Testing for improvement in prediction model performance

Neurofilament Light Chain and Risk of Stroke in Patients With Atrial Fibrillation

BACKGROUND

Biomarkers reflecting brain injury are not routinely used in risk assessment of stroke in atrial fibrillation (AF). Neurofilament light chain (NFL) is a novel biomarker released into blood after cerebral insults. We investigated the association between plasma concentrations of NFL, other biomarkers, and risk of stroke and death in patients with AF not receiving oral anticoagulation.

METHODS

For this observational study, baseline plasma samples were available from 3077 patients with AF randomized to aspirin in ACTIVE A (Atrial Fibrillation Clopidogrel Trial With Irbesartan for Prevention of Vascular Events; 2003 to 2008) and AVERROES (Apixaban Versus Acetylsalicylic Acid [ASA] to Prevent Stroke in Atrial Fibrillation Patients Who Have Failed or Are Unsuitable for Vitamin K Antagonist Treatment; 2007 to 2009). Median follow-up was 1.5 years. NFL was analyzed with a Single Molecule Array (Simoa). Associations with outcomes (total stroke or systemic embolism, ischemic stroke, cardiovascular death, and all-cause death) were explored with Cox regression models.

RESULTS

In the combined cohort, the median NFL level was 16.9 ng/L (interquartile range, 11.1-26.5 ng/L), the median age was 71 years, 58% were men, and 13% had a history of previous stroke. NFL was associated with older age, higher creatinine, lower body mass index, previous stroke, female sex, and diabetes but not cardiac rhythm. Higher NFL was associated with a higher risk of stroke or systemic embolism (n=206) independently of clinical characteristics (hazard ratio, 1.27 [95% CI, 1.10-1.46] per doubling of NFL) and other biomarkers (hazard ratio, 1.18 [95% CI, 1.01-1.37]) and including in patients without previous stroke (hazard ratio, 1.23 [95% CI, 1.02-1.48]). NFL was also independently associated with cardiovascular (n=219) and all-cause (n=311) death. The C index for stroke using only NFL was 0.642, on par with the currently used clinical risk scores. Addition of information on NFL improved discrimination in a model also including clinical information, NT-proBNP (N-terminal pro-B-type natriuretic peptide), and high-sensitivity cardiac troponin T, yielding a C index of 0.727.

CONCLUSIONS

NFL reflects overt and covert episodes of cerebral ischemia and improves risk assessment of stroke and death in patients with AF without oral anticoagulation, including in patients without previous stroke. The combination of NFL with information on age, history of stroke, and other biomarkers should be explored as a future avenue for stroke risk assessments in patients with AF.

Novel physiologic predictors of positive airway pressure effectiveness (NICEPAP) study: rationale, design and methods

PURPOSE

Continuous positive airway pressure (CPAP) is the primary therapy for obstructive sleep apnea (OSA); however the effectiveness of CPAP remains suboptimal. We describe the Novel PhysIologiC prEdictors of Positive Airway Pressure Effectiveness (NICEPAP) study. Its purpose is to determine whether physiological traits of OSA contribute to CPAP effectiveness.

METHODS

NICEPAP (NCT05067088) is a prospective, observational cohort study conducted at an academic sleep center. Adults newly diagnosed with OSA (n = 267) are assessed for OSA traits of loop gain, arousal threshold, pharyngeal collapsibility, and muscle compensation from baseline polysomnography. We perform a comprehensive assessment of covariates relevant to CPAP adherence, efficacy, and patient-centered outcomes. Participants are followed for 12 months. Primary outcomes include (1) CPAP adherence (hours/night), (2) CPAP efficacy (apneas-hypopneas/hour), and (3) quality of life at six months measured by objective CPAP data and Functional Outcomes of Sleep Questionnaire. Secondary outcomes include sleep quality, sleepiness, insomnia, and neurocognitive function.

RESULTS

Data on covariates, including demographics, sleep symptoms, medical history, medications, sleep quality, OSA and treatment self-efficacy, decisional balance, and socio-economic and social and partner support, are collected using validated instruments. The analysis for primary outcomes includes a generalized linear mixed model for an outcome (e.g., CPAP adherence) with OSA traits as exposures followed by the addition of relevant covariates.

CONCLUSION

The findings of the NICEPAP study will inform research aimed to enhance CPAP effectiveness. Understanding the role of physiological OSA traits in CPAP effectiveness is a crucial step toward a precision medicine approach to OSA.

Plasma proteomic signatures for type 2 diabetes mellitus and related traits in the UK Biobank cohort

Aims/hypothesis

The plasma proteome holds promise as a diagnostic and prognostic tool that can accurately reflect complex human traits and disease processes. We assessed the ability of plasma proteins to predict type 2 diabetes mellitus (T2DM) and related traits.

Methods

Clinical, genetic, and high-throughput proteomic data from three subcohorts of UK Biobank participants were analyzed for association with dual-energy x-ray absorptiometry (DXA) derived truncal fat (in the adiposity subcohort), estimated maximum oxygen consumption (VO2max) (in the fitness subcohort), and incident T2DM (in the T2DM subcohort). We used least absolute shrinkage and selection operator (LASSO) regression to assess the relative ability of non-proteomic and proteomic variables to associate with each trait by comparing variance explained (R2) and area under the curve (AUC) statistics between data types. Stability selection with randomized LASSO regression identified the most robustly associated proteins for each trait. The benefit of proteomic signatures (PSs) over QDiabetes, a T2DM clinical risk score, was evaluated through the derivation of delta (Δ) AUC values. We also assessed the incremental gain in model performance metrics using proteomic datasets with varying numbers of proteins. A series of two-sample Mendelian randomization (MR) analyses were conducted to identify potentially causal proteins for adiposity, fitness, and T2DM.

Results

Across all three subcohorts, the mean age was 56.7 years and 54.9% were female. In the T2DM subcohort, 5.8% developed incident T2DM over a median follow-up of 7.6 years. LASSO-derived PSs increased the R2 of truncal fat and VO2max over clinical and genetic factors by 0.074 and 0.057, respectively. We observed a similar improvement in T2DM prediction over the QDiabetes score [Δ AUC: 0.016 (95% CI 0.008, 0.024)] when using a robust PS derived strictly from the T2DM outcome versus a model further augmented with non-overlapping proteins associated with adiposity and fitness. A small number of proteins (29 for truncal adiposity, 18 for VO2max, and 26 for T2DM) identified by stability selection algorithms offered most of the improvement in prediction of each outcome. Filtered and clustered versions of the full proteomic dataset supplied by the UK Biobank (ranging between 600-1,500 proteins) performed comparably to the full dataset for T2DM prediction. Using MR, we identified 4 proteins as potentially causal for adiposity, 1 as potentially causal for fitness, and 4 as potentially causal for T2DM.

Conclusions/Interpretation

Plasma PSs modestly improve the prediction of incident T2DM over that possible with clinical and genetic factors. Further studies are warranted to better elucidate the clinical utility of these signatures in predicting the risk of T2DM over the standard practice of using the QDiabetes score. Candidate causally associated proteins identified through MR deserve further study as potential novel therapeutic targets for T2DM.

Predicting Risk of Colorectal Cancer After Adenoma Removal in a Large Community-Based Setting

INTRODUCTION

Colonoscopy surveillance guidelines categorize individuals as high or low risk for future colorectal cancer (CRC) based primarily on their prior polyp characteristics, but this approach is imprecise, and consideration of other risk factors may improve postpolypectomy risk stratification.

METHODS

Among patients who underwent a baseline colonoscopy with removal of a conventional adenoma in 2004-2016, we compared the performance for postpolypectomy CRC risk prediction (through 2020) of a comprehensive model featuring patient age, diabetes diagnosis, and baseline colonoscopy indication and prior polyp findings (i.e., adenoma with advanced histology, polyp size ≥10 mm, and sessile serrated adenoma or traditional serrated adenoma) with a polyp model featuring only polyp findings. Models were developed using Cox regression. Performance was assessed using area under the receiver operating characteristic curve (AUC) and calibration by the Hosmer-Lemeshow goodness-of-fit test.

RESULTS

Among 95,001 patients randomly divided 70:30 into model development (n = 66,500) and internal validation cohorts (n = 28,501), 495 CRC were subsequently diagnosed; 354 in the development cohort and 141 in the validation cohort. Models demonstrated adequate calibration, and the comprehensive model demonstrated superior predictive performance to the polyp model in the development cohort (AUC 0.71, 95% confidence interval [CI] 0.68-0.74 vs AUC 0.61, 95% CI 0.58-0.64, respectively) and validation cohort (AUC 0.70, 95% CI 0.65-0.75 vs AUC 0.62, 95% CI 0.57-0.67, respectively).

DISCUSSION

A comprehensive CRC risk prediction model featuring patient age, diabetes diagnosis, and baseline colonoscopy indication and polyp findings was more accurate at predicting postpolypectomy CRC diagnosis than a model based on polyp findings alone.

Validation of a clinical breast cancer risk assessment tool combining a polygenic score for all ancestries with traditional risk factors

PURPOSE

We previously described a combined risk score (CRS) that integrates a multiple-ancestry polygenic risk score (MA-PRS) with the Tyrer-Cuzick (TC) model to assess breast cancer (BC) risk. Here, we present a longitudinal validation of CRS in a real-world cohort.

METHODS

This study included 130,058 patients referred for hereditary cancer genetic testing and negative for germline pathogenic variants in BC-associated genes. Data were obtained by linking genetic test results to medical claims (median follow-up 12.1 months). CRS calibration was evaluated by the ratio of observed to expected BCs.

RESULTS

Three hundred forty BCs were observed over 148,349 patient-years. CRS was well-calibrated and demonstrated superior calibration compared with TC in high-risk deciles. MA-PRS alone had greater discriminatory accuracy than TC, and CRS had approximately 2-fold greater discriminatory accuracy than MA-PRS or TC. Among those classified as high risk by TC, 32.6% were low risk by CRS, and of those classified as low risk by TC, 4.3% were high risk by CRS. In cases where CRS and TC classifications disagreed, CRS was more accurate in predicting incident BC.

CONCLUSION

CRS was well-calibrated and significantly improved BC risk stratification. Short-term follow-up suggests that clinical implementation of CRS should improve outcomes for patients of all ancestries through personalized risk-based screening and prevention.

U-Turn Design Metatarsal Artery Flap: Reliable Solution in Distal Forefoot Defect

BACKGROUND

In distal forefoot defect, finding wound closure is challenging because of the distal site and small blood vessels involved. One possible resolution is the utilization of a metatarsal artery flap in a 'U-turn' design. This method offers several advantages, including its long length and a viable option for distal forefoot defect.

METHODS

Thirty-six patients with forefoot injuries from metatarsophalangeal (MTP) joint to distal interphalangeal (DIP) joint due to trauma were consecutively recruited and completed the study. Outcomes were analyzed descriptively, and risk prediction modeling for edge necrosis was performed.

RESULTS

The mean ± SD follow-up time was 27.3 months ±1.9. The median (IQR) MTP-to-DIP joint wound width and length were 1.8 (1.4, 3.0) and 3.2 cm (2.9, 6.2), respectively. The median (IQR) width, length, and width-to-length ratio flap dimensions were 3.6 (2.8, 6.0), 4.7 cm (4.3, 9.3), and 1.5 (1.2, 1.7), respectively. The mean ± SD operative time was 32.9 min ± 5.7. The median (IQR) intraoperative blood loss was 5.0 mL (4.0, 5.0). The mean ± SD hospital length of stay postoperatively was 4.0 days ±1.0. The mean ± SD Foot and Ankle Outcome Score and Foot Function Index were 64.1 ± 2.5 and 7.8% ± 3.3, respectively. All patients had good or excellent aesthetic satisfaction. Spontaneously resolving edge necrosis occurred in 13.9%. The mean ± SD time-to-start-ambulation was 1.7 weeks ±0.5. At the 2-year follow-up visit, all patients had reduced U-turn flap pivot point redundancy without shoe size impact, needing reoperation, or donor site morbidity. Edge necrosis was significantly associated with length-to-width ratio ( P = 0.014) but not with Foot and Ankle Outcome Score or Foot Function Index.

