Natriuretic Peptide Cut Points for Heart Failure Classification in Individuals With and Without Obesity.
Aug 20, 2025DOI: 10.1161/circheartfailure.125.013112Journal ArticleAHA
Written by: Mandana ChitsazanJuhi K ParekhLeah B KosyakovskySophie M NemethEmily S LauJames L JanuzziThomas J WangDaniel LevyChiadi E NdumeleElizabeth SelvinChristie M BallantyneBruce M PsatyJohn S GottdienerJorge R KizerChristopher R deFilippiNorrina B AllenRudolf A de BoerSanjiv J ShahJennifer E Ho
The prevalence of heart failure (HF) in the United States is projected to increase from 6.7 million in 2020 to more than 8 million by 2030, with estimates suggesting a further increase in HF prevalence by more than 30% to 2060, largely due to the aging population and accumulation of risk factors such as hypertension, diabetes, and the continued obesity epidemic.1,2 The definition and staging of HF were revised between the 2013 and subsequent 2022 American Heart Association (AHA)/American College of Cardiology (ACC)/Heart Failure Society of America (HFSA) HF guidelines.3,4 Notably, among other updates the definition of stage B HF (pre-HF) newly incorporated cardiovascular biomarkers, including elevated natriuretic peptide (NP) levels, as a new criterion in the 2022 guidelines.How the application of NP cutpoints may affect the prevalence and epidemiology of stage B HF, particularly in light of well-described NP deficiency observed among individuals with obesity, remains unclear.5,6 We previously demonstrated that applying the updated 2022 guidelines to individuals pooled across three community-based cohorts led to a significant up-classification of participants from the healthy and stage A HF (i.e., at risk) categories to stage B HF.7 Despite up-classification of more individuals, the relative risk of progression from stage B to stage C overt HF was similar. However, it is noteworthy that individuals classified as stage B HF based on the 2022 guidelines had a lower prevalence of obesity compared with the 2013 guidelines.Obesity is a recognized risk factor for HF, warranting at least stage A HF classification.4,8 Of note, individuals with obesity are known to have lower levels of NPs, yet the same NP cutpoints to define stage B HF are currently being applied across both obese and non-obese individuals.4,9 In this context, we hypothesized that the application of the updated 2022 HF criteria would disproportionately up-classify non-obese vs obese individuals to stage B HF, which would impair the utility of NP as a classification criterion in individuals with obesity. We leveraged five large community-based studies with comprehensive NP assessment and longitudinal ascertainment of incident HF. Further, we sought to investigate the performance of NPs with respect to the risk of future HF across obesity classes and to examine whether lower NP cutpoints might be considered among individuals with obesity to better inform potential opportunities for preventive strategies in the future.
METHODS
The data supporting the study findings, including clinical and biomarker data, can be obtained by application to dbGaP (https://www.ncbi.nlm.nih.gov/gap/or) or via formal requests to each cohort’s review board or Publications and Presentations committee.We conducted a restrospective pooled analysis using collected individual-level data from five prospective observational, community-based cohorts as described previously: the Framingham Heart Study (FHS) offspring cohort exam 6 (1995 −1996), the Multi-Ethnic Study of Atherosclerosis (MESA) baseline exam (2000 – 2002), the Cardiovascular Heart Study (CHS) baseline exam (1989 −1990; 1992 – 1993 for supplemental African American cohort), the Atherosclerosis Risk in Communities Study (ARIC) visit 2 (1990 −1992), and the Prevention of Renal and Vascular End-stage Disease (PREVEND) exam 1 (1997–1998).10–15 Individuals with prevalent HF (n=839), age less than 30 years at baseline exam (n=134), missing NP values (n=5091), and missing key covariates (n=295) were excluded, leaving 32,735 individuals for analysis. All participants provided written informed consent, and each participating institution obtained appropriate institutional review board approvals. FHS, MESA, CHS, and ARIC publications committees reviewed and approved our study protocol.All participants underwent a detailed medical history, physical examination, fasting laboratory test, and electrocardiogram. Cardiac imaging data were available for FHS, MESA, and CHS (n=11,311) participants. All risk factors and comorbidities were harmonized across the cohorts.7 Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters and expressed as kg/m2. Individuals were classified by BMI as normal weight (< 25 kg/m2), overweight (25 to < 30 kg/m2), obesity class 1 (30 to < 35 kg/m2), and obesity