The “Dilution Effect” of Augmented Renal Clearance on the Prognostic Utility of NT-proBNP in Patients with Severe Traumatic Brain Injury

Introduction

Severe traumatic brain injury (sTBI) remains a primary cause of mortality and long-term disability worldwide.^1,2 Accurate early-phase prognostication is imperative for guiding clinical decision-making and optimizing resource allocation. While neurological indices such as the Glasgow Coma Scale (GCS) are fundamental, the “brain-heart axis” has garnered increasing attention; specifically, the massive sympathetic surge following brain injury can precipitate neurogenic stunned myocardium and cardiac dysfunction. In this context, N-terminal pro-B-type natriuretic peptide (NT-proBNP), a sensitive marker of hemodynamic stress and myocardial strain, has emerged as a critical prognostic tool. Elevated NT-proBNP levels have been consistently correlate increased mortality and adverse functional outcomes in the TBI population.^3–5

Nevertheless, the diagnostic efficacy of circulating biomarkers is intrinsically dependent on the balance between their production and elimination. NT-proBNP is primarily cleared via renal excretion, making its serum concentration susceptible to variations in renal function.⁶ Augmented Renal Clearance (ARC)—defined as creatinine clearance >130 mL/min/1.73 m²—is a prevalent yet often overlooked phenomenon in the intensive care unit (ICU).^7,8 It is particularly common among younger sTBI patients who often exhibit a hyperdynamic response to trauma.^9,10

Despite its prevalence, the impact of ARC on renal-dependent biomarkers remains underappreciated. Theoretically, glomerular hyperfiltration in ARC could accelerate NT-proBNP elimination, leading to a “dilution effect” where serum levels are falsely low relative to actual cardiac stress. Consequently, standard prognostic cutoffs established in general populations may be misleading for sTBI patients with ARC, potentially causing clinicians to underestimate mortality risks.

To date, the interplay between ARC, serum NT-proBNP, and their combined prognostic value in sTBI remains fully elucidated. Therefore, this study investigates the influence of ARC on NT-proBNP levels and its prognostic validity in sTBI patients. Utilizing propensity score-matched (PSM) analysis to minimize baseline confounders, we sought to: (1) quantify the difference in NT-proBNP levels between ARC and non-ARC patients; (2) evaluate how ARC confounds and attenuates the predictive power of NT-proBNP for 28-day mortality; and (3) develop a novel, ARC-adjusted nomogram to recalibrate risk stratification. By addressing this diagnostic “blind spot,” our findings provide a refined framework for interpreting cardiac biomarkers in hypermetabolic states, ultimately enabling more precise risk assessment and personalized management for critically ill sTBI patients.

Materials and Methods

Study Design and Population

This retrospective cohort study was conducted at the First Affiliated Hospital with Nanjing Medical University between June 2020 and June 2024. The inclusion criteria were: (1) Age >18 years; (2) Diagnosis of severe TBI (GCS score≤8 on admission). Patients were excluded if they had: (1) a history of chronic heart failure (New York Heart Association class II–IV or Left Ventricular Ejection Fraction < 50%); (2) chronic kidney disease;¹¹ (3) Acute Kidney Injury;¹² (4) Incomplete clinical or laboratory data. The study protocol was approved by the Ethics Committee of the First Affiliated Hospital with Nanjing Medical University (2024-SR-077), with the requirement for informed consent waived due to the retrospective nature of the analysis.

Definition of Augmented Renal Clearance

ARC was defined as a measured creatinine clearance (CrCl) > 130 mL/min/1.73 m². CrCl was calculated from 24-hour urine collection using the standard formula: CrCl (mL/min) = [Urine Creatinine (μmol/L) × 24-hour Urine Volume (mL)]/[Serum Creatinine (μmol/L) × 1440 min].¹³ The calculated CrCl was normalized to a standard body surface area (BSA) of 1.73 m².