CONCLUSIONS

Metatarsal artery flap of U-turn design was reliable and was associated with a short recovery time, alternative resolution for forefoot area due to short operation time, minimal blood loss, short hospital length of stay, and excellent availability.

Comparing ancestry calibration approaches for a trans-ancestry colorectal cancer polygenic risk score

Background

Colorectal cancer (CRC) is a complex disease with monogenic, polygenic and environmental risk factors. Polygenic risk scores (PRS) are being developed to identify high polygenic risk individuals. Due to differences in genetic background, PRS distributions vary by ancestry, necessitating calibration.

Methods

We compared four calibration methods using the All of Us Research Program Whole Genome Sequence data for a CRC PRS previously developed in participants of European and East Asian ancestry. The methods contrasted results from linear models with A) the entire data set or an ancestrally diverse training set AND B) covariates including principal components of ancestry or admixture. Calibration with the training set adjusted the variance in addition to the mean.

Results

All methods performed similarly within ancestry with OR (95% C.I.) per s.d. change in PRS: African 1.5 (1.02, 2.08), Admixed American 2.2 (1.27, 3.85), European 1.6 (1.43, 1.89), and Middle Eastern 1.1 (0.71, 1.63). Using admixture and an ancestrally diverse training set provided distributions closest to standard Normal with accurate upper tail frequencies.

Conclusion

Although the PRS is predictive of CRC risk for most ancestries, its performance varies by ancestry. Post-hoc calibration preserves the risk prediction within ancestries. Training a calibration model on ancestrally diverse participants to adjust both the mean and variance of the PRS, using admixture as covariates, created standard Normal z-scores. These z-scores can be used to identify patients at high polygenic risk, and can be incorporated into comprehensive risk scores including other known risk factors, allowing for more precise risk estimates.

Validation of a Novel Clinical Dyspnea Scale - A Retrospective Pilot Study

Objective: to examine the validity of a novel dyspnea scale, Edmonton Dyspnea Inventory in idiopathic pulmonary fibrosis (IPF). Methods: Edmonton Dyspnea Inventory (EDI), is a clinical instrument to measure dyspnea severity with activities of daily living, exercise and rest using a numeric rating scale (0 -10). Consecutive IPF patients (2012-2018) with baseline MRC and EDI were included. To validate EDI, psychometric analysis was conducted. Correlations between EDI, MRC and lung function were examined. Group-based trajectory modeling was used to group patients based on dyspnea severity. Net Reclassification Improvement (NRI) was calculated to assess the improvement in 1-year mortality prediction by adding trajectory groups to MRC grade. Results: 100 consecutive IPF patients were identified; mean age 73 years (SD = 9) and 65% males; 73% were in MRC grades ≥3. Item analysis showed all 8 EDI components have excellent discrimination power with ability to differentiate patients with varying dyspnea severity. EDI has good internal consistency (Cronbach α = .92). Exploratory factor analysis showed a one-factor solution with loadings from .66 to .89 suggesting 8 EDI components measured essentially one dimension of dyspnea. All EDI components were correlated with MRC and some with lung function. Modeling data identified three EDI dyspnea severity groups with differing mortality (P = .009). The addition of EDI dyspnea severity groups to the MRC score improved 1-year mortality prediction (NRI = .66; 95% CI, .18-1.14). Conclusions: EDI is a valid dyspnea instrument, correlated with MRC and lung function. It can categorize IPF patients into 3 dyspnea severity groups associated with increased mortality. Key Message: We describe the development of a novel scale, Edmonton Dyspnea Inventory, that facilitates measurement of dyspnea severity in the context of daily activities in patients with IPF. The results indicate that the new instrument is valid and correlated to MRC. It identifies 3 categories of severity not recognized by MRC with impact on mortality. Knowledge of dyspnea severity can help triage patients and assign appropriate therapies.

The ASSESS-AKI Study found urinary epidermal growth factor is associated with reduced risk of major adverse kidney events

Biomarkers of tubular function such as epidermal growth factor (EGF) may improve prognostication of participants at highest risk for chronic kidney disease (CKD) after hospitalization. To examine this, we measured urinary EGF (uEGF) from samples collected in the Assessment, Serial Evaluation, and Subsequent Sequelae of Acute Kidney Injury (ASSESS-AKI) Study, a multi-center, prospective, observational cohort of hospitalized participants with and without AKI. Cox proportional hazards regression was used to investigate the association of uEGF/Cr at hospitalization, three months post-discharge, and the change between these time points with major adverse kidney events (MAKE): CKD incidence, progression, or development of kidney failure. Clinical findings were paired with mechanistic studies comparing relative Egf expression in mouse models of kidney atrophy or repair after ischemia-reperfusion injury. MAKE was observed in 20% of 1,509 participants over 4.3 years of follow-up. Each 2-fold higher level of uEGF/Cr at three months was associated with decreased risk of MAKE (adjusted hazards ratio 0.46, 95% confidence interval: 0.39-0.55). Participants with the highest increase in uEGF/Cr from hospitalization to three-month follow-up had a lower risk of MAKE (adjusted hazards ratio 0.52; 95% confidence interval: 0.36-0.74) compared to those with the least change in uEGF/Cr. A model using uEGF/Cr at three months combined with clinical variables yielded moderate discrimination for MAKE (area under the curve 0.73; 95% confidence interval: 0.69-0.77) and strong discrimination for kidney failure at four years (area under the curve 0.96; 95% confidence interval: 0.92-1.00). Accelerated restoration of Egf expression in mice was seen in the model of adaptive repair after injury, compared to a model of progressive atrophy. Thus, urinary EGF/Cr may be a biomarker of distal tubular health, with higher concentrations and increased uEGF/Cr post-discharge independently associated with reduced risk of MAKE in hospitalized patients.

Machine learning-based biomarker profile derived from 4210 serially measured proteins predicts clinical outcome of patients with heart failure

Aims

Risk assessment tools are needed for timely identification of patients with heart failure (HF) with reduced ejection fraction (HFrEF) who are at high risk of adverse events. In this study, we aim to derive a small set out of 4210 repeatedly measured proteins, which, along with clinical characteristics and established biomarkers, carry optimal prognostic capacity for adverse events, in patients with HFrEF.

Methods and results

In 382 patients, we performed repeated blood sampling (median follow-up: 2.1 years) and applied an aptamer-based multiplex proteomic approach. We used machine learning to select the optimal set of predictors for the primary endpoint (PEP: composite of cardiovascular death, heart transplantation, left ventricular assist device implantation, and HF hospitalization). The association between repeated measures of selected proteins and PEP was investigated by multivariable joint models. Internal validation (cross-validated c-index) and external validation (Henry Ford HF PharmacoGenomic Registry cohort) were performed. Nine proteins were selected in addition to the MAGGIC risk score, N-terminal pro-hormone B-type natriuretic peptide, and troponin T: suppression of tumourigenicity 2, tryptophanyl-tRNA synthetase cytoplasmic, histone H2A Type 3, angiotensinogen, deltex-1, thrombospondin-4, ADAMTS-like protein 2, anthrax toxin receptor 1, and cathepsin D. N-terminal pro-hormone B-type natriuretic peptide and angiotensinogen showed the strongest associations [hazard ratio (95% confidence interval): 1.96 (1.17-3.40) and 0.66 (0.49-0.88), respectively]. The multivariable model yielded a c-index of 0.85 upon internal validation and c-indices up to 0.80 upon external validation. The c-index was higher than that of a model containing established risk factors (P = 0.021).

Conclusion

Nine serially measured proteins captured the most essential prognostic information for the occurrence of adverse events in patients with HFrEF, and provided incremental value for HF prognostication beyond established risk factors. These proteins could be used for dynamic, individual risk assessment in a prospective setting. These findings also illustrate the potential value of relatively 'novel' biomarkers for prognostication.

Clinical Trial Registration

https://clinicaltrials.gov/ct2/show/NCT01851538?term=nCT01851538&draw=2&rank=1 24.

Exploring the added value of pretherapeutic MR descriptors in predicting breast cancer pathologic complete response to neoadjuvant chemotherapy

OBJECTIVES

To evaluate the association between pretreatment MRI descriptors and breast cancer (BC) pathological complete response (pCR) to neoadjuvant chemotherapy (NAC).

MATERIALS AND METHODS

Patients with BC treated by NAC with a breast MRI between 2016 and 2020 were included in this retrospective observational single-center study. MR studies were described using the standardized BI-RADS and breast edema score on T2-weighted MRI. Univariable and multivariable logistic regression analyses were performed to assess variables association with pCR according to residual cancer burden. Random forest classifiers were trained to predict pCR on a random split including 70% of the database and were validated on the remaining cases.

RESULTS

Among 129 BC, 59 (46%) achieved pCR after NAC (luminal (n = 7/37, 19%), triple negative (n = 30/55, 55%), HER2 + (n = 22/37, 59%)). Clinical and biological items associated with pCR were BC subtype (p < 0.001), T stage 0/I/II (p = 0.008), higher Ki67 (p = 0.005), and higher tumor-infiltrating lymphocytes levels (p = 0.016). Univariate analysis showed that the following MRI features, oval or round shape (p = 0.047), unifocality (p = 0.026), non-spiculated margins (p = 0.018), no associated non-mass enhancement (p = 0.024), and a lower MRI size (p = 0.031), were significantly associated with pCR. Unifocality and non-spiculated margins remained independently associated with pCR at multivariable analysis. Adding significant MRI features to clinicobiological variables in random forest classifiers significantly increased sensitivity (0.67 versus 0.62), specificity (0.69 versus 0.67), and precision (0.71 versus 0.67) for pCR prediction.

CONCLUSION

Non-spiculated margins and unifocality are independently associated with pCR and can increase models performance to predict BC response to NAC.

CLINICAL RELEVANCE STATEMENT

A multimodal approach integrating pretreatment MRI features with clinicobiological predictors, including tumor-infiltrating lymphocytes, could be employed to develop machine learning models for identifying patients at risk of non-response. This may enable consideration of alternative therapeutic strategies to optimize treatment outcomes.

KEY POINTS

• Unifocality and non-spiculated margins are independently associated with pCR at multivariable logistic regression analysis. • Breast edema score is associated with MR tumor size and TIL expression, not only in TN BC as previously reported, but also in luminal BC. • Adding significant MRI features to clinicobiological variables in machine learning classifiers significantly increased sensitivity, specificity, and precision for pCR prediction.

Assessing the clinical utility of biomarkers using the intervention probability curve (IPC)

BACKGROUND

Assessing the clinical utility of biomarkers is a critical step before clinical implementation. The reclassification of patients across clinically relevant subgroups is considered one of the best methods to estimate clinical utility. However, there are important limitations with this methodology. We recently proposed the intervention probability curve (IPC) which models the likelihood that a provider will choose an intervention as a continuous function of the probability, or risk, of disease.

OBJECTIVE

To assess the potential impact of a new biomarker for lung cancer using the IPC.

METHODS

The IPC derived from the National Lung Screening Trial was used to assess the potential clinical utility of a biomarker for suspected lung cancer. The summary statistics of the change in likelihood of intervention over the population can be interpreted as the expected clinical impact of the added biomarker.

RESULTS

The IPC analysis of the novel biomarker estimated that 8% of the benign nodules could avoid an invasive procedure while the cancer nodules would largely remain unchanged (0.1%). We showed the benefits of this approach compared to traditional reclassification methods based on thresholds.

CONCLUSIONS

The IPC methodology can be a valuable tool for assessing biomarkers prior to clinical implementation.

