Category-free Net Reclassification Index Research Articles

Von Willebrand factor improves risk prediction in addition to N-terminal pro-B-type natriuretic peptide in patients referred to coronary angiography and signs and symptoms of heart failure and preserved ejection fraction.

Heart failure with preserved ejection fraction (HFpEF) represents a growing health burden associated with substantial mortality and morbidity. Consequently, risk prediction is of highest importance. Endothelial dysfunction has been recently shown to play an important role in the complex pathophysiology of HFpEF. We therefore aimed to assess von Willebrand factor (vWF), a marker of endothelial damage, as potential biomarker for risk assessment in patients with HFpEF. Concentrations of vWF were assessed in 457 patients with HFpEF enrolled as part of the LUdwigshafen Risk and Cardiovascular Health (LURIC) study. All-cause mortality was observed in 40% of patients during a median follow-up time of 9.7 years. vWF significantly predicted mortality with a hazard ratio (HR) per increase of 1 SD of 1.45 (95% confidence interval, 1.26-1.68; P<0.001) and remained a significant predictor after adjustment for age, sex, body mass index, N-terminal pro-B-type natriuretic peptide (NT-proBNP), renal function, and frequent HFpEF-related comorbidities (adjusted HR per 1 SD, 1.22; 95% confidence interval, 1.05-1.42; P=0.001). Most notably, vWF showed additional prognostic value beyond that achievable with NT-proBNP indicated by improvements in C-Statistic (vWF×NT-proBNP: 0.65 versus NT-proBNP: 0.63; P for comparison, 0.004) and category-free net reclassification index (37.6%; P<0.001). vWF is an independent predictor of long-term outcome in patients with HFpEF, which is in line with endothelial dysfunction as potential mediator in the pathophysiology of HFpEF. In particular, combined assessment of vWF and NT-proBNP improved risk prediction in this vulnerable group of patients.

36

Introduction: Circulating concentrations of plasma inflammatory and apoptosis markers in critically ill patients with acute kidney injury requiring renal replacement therapy (RRT) are associated with RRT dependence and death. Whether these biomarkers predict clinical outcomes is unknown. Methods: The Biological Markers for Recovery of Kidney was a multicenter, prospective, nested, observational study conducted ancillary to the Veterans Affairs/Acute Renal Failure Trial Network study of 817 critically ill patients receiving RRT. Plasma markers were measured on day 1. Renal recovery was defined as alive and RRT independent by day 60. A parsimonious clinical model was determined using stepwise logistic regression and Least Absolute Shrinkage and Selection Operator (LASSO) based algorithms. We computed area under receiver operating characteristic curve (AUROC), integrated discrimination index (IDI), and category-free net reclassification index (CFNRI). Results: Of the 817 patients, 298 (36.5%) were alive and free of RRT and 415 (50.8%) died. The derivation (model building) and validation sets contained 394 and 423 subjects. In the validation dataset, a clinical model consisting of eight clinical variables (age, mechanical ventilation, arterial ph, bilirubin, platelets,mean arterial pressure, and Total ApacheII) predicted renal recovery and mortality (AUROC range: 0.74–0.77 and 0.75–0.80); with good calibration (P values range: 0.08–0.45); modest increment in IDI (ranges: 2%-5%, P values <0.05); and CFNRI (ranges: 27%-32%, p values <0.05). The AUROC of individual biomarkers were only modestly predictive of renal recovery, (range, 0.55–0.63) and mortality, (range 0.54–0.68). When added to the clinical model there was an increase in risk prediction for renal mortaltiy (AUROC, 0.78–0.82); but not for recovery (AUROC, 0.76–0.78); IDI (1%-2%: P value < 0.05) and CFNRI (27%-23%) for recovery and mortality. Conclusions: In a large cohort of severely ill patients with acute kidney injury requiring RRT, the addition of plasma inflammatory and apoptosis markers to a baseline clinical model resulted in a modest improvement in risk prediction of RRT independence and mortality.

Prognostic Usefulness of Left Ventricular Hypertrophy by Electrocardiography in Patients With Atrial Fibrillation (from the Randomized Evaluation of Long-Term Anticoagulant Therapy Study)

It is unknown whether left ventricular hypertrophy (LVH) diagnosis by electrocardiography improves risk stratification in patients with atrial fibrillation (AF). We investigated the prognostic impact of LVH diagnosis by electrocardiography in a large sample of anticoagulated patients with AF included in the Randomized Evaluation of Long-Term Anticoagulant Therapy (RE-LY) Study. We defined electrographic LVH (ECG-LVH) by strain pattern or Cornell voltage (R wave in aVL plus S wave in V3) >2.0 mV (women) or >2.4 mV (men). LVH prevalence was 22.7%. During a median follow-up of 2.0 years, 303 patients developed a stroke, 778 died (497 from cardiovascular causes), and 140 developed a myocardial infarction. LVH was associated with a greater risk of stroke (1.99% vs 1.32% per year, hazard ratio [HR] 1.51, 95% confidence interval [CI] 1.18 to 1.93, p <0.001), cardiovascular death (4.52% vs 1.80% per year, HR 2.56, 95% CI 2.14 to 3.06, p <0.0001), all-cause death (6.03% vs 3.11% per year, HR 1.95, 95% CI 1.68 to 2.26, p <0.0001), and myocardial infarction (1.11% vs 0.55% per year, HR 2.07, 95% CI 1.47 to 2.92, p <0.0001). In multivariate analysis, the prognostic value of LVH was additive to CHA2DS2-VASc score and other covariates. The category-free net reclassification index and integrated discrimination improvement increased significantly after adding LVH to multivariate models. In conclusion, our study demonstrates for the first time that ECG-LVH, a simple and easily accessible prognostic indicator, improves risk stratification in anticoagulated patients with AF.

