Inflammatory index-based nomogram for risk stratification of erectile dysfunction: a cross-sectional study with dual-cohort validation and mendelian randomization analysis

Introduction

Erectile dysfunction (ED), acknowledged as a pivotal challenge in male healthcare worldwide, shows markedly higher incidence rates in individuals affected by chronic metabolic disorders and those predisposed to cardiovascular complications (1). The pathogenesis of ED involves multifactorial interplay across physiological systems, including but not limited to vascular endothelial impairment, dysregulation of autonomic neural pathways, and variations in androgen concentrations (2). Clinically significant is ED’s dual manifestation: not only as a distinct sexual dysfunction entity but also as a precursor biomarker for systemic vascular pathology, demonstrating reciprocal relationships with coronary atherosclerosis and glucose metabolism disorders (3,4). Despite these associations, contemporary diagnostic paradigms predominantly depend on subjective symptom inventories [e.g., international index of erectile function-5 (IIEF-5)], lacking objective biomarkers for precise disease staging or treatment outcome prediction. This diagnostic gap highlights the urgent requirement for innovative biomarker discovery in ED management.

Subclinical inflammatory processes, recognized as central mediators of endothelial damage, compromise microvascular equilibrium in penile erectile tissues through mechanisms like pro-inflammatory cytokine secretion and redox imbalance amplification (5). Emerging evidence emphasizes the diagnostic potential of hematological inflammation indices, particularly composite parameters such as neutrophil-lymphocyte ratio (NLR) and lymphocyte-monocyte ratio (LMR). These integrative indices surpass individual leukocyte subtype counts in prognostic value, as they dynamically reflect the systemic inflammatory-anti-inflammatory equilibrium (6-8). While NLR-based predictive frameworks have been validated in cardiovascular and oncological contexts (9,10), their diagnostic utility in ED and hormonal interactions remains contentious. Current investigations predominantly derive from single-institution observational studies, lacking multicenter verification and mechanistic exploration to differentiate causal roles from epiphenomenal associations in ED progression.

Our investigation pioneers a tripartite methodological framework, synthesizing data from National Health and Nutrition Examination Survey (NHANES) with multicenter Chinese cohorts and Mendelian randomization (MR) analysis, to establish an evidence hierarchy encompassing observational correlation, clinical verification, and causal inference. Following rigorous inflammatory marker selection, we quantitatively examine non-linear associations between NLR/LMR indices and ED progression gradients. The MR component leverages Genome-Wide Association Study (GWAS) datasets, employing genetic instrumental variables to mitigate confounding and clarify causal links between immunocyte subpopulations, inflammatory mediators, and ED pathogenesis. This multidimensional analytical approach establishes a novel paradigm for investigating inflammatory cascades in ED while advancing methodologies for predictive model development. We present this article in accordance with the STROBE-MR reporting checklist (available at https://tau.amegroups.com/article/view/10.21037/tau-2025-346/rc).

Methods

Observation study design

Patients and data collection

The data used in this study were obtained from the NHANES, which is conducted annually by the National Center for Health Statistics (NCHS) of the Centers for Disease Control and Prevention (CDC) in the USA. The aim of the survey is to assess the health status and behaviors of the non-institutionalized population in the USA (11). NHANES employs a complex, multi-stage probability sampling design to collect representative population data (12). All NHANES procedures were approved under the USA “Human Research Subjects Protection Policy” and are reviewed and standardized annually by the NCHS Institutional Review Board. All participants provided informed consent, indicating their understanding of the study’s objectives.

In this analysis, data from two cycles of the NHANES survey, specifically 2001–2002 and 2003–2004, were chosen. The reason for this selection was that only these two cycles provided comprehensive data regarding ED. From 2001 to 2004, a total of 21,161 individuals took part in the NHANES survey. The diagnosis of ED was made using the self-reported questionnaire in the NHANES. All the participants were presented with the following question related to ED: “please describe your capacity to achieve and sustain an erection adequate for sexual intercourse”. The available response options were “never”, “occasionally”, “usually”, or “almost always”. Those participants who responded with “never” or “occasionally” were classified as having ED.

The covariates considered in this study encompassed various factors. These included age, ethnicity, body mass index (BMI), weight, waist measurement, marital situation (married/cohabiting or single), poverty-income ratio, educational attainment, exercise frequency, alcohol intake (drinker or non-drinker), exercise status (vigorous or moderate), smoking status (smoker or non-smoker), as well as the presence of hypertension, diabetes, cancer, stroke. Additionally, hematological parameters such as absolute lymphocyte count, absolute monocyte count, and absolute neutrophil count were also included as covariates. The exclusion criteria were as follows: females (n=10,860), males aged under 20 years (n=5,347), missing ED data (n=886), and missing covariate data (n=986). Ultimately, 3,082 participants were included in the study.

Validation data

Additionally, an independent dataset was used for external validation, comprising a Chinese dual-center validation cohort (from January 2015 to December 2024, from two hospitals: Tangdu Hospital and Xijing Hospital). The inclusion criteria for this cohort were: heterosexual orientation, complete medical records, normal hormone levels, and IIEF-5 score assessments. Exclusion criteria included: patients with spinal cord injury, concurrent neurological diseases, severe cardiovascular conditions, penile fibrosis, and patients currently receiving phosphodiesterase type 5 inhibitors (PDE-5i) treatment. ED was diagnosed for IIEF-5 scores <22 points (13). The study is registered on ClinicalTrials.gov with the number NCT06798350. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Medical Ethics Committee of the First Affiliated Hospital of the Air Force Medical University (Xijing Hospital) (approval No. KY20252014-C-1), and individual consent for this retrospective analysis was waived. Tangdu Hospital was also informed and agreed to the study. The patient screening flowcharts for both the observational study are presented in Figure 1.

Figure 1 Flowchart illustrating patient selection for this study. NHANES, National Health and Nutrition Examination Survey.

MR

MR analysis design

This investigation utilized GWAS-derived single-nucleotide polymorphisms (SNPs) as instrumental variables (IVs) to examine potential causal relationships among immune cell characteristics, inflammatory biomarkers, and ED (14). The analytical framework is illustrated in Figure 2. To assess potential reverse causation, we performed bidirectional MR analyses where ED served as the exposure variable while immune parameters and cytokine levels represented outcomes. Our approach adhered to three fundamental MR assumptions: (I) association assumption: Strong correlation between selected IVs and exposure factors; (II) independence assumption: IVs remain independent of potential confounders; (III) exclusion restrictions: IVs influence outcomes exclusively through the specified exposure pathways (15).

