fair lineups are superior to biased lineups. In the process of our proofs, we show that computational simulations have assumed, unrealistically, that all witnesses share exactly the same decision criteria. When criterial variance is included in computational models, differential filler siphoning emerges. The result proves dissociation between ROC curves and underlying discriminability: Higher ROC curves for lineups than for showups and for fair than for biased lineups despite no increase in underlying discriminability. (PsycINFO Database Record

Habitat selection following recent disturbance: model transferability with implications for management and conservation of moose (Alces alces)

Site-specific variation in relative habitat abundance and disturbance regimes may produce differences in habitat preferences of associated populations. An evaluation of the predictive power of habitat selection models across space would benefit our understanding of the reliability of models of selection and space use in predicting animal occurrence. We used presence–absence data from winter surveys of moose (Alces alces (L., 1758)) to estimate resource selection functions (RSFs) across two study sites using Far North Land Cover updated with recent disturbance from fire and timber harvest. Moose selected foraging habitat (e.g., deciduous land cover) and for increasing deciduous foliage cover (ΔNDVI, i.e., the difference in the normalized difference vegetation index). Snow depth negatively influenced habitat selection, likely due to increased predation risk and reduced movement and foraging efficiency. Models lost little predictive power when applied to another site based on comparison of receiver operating characteristic (ROC) curves. Our results corroborated the current body of knowledge concerning moose habitat selection, i.e., moose preferentially use forest stands dominated by deciduous species, but suggested that moose strongly avoided very recently disturbed areas. Minimal site-specific variation and ROC comparison suggests that RSFs may be extended into novel systems, given adequate consideration for habitat quality and abundance, thereby simplifying management needs of this important species.

Gray-Matter Volume Estimate Score: A Novel Semi-Automatic Method Measuring Early Ischemic Change on CT.

Background and PurposeWe developed a novel method named Gray-matter Volume Estimate Score (GRAVES), measuring early ischemic changes on Computed Tomography (CT) semi-automatically by computer software. This study aimed to compare GRAVES and Alberta Stroke Program Early CT Score (ASPECTS) with regards to outcome prediction and inter-rater agreement.MethodsThis was a retrospective cohort study. Among consecutive patients with ischemic stroke in the anterior circulation who received intra-arterial therapy (IAT), those with a readable pretreatment CT were included. Two stroke neurologists independently measured both the GRAVES and ASPECTS. GRAVES was defined as the percentage of estimated hypodense lesion in the gray matter of the ipsilateral hemisphere. Spearman correlation analysis, receiver operating characteristic (ROC) comparison test, and intra-class correlation coefficient (ICC) comparison tests were performed between GRAVES and ASPECTS.ResultsNinety-four subjects (age: 68.7±10.3; male: 54 [54.9%]) were enrolled. The mean GRAVES was 9.0±8.9 and the median ASPECTS was 8 (interquartile range, 6-9). Correlation between ASPECTS and GRAVES was good (Spearman’s rank correlation coefficient, 0.642; P<0.001). ROC comparison analysis showed that the predictive value of GRAVES for favorable outcome was not significantly different from that of ASPECTS (area under curve, 0.765 vs. 0.717; P=0.308). ICC comparison analysis revealed that inter-rater agreement of GRAVES was significantly better than that of ASPECTS (0.978 vs. 0.895; P<0.001).ConclusionsGRAVES had a good correlation with ASPECTS. GRAVES was as good as ASPECTS in predicting a favorable clinical outcome, but was better than ASPECTS regarding inter-rater agreement. GRAVES may be used to predict the outcome of IAT.

Voxel-based morphometry study of the insular cortex in female patients with current and remitted depression

ObjectiveWomen are more prone to major depressive disorders (MDDs) and the incidence of MDD in women is almost twice that of men. Insular cortex abnormalities are a common finding in neuroanatomical studies of patients with MDD. However, it remains largely unclear whether female MDD patients at different clinical stages show morphologic changes in a specific subregion of the insular cortex. Additionally, it is not understood if any subregion changes can be used as a state or trait marker of MDD, and whether the diagnostic performance of any marker is sufficient to identify MDD. MethodsNineteen right-handed current MDD (cMDD) female patients and 19 remitted MDD (rMDD) patients, as well as 19 healthy controls matched for age and educational level, were recruited into the study. By means of voxel-based morphometry (VBM), we investigated gray matter volume abnormalities in insular subregions among the three groups and further conducted region-of-interest (ROI)-based receiver operating characteristic (ROC) analyses. The data from these investigations were correlated with clinical data to confirm the effectiveness of the identified changes in the subregions in differentiating the three groups. ResultsBoth the cMDD and rMDD groups showed significantly decreased gray matter volumes in the left dorsal anterior insula compared to the healthy controls. The cMDD groups also showed decreased gray matter volumes in the right dorsal anterior insula relative to healthy controls. Further ROC comparisons demonstrated that the left dorsal anterior insula can effectively differentiate cMDD and rMDD groups from healthy controls. ConclusionsOur findings suggest that the volume changes in the left dorsal anterior insular cortex may be a trait-related marker of vulnerability to MDD and that the right dorsal anterior insular cortex may involve pathological changes of MDD.

