Abstract
Objective
This study aimed to comparatively evaluate the association between geriatric nutritional risk index (GNRI), prognostic nutritional index (PNI), and controlling nutritional status (CONUT) scores and long-term mortality in patients undergoing transcatheter aortic valve replacement (TAVR).
Method
This retrospective observational study included 262 patients who underwent TAVR for severe symptomatic aortic stenosis between January 2015 and December 2018. Patients were divided into two groups based on all-cause mortality development during 7.6 years of follow-up: survivors and those who died. Nutritional status was assessed using pre-admission laboratory and anthropometric data, as well as GNRI, PNI, and CONUT scores.
Results
The mean age was 79.0 ± 6.4 years, and 45% of the patients were female. During the follow-up period, 106 patients (40.5%) died. Malnutrition prevalence was found to be 79.8% according to GNRI, 63.7% according to PNI, and 60.7% according to CONUT. In univariate analysis, GNRI (HR: 0.974, p = 0.035), PNI (HR: 0.959, p = 0.007), and CONUT (HR: 1.140, p < 0.001) were significantly associated with mortality. However, in multivariate analysis, none of these scores remained as independent predictors. In ROC analysis, AUC values for GNRI, CONUT, and PNI were found to be 0.626, 0.610, and 0.590, respectively. In Kaplan-Meier analysis, survival was lower in patients with poor nutritional status (log-rank p < 0.001).
Conclusion
In patients undergoing TAVR, impaired nutritional status is associated with long-term mortality; However, GNRI, PNI, and CONUT scores are not clinically independent predictors of mortality. These scores may provide complementary information regarding nutritional status but showed limited discriminatory ability for mortality prediction.
Keywords: Transcatheter aortic valve replacement, Long-term follow-up, Nutritional status, Mortality
Introduction
Aortic stenosis (AS) is one of the most common valve diseases, particularly in the elderly population, and it has a progressive course. With the increase in life expectancy, the prevalence of degenerative aortic valve disease is increasing, and mortality remains a significant clinical problem in patients with severe symptomatic AS [1]. Currently, transcatheter aortic valve replacement (TAVR) is widely used as an effective and safe treatment option for patients unsuitable for surgical aortic valve replacement or those at high surgical risk [2]. Despite technological advances and increased experience, mortality after TAVR remains a significant clinical problem, especially in elderly and frail patients [3]. The patient population undergoing TAVR generally consists of elderly individuals with multiple comorbidities and frail conditions. Frailty syndrome is characterized by physical weakness, decreased reserve capacity, and increased sensitivity to stressors. Malnutrition is a significant component of this syndrome [4]. Malnutrition is particularly common among elderly patients and can negatively impact clinical outcomes through mechanisms such as weakened immune response, increased risk of infection, delayed healing, and decreased muscle mass [2]. In the context of cardiovascular diseases, malnutrition is associated with poor prognosis in conditions such as heart failure, coronary artery disease, atrial fibrillation, and peripheral artery disease [5–7].
Malnutrition is common among patients undergoing TAVR, and studies have reported that a significant proportion have varying degrees of nutritional deficiencies. For example, in the study by Ishizu et al., which included a large series of patients, malnutrition was detected in the vast majority of patients undergoing TAVR with at least one nutritional assessment score, and poor nutritional status was shown to be associated with increased mortality [5]. Similarly, various studies have reported that malnutrition is an important prognostic factor for short- and long-term mortality after TAVR [8]. Different objective indices have been developed to assess nutritional status. Among these, the Geriatric Nutritional Risk Index (GNRI) [9], Prognostic Nutritional Index (PNI) [10], and Controlling Nutritional Status (CONUT) [11] scores are among the most frequently used methods. These indices provide an objective assessment of nutritional status using parameters such as serum albumin, lymphocyte count, cholesterol level, and body weight [12]. These scores have been shown to have prognostic value in different cardiovascular diseases and have recently begun to be used in TAVR patients [2, 4, 12, 13].
A common feature of these studies is the comparison of nutritional indices with short-term mortality. However, a standard method for assessing nutritional status in TAVR patients has not yet been established, and studies comparing the prognostic value of different nutritional indices are limited [12].
In this context, assessing nutritional status in patients who have had TAVR and revealing its relationship with long-term mortality may be clinically important. Early determination of nutritional status can help identify high-risk patients and improve clinical outcomes through appropriate nutritional interventions. This study aims to evaluate the relationship between nutritional status and all-cause mortality during long-term follow-up in patients who have had TAVR.
Materials and methods
Study design and study population
This is a retrospective observational study evaluating the relationship between nutritional status and long-term all-cause mortality in patients undergoing TAVR. Our study included consecutive patients who underwent TAVR for severe symptomatic aortic stenosis at Dicle University Faculty of Medicine Hospital between January 2015 and December 2018. The indication for TAVR was evaluated for each patient by a multidisciplinary cardiac team comprising cardiology, cardiac surgery, and anesthesia specialists, according to the current ESC/EACTS or ACC/AHA guidelines [14, 15]. Patients at high risk for surgical aortic valve replacement or unsuitable for surgery were considered suitable for TAVR. In line with previous studies, patients without laboratory data to calculate nutritional indices, individuals with diseases that could affect nutritional parameters, such as hematological malignancy or active infection, and patients for whom follow-up data were unavailable were excluded from the study [5]. After applying the exclusion criteria, 262 patients were included in the study and divided into two groups based on mortality status. Patients who experienced all-cause mortality were defined as the non-survivor group (n = 106), while patients who were still alive at the end of the follow-up period were classified as the survivor group (n = 156). Our study was conducted in accordance with the principles of the Declaration of Helsinki. It was approved by the Clinical Research Ethics Committee of Gazi Yaşargil Training and Research Hospital, Health Sciences University, with decision number 101 dated 02.03.2026 [16].
Definitions
Hypertension, diabetes mellitus, atrial fibrillation, chronic kidney disease, coronary artery disease, and other comorbidities were defined according to established guideline criteria and documented medical history [17–19].