Figure 2 Flowchart of the MR analysis framework. IVs, instrumental variables; IVW, inverse-variance weighted; MFI, median fluorescence intensities; MR, Mendelian randomization; SNPs, single-nucleotide polymorphisms.

Data sources

For this study, immunological data from a large-scale GWAS were employed. The GWAS involved 3,757 individuals of European descent and characterized 731 immune phenotypes. High-density arrays were utilized to genotype around 22 million SNPs, covering various measurements such as absolute counts (AC) (n=118), relative counts (RC) (n=192), median fluorescence intensities (MFI) (n=389), and morphological parameters (MP) (n=32). The immune cell types included, among others, B cells, dendritic cells (DCs), mature T cells, monocytes, myeloid cells, as well as combinations of TBNK (T cells, B cells, natural killer cells) and Treg (16). The summary data of this immune-wide GWAS can be accessed publicly from the GWAS Catalog, with accession numbers ranging from GCST0001391 to GCST0002121. Another GWAS was carried out to explore the genetic variants related to 91 inflammatory cytokines. This comprehensive research aggregated data from 11 cohorts, involving a total of 14,824 European-ancestry participants. It conducted a genome-wide analysis of 91 protein quantitative trait loci (pQTL) associated with plasma proteins in these participants and measured plasma protein concentrations using the Olink Target 96 Inflammation Immunoassay panel (17). The accession numbers for this study were from GCST90274758 to GCST9027484817. The GWAS summary statistics for ED were obtained from the FinnGen Consortium Release 12 (R12), which was the latest comprehensive update of this Nordic biobank. As of November 4, 2024, the publicly accessible dataset consisted of 2,886 cases and 215,272 controls.

Selection of IVs

The following criteria were established for the selection of IVs in this study. (I) To identify IVs associated with each immune trait and inflammatory protein, a significance threshold of 1×10⁻⁵ was applied. For the results related to ED, this threshold was adjusted to 5×10⁻⁶ (18). (II) An F statistic greater than 10 was a prerequisite for each immune signature and inflammatory cytokine in the MR analysis. This ensures the strength of the association between IVs and the exposure variables. (III) To minimize the influence of related SNPs, all IVs underwent linkage disequilibrium (LD) trimming. The trimming was executed with parameters of r²=0.001 and a distance of 10,000 kb, utilizing the “TwoSampleMR” package (version 0.5.8) in R software. (IV) Palindromic SNPs were excluded from the study to avoid potential biases in the analysis. Notably, the screening criteria for IVs in the reverse MR analysis were consistent with those applied in the primary analysis, maintaining methodological consistency throughout the research process.

Statistical analysis

Baseline characteristics of the study participants were presented as weighted means ± standard deviations (SD) for continuous variables and frequencies (percentages) for categorical variables. Differences between the ED and non-ED groups were assessed using Student’s t-test for continuous variables and the Chi-squared test for categorical variables. The NLR and LMR were categorized into four quartiles, with the first quartile (Q1) designated as the reference group.

Multiple multivariable logistic regression models—unadjusted, minimally adjusted (Model I), and fully adjusted (Model II)—were employed to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for the associations between NLR/LMR and ED. Model I adjusted for age, sex, and race/ethnicity, while Model II further included education level, smoking status, alcohol consumption, BMI, diabetes, and hypertension. Restricted cubic spline (RCS) regression with three knots (10th, 50th, and 90th percentiles) was applied to examine nonlinear relationships between the Dietary Inflammatory Index (DII) and ED. Subgroup analyses stratified by age, sex, race/ethnicity, BMI, smoking status, alcohol consumption, diabetes, and hypertension were conducted to test potential interaction effects.

To identify key dietary predictors and mitigate multicollinearity, a least absolute shrinkage and selection operator (LASSO) regression model was implemented. This approach shrinks coefficients of variables with negligible contributions to the overall model, thereby enhancing predictive stability. A 10-fold cross-validation procedure was utilized for model evaluation and hyperparameter tuning. During validation, a λ-value curve was generated to visualize model performance across penalty parameters. The optimal λ-value was selected by adding one standard error to the minimum mean cross-validated error, balancing model parsimony and generalizability.

Additionally, a risk prediction nomogram incorporating critical ED-related variables was developed. Its discriminative performance was internally validated using the NHANES dataset and externally validated with a Chinese cohort. The nomogram-derived risk scores were applied to the validation sets to estimate individualized ED probabilities. Predictive accuracy was quantified using the area under the receiver operating characteristic (ROC) and area under the curve (AUC).

To investigate the causal relationships between immune cell phenotypes, inflammatory cytokines, and ED, this study subsequently computed the F-statistic for each SNP and removed those categorized as weak IVs (F <10). We then employed multiple MR approaches, including inverse-variance weighted (IVW), weighted median, weighted mode, and MR-Egger regression (19,20). The IVW method, which combines Wald ratios for each SNP under a fixed-effects model, was utilized as the primary analysis due to its superior statistical power (21). Additionally, allelic harmonization was implemented to standardize effect alleles and eliminate strand mismatches, thus enhancing the accuracy of causal inference. To address potential pleiotropic bias, we performed sensitivity analyses using approaches such as MR-Egger and MR-PRESSO to identify horizontal pleiotropy and verify the robustness of our results. All analyses were performed using the “TwoSampleMR” and “MR-PRESSO” packages in R software (version 4.3.1), with two-tailed P values <0.05 considered statistically significant.

Results

Observation study

Patient characteristics

Table 1 presents the baseline characteristics of 3,082 participants, comparing those with ED and those without (No-ED). Among the cohort, 902 (29.3%) were classified as ED, and 2,180 (70.7%) were in the No-ED group. Significant differences in inflammatory markers, specifically the NLR and LMR, were observed between the two groups. The ED group had a significantly higher NLR (2.53 vs. 2.14, P<0.001) and a lower LMR (3.50 vs. 3.91, P<0.001), indicating a pro-inflammatory state associated with ED.