Development and validation of a prediction model for evaluating extubation readiness in preterm infants

Successful early extubation has advantages not only in terms of short-term respiratory morbidities and survival but also in terms of long-term neurodevelopmental outcomes in preterm infants. However, no consensus exists regarding the optimal protocol or guidelines for extubation readiness in preterm infants. Therefore, the decision to extubate preterm infants was almost entirely at the attending physician's discretion. We identified robust and quantitative predictors of success or failure of the first planned extubation attempt before 36 weeks of post-menstrual age in preterm infants (<32 weeks gestational age) and developed a prediction model for evaluating extubation readiness using these predictors. Extubation success was defined as the absence of reintubation within 72 h after extubation. This observational cohort study used data from preterm infants admitted to the neonatal intensive care unit of Seoul National University Bundang Hospital in South Korea between July 2003 and June 2019 to identify predictors and develop and test a predictive model for extubation readiness. Data from preterm infants included in the Medical Informative Medicine for Intensive Care (MIMIC-III) database between 2001 and 2008 were used for external validation. From a machine learning model using predictors such as demographics, periodic vital signs, ventilator settings, and respiratory indices, the area under the receiver operating characteristic curve and average precision of our model were 0.805 (95% confidence interval [CI], 0.802-0.809) and 0.917, respectively in the internal validation and 0.715 (95% CI, 0.713-0.717) and 0.838, respectively in the external validation. Our prediction model (NExt-Predictor) demonstrated high performance in assessing extubation readiness in both internal and external validations.

Carotid Plaque Score for Stroke and Cardiovascular Risk Prediction in a Middle-Aged Cohort From the General Population

Background We aimed to explore the predictive value of the carotid plaque score, compared with the Systematic Coronary Risk Evaluation 2 (SCORE2) risk prediction algorithm, on incident ischemic stroke and major adverse cardiovascular events and establish a prognostic cutoff of the carotid plaque score. Methods and Results In the prospective ACE 1950 (Akershus Cardiac Examination 1950 study), carotid plaque score was calculated with ultrasonography at inclusion in 2012 to 2015. The largest plaque diameter in each extracranial segment of the carotid artery on both sides was scored from 0 to 3 points. The sum of points in all segments provided the carotid plaque score. The cohort was followed up by linkage to national registries for incident ischemic stroke and major adverse cardiovascular events (nonfatal ischemic stroke, nonfatal myocardial infarction, and cardiovascular death) throughout 2020. Carotid plaque score was available in 3650 (98.5%) participants, with mean±SD age of 63.9±0.64 years at inclusion. Only 462 (12.7%) participants were free of plaque, and and 970 (26.6%) had a carotid plaque score of >3. Carotid plaque score predicted ischemic stroke (hazard ratio [HR], 1.25 [95% CI, 1.15-1.36]) and major adverse cardiovascular events (HR, 1.21 [95% CI, 1.14-1.27]) after adjustment for SCORE2 and provided strong incremental prognostic information to SCORE2. The best cutoff value of carotid plaque score for ischemic stroke was >3, with positive predictive value of 2.5% and negative predictive value of 99.3%. Conclusions The carotid plaque score is a strong predictor of ischemic stroke and major adverse cardiovascular events, and it provides incremental prognostic information to SCORE2 for risk prediction. A cutoff score of >3 seems to be suitable to discriminate high-risk subjects. Registration Information clinicaltrials.gov. Identifier: NCT01555411.

Plasma proteomic signatures of a direct measure of insulin sensitivity in two population cohorts

AIMS/HYPOTHESIS

The euglycaemic-hyperinsulinaemic clamp (EIC) is the reference standard for the measurement of whole-body insulin sensitivity but is laborious and expensive to perform. We aimed to assess the incremental value of high-throughput plasma proteomic profiling in developing signatures correlating with the M value derived from the EIC.

METHODS

We measured 828 proteins in the fasting plasma of 966 participants from the Relationship between Insulin Sensitivity and Cardiovascular disease (RISC) study and 745 participants from the Uppsala Longitudinal Study of Adult Men (ULSAM) using a high-throughput proximity extension assay. We used the least absolute shrinkage and selection operator (LASSO) approach using clinical variables and protein measures as features. Models were tested within and across cohorts. Our primary model performance metric was the proportion of the M value variance explained (R2).

RESULTS

A standard LASSO model incorporating 53 proteins in addition to routinely available clinical variables increased the M value R2 from 0.237 (95% CI 0.178, 0.303) to 0.456 (0.372, 0.536) in RISC. A similar pattern was observed in ULSAM, in which the M value R2 increased from 0.443 (0.360, 0.530) to 0.632 (0.569, 0.698) with the addition of 61 proteins. Models trained in one cohort and tested in the other also demonstrated significant improvements in R2 despite differences in baseline cohort characteristics and clamp methodology (RISC to ULSAM: 0.491 [0.433, 0.539] for 51 proteins; ULSAM to RISC: 0.369 [0.331, 0.416] for 67 proteins). A randomised LASSO and stability selection algorithm selected only two proteins per cohort (three unique proteins), which improved R2 but to a lesser degree than in standard LASSO models: 0.352 (0.266, 0.439) in RISC and 0.495 (0.404, 0.585) in ULSAM. Reductions in improvements of R2 with randomised LASSO and stability selection were less marked in cross-cohort analyses (RISC to ULSAM R2 0.444 [0.391, 0.497]; ULSAM to RISC R2 0.348 [0.300, 0.396]). Models of proteins alone were as effective as models that included both clinical variables and proteins using either standard or randomised LASSO. The single most consistently selected protein across all analyses and models was IGF-binding protein 2.

CONCLUSIONS/INTERPRETATION

A plasma proteomic signature identified using a standard LASSO approach improves the cross-sectional estimation of the M value over routine clinical variables. However, a small subset of these proteins identified using a stability selection algorithm affords much of this improvement, especially when considering cross-cohort analyses. Our approach provides opportunities to improve the identification of insulin-resistant individuals at risk of insulin resistance-related adverse health consequences.

Impact of Artificial Intelligence System and Volumetric Density on Risk Prediction of Interval, Screen-Detected, and Advanced Breast Cancer

PURPOSE

Artificial intelligence (AI) algorithms improve breast cancer detection on mammography, but their contribution to long-term risk prediction for advanced and interval cancers is unknown.

METHODS

We identified 2,412 women with invasive breast cancer and 4,995 controls matched on age, race, and date of mammogram, from two US mammography cohorts, who had two-dimensional full-field digital mammograms performed 2-5.5 years before cancer diagnosis. We assessed Breast Imaging Reporting and Data System density, an AI malignancy score (1-10), and volumetric density measures. We used conditional logistic regression to estimate odds ratios (ORs), 95% CIs, adjusted for age and BMI, and C-statistics (AUC) to describe the association of AI score with invasive cancer and its contribution to models with breast density measures. Likelihood ratio tests (LRTs) and bootstrapping methods were used to compare model performance.

RESULTS

On mammograms between 2-5.5 years prior to cancer, a one unit increase in AI score was associated with 20% greater odds of invasive breast cancer (OR, 1.20; 95% CI, 1.17 to 1.22; AUC, 0.63; 95% CI, 0.62 to 0.64) and was similarly predictive of interval (OR, 1.20; 95% CI, 1.13 to 1.27; AUC, 0.63) and advanced cancers (OR, 1.23; 95% CI, 1.16 to 1.31; AUC, 0.64) and in dense (OR, 1.18; 95% CI, 1.15 to 1.22; AUC, 0.66) breasts. AI score improved prediction of all cancer types in models with density measures (PLRT values < .001); discrimination improved for advanced cancer (ie, AUC for dense volume increased from 0.624 to 0.679, Δ AUC 0.065, P = .01) but did not reach statistical significance for interval cancer.

CONCLUSION

AI imaging algorithms coupled with breast density independently contribute to long-term risk prediction of invasive breast cancers, in particular, advanced cancer.

Cardiovascular Disease Risk Assessment Using Traditional Risk Factors and Polygenic Risk Scores in the Million Veteran Program

Importance

Primary prevention of atherosclerotic cardiovascular disease (ASCVD) relies on risk stratification. Genome-wide polygenic risk scores (PRSs) are proposed to improve ASCVD risk estimation.

Objective

To determine whether genome-wide PRSs for coronary artery disease (CAD) and acute ischemic stroke improve ASCVD risk estimation with traditional clinical risk factors in an ancestrally diverse midlife population.

Design, Setting, and Participants

This was a prognostic analysis of incident events in a retrospectively defined longitudinal cohort conducted from January 1, 2011, to December 31, 2018. Included in the study were adults free of ASCVD and statin naive at baseline from the Million Veteran Program (MVP), a mega biobank with genetic, survey, and electronic health record data from a large US health care system. Data were analyzed from March 15, 2021, to January 5, 2023.

Exposures

PRSs for CAD and ischemic stroke derived from cohorts of largely European descent and risk factors, including age, sex, systolic blood pressure, total cholesterol, high-density lipoprotein (HDL) cholesterol, smoking, and diabetes status.

Main Outcomes and Measures

Incident nonfatal myocardial infarction (MI), ischemic stroke, ASCVD death, and composite ASCVD events.

Results

A total of 79 151 participants (mean [SD] age, 57.8 [13.7] years; 68 503 male [86.5%]) were included in the study. The cohort included participants from the following harmonized genetic ancestry and race and ethnicity categories: 18 505 non-Hispanic Black (23.4%), 6785 Hispanic (8.6%), and 53 861 non-Hispanic White (68.0%) with a median (5th-95th percentile) follow-up of 4.3 (0.7-6.9) years. From 2011 to 2018, 3186 MIs (4.0%), 1933 ischemic strokes (2.4%), 867 ASCVD deaths (1.1%), and 5485 composite ASCVD events (6.9%) were observed. CAD PRS was associated with incident MI in non-Hispanic Black (hazard ratio [HR], 1.10; 95% CI, 1.02-1.19), Hispanic (HR, 1.26; 95% CI, 1.09-1.46), and non-Hispanic White (HR, 1.23; 95% CI, 1.18-1.29) participants. Stroke PRS was associated with incident stroke in non-Hispanic White participants (HR, 1.15; 95% CI, 1.08-1.21). A combined CAD plus stroke PRS was associated with ASCVD deaths among non-Hispanic Black (HR, 1.19; 95% CI, 1.03-1.17) and non-Hispanic (HR, 1.11; 95% CI, 1.03-1.21) participants. The combined PRS was also associated with composite ASCVD across all ancestry groups but greater among non-Hispanic White (HR, 1.20; 95% CI, 1.16-1.24) than non-Hispanic Black (HR, 1.11; 95% CI, 1.05-1.17) and Hispanic (HR, 1.12; 95% CI, 1.00-1.25) participants. Net reclassification improvement from adding PRS to a traditional risk model was modest for the intermediate risk group for composite CVD among men (5-year risk >3.75%, 0.38%; 95% CI, 0.07%-0.68%), among women, (6.79%; 95% CI, 3.01%-10.58%), for age older than 55 years (0.25%; 95% CI, 0.03%-0.47%), and for ages 40 to 55 years (1.61%; 95% CI, -0.07% to 3.30%).

Conclusions and Relevance

Study results suggest that PRSs derived predominantly in European samples were statistically significantly associated with ASCVD in the multiancestry midlife and older-age MVP cohort. Overall, modest improvement in discrimination metrics were observed with addition of PRSs to traditional risk factors with greater magnitude in women and younger age groups.

A modified net reclassification improvement statistic

The continuous net reclassification improvement (NRI) statistic is a popular model change measure that was developed to assess the incremental value of new factors in a risk prediction model. Two prominent statistical issues identified in the literature call the utility of this measure into question: (1) it is not a proper scoring function and (2) it has a high false positive rate when testing whether new factors contribute to the risk model. For binary response regression models, these subjects are interrogated and a modification of the continuous NRI, guided by the likelihood-based score residual, is proposed to address these issues. Within a nested model framework, the modified NRI may be viewed as a distance measure between two risk models. An application of the modified NRI is illustrated using prostate cancer data.

Screening tools for predicting mortality of adults with suspected sepsis: an international sepsis cohort validation study

OBJECTIVES

We evaluated the performance of commonly used sepsis screening tools across prospective sepsis cohorts in the USA, Cambodia and Ghana.

DESIGN

Prospective cohort studies.

SETTING AND PARTICIPANTS

From 2014 to 2021, participants with two or more SIRS (Systemic Inflammatory Response Syndrome) criteria and suspected infection were enrolled in emergency departments and medical wards at hospitals in Cambodia and Ghana and hospitalised participants with suspected infection were enrolled in the USA. Cox proportional hazards regression was performed, and Harrell's C-statistic calculated to determine 28-day mortality prediction performance of the quick Sequential Organ Failure Assessment (qSOFA) score ≥2, SIRS score ≥3, National Early Warning Score (NEWS) ≥5, Modified Early Warning Score (MEWS) ≥5 or Universal Vital Assessment (UVA) score ≥2. Screening tools were compared with baseline risk (age and sex) with the Wald test.