Urine collection was initiated immediately following placement of an indwelling urinary catheter (typically within 1 hour of ICU admission). The exact 24-hour collection period was strictly timed and recorded. Total urine volume was measured using calibrated drainage containers. Upon completion of the collection, the entire urine volume was thoroughly mixed before a 20 mL aliquot was extracted for laboratory creatinine analysis. These procedures were executed by specialized ICU nursing staff to minimize sampling errors.^14,15

Propensity Score Matching

To minimize selection bias and isolate the effect of renal clearance on NT-proBNP levels, we performed Propensity Score Matching (PSM). A 1:1 nearest-neighbor matching algorithm with a caliper width of 0.1 was applied. The matching covariates included Age, Gender, APACHE II score, GCS and Albumin. Balance between groups was assessed using Standardized Mean Differences (SMD).

Statistical Analysis

Data management and analyses were performed using R software (version 4.2.3; The R Foundation for Statistical Computing). Continuous variables were expressed as median (interquartile range [IQR]) and compared between groups using the Mann–Whitney U-test. Categorical variables were summarized as frequencies (percentages) and compared using the Chi-square test. Serum NT-proBNP levels were log-transformed for regression analysis and visualization to normalize the distribution.

Univariate and multivariate logistic regression analyses were performed to identify independent predictors of 28-day mortality. Variables with P < 0.05 in univariate analysis were entered into the multivariate model using the Enter method. To prevent multicollinearity, the GCS score was intentionally excluded from the multivariable analysis as it is a fundamental component of the APACHE II score. Multicollinearity among independent variables was assessed using variance inflation factor (VIF), with VIF > 5 indicating significant multicollinearity. The prognostic performance of NT-proBNP was evaluated using Receiver Operating Characteristic (ROC) curves, and the Area Under the Curve (AUC) was compared between ARC and Non-ARC groups using the DeLong test. Optimal cutoffs were determined using the Youden index. Model improvement was quantified using Net Reclassification Improvement (NRI) and Integrated Discrimination Improvement (IDI).

A prognostic nomogram was established based on the independent predictors identified in the multivariate analysis. The comprehensive performance of the model was evaluated by assessing discrimination via the concordance index (C-index/AUC) and calibration through plots using 1,000 bootstrap resamples to compare predicted versus observed probabilities. Furthermore, the clinical utility of the nomogram was determined using Decision Curve Analysis (DCA) to estimate the net benefit across a range of threshold probabilities.

Analyses for mechanism validation (comparison of NT-proBNP levels and ROC curves between ARC and non-ARC groups) were performed in the propensity score-matched cohort (n=186). Multivariate logistic regression, model comparison, nomogram development, calibration, decision curve analysis, and subgroup analyses were performed in the original cohort (n=352).

Results

Baseline Characteristics and Propensity Score Matching

A total of 352 patients with severe TBI were included in the original cohort, as detailed in the study flowchart (Figure 1). The cohort comprised 141 (40.1%) patients in the ARC group and 211 (59.9%) in the Non-ARC group. Patients with ARC were significantly younger (median 57 vs 69 years, P < 0.001) and had lower APACHE II scores (17 vs 20, P < 0.001) compared to the Non-ARC group Notably, despite similar intracranial hemorrhage volumes, ARC patients presented with significantly lower median serum NT-proBNP levels (897.00 vs 1201.00pg/mL, P = 0.001) After 1:1 propensity score matching, 93 pairs (n=186) were generated. The matching process successfully balanced most baseline demographics, including Age, Gender, Hypertension, and APACHE II scores (Standardized Mean Differences < 0.1) (Table 1), with minor imbalances in GCS score adjusted through multivariate regression.

Figure 1 Flowchart of study participant selection. The study initially screened 490 patients with severe traumatic brain injury (TBI) admitted to the ICU between June 2020 and June 2024. After excluding 138 patients based on exclusion criteria (chronic heart/renal failure, missing data, or acute kidney injury), the original cohort consisted of 352 patients, including 211 in the Non-ARC group and 141 in the ARC group. Propensity score matching (PSM) was performed at a 1:1 ratio to balance baseline characteristics, resulting in a final matched cohort of 186 patients (93 in each group) for mechanism validation. The red box highlights the matched cohort (n=186) ultimately used for mechanism validation. Abbreviation: PSM, propensity score matching.

Table 1 Baseline Demographic and Clinical Characteristics of Patients with Severe Traumatic Brain Injury in the Original Cohort (n=352) and Matched Cohort (n=186)

The “Dilution Effect” of ARC on Serum NT-proBNP Levels

In the matched cohort, serum NT-proBNP levels remained significantly lower in the ARC group compared to the non-ARC group (896.10 vs 1296.00 pg/mL, P=0.009). This difference persisted after log transformation (P=0.0089 by Wilcoxon test), as illustrated in Figure 2, suggesting a potential association between ARC and reduced NT-proBNP levels consistent with a “dilution effect.”