class 2/3 (35 kg/m2 or higher).Participants were classified using the 2013 and 2022 guidelines criteria, as described in detail before.7In summary, the 2022 guidelines updated the definition of stage B (pre-HF) to include individuals with evidence of structural/functional heart disease or the presence of risk factors and increased biomarkers (troponin and NP) without current or previous signs/symptoms of HF.4 For stage C/D HF events, participants were followed longitudinally within each cohort, with a formal review of hospital and outpatient records and event adjudication according to established protocols.13,16–20 Details of event adjudication for each cohort are presented in Table S1.Brain natriuretic peptide (BNP) was measured in FHS.21 N-terminal pro-B-type natriuretic peptide (NT-proBNP) was measured in PREVEND, CHS, MESA, and ARIC (see Supplemental Methods).19,22–24 NP concentrations were natural logarithm (log)-transformed due to right-skewed distribution and then standardized to mean=0 and SD=1 within cohort to account for differences in assay, demographic, and clinical characteristics and subsequently pooled across cohorts for primary analyses. We set extreme outliers of NT-proBNP > 3300 pg/ml to 3300 pg/ml (5 SD above the mean) to limit the effect of extreme outliers. We selected NT-proBNP ≥ 125 pg/ml and BNP ≥ 35 pg/ml to define elevated NP levels, in accordance with definitions from the 2022 guidelines for stage B HF.4Baseline clinical characteristics are summarized by obesity class. Continuous variables are presented as mean ± standard deviation (SD) or median (first quartile, third quartile), as appropriate. Analysis of variance, Kruskal Wallis test, or chi-square tests were used to compare the distribution of baseline characteristics across obesity classes, as appropriate.To examine reclassification of HF stages between 2013 vs 2022 guidelines stratified by obesity class, we included participants with available imaging data as described in detail before.7In the whole available sample, we then utilized multivariable Cox proportional hazards models to examine the association of NPs with incident HF (stage C/D) across obesity classes after the exclusion of individuals with prevalent HF. Specifically, we examined whether BMI and obesity class modified the association of NP with HF risk using multiplicative interaction terms (NPBMI and NPobesity class, respectively). Clinical covariates included age, sex, race/ethnicity, smoking status, diabetes mellitus, dyslipidemia, hypertension, and history of myocardial infarction (MI). The models included a strata statement for the cohort to allow for varying baseline hazard functions while estimating hazard ratios (HR) across groups.Participants were followed for up to 15 years; follow-up was censored at the time of death, loss to follow-up, or at 15 years, whichever occurred first. We used Schoenfeld residuals to ascertain that the assumption of proportional hazards was met.In analyses stratified by obesity class, we examined the incremental utility of NP when added to a clinical model using Harrell’s C-Statistics, integrated discrimination improvement (IDI), and net reclassification index (NRI), including category-based NRI (with risk categories set at < 5%, 5% to < 10%, and ≥ 10%) and category-free NRI based on a >1% change in risk at 10 years.25,26 To assess the calibration, we examined predicted vs. observed probabilities of events, categorized by predicted risk groups of <5%, 5 to <10%, and ≥10% risk.27We then plotted Receiver Operating Characteristic (ROC) curves to estimate the optimal cutpoints for NPs to predict the risk of HF in normal, overweight, and obese groups. Due to the small number of events among individuals with class 1 and class 2/3 obesity, these groups were combined for ROC analyses. Since cardiac imaging was only available for the MESA, FHS, and CHS cohorts, analyses were focused on reclassification based on NP cutpoints without considering simultaneous imaging criteria. The dataset was randomly split into two parts; 60% of the dataset was used for deriving NP cutpoints. In the 40% subset, the newly derived cutpoints were applied to compare the risk of incident HF using Cox proportional hazards models. Specifically, the analysis compared the risk of incident HF for individuals up-classified based on new vs existing NP cutpoints compared with those at risk for HF (meeting risk factors specified as stage A).In exploratory analyses, we stratified the models by sex and also examined relative fat mass = ([64− (20 × height (m) / waist circumference (m)) + (12 × sex), with sex=0 (men), and sex=1 (women)]).28,29 Statistical analyses were performed using SAS software (SAS Institute 9.4 Inc.) and R (version 4.3.2). A two-sided p-value <0.05 was considered significant.