In addition to inflammatory markers, there were substantial differences in sociodemographic and health-related factors between the ED and No-ED groups. Age distribution was notably different, with a higher proportion of ED patients in the 60–79 years (56.3% vs. 16.6%) and 80+ years (17.8% vs. 2.1%) categories (P<0.001). The ED group also had a significantly higher prevalence of comorbid conditions, including hypertension (70.7% vs. 35.3%, P<0.001), diabetes (27.5% vs. 7.2%, P<0.001), and stroke (8.8% vs. 1.3%, P<0.001).

Lifestyle factors also differed significantly between the two groups, with higher smoking rates in the ED group (74.1% vs. 58.5%, P<0.001) and greater alcohol consumption in the No-ED group (82.7% vs. 64.1%, P<0.001). Additionally, the ED group had a larger waist circumference (104.13 vs. 98.81 cm, P<0.001) and lower levels of physical activity, with fewer individuals engaging in vigorous exercise (16.9% vs. 39.6%, P<0.001).

These results highlight the critical role of inflammation in the pathophysiology of ED. The elevated NLR and reduced LMR in the ED group strongly suggest that ED is associated with a pro-inflammatory state, which may contribute to the development and progression of the condition.

Association between ED and inflammatory index

As illustrated in Table 2, we conducted multivariate logistic regression analyses to explore the relationship between NLR, LMR, and ED. Model 1, which was unadjusted, demonstrated that an elevated NLR was significantly linked to an increased likelihood of ED (OR =1.31, 95% CI: 1.22–1.40, P<0.001). When sociodemographic factors such as age, ethnicity, family income-to-poverty ratio, education level, marital status, and BMI were accounted for in Model 2, the strength of the association diminished slightly but remained statistically significant (OR =1.16, 95% CI: 1.07–1.26, P<0.001). After further adjustments for health and lifestyle variables, including diabetes, hypertension, stroke, cancer, smoking habits, and alcohol intake, Model 3 still showed a significant correlation between NLR and ED (OR =1.15, 95% CI: 1.06–1.24, P=0.001). In the quartile analysis, compared to the reference group (Q1), the highest quartile (Q4) of NLR consistently demonstrated the strongest association with ED across all models. While this link weakened after additional adjustments, it remained statistically robust. Conversely, no significant association was found for Q2, and the association for Q3 lost its significance following adjustments.

Regarding LMR, Model 1 revealed a notable inverse relationship with ED (OR =0.83, 95% CI: 0.78–0.88, P<0.001). When sociodemographic factors were included in Model 2, this negative association became weaker but remained significant (OR =0.94, 95% CI: 0.88–0.99, P=0.03). In Model 3, after incorporating health and lifestyle factors, the negative association persisted (OR =0.93, 95% CI: 0.87–0.99, P=0.03). Additionally, RCS analysis was utilized to assess the potential nonlinear association between NLR, LMR, and ED (Figure 3). The findings indicated a linear relationship for both inflammatory markers with ED (nonlinearity P value for NLR =0.13; nonlinearity P value for LMR =0.20). In conclusion, elevated NLR appears to be a potential risk factor for ED, especially in the highest quartile, with the association remaining significant even after controlling for confounding factors. On the other hand, higher LMR was inversely correlated with ED, with the most pronounced effects observed in the lower quartiles, although its significance diminished after full adjustment.

Figure 3 RCS analysis of nonlinear associations between inflammatory biomarkers and the risk of ED based on Model 3. (A) NLR; (B) LMR. CI, confidence interval; ED, erectile dysfunction; LMR, lymphocyte-to-monocyte ratio; NLR, neutrophil-to-lymphocyte ratio; OR, odds ratio; RCS, restricted cubic spline.

Subgroup analyses

Figure 4A,4B illustrate the findings from subgroup analyses exploring the relationships between NLR, LMR, and ED. As depicted in Figure 4A, a higher NLR is strongly linked to an increased risk of ED in the majority of subgroups, with OR exceeding 1. This implies that NLR could function as an independent risk indicator for ED. Nevertheless, in some specific subgroups, such as particular age categories or underlying disease states, the CIs includes 1, indicating a lack of statistical significance. These results suggest that while the overall trend highlights NLR as a crucial predictor of ED risk, its impact may differ across subpopulations. Conversely, Figure 4B reveals a negative relationship between LMR and ED in most subgroups, with ORs below 1, suggesting that elevated LMR levels might lower ED risk. However, certain subgroups, such as those defined by specific age brackets or health statuses, also show CIs crossing 1, indicating that the observed relationship might not be statistically significant. This variability implies that the protective role of LMR against ED may depend on population-specific factors, but the general pattern indicates an association between higher LMR and reduced ED risk.

Figure 4 Subgroup analysis of associations between inflammatory biomarkers and the risk of ED. (A) NLR; (B) LMR. CI, confidence interval; ED, erectile dysfunction; GED, general educational development; LMR, lymphocyte-to-monocyte ratio; NLR, neutrophil-to-lymphocyte ratio; OR, odds ratio.

Development of the nomogram and performance assessment

To identify risk factors closely associated with ED, a risk prediction model was developed using LASSO penalized regression. This model incorporated all 18 covariates listed in Table 2 (Figure 5). In the LASSO regression, an L1 regularization term was added to the ordinary least squares regression, compressing certain coefficients to near-zero values. This process aimed to identify and retain the most significant features or variables while minimizing the risk of overfitting. By introducing the L1 regularization term to minimize the loss function, some coefficients were reduced to zero, effectively excluding the corresponding features (Figure 5A).

Figure 5 Development and validation of an ED risk prediction model using LASSO regression and nomogram. (A) LASSO coefficient profiles of 18 covariates, with the optimal λ value (vertical dashed line) selecting significant predictors. (B) Optimal λ selection via 10-fold cross-validation, minimizing mean cross-validated error (dotted line). (C) Nomogram incorporating eight predictive factors (including inflammatory markers) for ED risk estimation in individuals aged ≥20 years. (D) ROC curve of the nomogram in the training set (AUC =0.8553, 95% CI: 0.8407–0.8698), outperforming NLR or LMR alone or combined. (E) External validation in a multicentric cohort (n=9,161) from Tangdu and Xijing Hospitals, achieving an AUC of 0.7648 (95% CI: 0.7549–0.7747). AUC, area under the curve; CI, confidence interval; ED, erectile dysfunction; LASSO, least absolute shrinkage and selection operator; LMR, lymphocyte-to-monocyte ratio; NHANES, National Health and Nutrition Examination Survey; NLR, neutrophil-to-lymphocyte ratio; ROC, receiver operating characteristic.