Net reclassification indices for evaluating risk prediction instruments: a critical review.

Net reclassification indices have recently become popular statistics for measuring the prediction increment of new biomarkers. We review the various types of net reclassification indices and their correct interpretations. We evaluate the advantages and disadvantages of quantifying the prediction increment with these indices. For predefined risk categories, we relate net reclassification indices to existing measures of the prediction increment. We also consider statistical methodology for constructing confidence intervals for net reclassification indices and evaluate the merits of hypothesis testing based on such indices. We recommend that investigators using net reclassification indices should report them separately for events (cases) and nonevents (controls). When there are two risk categories, the components of net reclassification indices are the same as the changes in the true- and false-positive rates. We advocate the use of true- and false-positive rates and suggest it is more useful for investigators to retain the existing, descriptive terms. When there are three or more risk categories, we recommend against net reclassification indices because they do not adequately account for clinically important differences in shifts among risk categories. The category-free net reclassification index is a new descriptive device designed to avoid predefined risk categories. However, it experiences many of the same problems as other measures such as the area under the receiver operating characteristic curve. In addition, the category-free index can mislead investigators by overstating the incremental value of a biomarker, even in independent validation data. When investigators want to test a null hypothesis of no prediction increment, the well-established tests for coefficients in the regression model are superior to the net reclassification index. If investigators want to use net reclassification indices, confidence intervals should be calculated using bootstrap methods rather than published variance formulas. The preferred single-number summary of the prediction increment is the improvement in net benefit.

Adding electrocardiography improves cardiovascular risk assessment in the elderly

Purpose: To examine whether addition of electrocardiographic ST-T changes improves risk assessment in the elderly (65 years and above). Background: Increased life-expectancy makes accurate risk prediction in the elderly increasingly relevant. Traditional risk factors are important predictors in the younger population, but have less impact in the elderly. Conversely, ST-T changes are common in the elderly and have consistently been shown to be strong predictors of cardiovascular disease (CVD) Methods: Participants from a prospective cardiovascular cohort study attending at an age 65 years or above were included. ST-T changes were divided into three groups of increasing severity (minor/intermediate/major) based on the Minnesota Code. The primary end-point was fatal CVD-event and the secondary was non-fatal and fatal CVD-event. Results: A total of 7,201 persons were included with a median follow-up on 11.7 years. 2,210 reached the primary and 3,821 reached the secondary end-point. ST-T changes were frequently presented (18.2%) with increasing prevalence with increasing age. For the primary end-point, c-index improved from 0.707 (0.689-0.725) to 0.720 (0.702-0.738), p<0.001, the category-free net reclassification index (cNRI) was 29.2% (21.5%–36.9%), p<0.001 and integrated discrimination index (IDI) was 1.42% (0.86%–1.98%), p<0.001. For the secondary end-point, the c-index improved from 0.652 (0.640-0.664) to 0.663 (0.651-0.675), p<0.001, cNRI was 19.4% (13.7%-25.0%), p<0.001 and IDI was 1.42% (1.01%-1.83%), p<0.001. View this table: Cox regression model, HR (95% CI) Conclusion: Simple assessment of the existence of ST-T changes in the electrocardiogram is cheap and feasible and improves risk prediction significantly in the general, elderly population.

Prognostic value of chromogranin A in severe sepsis: data from the FINNSEPSIS study

To assess the prognostic information of chromogranin A (CgA), a marker associated with adrenergic tone and myocardial function, in patients with severe sepsis. CgA levels were measured at the time of study inclusion and 72 h later in 232 patients with severe sepsis recruited from 24 ICUs in Finland (FINNSEPSIS study). Sixty-five patients (28%) died during the index hospitalization. CgA levels at inclusion and after 72 h correlated with several established indices of risk in sepsis. Patients who died during the hospitalization had higher baseline CgA levels than hospital survivors: 14.0 (Q1-3, 7.4-27.4) versus 9.1 (5.9-15.8) nmol/l, P = 0.002, and after 72 h: 16.2 (9.0-31.1) versus 9.8 (6.0-18.0) nmol/l, P = 0.001. Prior cardiovascular disease (P = 0.04) and cardiovascular SOFA levels on day 3 (P = 0.03) were associated with higher CgA levels after 72 h by linear regression. CgA levels on study inclusion and after 72 h were independently associated with hospital mortality by logistic regression: OR (logarithmically transformed CgA levels) 1.95 (95% CI 1.01-3.77), P = 0.046 and OR 2.03 (95% CI 1.18-3.49), P = 0.01, respectively. The prognostic accuracy was comparable for CgA measurements and SAPS II score, and the addition of CgA measurements to the SAPS II score improved risk stratification of the patients as assessed by the category-free net reclassification index. A CgA level >6.6 nmol/l on study inclusion was associated with septic shock during the hospitalization. CgA levels measured during hospitalization for severe sepsis are associated with cardiovascular dysfunction and may provide additional prognostic information in patients with severe sepsis.