RESULTS

The cohorts included 567 participants (42.9% women) including 187 participants from Kumasi, Ghana, 200 participants from Takeo, Cambodia and 180 participants from Durham, North Carolina in the USA. The pooled mortality was 16.4% at 28 days. The mortality prediction accuracy increased from baseline risk with the MEWS (C-statistic: 0.63, 95% CI 0.58 to 0.68; p=0.002), NEWS (C-statistic: 0.68; 95% CI 0.64 to 0.73; p<0.001), qSOFA (C-statistic: 0.70, 95% CI 0.64 to 0.75; p<0.001), UVA score (C-statistic: 0.73, 95% CI 0.69 to 0.78; p<0.001), but not with SIRS (0.60; 95% CI 0.54 to 0.65; p=0.13). Within individual cohorts, only the UVA score in Ghana performed better than baseline risk (C-statistic: 0.77; 95% CI 0.71 to 0.83; p<0.001).

CONCLUSIONS

Among the cohorts, MEWS, NEWS, qSOFA and UVA scores performed better than baseline risk, largely driven by accuracy improvements in Ghana, while SIRS scores did not improve prognostication accuracy. Prognostication scores should be validated within the target population prior to clinical use.

Multiparametric Quantitative Imaging Biomarkers for Phenotype Classification: A Framework for Development and Validation

This manuscript is the third in a five-part series related to statistical assessment methodology for technical performance of multi-parametric quantitative imaging biomarkers (mp-QIBs). We outline approaches and statistical methodologies for developing and evaluating a phenotype classification model from a set of multiparametric QIBs. We then describe validation studies of the classifier for precision, diagnostic accuracy, and interchangeability with a comparator classifier. We follow with an end-to-end real-world example of development and validation of a classifier for atherosclerotic plaque phenotypes. We consider diagnostic accuracy and interchangeability to be clinically meaningful claims for a phenotype classification model informed by mp-QIB inputs, aiming to provide tools to demonstrate agreement between imaging-derived characteristics and clinically established phenotypes. Understanding that we are working in an evolving field, we close our manuscript with an acknowledgement of existing challenges and a discussion of where additional work is needed. In particular, we discuss the challenges involved with technical performance and analytical validation of mp-QIBs. We intend for this manuscript to further advance the robust and promising science of multiparametric biomarker development.

Presence of cells in the polyaneuploid cancer cell (PACC) state predicts the risk of recurrence in prostate cancer

BACKGROUND

The nonproliferating polyaneuploid cancer cell (PACC) state is associated with therapeutic resistance in cancer. A subset of cancer cells enters the PACC state by polyploidization and acts as cancer stem cells by undergoing depolyploidization and repopulating the tumor cell population after the therapeutic stress is relieved. Our aim was to systematically assess the presence and importance of this entity in men who underwent radical prostatectomy with curative intent to treat their presumed localized prostate cancer (PCa).

MATERIALS AND METHODS

Men with National Comprehensive Cancer Network intermediate- or high-risk PCa who underwent radical prostatectomy l from 2007 to 2015 and who did not receive neoadjuvant treatment were included. From the cohort of 2159 patients, the analysis focused on a subcohort of 209 patients and 38 cases. Prostate tissue microarrays (TMAs) were prepared from formalin-fixed, paraffin-embedded blocks of the radical prostatectomy specimens. A total of 2807 tissue samples of matched normal/benign and cancer were arrayed in nine TMA blocks. The presence of PACCs and the number of PACCs on each core were noted.

RESULTS

The total number of cells in the PACC state and the total number of cores with PACCs were significantly correlated with increasing Gleason score (p = 0.0004) and increasing Cancer of the Prostate Risk Assessment Postsurgical (CAPRA-S) (p = 0.004), but no other variables. In univariate proportional hazards models of metastasis-free survival, year of surgery, Gleason score (9-10 vs. 7-8), pathology stage, CAPRA-S, total PACCs, and cores positive for PACCs were all statistically significant. The multivariable models with PACCs that gave the best fit included CAPRA-S. Adding either total PACCs or cores positive for PACCs to CAPRA-S both significantly improved model fit compared to CAPRA-S alone.

CONCLUSION

Our findings show that the number of PACCs and the number of cores positive for PACCs are statistically significant prognostic factors for metastasis-free survival, after adjusting for CAPRA-S, in a case-cohort of intermediate- or high-risk men who underwent radical prostatectomy. In addition, despite the small number of men with complete data to evaluate time to metastatic castration-resistant PCa (mCRPC), the total number of PACCs was a statistically significant predictor of mCRPC in univariate analysis and suggested a prognostic effect even after adjusting for CAPRA-S.

A Framework for Evaluating the Technical Performance of Multiparameter Quantitative Imaging Biomarkers (mp-QIBs)

Multiparameter quantitative imaging incorporates anatomical, functional, and/or behavioral biomarkers to characterize tissue, detect disease, identify phenotypes, define longitudinal change, or predict outcome. Multiple imaging parameters are sometimes considered separately but ideally are evaluated collectively. Often, they are transformed as Likert interpretations, ignoring the correlations of quantitative properties that may result in better reproducibility or outcome prediction. In this paper we present three use cases of multiparameter quantitative imaging: i) multidimensional descriptor, ii) phenotype classification, and iii) risk prediction. A fourth application based on data-driven markers from radiomics is also presented. We describe the technical performance characteristics and their metrics common to all use cases, and provide a structure for the development, estimation, and testing of multiparameter quantitative imaging. This paper serves as an overview for a series of individual articles on the four applications, providing the statistical framework for multiparameter imaging applications in medicine.

Added prognostic value of Doppler ultrasound for adverse perinatal outcomes: A pooled analysis of three cohort studies

BACKGROUND

Fetal growth restriction (FGR) is an obstetric complication associated with adverse perinatal outcomes. Doppler ultrasound can improve perinatal outcomes through monitoring at-risk fetuses and helping time delivery.

AIM

To investigate the prognostic value of different Doppler ultrasound measurements for adverse perinatal outcomes.

MATERIALS

Individual participant data.

METHODS

We performed a pooled analysis on individual participant data. We compared six prognostic models using multilevel logistic regression, where each subsequent model added a new variable to a base model that included maternal characteristics. Estimated fetal weight (EFW) and four Doppler ultrasound measurements were added in turn: umbilical artery pulsatility index (UA PI), middle cerebral artery pulsatility index (MCA PI), cerebroplacental ratio (CPR), and mean uterine artery pulsatility index (mUtA PI). The primary outcome was a composite adverse perinatal outcome, defined as perinatal mortality, emergency caesarean delivery for fetal distress, or neonatal admission. Discriminative ability was quantified with area under the curve (AUC).

RESULTS

Three data sets (N = 3284) were included. Overall, the model that included EFW and UA PI improved AUC from 0.650 (95% CI 0.624-0.676) to 0.673 (95% CI 0.646-0.700). Adding more ultrasound measurements did not improve further the discriminative ability. In subgroup analysis, the addition of EFW and UA PI improved AUC in both preterm (AUC from 0.711 to 0.795) and small for gestational age pregnancies (AUC from 0.729 to 0.770), but they did not improve the models in term delivery or normal growth subgroups.

CONCLUSIONS

Umbilical artery pulsatility index added prognostic value for adverse perinatal outcomes to the already available information, but the combination of other Doppler ultrasound measurements (MCA PI, CPR or UtA PI) did not improve further prognostic performance.

Prognostic value of temporal patterns of global longitudinal strain in patients with chronic heart failure

Background

We investigated whether repeatedly measured global longitudinal strain (GLS) has incremental prognostic value over repeatedly measured left ventricular ejection fraction (LVEF) and N-terminal pro B-type natriuretic peptide (NT-proBNP), and a single "baseline" GLS value, in chronic heart failure (HF) patients.

Methods

In this prospective observational study, echocardiography was performed in 173 clinically stable chronic HF patients every six months during follow up. During a median follow-up of 2.7 years, a median of 3 (25th-75th percentile:2-4) echocardiograms were obtained per patient. The endpoint was a composite of HF hospitalization, left ventricular assist device, heart transplantation, cardiovascular death. We compared hazard ratios (HRs) for the endpoint from Cox models (used to analyze the first available GLS measurements) with HRs from joint models (which links repeated measurements to the time-to-event data).

Results

Mean age was 58 ± 11 years, 76% were men, 81% were in New York Heart Association functional class I/II, and all had LVEF < 50% (mean ± SD: 27 ± 9%). The endpoint was reached by 53 patients. GLS was persistently decreased over time in patients with the endpoint. However, temporal GLS trajectories did not further diverge in patients with versus without the endpoint and remained stable during follow-up. Both single measurements and temporal trajectories of GLS were significantly associated with the endpoint [HR per SD change (95%CI): 2.15(1.34-3.46), 3.54 (2.01-6.20)]. In a multivariable model, repeatedly measured GLS maintained its prognostic value while repeatedly measured LVEF did not [HR per SD change (95%CI): GLS:4.38 (1.49-14.70), LVEF:1.14 (0.41-3.23)]. The association disappeared when correcting for repeatedly measured NT-proBNP.

Conclusion

Temporal evolution of GLS was associated with adverse events, independent of LVEF but not independent of NT-proBNP. Since GLS showed decreased but stable values in patients with adverse prognosis, single measurements of GLS provide sufficient information for determining prognosis in clinical practice compared to repeated measurements, and temporal GLS patterns do not add prognostic information to NT-proBNP.

Additional value of chest CT AI-based quantification of lung involvement in predicting death and ICU admission for COVID-19 patients

Objectives

We evaluated the contribution of lung lesion quantification on chest CT using a clinical Artificial Intelligence (AI) software in predicting death and intensive care units (ICU) admission for COVID-19 patients.

Methods

For 349 patients with positive COVID-19-PCR test that underwent a chest CT scan at admittance or during hospitalization, we applied the AI for lung and lung lesion segmentation to obtain lesion volume (LV), and LV/Total Lung Volume (TLV) ratio. ROC analysis was used to extract the best CT criterion in predicting death and ICU admission. Two prognostic models using multivariate logistic regressions were constructed to predict each outcome and were compared using AUC values. The first model ("Clinical") was based on patients' characteristics and clinical symptoms only. The second model ("Clinical+LV/TLV") included also the best CT criterion.

Results

LV/TLV ratio demonstrated best performance for both outcomes; AUC of 67.8% (95% CI: 59.5 - 76.1) and 81.1% (95% CI: 75.7 - 86.5) respectively. Regarding death prediction, AUC values were 76.2% (95% CI: 69.9 - 82.6) and 79.9% (95%IC: 74.4 - 85.5) for the "Clinical" and the "Clinical+LV/TLV" models respectively, showing significant performance increase (+ 3.7%; p-value<0.001) when adding LV/TLV ratio. Similarly, for ICU admission prediction, AUC values were 74.9% (IC 95%: 69.2 - 80.6) and 84.8% (IC 95%: 80.4 - 89.2) respectively corresponding to significant performance increase (+ 10%: p-value<0.001).

Conclusions

Using a clinical AI software to quantify the COVID-19 lung involvement on chest CT, combined with clinical variables, allows better prediction of death and ICU admission.

Admission lactate level and the GRACE 2.0 score are independent and additive predictors of 30-day mortality of STEMI patients treated with primary PCI-Results of a real-world registry

BACKGROUND

In many of the risk estimation algorithms for patients with ST-elevation myocardial infarction (STEMI), heart rate and systolic blood pressure are key predictors. Yet, these parameters may also be altered by the applied medical treatment / circulatory support without concomitant improvement in microcirculation. Therefore, we aimed to investigate whether venous lactate level, a well-known marker of microcirculatory failure, may have an added prognostic value on top of the conventional variables of the "Global Registry of Acute Coronary Events" (GRACE) 2.0 model for predicting 30-day all-cause mortality of STEMI patients treated with primary percutaneous coronary intervention (PCI).