Based on the optimal λ-value determined from the LASSO regression (Figure 5B), eight predictive factors, including inflammatory markers, were ultimately selected. Using the final model, a nomogram was constructed to predict ED risk in individuals aged 20 years and older (Figure 5C). The nomogram achieved an AUC value of 0.8553 (95% CI: 0.8407–0.8698) (Figure 5D), outperforming the predictive ability of either NLR alone (AUC: 0.6101, 95% CI: 0.5877–0.6325), LMR alone (AUC: 0.6060, 95% CI: 0.5835–0.6285), or their combined use as predictive factors (AUC: 0.6201, 95% CI: 0.5924–0.6815), and also far exceeding the performance of clinical indicators alone (AUC: 0.5966, 95% CI: 0.5740–0.6191). This demonstrates that the nomogram provides superior predictive performance compared to using either inflammatory marker individually, in combination, or clinical indicators alone.

Validation of the nomogram

Furthermore, the model was further validated in an external multicentric validation cohort comprising datasets from Tangdu Hospital and Xijing Hospital in China. The validation cohort included 9,161 patients, with baseline characteristics revealing significant differences between the ED and non-ED groups (Table 3). The ED group exhibited higher age (61.11±14.96 vs. 42.66±12.06 years, P<0.001) and elevated inflammatory markers such as neutrophil count (4.81±3.20 vs. 3.42±0.99, P<0.001) and NLR (2.54±1.75 vs. 1.69±0.74, P<0.001). Additionally, the ED group had a higher proportion of patients aged 20–39 years (62.6% vs. 37.4%, P<0.001). The model achieved an AUC of 0.7648 (95% CI: 0.7549–0.7747) (Figure 5E), demonstrating stable predictive efficacy across both the training set and cross-center external validation cohorts. This highlights the robustness and generalizability of the model in diverse clinical settings.

MR analysis

Exploring the causal impact of inflammatory cytokines on ED

Detailed SNP information can be found in supplementary tables (available online: https://cdn.amegroups.cn/static/public/jtd-2025-896-1.xlsx, https://cdn.amegroups.cn/static/public/jtd-2025-896-2.xlsx). IVW analysis revealed multiple risk-associated immune phenotypes for ED, including IgD⁻ CD24⁻ B cell (OR =1.076, 95% CI: 1.011–1.146, P=0.02), CD33dim HLA DR⁺ CD11b⁺ AC (OR =1.066, 95 CI: 1.005–1.131, P=0.03), EM CD4⁺ T cell (OR =1.090, 95% CI: 1.019–1.167, P=0.01), and CD4⁺ AC (OR =1.087, 95% CI: 1.022–1.155, P=0.008), among other immunophenotypic profiles.

Conversely, the ORs for memory B cell AC (OR =0.943, 95% CI: 0.906–0.983, P=0.005), CD24⁺ CD27⁺ AC (OR =0.94, 95% CI: 0.903–0.978, P=0.002), CD39⁺ activated Treg %activated Treg (OR =0.87, 95% CI: 0.764–0.991, P=0.04), and memory B cell %lymphocyte (OR =0.94, 95% CI: 0.905–0.976, P=0.001) were all less than 1, indicating these cells act as protective factors against ED. The forest plot in Figure 6 illustrates the results of MR analysis, where 25 immune cell types were treated as exposure factors and ED served as the outcome. Comprehensive findings from these analyses are detailed in available online: https://cdn.amegroups.cn/static/public/jtd-2025-896-3.xlsx. MR analyses of inflammatory cell phenotypes (neutrophils, lymphocytes, monocytes) included in the 731 immune traits showed associations with ED risk that were consistent in direction with our observational findings on NLR and LMR, reinforcing a shared inflammatory pathway linking these markers to ED pathogenesis. Inverse MR analyses (ED as exposure, 25 immune cell phenotypes as outcomes) revealed no significant associations (all P>0.05), ruling out reverse causality and supporting immune cells as potential upstream factors in ED pathogenesis (see https://cdn.amegroups.cn/static/public/jtd-2025-896-4.xlsx). Methodologically robust estimates were observed for all immunophenotypes, with MR-Egger and MR-PRESSO analyses demonstrating absence of significant horizontal pleiotropy (all P>0.05; available online: https://cdn.amegroups.cn/static/public/jtd-2025-896-5.xlsx). Convergent evidence from IVW and MR-Egger approaches revealed no detectable heterogeneity (Cochran’s Q P>0.05; available online: https://cdn.amegroups.cn/static/public/jtd-2025-896-6.xlsx), supporting the validity of causal inference

Figure 6 Forest plot of MR results. CI, confidence interval; MR, Mendelian randomization; nSNP, number of single-nucleotide polymorphisms; OR, odds ratio.

Exploring the causal impact of inflammatory cytokines on ED

In follow-up analyses, a two-sample MR framework was employed to assess 91 inflammatory cytokines as exposure variables and ED as the outcome. This investigation uncovered causal associations between two specific inflammatory cytokines and ED. Using the IVW method, C-C motif chemokine 23 (CCL23) levels emerged as a risk factor for ED, with an OR of 1.135 (95% CI: 1.013–1.272, P=0.03). Using the IVW method, genetically predicted higher interleukin-8 (IL-8) levels were associated with a reduced risk of ED (OR =0.824, 95% CI: 0.700–0.970, P=0.02). Detailed MR results for these two cytokines are provided in the supplementary table (available online: https://cdn.amegroups.cn/static/public/jtd-2025-896-7.xlsx). Pleiotropy evaluations via MR-Egger and MR-PRESSO tests revealed P>0.05 for both cytokines (available online: https://cdn.amegroups.cn/static/public/jtd-2025-896-8.xlsx), indicating no evidence of pleiotropic effects. Heterogeneity analyses using IVW and MR-Egger methods further showed no significant heterogeneity across the two inflammatory cytokines (available online: https://cdn.amegroups.cn/static/public/jtd-2025-896-9.xlsx). Collectively, these findings suggest that inflammatory cytokines play a role in the pathogenesis of ED.