Can Hybrid SPECT-CT Overcome the Limitations Associated With Poor Imaging Properties of 131I-MIBG?

This study aimed to evaluate the incremental value of (131)I-MIBG hybrid SPECT-CT over planar scintigraphy (PS) and SPECT alone in patients with clinical or biochemical suspicion of pheochromocytoma. A total of 126 adrenals of 63 patients (mean [SD] age, 28.6 [15.7] years; male patients, n = 34; female patients, n = 29) with clinical or biochemical suspicion of pheochromocytoma were retrospectively evaluated. All patients had undergone (131)I-MIBG SPECT-CT of adrenal region. The PS, SPECT, and SPECT-CT images were independently evaluated by 2 nuclear medicine physicians with 6 years (R1) and 2 years (R2) experience and in separate sessions 1 week apart. A scoring scale of 1 to 5 was used, in which 1 is definitely abnormal, 2 is probably abnormal, 3 is indeterminate, 4 is probably normal, and 5 is definitely normal. Sensitivity, specificity, predictive values were calculated taking a score 2 or less as abnormal. With receiver operating characteristic (ROC) curve analysis, areas under the curve (AUC) were calculated for each modality and compared. Histopathology and/or clinical/imaging follow-up were taken as reference standard. Of the 126 adrenals evaluated, 29 were indeterminate on PS for R1 and 48 for R2, 39 were indeterminate on SPECT for both, and on SPECT-CT, 1 was indeterminate for R1 and 2 for R2. SPECT-CT correctly characterized 28 of 29 indeterminate adrenals on PS and 37 of 39 indeterminate adrenals on SPECT for R1. Similarly, for R2, SPECT-CT correctly characterized 45 of 48 indeterminate adrenals on PS and 33 of 39 indeterminate adrenals on SPECT. On ROC comparison, PS was inferior to SPECT (P = 0.040 for R1; P < 0.001 for R2) and SPECT-CT (P = 0.001 for R1; P < 0.001 for R2) for both the observers. Moreover, SPECT was inferior to SPECT-CT for both the observers (P = 0.017 for R1 and P = 0.001 for R2). Accuracy of SPECT-CT (R1, 97.6%; R2, 97.6%) was higher than PS (R1, 91.2%; R2, 84.1%) and SPECT (R1, 94.4%; R2, 86.5%). Interobserver agreement was highest for SPECT-CT (κ = 0.966) as compared with PS (κ = 0.815) and SPECT (κ = 0.826). I-MIBG hybrid SPECT-CT shows high sensitivity and specificity for characterizing adrenal lesions in patients with clinical or biochemical suspicion of pheochromocytoma and is superior to PS and SPECT alone. It will be especially useful in countries where (123)I-MIBG is not available.

Gene panels to help identify subgroups at high and low risk of coronary heart disease among those randomized to antihypertensive treatment