Clinical and demographic data
Clinical, laboratory, and echocardiographic data were retrospectively obtained from electronic records. TAVR procedures were performed in a hybrid operating room using standard techniques, with valve type and access route determined according to patient anatomy and clinical characteristics.
Nutritional status assessment
The nutritional status of patients was assessed using laboratory and anthropometric data obtained at hospital admission, before the procedure. To objectively determine nutritional status, three different nutritional indices commonly used in the literature were calculated:
Prognostic nutritional index (PNI)
PNI was calculated using the following formula:
[10].
PNI ≥ 45 was considered normal nutritional status, PNI 38-44.9 indicated mild-to-moderate malnutrition, and PNI < 38 indicated severe malnutrition.
Geriatric nutritional risk index (GNRI)
GNRI was calculated using the following formula:
[9].
Nutritional risk was categorized according to GNRI values as follows: GNRI > 98, indicating no nutritional risk; GNRI 92–98, indicating mild risk; GNRI 82–91, indicating moderate risk; and GNRI < 82, indicating severe nutritional risk.
CONUT Score
The CONUT score was calculated based on serum albumin level, total cholesterol level, and total lymphocyte count [11]. According to the total score, 0–1 points were considered normal nutritional status, 2–4 points mild malnutrition, 5–8 points moderate malnutrition, and 9–12 points severe malnutrition.
Follow-up and endpoints
The primary endpoint of the study was defined as long-term all-cause mortality after TAVR. Mortality data were obtained through hospital records, outpatient visits, and telephone interviews. Follow-up status and duration were available for all 262 patients in the supplied dataset. No patient was coded as lost to follow-up. Patients alive at last contact were censored at the last available follow-up date. Median follow-up was 383 days (IQR 131–1245; maximum 2749 days).
Statistical analysis
Statistical analyses were performed using IBM SPSS Statistics for Windows, Version 27.0 (IBM Corp., Armonk, NY, USA). The distribution of continuous variables was evaluated using the Kolmogorov–Smirnov test and histograms. Continuous variables showing a normal distribution were expressed as mean ± standard deviation, while variables not showing a normal distribution were expressed as median and interquartile range (IQR). Categorical variables were presented as numbers and percentages (%).
For comparisons between two groups, the Student t-test was used for continuous variables with a normal distribution, and the Mann–Whitney U test was used for variables not normally distributed. The chi-square test, or, where appropriate, the Fisher exact test, was used to compare categorical variables. Cox proportional hazards regression analysis was performed to evaluate factors associated with total mortality. First, variables were evaluated using univariate Cox regression analysis. Variables associated with mortality in univariate analysis (p < 0.10) and those considered clinically significant based on previous TAVR literature were considered for multivariate adjustment. Since there is strong clinical and biological overlap among renal function parameters (creatinine, urea, chronic kidney disease, and eGFR), only eGFR was retained in the final models to avoid redundancy and multicollinearity. Similarly, albumin was not included in multivariate models as it is a component of GNRI, PNI, and CONUT scores. STS score and sex were retained due to their proven prognostic significance in TAVR populations. The proportional hazards assumption was assessed for all multivariable Cox regression models using Schoenfeld residuals. Global and variable-specific tests were evaluated for each model. Results were reported as hazard ratios (HRs) and 95% confidence intervals (CIs).
Receiver operating characteristic (ROC) analysis was performed to evaluate the performance of nutritional indices in predicting mortality, and the results were expressed as the area under the curve (AUC). Optimal threshold values were determined using the Youden index, and the corresponding sensitivity and specificity values were reported. Survival analyses were performed using the Kaplan–Meier method, and differences between groups were evaluated using the log-rank test. In addition, the Harrell C-index was calculated to evaluate the prognostic performance of the clinical model. The change in the model’s discriminatory power with the addition of nutritional scores to the clinical model was evaluated. In all statistical analyses, a two-sided p < 0.05 value was considered statistically significant.
Results
Baseline characteristics
A total of 262 patients who underwent TAVR after meeting exclusion criteria were included in the study. The follow-up period was 7.6 years, and a total of 106 patients (40.5%) died. The mean age of the study population was 79.0 ± 6.4 years, and 45% were female. Compared to survivors, patients who died during the follow-up period were significantly older and had more advanced symptoms of heart failure. In particular, the NYHA functional class was significantly higher in non-survivors. In addition, chronic kidney disease and pulmonary edema were observed more frequently in patients who died. Detailed baseline characteristics of the study population are shown in Table 1. In Table 2, postoperative complications and postoperative treatments are summarized.
Table 1.