RESULTS
Of 32,735 participants included in the analysis, the mean age was 58 ±11 years, and 54% were women. A total of 35% were normal-weight, 40% were overweight, 17% had obesity class 1, and 8% had obesity class 2/3. For distribution of BMI across cohorts, see Figure S1. Comorbid conditions, including hypertension, diabetes, and dyslipidemia, were more prevalent among individuals with obesity compared with normal-weight individuals (P<0.0001 for all). For example, 42% of normal-weight vs. 77% of individuals with class 2/3 obesity had hypertension, and 5% of normal-weight vs. 27% of individuals with class 2/3 obesity had diabetes mellitus (Table 1).As expected, NP concentrations were highest among normal-weight individuals (Table S2). When examining the NP cutpoint used to classify stage B HF among the total sample (BNP ≥ 35 or NT-proBNP ≥ 125 pg/ml), 33% among normal weight, 39% among overweight, 19% with class 1 obesity, and 9% with class 2/3 obesity met elevated NP criteria (P=<0.0001 for between-group difference, see Table S2 for distribution of NP values across the obesity classes).Next, we applied the 2013 vs. 2022 criteria to classify HF stages among the subset of 11,311 participants who had available cardiac imaging (Table S3 for baseline characteristics). The application of the 2022 vs. 2013 criteria resulted in the up-classification of healthy and stage A individuals to stage B HF across all obesity classes. The proportion of individuals classified as stage B increased from 14% to 45% for normal weight, from 15% to 43% for overweight, from 13% to 40% for individuals with obesity class 1, and from 13% to 46% for individuals with obesity class 2/3. The majority of individuals who were up-classified to stage B met the elevated NP criterion (72%), and 49% were up-classified solely based on the elevated NP criterion. Figure 1 shows the distribution of 2022 vs. 2013 criteria for stage B HF across obesity classes. Specifically, 62% of those classified as stage B according to the 2022 guidelines in the normal weight group met the NP criterion, compared with 51% in the overweight, 47% in obesity class 1, and 42% in obesity class 2/3. Among individuals who were up-classified based on NP criterion alone (1,377 participants), 39% were in the normal-weight group, followed by 39% among overweight individuals, 15% among individuals with obesity class 1, and 7% for those with obesity class 2/3. Although atrial fibrillation (AF) is a known risk factor for HF, it is not independently included as a stage B HF criterion in the 2022 guidelines. Therefore, AF was not used for staging in our analysis. However, of the 283 participants with baseline AF, 199 (70%) were classified as stage B HF based on other qualifying criteria (i.e. imaging or biomarkers).Over a median follow-up of 13 (11.7–15) years, there were 3,077 incident HF events among the total sample. This included 760 events among individuals with normal weight, 1,231 in overweight, 685 in obesity class 1, and 401 in obesity class 2/3 groups. In multivariable Cox models, both higher NP concentrations and higher BMI were associated with a greater risk of incident HF (Table 2). A 1-SD higher NP was associated with a 2.1 fold higher hazard of incident HF (HR 2.12, 95% CI 2.08 – 2.17, P <0.0001), and a 1-SD higher BMI was associated with a 39% higher hazard of incident HF (HR 1.39, 95% CI 1.35 −1.43, P <0.0001).Importantly, while higher NP levels were associated with greater HF risk across the BMI spectrum, we found that this association was modified by BMI with similar results for obesity class (Pint <0.0001 for both). Specifically, the effect of NP on higher HF risk was more pronounced among individuals with lower BMI compared to individuals with higher BMI (Figure 2). For example, among individuals with a BMI of 20 kg/m2, the HR per 1-SD higher NP was 2.21 (95% CI 2.06 – 2.37), vs. among individuals with a BMI of 40 kg/m2, the HR was 1.75 (95% CI 1.61 – 1.91). When stratified by obesity class, the HR for NP among those with normal weight was 2.24 (95% CI 2.05 – 2.45) vs. 1.73 (95% CI 1.53 – 1.96) among those with class 2/3 obesity. We observed similar results in sex-stratified analyses, such that the association of NP with HF risk was greater among men and women with lower BMI (Table S4 and Figure S2). In exploratory analyses, we found that relative fat mass modified the association of NP with the risk of incident HF (Pint <0.0001), similar to the association of BMI and NPs.We next examined the incremental utility of NP when added to a clinical model in ascertaining the risk of incident HF stratified by obesity class. The addition of NPs significantly improved the c-statistic in all obesity classes (Table 3). For example, among normal weight, the addition of NP to a clinical model resulted in an increase in the c-statistic from 0.829 (95% CI 0.816–0.842) to 0.851 (95% CI 0.838–0.863, P<0.0001 for delta). Similarly, among obesity class 1, the addition of NP increased the c-statistic from 0.753 (95% CI 0.736–0.770) to 0.785 (95% CI 