Discussion

This study systematically elucidates the pivotal role of systemic inflammation and immune dysregulation in the pathogenesis of ED by integrating observational analysis, predictive modeling, and MR approaches. The findings demonstrate that an elevated NLR is significantly associated with an increased risk of ED (fully adjusted OR =1.15, P=0.001), whereas a higher LMR exhibits a protective effect (OR =0.93, P=0.03). Building on these insights, this study is the first to develop and validate a nomogram model based on inflammatory biomarkers (NLR/LMR). Internal validation within the NHANES cohort confirmed its robust predictive performance (AUC =0.8553, 95% CI: 0.8407–0.8698), substantially outperforming models that rely solely on NLR (AUC =0.6101, 95% CI: 0.5877–0.6325), LMR (AUC =0.6060, 95% CI: 0.5835–0.6285), or their combined use as predictors (AUC =0.6201, 95% CI: 0.5924–0.6815). Furthermore, the model significantly exceeded the predictive performance of models incorporating only conventional clinical parameters (AUC =0.5966, 95% CI: 0.5740–0.6191), thereby facilitating a shift from subjective assessment to objective quantification in ED diagnosis. MR analysis further identified 25 immune cell phenotypes causally associated with ED, along with two inflammatory cytokines exhibiting causal relationships with ED risk. Bidirectional MR analysis excluded reverse causality of ED on immune phenotypes and inflammatory cytokines, reinforcing the hypothesis that immune dysregulation is a driving factor in ED pathogenesis.

Inflammatory immune responses play a critical role in the development of ED. Current evidence suggests that inflammatory reactions may induce endothelial cell dysfunction, leading to vascular endothelial impairment that compromises blood flow dynamics and erectile processes (22). Furthermore, chronic inflammation contributes to vascular wall damage and fibrotic remodeling, which progressively undermine the normal engorgement capacity of penile cavernous tissue (23). Beyond vascular pathways, NLR/LMR may reflect testosterone deficiency or oxidative stress, which independently contribute to ED pathogenesis (24). Additionally, inflammatory cytokines have been shown to interfere with erectile function through dual mechanisms: disrupting neural signaling pathways and altering sex hormone homeostasis (25). Specifically, the inflammatory cascade generates a spectrum of cytokines and signaling molecules, including tumor necrosis factor-alpha (TNF-α), interleukin-1 (IL-1), and interleukin-6 (IL-6), which may disrupt the synthesis and release of sex hormones while simultaneously affecting neurotransmitter transmission and neural responsiveness during erection (24). Notably, despite these established pathophysiological connections, there remains a critical gap in clinical practice regarding validated inflammatory biomarkers for ED risk assessment and predictive modeling.

The NLR, derived from absolute neutrophil and lymphocyte counts, and the LMR, calculated from lymphocyte and monocyte counts, offer clinical advantages through their cost-effectiveness and routine accessibility in standard hematological testing (26). Pathophysiologically, elevated neutrophil counts reflect non-specific inflammatory activation, while reduced lymphocyte counts indicate relative immunosuppression, collectively defining NLR’s clinical significance (27). Compared to isolated neutrophil or lymphocyte parameters, NLR and LMR demonstrate superior predictive value as composite biomarkers integrating pro- and anti-inflammatory immune pathways (28). Clinical evidence has established that elevated NLR and reduced LMR independently correlate with adverse outcomes across multiple pathologies. For instance, in chronic kidney disease (CKD) patients, higher LMR demonstrates protective effects whereas increased NLR independently predicts all-cause mortality (27); diabetic populations show 23% increased mortality risk per NLR increment (HR =1.23, 95% CI: 1.11–1.36) (29); in non-alcoholic fatty liver disease (NAFLD) cohorts, each unit NLR elevation associates with 25% increased NAFLD risk (OR =1.25, 95% CI: 1.05–1.49), while LMR increments correspond to 39% risk elevation (OR =1.39, 95% CI: 1.14–1.69) (30).

In this research, we employed multivariable logistic regression analysis to explore the relationship between the LMR, the NLR, and ED susceptibility. The results demonstrated that a higher NLR correlated with an elevated ED risk (fully adjusted OR =1.15, P=0.001), whereas increased LMR was linked to a protective effect (OR =0.93, P=0.03). To ensure the reliability of these associations and promote their clinical applicability, we constructed a predictive nomogram incorporating inflammatory markers and key clinical factors. The model’s performance was assessed through ROC curve analysis. Internal validation using the NHANES dataset indicated robust predictive accuracy, with an AUC of 0.8553 (95% CI: 0.8407–0.8698), significantly surpassing models based only on NLR (AUC =0.6101, 95% CI: 0.5877–0.6325), LMR (AUC =0.6060, 95% CI: 0.5835–0.6285), or their combination (AUC =0.6201, 95% CI: 0.5924–0.6815). By integrating LMR, NLR, and six critical clinical parameters—marital status, age, hypertension, diabetes, alcohol consumption, and smoking status—the nomogram demonstrated significantly better predictive power compared to models relying solely on clinical features (AUC =0.5966, 95% CI: 0.5740–0.6191). These results not only underscore the stability of inflammatory biomarkers as ED risk predictors across different populations but also emphasize the broader role of inflammation in the disease’s pathogenesis. This suggests a promising tool for early diagnosis and tailored intervention in clinical settings. Neutrophils may play a key role in this process. As the most abundant type of leukocyte in peripheral blood (comprising 60–70%) (31,32), neutrophils are crucial for innate immunity, contributing to early inflammatory responses via chemotaxis and phagocytosis. Moreover, they have been linked to the progression of multiple cardiovascular disorders, including heart failure (33). Additionally, neutrophils contain high levels of myeloperoxidase, NADPH oxidase, and lipoxygenase, all of which contribute to oxidative stress and endothelial dysfunction, thereby worsening disease progression. This mechanism likely represents a central driver of inflammatory imbalance (34,35), aligning with our findings that elevated NLR is associated with an increased risk of ED. While ESR, CRP, and lipids are also valuable inflammatory/metabolic markers, NLR and LMR were prioritized for their ability to reflect pro/anti-inflammatory balance and data completeness in our cohorts. Future studies with standardized measurements of these markers will help clarify their comparative utility.