Baseline characteristics of the total population
| Variable | Total population (n = 262) |
Survivors (n = 156) |
Non-survivors (n = 106) |
p value |
|---|---|---|---|---|
| Age, (years) | 79.04 ± 6.41 | 78.21 ± 6.14 | 80.26 ± 6.63 | 0.012 |
| Male gender, n (%) | 122 (46.6%) | 75 (48.1%) | 47 (44.3%) | 0.639 |
| STS score | 8.52 ± 2.74 | 8.61 ± 2.81 | 8.38 ± 2.64 | 0.493 |
| Body mass index (kg/m2) | 21.89 ± 1.78 | 21.90 ± 1.88 | 21.87 ± 1.63 | 0.870 |
| NYHA Class | 3.35 ± 0.55 | 3.21 ± 0.51 | 3.56 ± 0.54 | < 0.001 |
| Pulmonary edema, n (%) | 9 (3.4%) | 2 (1.3%) | 7 (6.6%) | 0.048 |
| Comorbid diseases | ||||
| Hypertension, n (%) | 146 (55.7%) | 83 (53.2%) | 63 (59.4%) | 0.385 |
| Diabetes mellitus, n (%) | 66 (25.2%) | 38 (24.4%) | 28 (26.4%) | 0.817 |
| Dyslipidemia, n (%) | 69 (26.3%) | 44 (28.2%) | 25 (23.6%) | 0.490 |
| PCI history, n (%) | 87 (33.2%) | 55 (35.3%) | 32 (30.2%) | 0.471 |
| CABGO history, n (%) | 31 (11.8%) | 21 (13.5%) | 10 (9.4%) | 0.426 |
| Prosthetic heart valve, n (%) | 4 (1.5%) | 3 (1.9%) | 1 (0.9%) | 0.903 |
| Peripheral artery disease, n (%) | 7 (2.7%) | 4 (2.6%) | 3 (2.8%) | 1.000 |
| COPD, n (%) | 29 (11.1%) | 16 (10.3%) | 13 (12.3%) | 0.758 |
| Atrial fibrillation, n (%) | 56 (21.4%) | 30 (19.2%) | 26 (24.5%) | 0.383 |
| Cerebrovascular disease, n (%) | 4 (1.5%) | 2 (1.3%) | 2 (1.9%) | 1.000 |
| Chronic kidney failure, n (%) | 73 (27.9%) | 34 (21.8%) | 39 (36.8%) | 0.012 |
| Heart failure, n (%) | 103 (39.3%) | 66 (42.3%) | 37 (34.9%) | 0.282 |
| Anemia, n (%) | 138 (52.7%) | 71 (45.5%) | 67 (63.2%) | 0.007 |
| Smoking cigarette, n (%) | 65 (24.8%) | 37 (23.7%) | 28 (26.4%) | 0.726 |
| GNRI | 91.75 ± 7.45 | 93.16 ± 6.29 | 89.69 ± 8.51 | < 0.001 |
| PNI Score | 42.97 ± 6.24 | 43.96 ± 5.56 | 41.52 ± 6.90 | 0.003 |
| CONUT Score | 2.86 ± 2.36 | 2.44 ± 2.01 | 3.51 ± 2.71 | 0.001 |
| Total hospitalization duration (days)-IQR | 3 (5) | 3 (4) | 3 (6) | 0.023 |
| Echocardiography findings | ||||
| Low Flow Low Gradient, n (%) | 27 (10.3%) | 15 (9.6%) | 12 (11.3%) | 0.778 |
| Aortic Max Gradient, (mmHg) | 79.70 ± 16.15 | 80.23 ± 17.47 | 78.92 ± 14.02 | 0.501 |
| Aortic Mean Gradient, (mmHg) | 49.01 ± 10.30 | 49.22 ± 11.19 | 48.70 ± 8.87 | 0.676 |
| Aortic valve area (cm2) | 0.68 ± 0.18 | 0.67 ± 0.19 | 0.69 ± 0.18 | 0.266 |
| Ejection fraction (%) | 50.75 ± 11.80 | 50.16 ± 11.95 | 51.62 ± 11.57 | 0.329 |
| Left ventricular diastolic diameter,(cm) | 4.9 (0.7) | 4.9 (0.8) | 4.8 (0.8) | 0.393 |
| Left atrial diameter, (cm) | 4.47 ± 0.58 | 4.45 ± 0.57 | 4.51 ± 0.60 | 0.414 |
| Interventricular septum thickness, (cm) | 1.40 ± 0.18 | 1.42 ± 0.18 | 1.38 ± 0.17 | 0.084 |
| Moderate-to-severe MR, n (%) | 77 (29.4%) | 41 (26.3%) | 36 (34.0%) | 0.180 |
| Moderate-to-severe AR, n (%) | 31 (11.8%) | 23 (14.7%) | 8 (7.5%) | 0.129 |
| Moderate-to-severe TR, n (%) | 59 (22.5%) | 29 (18.6%) | 30 (28.3%) | 0.076 |
| Pulmonary artery pressure, (mmHg) | 41 (20) | 40 (23) | 45 (17) | 0.034 |
| Laboratory findings | ||||
| White blood cell, (×10³/µL) | 8.35 ± 2.84 | 8.37 ± 2.88 | 8.31 ± 2.79 | 0.875 |
| Hemoglobin, (g/dL) | 12.14 ± 1.93 | 12.23 ± 1.97 | 11.99 ± 1.87 | 0.311 |
| Neutrophil, (×10³/µL) | 5.84 ± 2.82 | 5.89 ± 3.08 | 5.77 ± 2.39 | 0.708 |
| Platelet, (×10³/µL) | 238.1 ± 83 | 236.5 ± 85.4 | 240.6 ± 79.7 | 0.694 |
| Lymphocyte, (×10³/µL) | 1.67 ± 0.68 | 1.67 ± 0.65 | 1.67 ± 0.73 | 0.958 |
| GFR, (ml/min/1.73m2) | 72.25 ± 26.4 | 75.61 ± 24.2 | 67.21 ± 28.8 | 0.015 |
| Urea, (mg/dL)-IQR | 45.7 (27) | 43.8 (20.5) | 48 (38) | 0.002 |
| Creatinine, (mg/dL)-IQR | 0.85 (0.4) | 0.83 (0.37) | 0.93 (0.55) | 0.010 |
| Albumin, (g/dL) | 3.46 ± 0.48 | 3.56 ± 0.41 | 3.32 ± 0.55 | < 0.001 |
| Glucose, (mg/dL)-IQR | 118 (49) | 118 (46) | 118 (55) | 0.430 |
| Sodium, (mmol/L) | 136.11 ± 7.17 | 136.11 ± 8.69 | 136.10 ± 4.04 | 0.988 |
| Total cholesterol, (mg/dL) | 175.8 ± 47.7 | 178.4 ± 45.5 | 171.9 ± 50.9 | 0.309 |
AR Aortic regurgitation, CABG Coronary artery bypass grafting, CONUT Controlling nutritional status, COPD Chronic obstructive pulmonary disease, CVD Cerebrovascular disease, GFR Glomerular filtration rate, GNRI Geriatric nutritional risk index, LBBB Left bundle branch block, MR Mitral regurgitation, NYHA New York Heart Association, PCI Percutaneous coronary intervention, PNI Prognostic nutritional index, STS Society of Thoracic Surgeons, TAVR Transcatheter aortic valve replacement, TR Tricuspid regurgitation
Table 2.