Figure 2 Comparison of serum NT-proBNP levels between Non-ARC and ARC groups in the matched cohort (n=186). The boxplot displays the distribution of log-transformed NT-proBNP levels.6 Patients in the ARC group exhibited significantly lower NT-proBNP levels compared to the Non-ARC group (P = 0.0089). The horizontal line within the box represents the median, and the box edges represent the interquartile range. ** indicates p<0.001.

Risk Factors for 28-Day Mortality

In the univariate analysis, ARC status was associated with significantly reduced odds of 28-day mortality (OR= 0.264, P<0.001), while elevated NT-proBNP was associated with increased risk (OR=1.035 per 100 pg/mL, P<0.001). Other significant risk factors included hypertension (OR=2.169), higher APACHE II scores (OR=1.104), and lower GCS scores (OR=0.792). In the multivariate logistic regression model, both ARC and NT-proBNP remained independent predictors. ARC was independently protective (OR=0.446, P=0.007), whereas NT-proBNP remained an independent risk factor for mortality (OR=1.031, P<0.001). Hypertension (P=0.025) and APACHE II score (P=0.005) also retained significance (Table 2). Collinearity diagnostics using variance inflation factor (VIF) confirmed no significant multicollinearity among the independent variables in the multivariate model (all VIF < 2). ARC status VIF = 1.37; NT-proBNP VIF = 1.15; APACHE II VIF = 1.13; Hypertension VIF=1.03.

Table 2 Univariate and Multivariate Logistic Regression Analysis of Risk Factors for 28-Day Mortality in Patients with Severe Traumatic Brain Injury

Impact of ARC on the Predictive Performance of NT-proBNP

We evaluated the prognostic accuracy of NT-proBNP for 28-day mortality using ROC analysis. As shown in Figure 3, NT-proBNP demonstrated excellent discriminative ability in non-ARC patients (AUC=0.833, 95% CI:0.763–0.904). However, its predictive performance was significantly attenuated in ARC patients (AUC=0.637, 95% CI: 0.513–0.761) (P<0.001 by DeLong test). Furthermore, the optimal cutoff value derived from the Youden index in the non-ARC group (986 pg/mL) strongly stratified mortality risk (OR=23.29, P<0.001), whereas the corresponding cutoff in the ARC subgroup (847.26 pg/mL) failed to achieve statistical significance (OR=2.00, P=0.144) (Table 3).

Figure 3 Receiver operating characteristic (ROC) curves for the prognostic value of NT-proBNP (n=186). ROC analysis was performed to evaluate the predictive ability of NT-proBNP for 28-day mortality in the matched cohort. The area under the curve (AUC) was significantly higher in the Non-ARC group (AUC = 0.833; solid blue line) compared to the ARC group (AUC = 0.637; dashed red line), suggesting a reduced prognostic utility of NT-proBNP in the presence of ARC. Abbreviation: ROC, receiver operating characteristic.

Table 3 Optimal Cutoff Values of NT-proBNP for Predicting 28-Day Mortality Stratified by ARC Status

Subgroup Analysis and Interaction Effects

Subgroup analysis was performed in the original cohort (n=352) to explore potential effect modification by clinical characteristics (Figure 4). The interaction tests revealed that the dilution effect of ARC on NT-proBNP prognostic utility was more pronounced in patients aged <65 years (P for interaction=0.024), male sex (P for interaction=0.01), those without hypertension (P for interaction=0.002), and those with APACHE II scores <20 (P for interaction=0.014). In contrast, no significant interaction was observed for other clinical parameters.

Figure 4 Forest plot of subgroup analysis for the association between NT-proBNP and mortality (n=352). The plot shows the odds ratios (OR) and 95% confidence intervals (CI) for the association between NT-proBNP and 28-day mortality across various subgroups, including Age, Sex, Hypertension status, and APACHE II score. The P-value for interaction indicates whether the effect size differs significantly across subgroups. Abbreviations: OR, odds ratio; CI, confidence interval.