METHODS

In a prospective single-center registry study conducted from May 2020 through April 2021, we analyzed data of 323 cases. Venous blood gas analysis was performed in all patients at admission. Nested logistic regression models were built using the GRACE 2.0 score alone (base model) and with the addition of venous lactate level (expanded model) with 30-day all-cause mortality as primary outcome measure. Difference in model performance was analyzed by the likelihood ratio (LR) test and the integrated discrimination improvement (IDI). Independence of the predictors was evaluated by the variance inflation factor (VIF). Discrimination and calibration was characterized by the c-statistic and calibration intercept / slope, respectively.

RESULTS

Addition of lactate level to the GRACE 2.0 score improved the predictions of 30-day mortality significantly as assessed by both LR test (LR Chi-square = 8.7967, p = 0.0030) and IDI (IDI = 0.0685, p = 0.0402), suggesting that the expanded model may have better predictive ability than the GRACE 2.0 score. Furthermore, the VIF was 1.1203, indicating that the measured lactate values were independent of the calculated GRACE 2.0 scores.

CONCLUSIONS

Our results suggest that admission venous lactate level and the GRACE 2.0 score may be independent and additive predictors of 30-day all-cause mortality of STEMI patients treated with primary PCI.

Stratification in systemic sclerosis according to autoantibody status versus skin involvement: a study of the prospective EUSTAR cohort

BACKGROUND

The current subclassification of systemic sclerosis into cutaneous subtypes does not fully capture the heterogeneity of the disease. We aimed to compare the performances of stratification into LeRoy's cutaneous subtypes versus stratification by autoantibody status in systemic sclerosis.

METHODS

For this cohort study, we assessed people with systemic sclerosis in the multicentre international European Scleroderma Trials and Research (EUSTAR) database. Individuals positive for systemic-sclerosis autoantibodies of two specificities were excluded, and remaining individuals were classified by cutaneous subtype, according to their systemic sclerosis-specific autoantibodies, or both. We assessed the performance of each model to predict overall survival, progression-free survival, disease progression, and different organ involvement. The three models were compared by use of the area under the curve (AUC) of the receiver operating characteristic and the net reclassification improvement (NRI). Missing data were imputed.

FINDINGS

We assessed the database on July 26, 2019. Of 16 939 patients assessed for eligibility, 10 711 patients were included: 1647 (15·4%) of 10 709 were male, 9062 (84·6%) were female, mean age was 54·4 (SD 13·8) years, and mean disease duration was 7·9 (SD 8·2) years. Information regarding cutaneous subtype was available for 10 176 participants and antibody data were available for 9643 participants. In the prognostic analysis, there was no difference in AUC for overall survival (0·82, 95% CI 0·81-0·84 for cutaneous only vs 0·84, 0·82-0·85 for antibody only vs 0·84, 0·83-0·86 for combined) or for progression-free survival (0·70, 0·69-0·71 vs 0·71, 0·70-0·72 vs 0·71, 0·70-0·72). However, at 4 years the NRI showed substantial improvement for the antibody-only model compared with the cutaneous-only model in prediction of overall survival (0·57, 0·46-0·71 for antibody only vs 0·29, 0·19-0·39 for cutaneous only) and disease progression (0·36, 0·29-0·46 vs 0·21, 0·14-0·28). The antibody-only model did better than the cutaneous-only model in predicting renal crisis (AUC 0·72, 0·70-0·74 for antibody only vs 0·66, 0·64-0·69 for cutaneous only) and lung fibrosis leading to restrictive lung function (AUC 0·76, 0·75-0·77 vs 0·71, 0·70-0·72). The combined model improved the prediction of digital ulcers and elevated systolic pulmonary artery pressure, but did poorly for cardiac involvement.

INTERPRETATION

The autoantibody-only model outperforms cutaneous-only subsetting for risk stratifying people with systemic sclerosis in the EUSTAR cohort. Physicians should be aware of these findings at the time of decision making for patient management.

FUNDING

World Scleroderma Foundation.

Deep learning predicts all-cause mortality from longitudinal total-body DXA imaging

Background

Mortality research has identified biomarkers predictive of all-cause mortality risk. Most of these markers, such as body mass index, are predictive cross-sectionally, while for others the longitudinal change has been shown to be predictive, for instance greater-than-average muscle and weight loss in older adults. And while sometimes markers are derived from imaging modalities such as DXA, full scans are rarely used. This study builds on that knowledge and tests two hypotheses to improve all-cause mortality prediction. The first hypothesis is that features derived from raw total-body DXA imaging using deep learning are predictive of all-cause mortality with and without clinical risk factors, meanwhile, the second hypothesis states that sequential total-body DXA scans and recurrent neural network models outperform comparable models using only one observation with and without clinical risk factors.

Methods

Multiple deep neural network architectures were designed to test theses hypotheses. The models were trained and evaluated on data from the 16-year-long Health, Aging, and Body Composition Study including over 15,000 scans from over 3000 older, multi-race male and female adults. This study further used explainable AI techniques to interpret the predictions and evaluate the contribution of different inputs.

Results

The results demonstrate that longitudinal total-body DXA scans are predictive of all-cause mortality and improve performance of traditional mortality prediction models. On a held-out test set, the strongest model achieves an area under the receiver operator characteristic curve of 0.79.

Conclusion

This study demonstrates the efficacy of deep learning for the analysis of DXA medical imaging in a cross-sectional and longitudinal setting. By analyzing the trained deep learning models, this work also sheds light on what constitutes healthy aging in a diverse cohort.

Contribution of Genome-Wide Polygenic Score to Risk of Coronary Artery Disease in Childhood Cancer Survivors

Background

Adverse cardiovascular outcomes such as coronary artery disease (CAD) are the leading noncancer causes of morbidity and mortality among childhood cancer survivors.

Objectives

The aim of this study was to assess the role of a genome-wide polygenic score (GPS) for CAD, well validated in the general population, and its interplay with cancer-related risk factors among childhood cancer survivors.

Methods

In a cohort study of 2,472 5-year childhood cancer survivors from the St. Jude Lifetime Cohort, the association between the GPS and the risk of CAD was performed using Cox regression models adjusted for age at cancer diagnosis, sex, cumulative dose of anthracyclines, and mean heart radiation dose.

Results

Among survivors of European ancestry, the GPS was significantly associated with the risk of CAD (HR per 1 SD of the GPS: 1.25; 95% CI: 1.04-1.49; P = 0.014). Compared with the first tertile, survivors in the upper tertile had a greater risk of CAD (1.51-fold higher HR of CAD [95% CI: 0.96-2.37; P = 0.074]), although the difference was not statistically significant. The GPS-CAD association was stronger among survivors diagnosed with cancer at age <10 years exposed to >25 Gy heart radiation (HR top vs. bottom tertile of GPS: 15.49; 95% CI: 5.24-45.52; P_trend = 0.005) but not among those diagnosed at age ≥10 years (P_trend ≥ 0.77) and not among those diagnosed at age <10 years exposed to ≤25 Gy heart radiation (P_trend = 0.23). Among high-risk survivors, defined by an estimated relative hazard ≥3.0 from fitted Cox models including clinical risk factors alone, the cumulative incidence of CAD at 40 years from diagnosis was 29% (95% CI: 13%-45%). After incorporating the GPS into the model, the cumulative incidence increased to 48% (95% CI: 26%-69%).

Conclusions

Childhood cancer survivors are at risk for premature CAD. A GPS may help identify those who may benefit from targeted screening and personalized preventive interventions.

Are serum estrogen concentrations associated with menopausal symptom bother among postmenopausal women? Baseline results from two MsFLASH clinical trials

OBJECTIVES

To evaluate whether single measurements of serum estradiol (E₂), estrone (E₁) and sex hormone-binding globulin (SHBG) concentration distinguishes between women with and without menopausal symptom bother.

STUDY DESIGN

We analyzed baseline data from two clinical trials conducted in 2012-2017: MsFLASH 03 (178 peri-/post-menopausal women aged 40-62 years with bothersome vasomotor symptoms, mean age 54) and MsFLASH 05 (181 post-menopausal women aged 45-70 years with moderate-to-severe vulvovaginal symptoms, mean age 61).

MAIN OUTCOME MEASURES

Symptom bother (hot flushes or flashes, night sweats, sweating, aching in muscles and joints, change in sexual desire, vaginal dryness during intercourse, and avoiding intimacy) in the past month was assessed using the Menopause-Specific Quality of Life questionnaire. Using logistic regression, we calculated the area under the receiver operating characteristic curve (AUC) values for E₁, E₂, and SHBG concentration in relation to being at least somewhat bothered (symptom bother score ≥3) by each symptom within each trial study population.

RESULTS

AUC values (95% confidence interval) ranged between 0.51 (0.41-0.60) and 0.62 (0.53, 0.72) for MsFLASH 03 and between 0.51 (0.42, 0.59) and 0.64 (0.53, 0.75) for MsFLASH 05. There was little evidence of associations between serum hormone levels and bother by a given menopausal symptom.

CONCLUSION

These findings do not support the clinical utility of a single measurement of serum of E₁, E₂, or SHBG concentrations in differentiating between women who are bothered by a given menopausal symptom and those who are not.

Bioactive adrenomedullin a prognostic biomarker in patients with mild to moderate dyspnea at the emergency department: an observational study

Acute dyspnea with underlying congestion is a leading cause of emergency department (ED) visits with high rates of hospitalization. Adrenomedullin is a vasoactive neuropeptide hormone secreted by the endothelium that mediates vasodilation and maintains vascular integrity. Plasma levels of biologically active adrenomedullin (bio-ADM) predict septic shock and vasopressor need in critically ill patients and are associated with congestion in patients with acute heart failure (HF) but the prognostic value in unselected dyspneic patients at the ED is unknown. The purpose of this study is to test if bio-ADM predicts adverse outcomes when sampled in patients with acute dyspnea at presentation to the ED. In this single-center prospective observational study, we included 1402 patients from the ADYS (Acute DYSpnea at the Emergency Department) cohort in Malmö, Sweden. We fitted logistic regression models adjusted for sex, age, N-terminal pro-B-type natriuretic peptide (NT-proBNP), creatinine, and C-reactive protein (CRP) to associate bio-ADM plasma levels to mortality, hospitalization, intravenous (IV) diuretic treatment and HF diagnosis. Using receiver operating characteristic (ROC) curve analysis we evaluated bio-ADM discrimination for these outcomes compared to a reference model (sex, age, NT-proBNP, creatinine, and CRP). Model performance was compared by performing a likelihood ratio test on the deviances of the models. Bio-ADM (per interquartile range from median) predicts both 90-day mortality [odds ratio (OR): 1.5, 95% confidence interval (CI) 1.2-2.0, p < 0.002] and hospitalization (OR: 1.5, 95% CI 1.2-1.8, p < 0.001) independently of sex, age, NT-proBNP, creatinine, and CRP. Bio-ADM statistically significantly improves the reference model in predicting mortality (added χ² 9.8, p = 0.002) and hospitalization (added χ² 14.1, p = 0.0002), and is associated with IV diuretic treatment and HF diagnosis at discharge. Plasma levels of bio-ADM sampled at ED presentation in acutely dyspneic patients are independently associated with 90-day mortality, hospitalization and indicate the need for decongestive therapy.