Postoperative complications and medications
| Variable | Total population(n = 262) | Survivors(n = 156) | Non-survivors(n = 106) | p value | |||
|---|---|---|---|---|---|---|---|
| Post-TAVR Complications | |||||||
| Permanent pacemaker implantation, n (%) | 21 (8.0%) | 9 (5.8%) | 12 (11.3%) | 0.164 | |||
| CVD development, n (%) | 8 (3.1%) | 3 (1.9%) | 5 (4.7%) | 0.355 | |||
| Cardiac tamponade, n (%) | 7 (2.7%) | 0 (0.0%) | 7 (6.6%) | 0.004 | |||
| In-hospital arrhythmia, n (%) | 45 (17.2%) | 21 (13.5%) | 24 (22.6%) | 0.077 | |||
| Acute kidney injury, n (%) | 14 (5.3%) | 5 (3.2%) | 9 (8.5%) | 0.112 | |||
| Major bleeding, n (%) | 14 (5.3%) | 1 (0.6%) | 13 (12.3%) | < 0.001 | |||
| Myocardial infarction, n (%) | 3 (1.1%) | 2 (1.3%) | 1 (0.9%) | 1.000 | |||
| Coronary obstruction, n (%) | 1 (0.4%) | 0 (0.0%) | 1 (0.9%) | 0.846 | |||
| LBBB development, n (%) | 88 (33.6%) | 48 (30.8%) | 40 (37.7%) | 0.250 | |||
| After TAVR Medications | |||||||
| ASA or P2Y12 Inhibitors, n (%) | 238 (90.8%) | 150 (96.2%) | 88 (83.0%) | 0.489 | |||
| Warfarin, n (%) | 12 (4.6%) | 9 (5.8%) | 3 (2.8%) | 0.389 | |||
| Oral anticoagulants, n (%) | 37 (14.1%) | 22 (14.1%) | 15 (14.2%) | 0.748 | |||
ASA Acetylsalicylic acid, CVD Cerebrovascular disease, TAVR Transcatheter aortic valve replacement
A Venn diagram was created to schematically represent the overlap of malnutrition classifications based on GNRI, PNI, and CONUT scores (Fig. 1). Malnutrition was detected in 79.8% of patients according to the GNRI (< 98) criterion, 63.7% according to the PNI (< 45) criterion, and 60.7% according to the CONUT (≥ 2) criterion. A total of 126 patients (48.1%) were classified as malnourished by all three indices, and the mortality rate in this group was 45.2%.
Fig. 1.
Venn diagram showing the overlap of malnutrition classifications according to GNRI, PNI, and CONUT scores
Malnutrition prevalence
Malnutrition was common in the study population, and prevalence rates varied by nutritional index. According to the Geriatric Nutritional Risk Index (GNRI) classification, 121 patients (46.2%) had a GNRI < 92, indicating moderate-to-severe malnutrition. Eighty-nine patients (34.0%) had a GNRI in the 92–98 range, indicating mild malnutrition risk, while 52 patients (19.8%) had a GNRI ≥ 98 and were considered to have normal nutritional status. When mortality rates were examined according to GNRI categories, 62 deaths (51.2%) were observed in the GNRI < 92 group, 28 deaths (31.5%) in the GNRI 92–98 group, and 16 deaths (30.8%) in the GNRI ≥ 98 group. It was noteworthy that the mortality rate was highest in patients with a GNRI < 92.
According to the PNI classification, 52 patients (19.8%) had severe malnutrition (PNI < 38), 115 patients (43.9%) had mild-moderate malnutrition (PNI 38–45), and 95 patients (36.3%) had normal nutritional status (PNI ≥ 45). Mortality rates by PNI category were 34 deaths (65.4%) in the PNI < 38 group, 38 deaths (64.5%) in the PNI 38–45 group, and 34 deaths (35.8%) in the PNI ≥ 45 group. The highest mortality rate was observed in patients with severe malnutrition.
According to the CONUT score, 83 patients (34.3%) had normal nutritional status (CONUT 0–1), 111 patients (45.9%) had mild malnutrition (CONUT 2–4), 39 patients (16.1%) had moderate malnutrition (CONUT 5–8), and 9 patients (3.7%) had severe malnutrition (CONUT 9–12). Mortality rates increased with increasing CONUT category. Accordingly, 25 deaths (30.1%) were recorded in the CONUT 0–1 group, 42 deaths (37.8%) in the CONUT 2–4 group, 22 deaths (56.4%) in the CONUT 5–8 group, and 7 deaths (77.8%) in the CONUT 9–12 group. Patients with severe malnutrition, according to the CONUT score, had the highest risk of mortality.
Univariable and multivariable cox regression analysis
In the univariable Cox regression analysis, all three nutritional indices were significantly associated with mortality. Specifically, the hazard ratio (HR) for GNRI was 0.974 (95% CI 0.951–0.998; p = 0.035), for PNI HR was 0.959 (95% CI 0.931–0.989; p = 0.007), and for CONUT HR was 1.140 (95% CI 1.057–1.231; p < 0.001). Lower GNRI and PNI values and higher CONUT scores were associated with an increased risk of mortality. The details of the univariable analysis results are presented in Table 3.
Table 3.