Given the extensive inclusion of immune cell phenotypes within LMR and NLR, this study further employed MR analysis to elucidate the causal relationships between immune phenotypes, inflammatory factors, and ED. Notably, immunophenotypes including IgD⁻ CD24⁻ B cells, CD33dim HLA DR⁺ CD11b⁺ AC, EM CD4⁺ T cells, and CD4⁺ AC demonstrated significant risk associations with ED (OR >1.00, P<0.05). Conversely, protective effects were observed for memory B cells, CD24⁺ CD27⁺ AC, and CD39⁺ activated regulatory T cells (OR <1.00, P<0.05). Mechanistically, memory B cells—central effectors of adaptive immunity—orchestrate immune homeostasis through unique metabolic reprogramming (e.g., enhanced oxidative phosphorylation capacity) and tissue-residency predisposition, enabling sustained immunoregulatory functions in chronic disease microenvironments. Our MR analysis revealed both Memory B cell Absolute Count and Memory B cell %lymphocyte as protective factors against ED. These cells likely maintain endothelial homeostasis via non-canonical mechanisms, particularly through secretion of anti-inflammatory cytokines IL-10 and TGF-β, which inhibit monocyte/macrophage overactivation while suppressing pro-inflammatory mediators TNF-α and IL-6 (36). Emerging evidence suggests IL-10 ameliorates chronic low-grade inflammation in corporal tissue (37), potentially enhancing endothelial nitric oxide synthase (eNOS) activity to normalize NO-cGMP signaling and improve cavernosal smooth muscle relaxation (38). The CD33dim HLA DR⁺ CD11b⁺ AC phenotype, identified as a risk factor, manifests multifactorial pathogenesis. The HLA DR component (an MHC class II receptor mediating exogenous antigen presentation to CD4⁺ T cells) correlates with lymphocyte infiltration intensity. Pathologically elevated serum CRP levels associate with increased CD11b⁺/CD33⁺/HLA-DR- myeloid cells (31), aligning with our findings. Bidirectional MR analyses excluded reverse causation between ED and these immunophenotypes, substantiating the causal paradigm of immune dysregulation driving ED pathogenesis.

At the level of inflammatory cytokines, CCL23 was identified as a risk factor for ED (OR =1.135, P=0.03). CCL23 contributes to the onset and progression of various inflammatory diseases, such as rheumatoid arthritis, chronic rhinosinusitis, chronic renal insufficiency, and systemic sclerosis, as documented in prior studies (39-41). Functionally, circulating CCL23 interacts with CC chemokine receptor 1 (CCR1), facilitating immune cell recruitment into inflamed microenvironments. This CCL23-CCR1 axis activation promotes the secretion of additional pro-inflammatory cytokines, including macrophage inflammatory protein (MIP)-1α, IL-1β, and tumor necrosis factor-α (TNF-α), thereby aggravating cavernosal smooth muscle damage (42,43).

Conversely, IL-8 exhibited a protective effect against ED (OR =0.824, P=0.02). Mechanistically, IL-8 enhances penile hemodynamics through two pathways: (I) reducing phosphorylated myosin light chain (p-MLC) formation via inhibition of the RhoA/ROCK signaling cascade, thereby lowering calcium sensitivity in cavernosal smooth muscle; (II) enhancing the responsiveness of large-conductance calcium-activated potassium (BKCa) channels to nitric oxide (NO), synergistically improving smooth muscle relaxation. This dual mechanism helps maintain vascular homeostasis by balancing vasodilation and vasoconstriction in corpora tissue. Generally, CCL23 may promote endothelial damage via CCR1-mediated recruitment of pro-inflammatory macrophages, impairing NO production. Conversely, IL-8 enhances BKCa channel sensitivity to NO, improving cavernosal smooth muscle relaxation.

Several limitations merit consideration. First, the cross-sectional design of the NHANES dataset constrains causal inference, though MR analyses partially addressed this issue. Second, the cross-sectional design and self-reported ED may underestimate prevalence in younger cohorts and limit causal inference. Third, residual confounding by unmeasured psychosocial factors (e.g., relationship stress) cannot be excluded. Fourth, the Chinese validation cohort used retrospective data, which may introduce selection bias due to variations in medical record completeness and inclusion criteria. Prospective studies with standardized data collection are needed to confirm the nomogram’s generalizability. Fifth, due to the lack of data on acute infections, recent stress, and other transient states in the NHANES database, we were unable to quantify their potential impact on NLR/LMR. This represents a limitation of the current analysis, which we aim to address in future prospective studies by incorporating systematic assessments of such factors. Temporal differences between the NHANES [2001–2004] and Chinese [2015–2024] cohorts represent a potential limitation. However, core measurements (ED diagnosis, NLR/LMR calculation) were consistent, and multivariable models adjusted for time-sensitive confounders (e.g., comorbidity management). The consistent direction and magnitude of associations across cohorts support the robustness of findings, with future work planned to include temporally aligned datasets. Missing data (ED status and covariates) were handled via complete-case analysis, as missingness was random and did not bias baseline characteristics. While this is a limitation, our sample size remained statistically robust, and results align with standard practices in NHANES research. Future mechanistic investigations employing single-cell RNA sequencing and animal models are essential to elucidate how these immunophenotypes regulate eNOS activity and smooth muscle function. Lastly, further validation is required to determine the generalizability of these findings beyond European and East Asian populations.