Polysomnographic Phenotypes of Obstructive Sleep Apnea and Incident Type 2 Diabetes: Results from the DREAM Study

Rationale: Obstructive sleep apnea (OSA) is associated with cardiovascular disease and incident type 2 diabetes (T2DM). Seven OSA phenotypes, labeled on the basis of their most distinguishing polysomnographic features, have been shown to be differentially associated with incident cardiovascular disease. However, little is known about the relevance of polysomnographic phenotypes for the risk of T2DM. Objectives: To assess whether polysomnographic phenotypes are associated with incident T2DM and to compare the predictive value of baseline polysomnographic phenotypes with the Apnea-Hypopnea Index (AHI) for T2DM. Methods: The study included 840 individuals without baseline diabetes from a multisite observational U.S. veteran cohort who underwent OSA evaluation between 2000 and 2004, with follow-up through 2012. The primary outcome was incident T2DM, defined as no diagnosis at baseline and a new physician diagnosis confirmed by fasting blood glucose >126 mg/dL during follow-up. Relationships between the seven polysomnographic phenotypes (1. mild, 2. periodic limb movements of sleep [PLMS], 3. non-rapid eye movement and poor sleep, 4. rapid eye movement and hypoxia, 5. hypopnea and hypoxia, 6. arousal and poor sleep, and 7. combined severe) and incident T2DM were investigated using Cox proportional hazards regression and competing risk regression models with and without adjustment for baseline covariates. Likelihood ratio tests were conducted to compare the predictive value of the phenotypes with the AHI. Results: During a median follow-up period of 61 months, 122 (14.5%) patients developed incident T2DM. After adjustment for baseline sociodemographics, fasting blood glucose, body mass index, comorbidities, and behavioral risk factors, hazard ratios among persons with "hypopnea and hypoxia" and "PLMS" phenotypes as compared with persons with "mild" phenotype were 3.18 (95% confidence interval [CI], 1.53-6.61] and 2.26 (95% CI, 1.06-4.83) for incident T2DM, respectively. Mild OSA (5 ⩽ AHI < 15) (vs. no OSA) was directly associated with incident T2DM in both unadjusted and multivariable-adjusted regression models. The addition of polysomnographic phenotypes, but not AHI, to known T2DM risk factors greatly improved the predictive value of the computed prediction model. Conclusions: Polysomnographic phenotypes "hypopnea and hypoxia" and "PLMS" independently predict risk of T2DM among a predominantly male veteran population. Polysomnographic phenotypes improved T2DM risk prediction comared with the use of AHI.

Integrated Biomarkers for the Management of Indeterminate Pulmonary Nodules

Rationale: Patients with indeterminate pulmonary nodules (IPNs) at risk of cancer undergo high rates of invasive, costly, and morbid procedures. Objectives: To train and externally validate a risk prediction model that combined clinical, blood, and imaging biomarkers to improve the noninvasive management of IPNs. Methods: In this prospectively collected, retrospective blinded evaluation study, probability of cancer was calculated for 456 patient nodules using the Mayo Clinic model, and patients were categorized into low-, intermediate-, and high-risk groups. A combined biomarker model (CBM) including clinical variables, serum high sensitivity CYFRA 21-1 level, and a radiomic signature was trained in cohort 1 (n = 170) and validated in cohorts 2-4 (total n = 286). All patients were pooled to recalibrate the model for clinical implementation. The clinical utility of the CBM compared with current clinical care was evaluated in 2 cohorts. Measurements and Main Results: The CBM provided improved diagnostic accuracy over the Mayo Clinic model with an improvement in area under the curve of 0.124 (95% bootstrap confidence interval, 0.091-0.156; P < 2 × 10-16). Applying 10% and 70% risk thresholds resulted in a bias-corrected clinical reclassification index for cases and control subjects of 0.15 and 0.12, respectively. A clinical utility analysis of patient medical records estimated that a CBM-guided strategy would have reduced invasive procedures from 62.9% to 50.6% in the intermediate-risk benign population and shortened the median time to diagnosis of cancer from 60 to 21 days in intermediate-risk cancers. Conclusions: Integration of clinical, blood, and image biomarkers improves noninvasive diagnosis of patients with IPNs, potentially reducing the rate of unnecessary invasive procedures while shortening the time to diagnosis.

Development and Internal Validation of a Preoperative Prediction Model for Sentinel Lymph Node Status in Breast Cancer: Combining Radiomics Signature and Clinical Factors

Purpose

To develop and internally validate a nomogram combining radiomics signature of primary tumor and fibroglandular tissue (FGT) based on pharmacokinetic dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and clinical factors for preoperative prediction of sentinel lymph node (SLN) status in breast cancer patients.

Methods

This study retrospectively enrolled 186 breast cancer patients who underwent pretreatment pharmacokinetic DCE-MRI with positive (n = 93) and negative (n = 93) SLN. Logistic regression models and radiomics signatures of tumor and FGT were constructed after feature extraction and selection. The radiomics signatures were further combined with independent predictors of clinical factors for constructing a combined model. Prediction performance was assessed by receiver operating characteristic (ROC), calibration, and decision curve analysis. The areas under the ROC curve (AUCs) of models were corrected by 1,000-times bootstrapping method and compared by Delong's test. The added value of each independent model or their combinations was also assessed by net reclassification improvement (NRI) and integrated discrimination improvement (IDI) indices. This report referred to the "Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis" (TRIPOD) statement.

Results

The AUCs of the tumor radiomic model (eight features) and the FGT radiomic model (three features) were 0.783 (95% confidence interval [CI], 0.717-0.849) and 0.680 (95% CI, 0.604-0.757), respectively. A higher AUC of 0.799 (95% CI, 0.737-0.862) was obtained by combining tumor and FGT radiomics signatures. By further combining tumor and FGT radiomics signatures with progesterone receptor (PR) status, a nomogram was developed and showed better discriminative ability for SLN status [AUC 0.839 (95% CI, 0.783-0.895)]. The IDI and NRI indices also showed significant improvement when combining tumor, FGT, and PR compared with each independent model or a combination of any two of them (all p < 0.05).

Conclusion

FGT and clinical factors improved the prediction performance of SLN status in breast cancer. A nomogram integrating the DCE-MRI radiomics signature of tumor and FGT and PR expression achieved good performance for the prediction of SLN status, which provides a potential biomarker for clinical treatment decision-making.

Study of Diffusion Weighted Imaging Derived Diffusion Parameters as Biomarkers for the Microenvironment in Gliomas

Objectives

To explore the efficacy of diffusion weighted imaging (DWI)-derived metrics under different models as surrogate indicators for molecular biomarkers and tumor microenvironment in gliomas.

Methods

A retrospective study was performed for 41 patients with gliomas. The standard apparent diffusion coefficient (ADC_st) and ADC under ultra-high b values (ADC_uh) (b values: 2500 to 5000 s/mm²) were calculated based on monoexponential model. The fraction of fast diffusion (f), pseudo ADC (ADC_fast) and true ADC (ADC_slow) were calculated by bi-exponential model (b values: 0 to 2000 s/mm²). The apparent diffusional kurtosis (K_app) was derived from the simplified diffusion kurtosis imaging (DKI) model (b values: 200 to 3000 s/mm²). Potential correlations between DWI parameters and immunohistological indices (i.e. Aquaporin (AQP)1, AQP4, AQP9 and Ki-67) were investigated and DWI parameters were compared between high- and low-grade gliomas, and between tumor center and peritumor. Receiver operator characteristic (ROC) curve and area under the curve (AUC) were calculated to determine the performance of independent or combined DWI parameters in grading gliomas.

Results

The ADC_slow and ADC_uh at tumor center showed a stronger correlation with Ki-67 than other DWI metrics. The ADC_st, ADC_slow and ADC_uh at tumor center presented correlations with AQP1 and AQP4 while AQP9 did not correlate with any DWI metric. K_app showed a correlation with Ki-67 while no significant correlation with AQPs. ADC_st (p < 0.001) and ADC_slow (p = 0.001) were significantly lower while the ADC_uh (p = 0.006) and K_app (p = 0.005) were significantly higher in the high-grade than in the low-grade gliomas. ADC_st, f, ADC_fast, ADC_slow, ADC_uh, K_app at the tumor center had significant differences with those in peritumor when the gliomas grade became high (p < 0.05). Involving ADC_uh and K_app simultaneously into an independent ADC_st model (AUC = 0.833) could further improve the grading performance (ADC_st+ADC_uh+K_app: AUC = 0.923).

Conclusion

Different DWI metrics fitted within different b-value ranges (low to ultra-high b values) have different efficacies as a surrogate indicator for molecular expression or microstructural complexity in gliomas. Further studies are needed to better explain the biological meanings of these DWI parameters in gliomas.

Free Hemoglobin Ratio as a Novel Biomarker of Acute Kidney Injury After On-Pump Cardiac Surgery: Secondary Analysis of a Randomized Controlled Trial

BACKGROUND

Cardiac surgery with cardiopulmonary bypass (CPB) is associated with a high risk of postoperative acute kidney injury (AKI). Due to limitations of current diagnostic strategies, we sought to determine whether free hemoglobin (fHb) ratio (ie, levels of fHb at the end of CPB divided by baseline fHb) could predict AKI after on-pump cardiac surgery.

METHODS

This is a secondary analysis of a randomized controlled trial comparing the effect of nitric oxide (intervention) versus nitrogen (control) on AKI after cardiac surgery (NCT01802619). A total of 110 adult patients in the control arm were included. First, we determined whether fHb ratio was associated with AKI via multivariable analysis. Second, we verified whether fHb ratio could predict AKI and incorporation of fHb ratio could improve predictive performance at an early stage, compared with prediction using urinary biomarkers alone. We conducted restricted cubic spline in logistic regression for model development. We determined the predictive performance, including area under the receiver-operating-characteristics curve (AUC) and calibration (calibration plot and accuracy, ie, number of correct predictions divided by total number of predictions). We also used AUC test, likelihood ratio test, and net reclassification index (NRI) to compare the predictive performance between competing models (ie, fHb ratio versus neutrophil gelatinase-associated lipocalin [NGAL], N-acetyl-β-d-glucosaminidase [NAG], and kidney injury molecule-1 [KIM-1], respectively, and incorporation of fHb ratio with NGAL, NAG, and KIM-1 versus urinary biomarkers alone), if applicable.

RESULTS

Data stratified by median fHb ratio showed that subjects with an fHb ratio >2.23 presented higher incidence of AKI (80.0% vs 49.1%; P = .001), more need of renal replacement therapy (10.9% vs 0%; P = .036), and higher in-hospital mortality (10.9% vs 0%; P = .036) than subjects with an fHb ratio ≤2.23. fHb ratio was associated with AKI after adjustment for preestablished factors. fHb ratio outperformed urinary biomarkers with the highest AUC of 0.704 (95% confidence interval [CI], 0.592-0.804) and accuracy of 0.714 (95% CI, 0.579-0.804). Incorporation of fHb ratio achieved better discrimination (AUC test, P = .012), calibration (likelihood ratio test, P < .001; accuracy, 0.740 [95% CI, 0.617-0.832] vs 0.632 [95% CI, 0.477-0.748]), and significant prediction increment (NRI, 0.638; 95% CI, 0.269-1.008; P < .001) at an early stage, compared with prediction using urinary biomarkers alone.

CONCLUSIONS

Results from this exploratory, hypothesis-generating retrospective, observational study shows that fHb ratio at the end of CPB might be used as a novel, widely applicable biomarker for AKI. The use of fHb ratio might help for an early detection of AKI, compared with prediction based only on urinary biomarkers.

Predictive Added Value of Selected Plasma Lipids to a Re-estimated Minimal Risk Tool

Background: Lipidomics is emerging for biomarker discovery in cardiovascular disease, and circulating lipids are increasingly incorporated in risk models to predict cardiovascular events. Moreover, specific classes of lipids, such as sphingomyelins, ceramides, and triglycerides, have been related to coronary artery disease (CAD) severity and plaque characteristics. To avoid unnecessary testing, it is important to identify individuals at low CAD risk. The only pretest model available so far to rule out the presence of coronary atherosclerosis in patients with chest pain, but normal coronary arteries, is the minimal risk tool (MRT).

Aim: Using state-of-the-art statistical methods, we aim to verify the additive predictive value of a set of lipids, derived from targeted plasma lipidomics of suspected CAD patients, to a re-estimated version of the MRT for ruling out the presence of coronary atherosclerosis assessed by coronary CT angiography (CCTA).

Methods: Two hundred and fifty-six subjects with suspected stable CAD recruited from five European countries within H2020-SMARTool, undergoing CCTA and blood sampling for clinical biochemistry and lipidomics, were selected. The MRT was validated by regression methods and then re-estimated (reMRT). The reMRT was used as a baseline model in a likelihood ratio test approach to assess the added predictive value of each lipid from 13 among ceramides, triglycerides, and sphingomyelins. Except for one lipid, the analysis was carried out on more than 240 subjects for each lipid. A sensitivity analysis was carried out by considering two alternative models developed on the cohort as baseline models.