Univariable cox regression for total mortality
| Variable | HR (95% CI) | p value |
|---|---|---|
| Age | 1.027 (0.997–1.059) | 0.075 |
| Male gender | 1.175 (0.799–1.730) | 0.412 |
| PNI | 0.959 (0.931–0.989) | 0.007 |
| GNRI | 0.974 (0.951–0.998) | 0.035 |
| CONUT | 1.140 (1.057–1.231) | < 0.001 |
| STS Score | 0.975 (0.909–1.046) | 0.482 |
| Albumin | 0.616 (0.427–0.889) | 0.010 |
| Creatinine | 1.310 (1.112–1.543) | 0.001 |
| Chronic kidney failure | 1.892 (1.267–2.824) | 0.002 |
| GFR | 0.989 (0.981–0.996) | 0.004 |
| NYHA class | 1.669 (1.156–2.409) | 0.006 |
| Pulmonary edema | 3.425 (1.584–7.407) | 0.002 |
| Body mass index | 1.086 (0.974–1.211) | 0.137 |
| Cerebrovascular disease | 3.309 (0.807–13.566) | 0.096 |
| Heart failure | 0.715 (0.479–1.067) | 0.100 |
| Ejection fraction | 1.007 (0.990–1.024) | 0.443 |
| Hypertension | 1.150 (0.780–1.696) | 0.481 |
| Diabetes mellitus | 1.129 (0.732–1.741) | 0.583 |
| CABG history | 0.620 (0.323–1.192) | 0.152 |
| Atrial fibrillation | 1.327 (0.850–2.073) | 0.214 |
| Dyslipidemia | 0.713 (0.454–1.121) | 0.143 |
| Anemia | 1.990 (1.338–2.961) | < 0.001 |
| Urea | 1.013 (1.007–1.019) | < 0.001 |
| Potassium | 1.946 (1.421–2.666) | < 0.001 |
| Aspartate aminotransferase | 1.006 (0.998–1.014) | 0.133 |
| Alanine aminotransferase | 1.002 (0.991–1.014) | 0.702 |
| Hematocrit | 0.995 (0.962–1.029) | 0.780 |
| Monocyte | 1.575 (0.726–3.419) | 0.251 |
| Lymphocyte | 0.837 (0.616–1.137) | 0.255 |
| Hemoglobin | 0.947 (0.860–1.042) | 0.263 |
| Total cholesterol | 0.997 (0.992–1.001) | 0.146 |
| Smoking cigarette | 1.281 (0.827–1.985) | 0.267 |
Multivariable Cox regression models were constructed by adjusting for clinically relevant baseline variables. Variables that were significant in the univariate analysis or deemed clinically important based on prior literature were included in the multivariable model: age, sex, STS score, GFR, potassium, anemia, and pulmonary edema. Nutritional indices were included in the analyses individually to avoid multicollinearity. Modeling analyses were performed. In the adjusted models, HR for GNRI was 0.979 (95% CI 0.954–1.004; p = 0.104), HR for PNI was 0.968 (95% CI 0.936–1.001; p = 0.061), and HR for CONUT was 1.085 (95% CI 0.991–1.188; p = 0.078). Although nutritional indices were associated with mortality in the univariable analysis, this association weakened after adjusting for baseline clinical variables, and no index remained an independent predictor. Detailed results of the multivariable Cox regression analysis are presented in Table 4. In the Harrell C-index analysis, the clinical model’s discriminative power was 0.694. With the addition of nutritional scores to the model, the C-index increased to 0.700 for all three scores (GNRI: 0.694 → 0.700; PNI: 0.694 → 0.700; CONUT: 0.694 → 0.700). Addition of the nutritional scores resulted in only a minimal numerical increase in Harrell’s C-index. Therefore, the incremental prognostic value of these indices beyond established clinical variables appears limited. The proportional hazards assumption was not violated in the multivariable Cox regression models. Global Schoenfeld residual tests were nonsignificant for the GNRI model, PNI model, and CONUT model, supporting the validity of the Cox regression estimates. Still, no score showed a significant superiority over the others. Multivariable Cox regression analyses were also schematized in the form of a Forest-Plot graph (Figs. 2, 3 and 4).
Table 4.
Multivariable cox regression analyses
| Variable | Hazard Ratio | 95% CI | p-value |
|---|---|---|---|
| GNRI Model | |||
| Age | 1.002 | 0.966–1.038 | 0.929 |
| Male gender | 1.750 | 1.130–2.710 | 0.012 |
| STS score | 0.969 | 0.904–1.040 | 0.386 |
| Potassium | 1.915 | 1.361–2.694 | < 0.001 |
| Anemia | 2.025 | 1.300–3.156 | 0.002 |
| Pulmonary edema | 5.795 | 2.313–14.523 | < 0.001 |
| eGFR | 0.995 | 0.986–1.003 | 0.199 |
| GNRI | 0.977 | 0.950–1.004 | 0.091 |
| PNI Model | |||
| Age | 1.002 | 0.966–1.039 | 0.919 |
| Male gender | 1.628 | 1.057–2.509 | 0.027 |
| STS score | 0.970 | 0.904–1.040 | 0.392 |
| Potassium | 1.935 | 1.372–2.728 | < 0.001 |
| Anemia | 1.939 | 1.237–3.039 | 0.004 |
| Pulmonary edema | 6.452 | 2.581–16.128 | < 0.001 |
| eGFR | 0.996 | 0.987–1.005 | 0.415 |
| PNI | 0.964 | 0.929–1.001 | 0.054 |
| CONUT Model | |||
| Age | 1.005 | 0.970–1.042 | 0.765 |
| Male gender | 1.576 | 1.020–2.435 | 0.040 |
| STS score | 0.974 | 0.909–1.044 | 0.454 |
| Potassium | 1.906 | 1.361–2.668 | < 0.001 |
| Anemia | 1.859 | 1.173–2.945 | 0.008 |
| Pulmonary edema | 6.745 | 2.692–16.896 | < 0.001 |
| eGFR | 0.996 | 0.987–1.005 | 0.344 |
| CONUT | 1.092 | 0.996–1.198 | 0.062 |
Fig. 2.
Forest-plot model for GNRI
Fig. 3.
Forest-plot model for PNI
Fig. 4.