0.768–0.801, P=0.0002 for delta). Addition of NP to the clinical model resulted in a category-free NRI (>1%) of 36% in individuals with normal BMI, 32% with overweight, 46% with obesity class 1, and 37% with obesity class 2/3, with other metrics shown in Table 3. When examining the calibration of the model, we found that predicted vs. observed probabilities appeared consistent across all obesity classes, with and without the addition of NP (Table S5–S8).We used ROC curve analysis to determine the optimal NP cutpoints for predicting the 15-year risk of HF across obesity classes using a random sample of 60% of the total (Table S9). The optimal NT-proBNP cutpoint for predicting HF was 109 pg/mL (95% CI 80 – 157) among normal-weight individuals. In comparison, the cutpoints were lower for overweight individuals (77 pg/mL, 95% CI 54 – 91) and those with obesity (80 pg/mL, 95% CI 53 –121).Use of newly derived cutpoints in the remaining held out 40% of the dataset resulted in the up-classification of 1,115 individuals (9%) based on elevated NPs. Over 15 years, 182 incident HF events occurred among these individuals. Compared with those at risk for HF with low NP, the HR for incident HF in individuals meeting newly derived elevated NP cutpoints specific to obesity class was 1.97 (95% CI 1.65–2.35, P < 0.0001). As a comparison, individuals classified as having elevated NP using the single NP cutpoint per current guidelines had HR for incident HF of 2.75 (95% CI 2.39–3.17, P < 0.0001).When applying the new cutpoints to the held-out 40% of the dataset (Table 4), n=3,201 (24%) met elevated NP criteria based on obesity class-specific cutpoints, whereas the use of the single NP cutpoint used in the 2022 guidelines resulted in up-classification of 2,086 individuals (16%). There were 700 events among those with elevated NP, and 518 events among those without elevated NP based on newly derived cutpoints. When comparing those with elevated NP to those at-risk for HF (based on stage A criteria), the HR for incident HF among individuals upclassified using the newly derived NP cutpoints specific to obesity class was 2.35 (95% CI 2.06–2.69, P < 0.0001). Similarly, the HR for incident HF among individuals upclassified using the 2022 guidelines single NP cutpoint was 2.21 (95% CI 1.94–2.52, P < 0.0001).
DISCUSSION
We leveraged five pooled and harmonized longitudinal community-based studies with individual-level data to examine the use of NPs to quantify the risk of HF across obesity classes. Our principal findings are as follows: first, the application of NP cutpoints to classify stage B HF according to the 2022 guidelines resulted in the up-classification of individuals in all obesity classes. While the majority of up-classifications were based on NP cutpoints, fewer individuals with overweight and obesity were up-classified based on NP criterion compared with normal-weight individuals. Second, we found that BMI modified the association of NP with subsequent HF risk, such that NPs portended a greater risk of HF among individuals with normal weight vs. those with obesity. Third, the addition of NP to clinical covariates improved model discrimination among individuals in obesity class 2/3, similar to other obesity classes. Lastly, ROC analyses suggest that lower obesity class-specific NP cutpoints may be appropriate for individuals with obesity compared with those of normal weight, with greater numbers of individuals upclassified in risk alongside similar hazard ratios of future HF compared with a single NP cutpoint.The updated 2022 vs. 2013 AHA/ACC HF guidelines provide further guidance on the classification of pre-HF (stage B) with the new addition of the NP criterion. However, NP deficiency has been well-described among those with obesity, and the effect of a single NP cutpoint across obesity classes in the reclassification of HF remains unclear.4 In light of the increasing prevalence of HF, as well as evolving medical therapies such as glucagon-like peptide-1 receptor agonists (GLP-1RA) and sodium-glucose co-transporter-2 inhibitors (SGLT2i) that may prevent the onset or progression of stage C/D HF in high-risk individuals, including those with obesity, diabetes, and chronic kidney disease, the identification of high-risk individuals has become clinically more important than ever.30–34Previous population-based studies have persistently shown an inverse relationship between NP levels and BMI.5,6 Lower NP levels in individuals with obesity have also been observed in individuals with known HF. The underlying mechanisms of natriuretic peptide (NP) deficiency in obesity remain unclear, with the initial hypothesis of overexpression of the C-type natriuretic peptide receptor (NPR-C) and increased peripheral clearance of BNP being insufficient to explain low NT-proBNP levels in individuals with obesity.5,35 Other proposed hypotheses include decreased production/secretion of NPs, a potential role for androgens in lowering NP levels by promoting the development of lean mass, adipose tissue inflammation causing local NP receptor