Results: In 205 subjects, coronary atherosclerosis ranged from minimal lesions to overt obstructive CAD, while in 51 subjects (19.9%) the coronary arteries were intact. Four triglycerides and seven sphingomyelins were significantly (p < 0.05) and differentially expressed in the two groups and, at a lesser extent, one ceramide (p = 0.067). The probability of being at minimal risk was significantly better estimated by adding either Cer(d18:1/16:0) (p = 0.01), SM(40:2) (p = 0.04), or SM(41:1) at a lesser extent (p = 0.052) to reMRT than by applying the reMRT alone. The sensitivity analysis confirmed the relevance of these lipids. Furthermore, the addition of SM(34:1), SM(38:2), SM(41:2), and SM(42:4) improved the predictive performance of at least one of the other baseline models. None of the selected triglycerides was found to provide an added value.

Conclusions: Plasma lipidomics can be a promising source of diagnostic and prognostic biomarkers in cardiovascular disease, exploitable not only to assess the risk of adverse events but also to identify subjects without coronary atherosclerosis, thus reducing unnecessary further testing in normal subjects.

A relationship between the incremental values of area under the ROC curve and of area under the precision-recall curve

BACKGROUND

Incremental value (IncV) evaluates the performance change between an existing risk model and a new model. Different IncV metrics do not always agree with each other. For example, compared with a prescribed-dose model, an ovarian-dose model for predicting acute ovarian failure has a slightly lower area under the receiver operating characteristic curve (AUC) but increases the area under the precision-recall curve (AP) by 48%. This phenomenon of disagreement is not uncommon, and can create confusion when assessing whether the added information improves the model prediction accuracy.

METHODS

In this article, we examine the analytical connections and differences between the AUC IncV (ΔAUC) and AP IncV (ΔAP). We also compare the true values of these two IncV metrics in a numerical study. Additionally, as both are semi-proper scoring rules, we compare them with a strictly proper scoring rule: the IncV of the scaled Brier score (ΔsBrS) in the numerical study.

RESULTS

We demonstrate that ΔAUC and ΔAP are both weighted averages of the changes (from the existing model to the new one) in separating the risk score distributions between events and non-events. However, ΔAP assigns heavier weights to the changes in higher-risk regions, whereas ΔAUC weights the changes equally. Due to this difference, the two IncV metrics can disagree, and the numerical study shows that their disagreement becomes more pronounced as the event rate decreases. In the numerical study, we also find that ΔAP has a wide range, from negative to positive, but the range of ΔAUC is much smaller. In addition, ΔAP and ΔsBrS are highly consistent, but ΔAUC is negatively correlated with ΔsBrS and ΔAP when the event rate is low.

CONCLUSIONS

ΔAUC treats the wins and losses of a new risk model equally across different risk regions. When neither the existing or new model is the true model, this equality could attenuate a superior performance of the new model for a sub-region. In contrast, ΔAP accentuates the change in the prediction accuracy for higher-risk regions.

Performance of the IBIS/Tyrer-Cuzick model of breast cancer risk by race and ethnicity in the Women's Health Initiative

BACKGROUND

The IBIS/Tyrer-Cuzick model is used clinically to guide breast cancer screening and prevention, but was developed primarily in non-Hispanic White women. Little is known about its long-term performance in a racially/ethnically diverse population.

METHODS

The Women's Health Initiative study enrolled postmenopausal women from 1993-1998. Women were included who were aged <80 years at enrollment with no prior breast cancer or mastectomy and with data required for IBIS/Tyrer-Cuzick calculation (weight; height; ages at menarche, first birth, and menopause; menopausal hormone therapy use; and family history of breast or ovarian cancer). Calibration was assessed by the ratio of observed breast cancer cases to the number expected by the IBIS/Tyrer-Cuzick model (O/E; calculated as the sum of cumulative hazards). Differential discrimination was tested for by self-reported race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, Asian or Pacific Islander, and American Indian or Alaskan Native) using Cox regression. Exploratory analyses, including simulation of a protective single-nucleotide polymorphism (SNP), rs140068132 at 6q25, were performed.

RESULTS

During follow-up (median 18.9 years, maximum 23.4 years), 6783 breast cancer cases occurred among 90,967 women. IBIS/Tyrer-Cuzick was well calibrated overall (O/E ratio = 0.95; 95% CI, 0.93-0.97) and in most racial/ethnic groups, but overestimated risk for Hispanic women (O/E ratio = 0.75; 95% CI, 0.62-0.90). Discrimination did not differ by race/ethnicity. Exploratory simulation of the protective SNP suggested improved IBIS/Tyrer-Cuzick calibration for Hispanic women (O/E ratio = 0.80; 95% CI, 0.66-0.96).

CONCLUSIONS

The IBIS/Tyrer-Cuzick model is well calibrated for several racial/ethnic groups over 2 decades of follow-up. Studies that incorporate genetic and other risk factors, particularly among Hispanic women, are essential to improve breast cancer-risk prediction.

Nested logistic regression models and ΔAUC applications: Change-point analysis

The area under the receiver operating characteristic curve (AUC) is one of the most popular measures for evaluating the performance of a predictive model. In nested models, the change in AUC (ΔAUC) can be a discriminatory measure of whether the newly added predictors provide significant improvement in terms of predictive accuracy. Recently, several authors have shown rigorously that ΔAUC can be degenerate and its asymptotic distribution is no longer normal when the reduced model is true, but it could be the distribution of a linear combination of some χ12 random variables [1,2]. Hence, the normality assumption and existing variance estimate cannot be applied directly for developing a statistical test under the nested models. In this paper, we first provide a brief review on the use of ΔAUC for comparing nested logistic models and the difficulty of retrieving the reference distribution behind. Then, we present a special case of the nested logistic regression models that the newly added predictor to the reduced model contains a change-point in its effects. A new test statistic based on ΔAUC is proposed in this setting. A simple resampling scheme is proposed to approximate the critical values for the test statistic. The inference of the change-point parameter is done via m-out-of-n bootstrap. Large-scale simulation is conducted to evaluate the finite-sample performance of the ΔAUC test for the change-point model. The proposed method is applied to two real-life datasets for illustration.

Cervical Cancer Stem-Like Cell Transcriptome Profiles Predict Response to Chemoradiotherapy

Cervical cancer (CC) represents a major global health issue, particularly impacting women from resource constrained regions worldwide. Treatment refractoriness to standard chemoradiotheraphy has identified cancer stem cells as critical coordinators behind the biological mechanisms of resistance, contributing to CC recurrence. In this work, we evaluated differential gene expression in cervical cancer stem-like cells (CCSC) as biomarkers related to intrinsic chemoradioresistance in CC. A total of 31 patients with locally advanced CC and referred to Mário Penna Institute (Belo Horizonte, Brazil) from August 2017 to May 2018 were recruited for the study. Fluorescence-activated cell sorting was used to enrich CD34+/CD45- CCSC from tumor biopsies. Transcriptome was performed using ultra-low input RNA sequencing and differentially expressed genes (DEGs) using Log2 fold differences and adjusted p-value < 0.05 were determined. The analysis returned 1050 DEGs when comparing the Non-Responder (NR) (n=10) and Responder (R) (n=21) groups to chemoradiotherapy. These included a wide-ranging pattern of underexpressed coding genes in the NR vs. R patients and a panel of lncRNAs and miRNAs with implications for CC tumorigenesis. A panel of biomarkers was selected using the rank-based AUC (Area Under the ROC Curve) and pAUC (partial AUC) measurements for diagnostic sensitivity and specificity. Genes overlapping between the 21 highest AUC and pAUC loci revealed seven genes with a strong capacity for identifying NR vs. R patients (ILF2, RBM22P2, ACO16722.1, AL360175.1 and AC092354.1), of which four also returned significant survival Hazard Ratios. This study identifies DEG signatures that provide potential biomarkers in CC prognosis and treatment outcome, as well as identifies potential alternative targets for cancer therapy.

Cardiac troponin I measured with a very high sensitivity assay predicts subclinical carotid atherosclerosis: The Akershus Cardiac Examination 1950 Study

AIMS

Concentrations of cardiac troponin I (cTnI) are associated with incident ischemic stroke and predict the presence and severity of coronary atherosclerosis. Accordingly, we hypothesized that concentrations of cTnI measured with a very high sensitivity (hs-) assay would be associated with subclinical stages of carotid atherosclerosis in the general population.

METHODS

We measured hs-cTnI on the Singulex Clarity System in 1745 women and 1666 men participating in the prospective observational Akershus Cardiac Examination 1950 Study. All study participants were free from known coronary heart disease and underwent extensive cardiovascular phenotyping at baseline, including carotid ultrasound. We quantified carotid atherosclerosis by the carotid plaque score, carotid intima-media thickness (cIMT) and the presence of hypoechoic plaques.

RESULTS

Concentrations of hs-cTnI were measurable in 99.8% of study participants and were significantly associated with increased carotid plaque score (odds ratio for quartile 4 of hs-cTnI 1.59, 95% CI 1.22 to 2.07, p for trend < 0.001) and cIMT (odds ratio for quartile 4 of hs-cTnI 1.57, 95% CI 1.02 to 2.42, p for trend = 0.036), but not with the presence of hypoechoic plaques. hs-cTnI concentrations significantly improved reclassification and discrimination models in predicting carotid plaques when added to cardiovascular risk factors, no improvements were evident in predicting cIMT or hypoechoic plaques.

CONCLUSION

Concentrations of cTnI measured with a very high sensitivity assay are predictive of carotid atherosclerotic burden, a phenomenon likely attributable to common risk factors of subclinical myocardial injury, coronary and carotid atherosclerosis.

Clinical Application of High-Sensitivity Troponin Testing in the Atherosclerotic Cardiovascular Disease Framework of the Current Cholesterol Guidelines

Importance

The 2018 American Heart Association/American College of Cardiology (AHA/ACC) cholesterol management guidelines identified 2 distinct groups of patients with atherosclerotic cardiovascular disease (ASCVD) prompting different treatment recommendations.

Objective

To investigate whether the addition of high-sensitivity troponin (hsTn) testing to guideline-derived ASCVD risk can improve risk classification and downstream treatment recommendations.

Design, Setting, and Participants

A prospective cohort biomarker substudy was performed that included 8635 patients enrolled in the Prevention of Cardiovascular Events in Patients with Prior Heart Attack Using Ticagrelor Compared to Placebo on a Background of Aspirin-Thrombolysis in Myocardial Infarction 54 (PEGASUS-TIMI 54) trial. Patients were assigned to risk groups of either very high-risk ASCVD or lower-risk ASCVD based on their cardiovascular history and comorbidities, in line with the 2018 AHA/ACC cholesterol management guidelines criteria. Patients were also classified on the basis of hsTnI level (ARCHITECT assay; Abbott) using cut points of 2 ng/L (limit of detection) and 6 ng/L (risk threshold), followed by joint classification on the basis of clinical features and hsTnI level. The setting was a nested prospective cohort study in a completed multinational trial. Participants were all patients who had a myocardial infarction 1 to 3 years before enrollment, were at least 50 years of age, and had at least 1 high-risk feature. The study dates were October 2010 to December 2014. The dates of analysis were June 2019 to January 2020.

Main Outcomes and Measures

The primary end point was a composite of cardiovascular death, myocardial infarction, or stroke.

Results

Among 8635 patients enrolled in the PEGASUS-TIMI 54 trial, the median age was 65 years (interquartile range, 58-71 years), and 6614 (76.6%) were men; 8340 (96.6%) were White individuals and 176 (2.0%) were Black individuals. Patients meeting clinical criteria for the very high-risk ASCVD group had a primary end point 3-year event rate of 8.8% compared with 5.0% in the lower-risk ASCVD group (hazard ratio, 2.01; 95% CI, 1.58-2.57; P < .001). When patients in the very high-risk ASCVD group were further risk stratified by hsTnI level, 614 of 6789 patients (9.0%) with an undetectable hsTnI level had a 3-year event rate of 2.7% (<1% per year), which was less than the overall rate in the lower-risk ASCVD group. Analogously, in the lower-risk ASCVD group, 417 of 1846 patients (22.6%) with an hsTnI level exceeding 6 ng/L had an event rate of 9.1%, comparable to the overall rate in the very high-risk ASCVD group. The addition of hsTnI to guideline-derived ASCVD risk led to a net reclassification index at event rate of 0.15 (95% CI, 0.10-0.21). Overall, use of hsTnI reclassified 1031 of 8635 patients (11.9%) (1 in 11 with very high-risk ASCVD and 1 in 4 with lower-risk ASCVD).