Forest-plot model for CONUT
To evaluate the predictive power of nutritional indices on total mortality, ROC analysis was performed. The discriminative performance of nutritional scores was moderate. The area under the curve (AUC) was 0.626 (95% CI 0.550–0.695) for GNRI, 0.610 (95% CI 0.534–0.681) for CONUT, and 0.590 (95% CI 0.516–0.663) for PNI. Although GNRI showed the highest discriminative power among the three, its overall predictive performance remained moderate. A comparison of the ROC curves is presented in Fig. 5. In the ROC analysis, the optimal threshold values were determined as 39.00 for PNI (sensitivity 35.4%, specificity 89.0%, p = 0.022), 92.20 for GNRI (sensitivity 62.5%, specificity 60.3%, p = 0.001), and 5.00 for CONUT (sensitivity 30.2%, specificity 87.0%, p = 0.003), respectively. Although the optimal threshold values showed relatively high specificity, the sensitivity remained modest; this reflects the limited discriminatory performance of nutritional indices in predicting mortality.
Fig. 5.
ROC analyses of all nutritional tools
Kaplan-Meier survival curves showed a significant difference in survival by nutritional status. Patients with lower GNRI tertiles showed significantly worse survival compared to patients with normal GNRI values (log-rank p < 0.001). Similarly, PNI tertiles showed significant differences in survival, with the lowest PNI tertile exhibiting the worst outcomes (log-rank p < 0.001). Likewise, the probability of survival progressively decreased with increasing CONUT score tertiles, indicating worse outcomes in patients with more severe malnutrition (log-rank p < 0.001). Kaplan-Meier survival curves are presented in Fig. 6.
Fig. 6.
Kaplan-Meier analyses for all nutritional statement tools for all cause mortality
Discussion
In this study, we examined the prevalence of malnutrition and its long-term prognostic impact in patients undergoing TAVR, as assessed using three existing indices. The results of our study show that malnutrition is common among TAVR patients and that nutritional status indicators may be significantly associated with mortality. The prevalence of malnutrition in the study population was high, though it varied by assessment method. According to GNRI, 209 (79.8%) patients had malnutrition; according to PNI, 167 (63.7%); and according to the CONUT score, 159 (60.7%). We believe that the most important reason for such high malnutrition rates is that TAVR patients are a high-risk population for nutritional deficiencies due to their advanced age, frailty, chronic inflammation, and multiple comorbidities. Another important finding of our study is that all GNRI, PNI, and CONUT scores, which assess nutritional status, were significantly associated with mortality in univariable Cox regression analysis. Lower GNRI and PNI values and higher CONUT scores were observed to be associated with an increased risk of all-cause mortality. One of the striking findings of our study is that although GNRI, PNI, and CONUT scores were significantly associated with mortality in univariate analysis, they did not remain independent predictors in multivariate analysis.
We believe there may be several possible reasons for this. First, it may be a significant factor that these nutritional scores largely include similar biochemical and clinical parameters. These indices include albumin, total cholesterol, lymphocyte count, and body weight, and therefore, a high level of correlation is likely among them. This collinearity may have weakened the independent effect of the scores in multivariate models. Although we performed analyses by modeling each indicator separately to avoid this, we did not prevent it. In addition, some clinical variables that were more strongly associated with mortality in our study may have influenced this result. In particular, variables such as potassium level, presence of anemia, pulmonary edema, and gender were found to have more significant effects on mortality. These variables may be stronger predictors of prognosis because they more directly reflect patients’ cardiac reserve, renal function, and hemodynamic status. Another potential explanation for the attenuation of the associations in multivariable analyses is overadjustment. GNRI, PNI, and CONUT are composite indices derived from laboratory and anthropometric parameters that reflect several aspects of patients’ clinical condition. Therefore, adjustment for variables capturing related biological domains may partially remove the prognostic information contained within these nutritional indices. Consequently, the lack of independent significance should not be interpreted as evidence that nutritional status lacks prognostic relevance, but rather that its prognostic contribution overlaps with other established clinical risk factors.
Although not independent predictors in multivariable analysis, nutritional indices demonstrated only modest discrimination (AUC: GNRI 0.63, CONUT 0.61, PNI 0.59) and may aid risk assessment when considered with other clinical factors. These results are consistent with previous studies highlighting the negative impact of malnutrition on clinical outcomes in TAVR patients [2–5, 8, 12, 13, 20, 21]. Perhaps the most significant difference from previous studies is the sharing of long-term results over 7 years. In the study by Sihag V. et al., GNRI, PNI, and CONUT scores were all independent predictors of death in patients undergoing TAVR, and malnutrition was significantly more common in this patient group compared to the normal population [8]. Similarly, Dong M. et al., in their meta-analysis, showed that malnutrition increased the risk of all-cause mortality and could predict adverse postoperative outcomes [2]. A large retrospective national cohort study involving approximately 106,000 TAVR patients demonstrated that malnutrition is a significant independent predictor of postoperative mortality, complications, and rehospitalization, and has stronger prognostic value than 1-year mortality, particularly for predicting long-term clinical outcomes [20]. Our clinical experience also suggests that the impact of malnutrition, especially on long-term mortality, may be even more pronounced. Therefore, in our study, we evaluated long-term follow-up results to better elucidate the prognostic value of nutritional indices.
In the study by Küçükosmanoğlu et al., pairwise comparisons of ROC curves showed that the GNRI score significantly better predicted mortality than the PNI and CONUT scores (area under the curve [AUC] 0.898 for GNRI, 0.664 for PNI, and 0.722 for CONUT, respectively; p < 0.001 for all comparisons). According to Cox proportional hazards analysis, GNRI (hazard ratio [HR]: 0.898) and STS score (HR: 1.359) were independently associated with 1-year mortality (p < 0.05 for all analyses). In our study, similarly, although GNRI yielded better results, this was not statistically significant [4].
Previous studies have suggested that adding malnutrition and frailty assessments to traditional preoperative risk models can improve mortality prediction [20–27]. In our study, we tested three nutritional scoring systems for their predictive ability for long-term all-cause mortality. Although the addition of nutritional scores slightly increased model discrimination, the magnitude of this improvement was small. Therefore, these findings do not support a substantial enhancement in predictive performance beyond conventional clinical risk factors.
These findings indicate that although nutritional scores are not strong predictors on their own, they can be useful tools in clinical risk assessment. Especially since they are based on readily available laboratory parameters and easy to calculate, these scores can be considered complementary indicators for risk stratification in TAVR patients.