dysfunction and BNP deficiency, and enhanced BNP degradation via the neprilysin pathway among individuals with obesity.5,36–38 In addition, known sex differences in NP may account for some of the observed differences.39 In sum, obesity has broad-based effects on the production of BNP and may enhance its clearance.40We now show that the relative changes in the proportion of individuals up-classified to stage B HF according to the 2022 NP criteria were lower among individuals with obesity compared with normal-weight individuals despite a higher prevalence of comorbidities, including hypertension, diabetes, and chronic kidney disease, which typically associate with higher NP levels. These findings suggest a potential gap in the detection of individuals with obesity at high risk of subsequent HF. Given the increased focus on the treatment of obesity as a key risk factor for a wide range of cardiovascular diseases, including HF, our findings are of particular relevance.Consistent with prior studies, we found that BMI and NP concentrations are each associated with incident HF.7,19,41 Importantly, we demonstrate a significant interaction between NP and BMI, such that the HF risk associated with higher NP is greater among individuals with lower BMI and progressively attenuates with higher BMI. We find that NP concentrations are lower among those with obesity, and a given change in NP conveys the greatest increase in relative HF risk among lean individuals. This may reflect the known NP deficiency and lower baseline NP values among those with obesity, such that a similar increase in NP carries less prognostic information in this group. Additionally, we acknowledge that our findings may be subject to potential unmeasured confounders or collider bias in the context of the well-described ‘obesity paradox’. Overall, our study adds to the growing evidence of the complex interplay between metabolic factors and cardiovascular risk.8 Specifically, the interaction between NP levels and BMI has implications for HF risk stratification in clinical practice and suggests that BMI-specific NP cutpoints may need to be considered to optimally assess HF risk across the BMI spectrum.We also demonstrated that lower NP cutpoints among individuals with obesity may allow appropriate reclassification of a greater number of individuals as stage B HF compared to the current guideline-recommended cutpoints. Notably, individuals reclassified to stage B HF using our newly derived cutpoints — who would otherwise be classified as stage A HF according to the 2022 guidelines — had similar hazard of incident HF compared with existing classification using a single cutpoint. These findings underscore the potential clinical implications of adopting BMI-adjusted NP cutpoints to improve early risk identification and optimize management strategies for individuals with obesity.Our study has several limitations. First, in our stratified analysis by obesity class, there were fewer incident HF events among individuals with obesity class 2/3 compared with other groups, limiting the power to detect associations with future HF across stratified analyses. For this reason, obesity class 1 and 2/3 were combined for the ROC analysis to improve the stability of estimated NP cutpoints. Moreover, given the modest number of events among individuals with obesity, BMI-specific NP cutpoints need to be validated in future studies. Second, given a predominantly white sample, the generalizability of our findings may be limited. Third, we acknowledge that BMI is an imperfect measure of adiposity, though we observed similar results with relative fat mass analyses. Fourth, we recognize significant improvements in the management of both HF and obesity over the past decade, which may affect secular trends in the progression to HF among individuals with obesity that have not been accounted for in our analyses. Lastly, we ascertained NP levels at one point in time, and it is unclear whether repeated measurements of NP would provide additional insights into the progression of HF in individuals with obesity.In sum, the findings from this pooled analysis of five community-based cohorts showed that use of a single NP cutpoint as outlined in current HF guidelines resulted in fewer individuals with obesity being upclassified to stage B HF compared with normal-weight individuals. We further show that NPs carry prognostic information across all obesity classes, and that BMI modified the association of NP on future HF risk, with elevated NPs indicating a greater risk of HF among individuals with normal weight compared with individuals with obesity. Finally, a lower NP cutpoint may better suit HF risk assessment in individuals with obesity. Overall, our findings provide epidemiologic evidence supporting the use of BMI-specific NP cutpoints to improve HF risk stratification and guide early prevention strategies in individuals with obesity. Further studies are needed to validate the clinical utility of BMI-specific cutpoints for HF risk assessment and classification.