Model Development, Comparison, and Clinical Utility

To assess the added value of integrating ARC status into risk prediction, we compared four models using AUC, Net Reclassification Improvement (NRI), and Integrated Discrimination Improvement (IDI) (Table 4). Model 1 (age and gender) showed limited discriminative ability with an AUC of 0.585 (95% CI: 0.530–0.651). The addition of APACHE II score and hypertension in Model 2 significantly improved the AUC to 0.682 (95% CI: 0.624–0.740, P=0.016 vs Model 1). A substantial increase in model performance was observed in Model 3 with the inclusion of log10NT-ProBNP, reaching an AUC of 0.811 (95% CI: 0.771–0.857, P<0.001 vs Model 2).

Table 4 Comparison of Model Performance (AUC, NRI, and IDI) for Predicting 28-Day Mortality

Notably, although the further addition of ARC status in Model 4 yielded a modest increase in AUC to 0.856 (95% CI:0.790–0.881) and the DeLong test showed no statistically significant difference compared to Model 3 (P=0.317), the reclassification indices demonstrated significant incremental value. Specifically, the category-free NRI for Model 4 versus Model 3 was 0.590 (95% CI: 0.397–0.848), and the IDI was 0.066 (95% CI:0.027–0.126). These results indicate that incorporating ARC status significantly improves the accuracy of risk stratification beyond traditional clinical factors and BNP levels.

Based on the identified independent predictors, a prognostic nomogram was constructed to facilitate clinical application (Figure 5A). The model integrates the four confirmed independent predictors: ARC status, log10NT-proBNP, APACHE II score, and Hypertension. The calibration curve demonstrated excellent agreement between the nomogram-predicted probabilities and the actual observed mortality rates, with the bias-corrected line closely tracking the ideal reference line (Figure 5B). Decision Curve Analysis (DCA) showed that the new nomogram (incorporating ARC status) provided a higher net benefit across a broad range of threshold probabilities compared to the baseline model (without ARC) or simple “treat-all”/“treat-none” strategies (Figure 5C). These findings suggest the potential clinical utility of the ARC-adjusted model for decision-making in sTBI management.

Figure 5 Construction, calibration, and clinical utility of the predictive nomogram for 28-day mortality (n=352). (A) The prognostic nomogram integrating ARC status, Age, log-transformed NT-proBNP, APACHE II score and Hypertension. To use the nomogram, locate the patient’s value on each variable axis, draw a vertical line upwards to the “Points” axis to determine the score, sum the points for all variables to obtain the “Total Points,” and draw a vertical line downwards to estimate the 28-day mortality probability. (B) Calibration curve of the nomogram. The x-axis represents the predicted mortality risk, and the y-axis represents the actual observed risk. The diagonal dashed line represents a perfect prediction by an ideal model. The solid line represents the bias-corrected performance of the nomogram (bootstrapping B=1000), indicating good agreement between predicted and observed outcomes. (C) Decision Curve Analysis (DCA) evaluating the clinical usefulness of the predictive models. The y-axis measures the net benefit. The blue solid line represents the “New Model” (incorporating ARC status), while the Orange dashed line represents the “Baseline Model” (without ARC status). The horizontal black line assumes no intervention (Treat None), and the gray line assumes intervention for all patients (Treat All). The New Model demonstrates an improved net benefit across the majority of threshold probabilities compared to the Baseline Model, suggesting potential incremental clinical utility.

Discussion

To our knowledge, this is the first study to systematically quantify the “dilution effect” of ARC on serum NT-proBNP levels and its consequential impact on mortality prediction in severe TBI. Our findings reveal a critical “diagnostic blind spot”: in nearly 40% of our cohort (the ARC group), the classic “brain-heart axis” signal appeared to be masked by renal hyperfiltration. While previous studies have established elevated NT-proBNP as a robust predictor of poor outcomes,^6,16 our data suggest that this dogma holds true only when renal function is “normal” or impaired. In the presence of ARC, the prognostic utility of NT-proBNP is significantly compromised (AUC dropped from 0.833 to 0.637), leading to a dangerous underestimation of mortality risk if standard reference ranges are applied.