Limitations
Our study has several limitations. Perhaps most importantly, it is a retrospective study with a limited number of patients. Therefore, unrecorded confounding factors and changes may have independently influenced the findings, and the causal relationship is debatable. Furthermore, due to its retrospective nature, we cannot completely rule out bias. Nutritional indices were calculated solely based on laboratory values at admission; changes occurring during follow-up were not evaluated, and dynamic changes in nutritional status were not analyzed. Data on patients’ nutritional interventions or dietary treatments are unavailable. Our study only included patients from one region of Türkiye. Therefore, our results cannot be generalized. Although the number of events was adequate for the multivariable models according to accepted events-per-variable criteria, the borderline significance observed for PNI and CONUT suggests that a Type II error cannot be completely excluded. Therefore, larger prospective studies are warranted to confirm the independent prognostic value of these nutritional indices.
Additionally, we used only three of the many scoring systems developed to assess malnutrition. Our results may have been affected because statin use or the use of medications that could affect outcomes were not excluded. Furthermore, we could not completely exclude inflammatory and hematological diseases that could affect the parameters included in the scoring systems. Specific causes of death during the follow-up period were not analyzed in detail. The study only evaluated all-cause mortality.
Conclusion
Our study found that impaired nutritional status, assessed by GNRI, PNI, and CONUT scores, was associated with increased long-term mortality in patients undergoing TAVR. While these indices showed significant associations with mortality in univariate analyses, they did not remain independent predictors after multivariable adjustment. These findings suggest that nutritional status reflects patients’ overall clinical condition rather than serving as a prognostic determinant on its own. Therefore, assessing nutritional status is beneficial as part of a comprehensive evaluation of patients scheduled for TAVR.
Authors’ contributions
Author ContributionsSerdar Söner, Erkan Baysal, and Metin Okşul contributed to the conception and design of the study.Serdar Söner, Erkan Baysal, Öner Avınca, and Gamze Yeter Arslan were responsible for data collection.Tuncay Güzel, Hülya Tosun Söner, Abdülkadir Arpa, Harun Elmast, Umut Erdolu, Celal Öztekin, Mehmet Sait Coşkun, Ahmet Taş, and Faruk Ertaş contributed to data analysis and interpretation.Serdar Söner drafted the manuscript.Metin Okşul and Faruk Ertaş critically revised the manuscript for important intellectual content.All authors read and approved the final manuscript.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Data availability
The datasets generated and/or analyzed during the current study are not publicly available due to institutional and patient privacy regulations but are available from the corresponding author on reasonable request.
Declarations
Ethics approval and consent to participate
This study was conducted in accordance with the principles of the Declaration of Helsinki. Ethical approval was obtained from the Clinical Research Ethics Committee of Gazi Yaşargil Training and Research Hospital, University of Health Sciences (Decision No: 101, Date: 02.03.2026). Due to the retrospective design of the study, the requirement for informed consent was waived by the ethics committee.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Osnabrugge RL, Mylotte D, Head SJ, Van Mieghem NM, Nkomo VT, LeReun CM, et al. Aortic stenosis in the elderly: disease prevalence and number of candidates for transcatheter aortic valve replacement: a meta-analysis and modeling study. J Am Coll Cardiol. 2013;62(11):1002–12. 10.1016/j.jacc.2013.05.015. [DOI] [PubMed] [Google Scholar]
- 2.Dong M, Cheng J, Gong L, Xiao Y, Shao S, Song J. Malnutrition predicts adverse outcomes after transcatheter aortic valve replacement: a systematic review and meta-analysis. Anatol J Cardiol. 2023;27(5):240–8. 10.5152/AnatolJCardiol.2023.23119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Lee K, Ahn JM, Kang DY, Ko E, Kwon O, Lee PH, et al. Nutritional status and risk of all-cause mortality in patients undergoing transcatheter aortic valve replacement: assessment using the geriatric nutritional risk index and the controlling nutritional status score. Clin Res Cardiol. 2020;109(2):161–71. 10.1007/s00392-019-01497-9. [DOI] [PubMed] [Google Scholar]
- 4.Kucukosmanoglu M, Kilic S, Urgun OD, Sahin S, Yildirim A, Sen O, et al. Impact of objective nutritional indexes on 1-year mortality after transcatheter aortic valve implantation: a prospective observational cohort study. Acta Cardiol. 2021;76(4):402–9. 10.1080/00015385.2020.1757854. [DOI] [PubMed] [Google Scholar]
- 5.Ishizu K, Shirai S, Tashiro H, Kitano K, Tabata H, Nakamura M, et al. Prevalence and prognostic significance of malnutrition in older Japanese adults at high surgical risk undergoing transcatheter aortic valve implantation. J Am Heart Assoc. 2022;11(19):e026294. 10.1161/JAHA.122.026294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Söner S, Güzel T, Aktan A, Kılıç R, Arslan B, Demir M, et al. Predictive value of nutritional scores in non-valvular atrial fibrillation patients: insights from the AFTER-2 study. Nutr Metab Cardiovasc Dis. 2025;35(3):103794. 10.1016/j.numecd.2024.103794. [DOI] [PubMed] [Google Scholar]