Conclusions and Relevance

The findings of this cohort substudy suggest that a strategy incorporating hsTn into a guideline-derived ASCVD risk algorithm provides enhanced risk stratification and reclassifies 11.9% of patients into a more appropriate risk group. This application of hsTn testing might be used to optimize the care of patients with ASCVD.

Whole-Lesion DCE-MRI Intensity Histogram Analysis for Diagnosis in Patients with Suspected Lung Cancer

RATIONALE AND OBJECTIVES

To explore the diagnostic value of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) intensity histogram metrics, relative to time intensity curve (TIC)-derived metrics, in patients with suspected lung cancer.

MATERIALS AND METHODS

This retrospective study enrolled 49 patients with suspected lung cancer on routine CT imaging who underwent DCE-MRI scans and had final histopathologic diagnosis. Three TIC-derived metrics (maximum enhancement ratio, peak time [T_max] and slope) and eight intensity histogram metrics (volume, integral, maximum, minimum, median, coefficient of variation [CoV], skewness, and kurtosis) were extracted from DCE-MRI images. TIC-derived and intensity histogram metrics were compared between benignity versus malignancy using the Wilcoxon rank-sum test. Associations between imaging metrics and malignancy risk were assessed by univariate and multivariate logistic regression odds ratios (ORs).

RESULTS

There were 33 malignant lesions and 16 benign lesions based on histopathology. Lower CoV (OR = 0.2 per 1-SD increase, p = 0.0006), lower T_max (OR = 0.4 per 1-SD increase, p = 0.005), and steeper slope (OR = 2.4 per 1-SD increase, p = 0.010) were significantly associated with increased risk of malignancy. Under multivariate analysis, CoV was significantly independently associated with malignancy likelihood after accounting for either T_max (OR = 0.3 per 1-SD increase, p = 0.007) or slope (OR = 0.3 per 1-SD increase, p = 0.011).

CONCLUSION

This initial study found that DCE-MRI CoV was independently associated with malignancy in patients with suspected lung cancer. CoV has the potential to help diagnose indeterminate pulmonary lesions and may complement TIC-derived DCE-MRI metrics. Further studies are warranted to validate the diagnostic value of DCE-MRI intensity histogram analysis.

Validation of the adjusted multi-biomarker disease activity score as a prognostic test for radiographic progression in rheumatoid arthritis: a combined analysis of multiple studies

BACKGROUND

The multi-biomarker disease activity (MBDA) test measures 12 serum protein biomarkers to quantify disease activity in RA patients. A newer version of the MBDA score, adjusted for age, sex, and adiposity, has been validated in two cohorts (OPERA and BRASS) for predicting risk for radiographic progression. We now extend these findings with additional cohorts to further validate the adjusted MBDA score as a predictor of radiographic progression risk and compare its performance with that of other risk factors.

METHODS

Four cohorts were analyzed: the BRASS and Leiden registries and the OPERA and SWEFOT studies (total N = 953). Treatments included conventional DMARDs and anti-TNFs. Associations of radiographic progression (ΔTSS) per year with the adjusted MBDA score, seropositivity, and clinical measures were evaluated using linear and logistic regression. The adjusted MBDA score was (1) validated in Leiden and SWEFOT, (2) compared with other measures in all four cohorts, and (3) used to generate curves for predicting risk of radiographic progression.

RESULTS

Univariable and bivariable analyses validated the adjusted MBDA score and found it to be the strongest, independent predicator of radiographic progression (ΔTSS > 5) compared with seropositivity (rheumatoid factor and/or anti-CCP), baseline TSS, DAS28-CRP, CRP SJC, or CDAI. Neither DAS28-CRP, CDAI, SJC, nor CRP added significant information to the adjusted MBDA score as a predictor, and the frequency of radiographic progression agreed with the adjusted MBDA score when it was discordant with these measures. The rate of progression (ΔTSS > 5) increased from < 2% in the low (1-29) adjusted MBDA category to 16% in the high (45-100) category. A modeled risk curve indicated that risk increased continuously, exceeding 40% for the highest adjusted MBDA scores.

CONCLUSION

The adjusted MBDA score was validated as an RA disease activity measure that is prognostic for radiographic progression. The adjusted MBDA score was a stronger predictor of radiographic progression than conventional risk factors, including seropositivity, and its prognostic ability was not significantly improved by the addition of DAS28-CRP, CRP, SJC, or CDAI.

The added value of new covariates to the brier score in cox survival models

Calibration is an important measure of the predictive accuracy for a prognostic risk model. A widely used measure of calibration when the outcome is survival time is the expected Brier score. In this paper, methodology is developed to accurately estimate the difference in expected Brier scores derived from nested survival models and to compute an accompanying variance estimate of this difference. The methodology is applicable to time invariant and time-varying coefficient Cox survival models. The nested survival model approach is often applied to the scenario where the full model consists of conventional and new covariates and the subset model contains the conventional covariates alone. A complicating factor in the methodologic development is that the Cox model specification cannot, in general, be simultaneously satisfied for nested models. The problem has been resolved by projecting the properly specified full survival model onto the lower dimensional space of conventional markers alone. Simulations are performed to examine the method's finite sample properties and a prostate cancer data set is used to illustrate its application.

Relationship between baseline cardiac biomarkers and cardiovascular death or hospitalization for heart failure with and without sodium-glucose co-transporter 2 inhibitor therapy in DECLARE-TIMI 58

AIMS

Dapagliflozin reduced the risk of the composite of cardiovascular (CV) death or hospitalization for heart failure (HHF) in patients with type 2 diabetes mellitus in DECLARE-TIMI 58. We hypothesized that baseline N-terminal pro B-type natriuretic peptide (NT-proBNP) and high-sensitivity troponin T (hsTnT) levels would help identify patients who are at higher baseline risk and we describe the treatment effects of dapagliflozin in patients according to their baseline NT-proBNP and hsTnT levels.

METHODS AND RESULTS

This was a pre-specified biomarker study from DECLARE-TIMI 58, a randomized, double-blind, placebo-controlled CV outcomes trial of dapagliflozin. Baseline NT-proBNP and hsTnT levels were measured in the TIMI Clinical Trials Laboratory in 14 565 patients. Among the included patients, 9143 patients (62.8%) were male, 1464 (10.1%) had a history of heart failure and the mean age was 63.9 years. The median baseline NT-proBNP and hsTnT levels were 75 pg/mL [interquartile range (IQR) 35-165] and 10.2 pg/mL (IQR 6.9-15.5), respectively. Patients with higher NT-proBNP and hsTnT quartiles had higher rates of CV death/HHF (Q4 vs. Q1: NT-proBNP: 4-year Kaplan-Meier event rates 13.7% vs. 1.0%; hsTnT: 11.8% vs. 1.4%; P-trend <0.001). Dapagliflozin consistently reduced the relative risk of CV death/HHF regardless of baseline NT-proBNP (P-interaction 0.72) or hsTnT quartiles (P-interaction 0.93). Given their higher baseline risk, patients with NT-proBNP and/or hsTnT levels above the median derived larger absolute risk reductions with dapagliflozin (NT-proBNP 1.9% vs. 0%, P-interaction 0.010; hsTnT 1.8% vs. 0.1%, P-interaction 0.026).

CONCLUSION

Patients with type 2 diabetes mellitus and higher NT-proBNP or hsTnT levels are at increased risk of CV death and HHF. Dapagliflozin reduced the relative risk of CV death/HHF irrespective of NT-proBNP and hsTnT levels, with greater absolute risk reductions seen in patients with higher baseline biomarker levels.

Traditional Cardiovascular Risk Factors and Individual Prediction of Cardiovascular Events in Childhood Cancer Survivors

BACKGROUND

Childhood cancer survivors have an increased risk of heart failure, ischemic heart disease, and stroke. They may benefit from prediction models that account for cardiotoxic cancer treatment exposures combined with information on traditional cardiovascular risk factors such as hypertension, dyslipidemia, and diabetes.

METHODS

Childhood Cancer Survivor Study participants (n = 22 643) were followed through age 50 years for incident heart failure, ischemic heart disease, and stroke. Siblings (n = 5056) served as a comparator. Participants were assessed longitudinally for hypertension, dyslipidemia, and diabetes based on self-reported prescription medication use. Half the cohort was used for discovery; the remainder for replication. Models for each outcome were created for survivors ages 20, 25, 30, and 35 years at the time of prediction (n = 12 models).

RESULTS

For discovery, risk scores based on demographic, cancer treatment, hypertension, dyslipidemia, and diabetes information achieved areas under the receiver operating characteristic curve and concordance statistics 0.70 or greater in 9 and 10 of the 12 models, respectively. For replication, achieved areas under the receiver operating characteristic curve and concordance statistics 0.70 or greater were observed in 7 and 9 of the models, respectively. Across outcomes, the most influential exposures were anthracycline chemotherapy, radiotherapy, diabetes, and hypertension. Survivors were then assigned to statistically distinct risk groups corresponding to cumulative incidences at age 50 years of each target outcome of less than 3% (moderate-risk) or approximately 10% or greater (high-risk). Cumulative incidence of all outcomes was 1% or less among siblings.

CONCLUSIONS

Traditional cardiovascular risk factors remain important for predicting risk of cardiovascular disease among adult-age survivors of childhood cancer. These prediction models provide a framework on which to base future surveillance strategies and interventions.

Blood-based biomarkers for prediction of intracranial hemorrhage and outcome in patients with moderate or severe traumatic brain injury

BACKGROUND

Early identification of traumatic intracranial hemorrhage (ICH) has implications for triage and intervention. Blood-based biomarkers were recently approved by the Food and Drug Administration (FDA) for prediction of ICH in patients with mild traumatic brain injury (TBI). We sought to determine if biomarkers measured early after injury improve prediction of mortality and clinical/radiologic outcomes compared with Glasgow Coma Scale (GCS) alone in patients with moderate or severe TBI (MS-TBI).

METHODS

We measured glial fibrillary acidic protein (GFAP), ubiquitin C-terminal hydrolase L1 (UCH-L1), and microtubule-associated protein-2 (MAP-2) on arrival to the emergency department (ED) in patients with blunt TBI enrolled in the placebo arm of the Prehospital TXA for TBI Trial (prehospital GCS score, 3-12; SPB, > 90). Biomarkers were modeled individually and together with prehospital predictor variables [PH] (GCS score, age, sex). Data were divided into a training data set and test data set for model derivation and evaluation. Models were evaluated for prediction of ICH, mass lesion, 48-hour and 28-day mortality, and 6-month Glasgow Outcome Scale-Extended (GOS-E) and Disability Rating Scale (DRS). Area under the curve (AUC) was evaluated in test data for PH alone, PH + individual biomarkers, and PH + three biomarkers.

RESULTS

Of 243 patients with baseline samples (obtained a median of 84 minutes after injury), prehospital GCS score was 8 (interquartile range, 5-10), 55% had ICH, and 48-hour and 28-day mortality were 7% and 13%, respectively. Poor neurologic outcome at 6 months was observed in 34% based on GOS-E of 4 or less, and 24% based on DRS greater than or equal to7. Addition of each biomarker to PH improved AUC in the majority of predictive models. GFAP+PH compared with PH alone significantly improved AUC in all models (ICH, 0.82 vs. 0.64; 48-hour mortality, 0.84 vs. 0.71; 28-day mortality, 0.84 vs. 0.66; GOS-E, 0.78 vs. 0.69; DRS, 0.84 vs. 0.81, all p < 0.001).

CONCLUSION

Circulating blood-based biomarkers may improve prediction of neurological outcomes and mortality in patients with MS-TBI over prehospital characteristics alone. Glial fibrillary acidic protein appears to be the most promising. Future evaluation in the prehospital setting is warranted.

LEVEL OF EVIDENCE

Prospective, Prognostic and Epidemiological, level II.