The inverse relationship between renal function and natriuretic peptides is well-documented in renal failure, where reduced GFR leads to biomarker accumulation. Our study explores the opposite spectrum—renal hyperfiltration. We hypothesize a mechanism of “Pharmacokinetic Decoupling”: in the hyperdynamic phase of severe TBI, the massive catecholamine surge and young age may drive a supraphysiological increase in GFR.^15,17 This phenomenon creates a divergence between biological production and serum concentration. As elucidated by Udy et al,¹⁷ the TBI-induced surge in Atrial Natriuretic Peptide (ANP) acts as a physiological driver to increase glomerular filtration. If this mechanism holds, NT-proBNP—being metabolically inert and exclusively dependent on renal excretion—may become disproportionately cleared, potentially explaining the observed association between ARC and lower NT-proBNP levels. However, this proposed mechanism remains hypothetical and requires confirmation through prospective pharmacokinetic studies that directly measure renal clearance of NT-proBNP.

The most clinically relevant finding is the failure of conventional cutoffs. In our Non-ARC cohort, a cutoff of 986 pg/mL effectively stratified high-risk patients. However, in the ARC subgroup, even much lower levels failed to reach statistical significance for mortality prediction. This implies that for a young TBI patient with ARC, a “normal” NT-proBNP level should not be reassuring. This necessitates a paradigm shift in biomarker interpretation: renal function should be viewed as a continuous variable that modulates biomarker “sensitivity.” It is also important to note that although the optimal NT-proBNP cut-off value in the ARC group (847.26 pg/mL) was similar to that of the total population, the clinical interpretation differs fundamentally. The total population cut-off represents a weighted average of two distinct subgroups. In ARC patients, the same numerical value may reflect a higher underlying mortality risk due to biomarker dilution. Thus, the ARC-adjusted model serves as a recalibration tool to ensure that high-risk ARC patients are not overlooked by traditional cutoffs.

A key finding is the significant incremental value of ARC status beyond BNP for predicting mortality. While the AUC increase from Model 3 to Model 4 was modest (0.811 to 0.856, P=0.317), the reclassification metrics demonstrated superior utility: the category-free NRI was 0.590 (95% CI: 0.397–0.848) and the IDI was 0.066 (95% CI: 0.027–0.126). This divergence between AUC and reclassification indices is well-documented. As noted by Pencina et al,¹⁸ when a baseline model is already strong (AUC > 0.8), the AUC metric becomes insensitive to new markers. In such cases, NRI and IDI offer a more clinically relevant assessment. Our significant NRI indicates that adding ARC status correctly shifted the risk prediction direction for approximately 59.0% of patients compared to using BNP alone. These results confirm that ARC status is a critical, independent prognostic indicator for mortality in ICU patients. Our nomogram (Figure 5A) operationalizes this concept by integrating ARC status as a correction factor. These findings suggest that the ARC-adjusted nomogram may provide incremental value, though further prospective multicenter validation is required to confirm its clinical utility.

Our study has several limitations. First, as a single-center retrospective study, selection bias cannot be entirely ruled out, although propensity score matching was used to mitigate this. Second, we relied on creatinine clearance (CrCl) estimated by the Cockcroft-Gault formula or urinary creatinine collection to define ARC; while practical, measured GFR using exogenous markers would be more precise. Third, we assessed baseline NT-proBNP levels; dynamic changes in BNP and renal function over time might provide additional prognostic information. Fourth, the observational nature of our study precludes definitive conclusions about causality; the observed association between ARC and reduced NT-proBNP levels suggests but does not prove a direct “dilution effect” mediated by renal hyperfiltration. Fifth, the subgroup analyses were exploratory and not adjusted for multiple testing, thus these findings should be interpreted with caution and require validation in future studies. Finally, our model requires external validation in multi-center cohorts to confirm its generalizability.

Conclusion

In conclusion, our findings suggest that ARC may exert a “dilution effect” on serum NT-proBNP levels in patients with severe TBI, which potentially limits the reliability of conventional prognostic cutoffs. The integration of ARC status into a prognostic nomogram appears to account for this pharmacokinetic influence, providing enhanced mortality risk stratification. Clinicians should consider that in hyperdynamic TBI patients, a relatively “low” NT-proBNP level does not necessarily equate to low clinical risk, as it may partly reflect accelerated renal elimination. However, prospective multi-center studies are warranted to validate the clinical utility of this ARC-adjusted nomogram before it can be recommended for routine application.