- 7.Söner S, Güzel T, Aktan A, Kılıç R, Söner HT, Demir M, et al. Prognostic value of hemoglobin, albumin, lymphocyte, platelet (HALP) score in patients with non-valvular atrial fibrillation: insights from the AFTER-2 study. BMC Cardiovasc Disord. 2025;25(1):528. 10.1186/s12872-025-03528-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Sihag V, Li W, Hagar A, Peng Y, Feng Y, Bhushan S, et al. The impact of nutritional status on the outcome of transcatheter aortic valve implantation. Heart Surg Forum. 2022;25(2):E267–72. 10.1532/hsf.4547. [DOI] [PubMed] [Google Scholar]
- 9.Bouillanne O, Morineau G, Dupont C, Coulombel I, Vincent JP, Nicolis I, et al. Geriatric nutritional risk index: a new index for evaluating at-risk elderly medical patients. Am J Clin Nutr. 2005;82(4):777–83. 10.1093/ajcn/82.4.777. [DOI] [PubMed] [Google Scholar]
- 10.Buzby GP, Mullen JL, Matthews DC, Hobbs CL, Rosato EF. Prognostic nutritional index in gastrointestinal surgery. Am J Surg. 1980;139(1):160–7. 10.1016/0002-9610(80)90246-9. [DOI] [PubMed] [Google Scholar]
- 11.Ignacio de Ulíbarri J, González-Madroño A, de Villar NG, González P, González B, Mancha A, et al. CONUT: a tool for controlling nutritional status. Nutr Hosp. 2005;20(1):38–45. [PubMed] [Google Scholar]
- 12.Okuno T, Koseki K, Nakanishi T, Sato K, Ninomiya K, Tomii D, et al. Evaluation of objective nutritional indexes as predictors of one-year outcomes after transcatheter aortic valve implantation. J Cardiol. 2019;74(1):34–9. 10.1016/j.jjcc.2019.02.002. [DOI] [PubMed] [Google Scholar]
- 13.Mas-Peiro S, Hoffmann J, Seppelt PC, De Rosa R, Murray MI, Walther T, et al. Value of prognostic nutritional index for survival prediction in transcatheter aortic valve replacement compared to other common nutritional indexes. Acta Cardiol. 2021;76(6):615–22. 10.1080/00015385.2020.1757854. [DOI] [PubMed] [Google Scholar]
- 14.Vahanian A, Alfieri O, Andreotti F, Antunes MJ, Barón-Esquivias G, Baumgartner H, et al. Guidelines on the management of valvular heart disease (version 2012). Eur J Cardiothorac Surg. 2012;42(4):S1–44. 10.1093/ejcts/ezs455. [DOI] [PubMed] [Google Scholar]
- 15.Nishimura RA, Otto CM, Bonow RO, Carabello BA, Erwin JP III, Guyton RA, et al. 2014 AHA/ACC guideline for the management of patients with valvular heart disease. J Am Coll Cardiol. 2014;63(22):e57–185. 10.1016/j.jacc.2014.02.536. [DOI] [PubMed] [Google Scholar]
- 16.World Medical Association. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA. 2013;310(20):2191–4. 10.1001/jama.2013.281053. [DOI] [PubMed] [Google Scholar]
- 17.American Diabetes Association. Standards of medical care in diabetes—2013. Diabetes Care. 2013;36(Suppl 1):S4–10. 10.2337/dc13-S004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Echouffo-Tcheugui JB, Daya N, Ndumele CE, Matsushita K, Hoogeveen RC, Ballantyne CM, et al. Diabetes, GDF-15 and incident heart failure: the ARIC study. Diabetologia. 2022;65(6):955–63. 10.1007/s00125-022-05666-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF III, Feldman HI, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150(9):604–12. 10.7326/0003-4819-150-9-200905050-00006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Emami S, Rudasill S, Bellamkonda N, Sanaiha Y, Cale M, Madrigal J, et al. Impact of malnutrition on outcomes following transcatheter aortic valve implantation (from a national cohort). Am J Cardiol. 2020;125(7):1096–101. 10.1016/j.amjcard.2019.12.040. [DOI] [PubMed] [Google Scholar]
- 21.Honda Y, Yamawaki M, Shigemitsu S, Kenji M, Tokuda T, Tsutumi M, et al. Prognostic value of objective nutritional status after transcatheter aortic valve replacement. J Cardiol. 2019;73(5):401–7. 10.1016/j.jjcc.2018.10.013. [DOI] [PubMed] [Google Scholar]
- 22.Steinvil A, Buchanan KD, Kiramijyan S, Bond E, Rogers T, Koifman E, et al. Utility of an additive frailty tests index score for mortality risk assessment following transcatheter aortic valve replacement. Am Heart J. 2018;200:11–6. 10.1016/j.ahj.2018.02.012. [DOI] [PubMed] [Google Scholar]
- 23.Rosenhek R, Iung B, Tornos P, Antunes MJ, Prendergast BD, Otto CM, et al. Assessing the risk of interventions in patients with valvular heart disease. Eur Heart J. 2012;33(7):822–8. 10.1093/eurheartj/ehr311. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Afilalo J, Mottillo S, Eisenberg MJ, Alexander KP, Noiseux N, Perrault LP, et al. Addition of frailty and disability to cardiac surgery risk scores identifies elderly patients at high risk of mortality or major morbidity. Circ Cardiovasc Qual Outcomes. 2012;5(2):222–8. 10.1161/CIRCOUTCOMES.111.963348. [DOI] [PubMed] [Google Scholar]
- 25.Alfredsson J, Stebbins A, Brennan JM, Matsouaka R, Afilalo J, Peterson ED, et al. Gait speed predicts 30-day mortality after transcatheter aortic valve replacement. Circulation. 2016;133(14):1351–9. 10.1161/CIRCULATIONAHA.115.020279. [DOI] [PubMed] [Google Scholar]
- 26.Koifman E, Kiramijyan S, Negi SI, Didier R, Escarcega RO, Minha S, et al. Body mass index association with survival in severe aortic stenosis patients undergoing transcatheter aortic valve replacement. Catheter Cardiovasc Interv. 2016;88(1):118–24. 10.1002/ccd.26232. [DOI] [PubMed] [Google Scholar]
- 27.Koifman E, Magalhaes MA, Ben-Dor I, Kiramijyan S, Escarcega RO, Fang C, et al. Impact of pre-procedural serum albumin levels on outcome of patients undergoing transcatheter aortic valve replacement. Am J Cardiol. 2015;115(9):1260–4. 10.1016/j.amjcard.2015.02.006. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets generated and/or analyzed during the current study are not publicly available due to institutional and patient privacy regulations but are available from the corresponding author on reasonable request.