This study systematically highlights the critical role of immune dysregulation and inflammation in ED pathogenesis by integrating cross-sectional data analysis, multi-center clinical validation, multivariate regression modeling, and MR methodologies. Our results establish NLR and LMR as clinically relevant inflammatory biomarkers for ED risk assessment, while the LASSO regression-derived nomogram offers a validated predictive tool for clinical application. In clinical practice, the nomogram can be used to calculate ED risk scores. Doctors can quickly obtain the ED risk probability of patients (such as low-risk <30%, medium risk 30–50%, high-risk >50%) by conducting simple inquiries (age, smoking/drinking history, underlying medical history) and routine blood tests (obtaining NLR, LMR), matching the corresponding scores of each indicator on the nomogram and summarizing them. Notably, the observed associations between NLR/LMR and ED in the NHANES cohort are based on cross-sectional data, thus remaining correlational rather than causal. Future prospective studies are needed to confirm whether dynamic changes in NLR/LMR can predict ED onset or progression. Moving forward, research should focus on elucidating the mechanistic contributions of specific inflammatory mediators and immune phenotypes to ED progression, paving the way for targeted immunomodulatory interventions. Emphasis should be placed on translational studies investigating therapeutic modulation of NLR/LMR dynamics and memory B cell-mediated immune regulation, which may collectively restore endothelial function and improve cavernosal hemodynamics in ED patients.

Conclusions

This study is a pioneering effort in developing and validating a nomogram model based on the inflammatory index (NLR/LMR), representing a significant shift from subjective evaluation to objective quantification in the diagnosis of ED, thereby improving predictive accuracy. Additionally, our research has revealed potential links between immune phenotypes, inflammatory factors, and ED, offering new insights into the pathophysiology of the condition. The risk stratification system derived from our model holds great promise in providing strong theoretical support and practical guidance for clinical management and personalized decision-making in ED treatment.

Acknowledgments

The authors gratefully acknowledge Jiaxin Kang for technical assistance.

Footnote

Reporting Checklist: The authors have completed the STROBE-MR reporting checklist. Available at https://tau.amegroups.com/article/view/10.21037/tau-2025-346/rc

Data Sharing Statement: Available at https://tau.amegroups.com/article/view/10.21037/tau-2025-346/dss

Peer Review File: Available at https://tau.amegroups.com/article/view/10.21037/tau-2025-346/prf