To identify panels of genetic variants that predict treatment-related coronary heart disease (CHD) outcomes in hypertensive patients on one of four different classes of initial antihypertensive treatment. The goal was to identify subgroups of individuals on the basis of their genetic profile who benefit most from a particular treatment. Candidate genetic variants (n=78) were genotyped in 39 114 participants from Genetics of Hypertension Associated Treatment study, ancillary to Antihypertensive and Lipid Lowering Treatment to Prevent Heart Attack Trial. Antihypertensive and Lipid Lowering Treatment to Prevent Heart Attack Trial randomized hypertensive participants (≥55 years) to one of four treatments (amlodipine, chlorthalidone, doxazosin, lisinopril). The primary outcome was fatal CHD or nonfatal myocardial infarction (mean follow-up=4.9 years). A pharmacogenetic panel was derived within each of the four treatment groups. Receiver-operating characteristic (ROC) curves estimated the discrimination rate between those with and without a CHD event, on the basis of the addition of the genetic panel risk score. For each treatment group, we identified a panel of genetic variants that collectively improved the prediction of CHD to a small but statistically significant extent. Chlorthalidone (A): NOS3 rs3918226; SELE rs5361; ICAM1 rs1799969; AGT rs5051; GNAS rs7121; ROC comparison, P=0.004; Amlodipine (B): MMP1 rs1799750; Factor5 (F5) rs6025; NPPA rs5065; PDE4D rs6450512; MMP9 rs2274756; ROC comparison, P=0.006; Lisinopril (C): AGT rs5051; PON1 rs705379; MMP12 rs652438; F12 rs1801020; GP1BA rs6065; PDE4D rs27653; ROC comparison, P=0.01; Doxazosin (D): F2 rs1799963; PAI1 rs1799768; MMP7 rs11568818; AGT rs5051; ACE rs4343; MMP2 rs243865; ROC comparison, P=0.007. Each panel was tested for a pharmacogenetic effect; panels A, B, and D showed such evidence (P=0.009, 0.006, and 0.001, respectively) and panel C did not (P=0.09). Because each panel was associated with CHD in a specific treatment group but not the others, this research provides evidence that it may be possible to use gene panel scores as a tool to better assess antihypertensive treatment choices to reduce CHD risk in hypertensive individuals.

Detection of granularity in dermoscopy images of malignant melanoma using color and texture features

Granularity, also called peppering and multiple blue-grey dots, is defined as an accumulation of tiny, blue-grey granules in dermoscopy images. Granularity is most closely associated with a diagnosis of malignant melanoma. This study analyzes areas of granularity with color and texture measures to discriminate granularity in melanoma from similar areas in non-melanoma skin lesions. The granular areas in dermoscopy images of 74 melanomas and 14 melanomas in situ were identified and manually selected. For 200 non-melanoma dermoscopy images, those areas which most closely resembled granularity in color and texture were similarly selected. Ten texture and twenty-two color measures were studied. The texture measures consisted of the average and range of energy, inertia, correlation, inverse difference, and entropy. The color measures consisted of absolute and relative RGB averages, absolute and relative RGB chromaticity averages, absolute and relative G/B averages, CIE X, Y, Z, X/Y, X/Z and Y/Z averages, R variance, and luminance. These measures were calculated for each granular area of the melanomas and the comparable areas in the non-melanoma images. Receiver operating characteristic (ROC) curve analysis showed that the best separation of melanoma images from non-melanoma images by granular area features was obtained with a combination of color and texture measures. Comparison of ROC results showed greater separation of melanoma from benign lesions using relative color than using absolute color. Statistical analysis showed that the four most significant measures of granularity in melanoma are two color measures and two texture measures averaged over the spots: relative blue, relative green, texture correlation, and texture energy range. The best feature set, utilizing texture and relative color measures, achieved an accuracy of 96.4% based on area under the receiver operating characteristic curve.

Esophageal atresia: Prognostic classification revisited

Although the Spitz classification is the most widely used prognostic classification for esophageal atresia and/or tracheoesophageal fistula (EA), its discrimination ability remains unclear. We sought to develop a more accurate prognostic classification for EA. The records of 121 consecutive infants with EA (1980-2005) were reviewed. The independent variables included 6 clinical characteristics, and the dependent variables were survival and mortality. Stepwise logistic regression analysis was used to construct models predicting mortality and create a revised prognostic classification. The discrimination abilities of the revised classification and the Spitz classification were compared using receiver-operating characteristic (ROC) curves. Birth weight and the presence of major cardiac anomalies were significant prognostic factors for mortality, and major cardiac anomalies affected mortality more than birth weight. The ROC curve for birth weight suggested that 2,000 g was an appropriate cutoff point. The Spitz classification was revised as follows: the revised class I (low-risk group) consisted of patients without major cardiac anomalies and birth weight >2,000 g; class II (moderate-risk group) consisted of patients without major cardiac abnormalities and birth weight <2,000 g; class III (relatively high-risk group) consisted of patients with major cardiac anomalies and birth weight >2,000 g; and class IV (high-risk group) consisted of patients with major cardiac anomalies and birth weight <2,000 g. The ROC comparisons showed that the revised classification provided a significant improvement (P = .049). This revised classification can improve the stratification of EA patients and be a useful predictor of survival.