Funding: This work was supported by the National Natural Science Foundation of China (Nos. 82201774 and 82220108004) and the China Postdoctoral Science Foundation (Nos. 2024M754265 and 2024T171196).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tau.amegroups.com/article/view/10.21037/tau-2025-346/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Medical Ethics Committee of the First Affiliated Hospital of the Air Force Medical University (Xijing Hospital) (approval No. KY20252014-C-1), and individual consent for this retrospective analysis was waived. Tangdu Hospital was also informed and agreed to the study.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Shamloul R, Ghanem H. Erectile dysfunction. Lancet 2013;381:153-65. [Crossref] [PubMed]
2. Abu-Soud HM, Camp OG, Ramadoss J, et al. Regulation of nitric oxide generation and consumption. Int J Biol Sci 2025;21:1097-109. [Crossref] [PubMed]
3. Köhler TS, Kloner RA, Rosen RC, et al. The Princeton IV Consensus Recommendations for the Management of Erectile Dysfunction and Cardiovascular Disease. Mayo Clin Proc 2024;99:1500-17. [Crossref] [PubMed]
4. Götzinger F, Kunz M, Lauder L, et al. New ways of mitigating aldosterone in cardiorenal disease. Eur Heart J Cardiovasc Pharmacother 2024;10:557-65. [Crossref] [PubMed]
5. Wang S, Wang Z, Zang Z, et al. A Mitochondrion-Targeting Piezoelectric Nanosystem for the Treatment of Erectile Dysfunction via Autophagy Regulation. Adv Mater 2025;37:e2413287. [Crossref] [PubMed]
6. Shi C, Cao H, Zeng G, et al. The relationship between complete blood cell count-derived inflammatory biomarkers and benign prostatic hyperplasia in middle-aged and elderly individuals in the United States: Evidence from NHANES 2001-2008. PLoS One 2024;19:e0306860. [Crossref] [PubMed]
7. Ferro M, Caputo VF, Barone B, et al. Lymphocyte to Monocyte Ratio: A New Independent Prognostic Factor in Bladder Cancer Progression? Front Oncol 2021;11:754649. [Crossref] [PubMed]
8. Guo B, Liu X, Si Q, et al. Associations of CBC-Derived inflammatory indicators with sarcopenia and mortality in adults: evidence from Nhanes 1999 ∼ 2006. BMC Geriatr 2024;24:432. [Crossref] [PubMed]
9. Chi Y, Jiang H, Yin Y, et al. Macrophage Signaling Pathways in Health and Disease: From Bench to Bedside Applications. MedComm (2020) 2025;6:e70256.
10. Xiao Y, Tang B, Wang J, et al. Study of the interaction between cardiometabolic index and inflammatory index on the risk of prostate cancer development. Front Immunol 2025;16:1591879. [Crossref] [PubMed]
11. Mao W, Wu J, Zhang Z, et al. Neutrophil-lymphocyte ratio acts as a novel diagnostic biomarker for kidney stone prevalence and number of stones passed. Transl Androl Urol 2021;10:77-86. [Crossref] [PubMed]
12. Mao W, Hu Q, Chen S, et al. Polyfluoroalkyl chemicals and the risk of kidney stones in US adults: A population-based study. Ecotoxicol Environ Saf 2021;208:111497. [Crossref] [PubMed]
13. van Kollenburg RAA, de Bruin DM, Wijkstra H. Validation of the Electronic Version of the International Index of Erectile Function (IIEF-5 and IIEF-15): A Crossover Study. J Med Internet Res 2019;21:e13490. [Crossref] [PubMed]
14. Sekula P, Del Greco M F, Pattaro C, et al. Mendelian Randomization as an Approach to Assess Causality Using Observational Data. J Am Soc Nephrol 2016;27:3253-65. [Crossref] [PubMed]
15. Bowden J, Holmes MV. Meta-analysis and Mendelian randomization: A review. Res Synth Methods 2019;10:486-96. [Crossref] [PubMed]
16. Orrù V, Steri M, Sidore C, et al. Complex genetic signatures in immune cells underlie autoimmunity and inform therapy. Nat Genet 2020;52:1036-45. [Crossref] [PubMed]
17. Zhao JH, Stacey D, Eriksson N, et al. Genetics of circulating inflammatory proteins identifies drivers of immune-mediated disease risk and therapeutic targets. Nat Immunol 2023;24:1540-51. [Crossref] [PubMed]
18. Davies NM, Holmes MV, Davey Smith G. Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ 2018;362:k601. [Crossref] [PubMed]
19. Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol 2015;44:512-25. [Crossref] [PubMed]
20. Bowden J, Davey Smith G, Haycock PC, et al. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genet Epidemiol 2016;40:304-14. [Crossref] [PubMed]
21. Burgess S, Butterworth A, Thompson SG. Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol 2013;37:658-65. [Crossref] [PubMed]
22. Luan J, Yu M, Gu Q, et al. Fatty acid synthase inhibition improves hypertension-induced erectile dysfunction by suppressing oxidative stress and NLRP3 inflammasome-dependent pyroptosis through activating the Nrf2/HO-1 pathway. Front Immunol 2024;15:1532021. [Crossref] [PubMed]
23. Liu D, Ghani D, Wain J, et al. Concomitant elevated serum levels of tenascin, MMP-9 and YKL-40, suggest ongoing remodeling of the heart up to 3 months after cardiac surgery after normalization of the revascularization markers. Eur J Med Res 2022;27:208. [Crossref] [PubMed]
24. Cantone E, Massanova M, Crocetto F, et al. The relationship between obstructive sleep apnoea and erectile dysfunction: An underdiagnosed link? A prospective cross-sectional study. Andrologia 2022;54:e14504. [Crossref] [PubMed]
25. He X, Li Y, Deng B, et al. The PI3K/AKT signalling pathway in inflammation, cell death and glial scar formation after traumatic spinal cord injury: Mechanisms and therapeutic opportunities. Cell Prolif 2022;55:e13275. [Crossref] [PubMed]
26. Adane T, Melku M, Worku YB, et al. The Association between Neutrophil-to-Lymphocyte Ratio and Glycemic Control in Type 2 Diabetes Mellitus: A Systematic Review and Meta-Analysis. J Diabetes Res 2023;2023:3117396. [Crossref] [PubMed]
27. Yu R, Zhang L, Zhang J, et al. The diagnostic and prognostic role of novel biomarkers in anti-neutrophil cytoplasmic antibody-associated vasculitis. Front Immunol 2025;16:1588287. [Crossref] [PubMed]
28. Chen Y, Nie Y, Wu J, et al. Association between systemic inflammatory indicators with the survival of chronic kidney disease: a prospective study based on NHANES. Front Immunol 2024;15:1365591. [Crossref] [PubMed]
29. Dong G, Gan M, Xu S, et al. The neutrophil-lymphocyte ratio as a risk factor for all-cause and cardiovascular mortality among individuals with diabetes: evidence from the NHANES 2003-2016. Cardiovasc Diabetol 2023;22:267. [Crossref] [PubMed]
30. Liu K, Tang S, Liu C, et al. Systemic immune-inflammatory biomarkers (SII, NLR, PLR and LMR) linked to non-alcoholic fatty liver disease risk. Front Immunol 2024;15:1337241. [Crossref] [PubMed]
31. Xiang J, He L, Li D, et al. Value of the systemic immune-inflammation index in predicting poor postoperative outcomes and the short-term prognosis of heart valve diseases: a retrospective cohort study. BMJ Open 2022;12:e064171. [Crossref] [PubMed]
32. Shao S, Yang Q, Pan R, et al. Interaction of Severe Acute Respiratory Syndrome Coronavirus 2 and Diabetes. Front Endocrinol (Lausanne) 2021;12:731974. [Crossref] [PubMed]
33. Langseth MS, Andersen GØ, Husebye T, et al. Neutrophil extracellular trap components and myocardial recovery in post-ischemic acute heart failure. PLoS One 2020;15:e0241333. [Crossref] [PubMed]
34. Dubey N, Wu CY, Zubiri L, et al. Predictors of Long-Term Survival in Patients With Immune Checkpoint Inhibitor-Associated Myocarditis. J Am Heart Assoc 2025;14:e038719. [Crossref] [PubMed]
35. Yuksel BT, Isık M, Tanyeli O, et al. Effect of Systemic Immune-Inflammation Index on Prognosis in Non-Functional Mitral Regurgitation Patients Undergoing Isolated Mitral Valve Replacement. Braz J Cardiovasc Surg 2025;40:e20230362. [Crossref] [PubMed]
36. Hasan MM, Nair SS, O'Leary JG, et al. Implication of TIGIT(+) human memory B cells in immune regulation. Nat Commun 2021;12:1534. [Crossref] [PubMed]
37. Mao Y, Zha Y, Zang Y, et al. Isorhamnetin improves diabetes-induced erectile dysfunction in rats through activation of the PI3K/AKT/eNOS signaling pathway. Biomed Pharmacother 2024;177:116987. [Crossref] [PubMed]
38. El Assar M, La Fuente JM, Sosa P, et al. PKC Inhibition Improves Human Penile Vascular Function and the NO/cGMP Pathway in Diabetic Erectile Dysfunction: The Role of NADPH Oxidase. Int J Mol Sci 2024;25:3111. [Crossref] [PubMed]
39. Oliveira C, Temesgen-Oyelakin Y, Naqi M, et al. A Multiomic Analysis to Identify Drivers of Subclinical Vascular Disease in Systemic Lupus Erythematosus. Arthritis Rheumatol 2024;76:1501-11. [Crossref] [PubMed]
40. Goonewardena SN, Chen Q, Tate AM, et al. Monocyte-Mediated Thrombosis Linked to Circulating Tissue Factor and Immune Paralysis in COVID-19. Arterioscler Thromb Vasc Biol 2024;44:1124-34. [Crossref] [PubMed]
41. Hao W, Li W, Wang X, et al. Profiles of cytokines in patients with antineutrophil cytoplasmic antibody-associated vasculitis. Front Immunol 2024;15:1428044. [Crossref] [PubMed]
42. Kim J, Kim YS, Ko J. CK beta 8/CCL23 induces cell migration via the Gi/Go protein/PLC/PKC delta/NF-kappa B and is involved in inflammatory responses. Life Sci 2010;86:300-8. [Crossref] [PubMed]
43. Wang L, Wei C, Jing J, et al. Associations of dietary inflammatory index and composite dietary antioxidant index with erectile dysfunction and the mediating role of metabolic dysregulation: a cross-sectional analysis of NHANES 2001-2004 data. Front Nutr 2025;12:1538874. [Crossref] [PubMed]