WHAT IS ALREADY KNOWN ON THIS TOPIC

Prognostic nutritional index (PNI) is a widely used prognostic indicator for various chronic diseases in recent years.

Malnutrition has been associated with diabetes risk and mortality in several small studies.

WHAT THIS STUDY ADDS

Using nationally representative data on US adults, we found that lower serum PNI levels were associated with higher all-cause mortality and cardiovascular disease mortality.

These findings suggest that maintaining an appropriate range of serum PNI status may reduce the risk of death in patients with diabetes.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

Our findings highlight the importance of malnutrition as reflected by PNI in patients with type 2 diabetes (T2D).

Future studies are needed to assess the clinical benefit of prognostic nutritional indices and thus the potential to reduce all-cause and cardiovascular deaths in patients with T2D.

Introduction

Currently, there are approximately 537 million adults worldwide with diabetes, over 90% of whom suffer from type 2 diabetes mellitus (T2DM), contributing to a prevalence rate that has reached as high as 10.5% among adults.1 Individuals with diabetes face a two to four times greater risk of cardiovascular disease (CVD) and mortality compared with non-diabetics.2 As a result, it is of utmost importance to identify prognostic factors that have the potential to hinder or postpone diabetic complications and mortality.

Serum albumin (ALB) is a widely used indicator to assess nutritional risk. ALB, the most abundant protein synthesized by the liver and found in serum, is decreased in both malnutrition and inflammation.3 Previous studies have found that low ALB levels in healthy adults are associated with an increased risk of diabetes.4 5 Available studies suggest that inflammatory responses due to immune dysfunction play an important role in the development and progression of diabetic complications, with lymphocytes playing a crucial role.6 7 In contrast, the prognostic nutritional index (PNI) is a simple and objective nutritional risk indicator calculated from serum ALB and total lymphocyte count.8 Therefore, we hypothesized that PNI is closely related to the development of T2DM.

In order to bridge these gaps in knowledge, our objective was to conduct a systematic investigation into the potential link between serum PNI levels and mortality rates, encompassing both all-cause and cardiovascular mortality. This analysis was carried out using a nationally representative sample of individuals diagnosed with diabetes mellitus within the USA.

Research design and methodsStudy population

The National Health and Nutrition Examination Survey (NHANES) is a continuous research initiative aimed at gathering data on the nutrition and health status of both adults and children in the USA. This survey employs a carefully planned stratified, multistage probability design to ensure that the sample selected represents the entire US population accurately. Data were obtained through structured interviews with individuals at home, health screenings at mobile health screening centers, and laboratory sample analysis.9 For this study, we used data from NHANES 1999–2018 that provided information on serum PNI. We included patients aged >20 years with T2DM, resulting in a study sample of 7556 subjects. T2DM was defined as a physician self-reported diagnosis of diabetes mellitus with insulin or oral hypoglycemic agents and fasting glucose >7.0 mmol/L (126 mg/dL) or glycated hemoglobin A1c (HbA1c) >6.5%. We excluded participants who were pregnant (n=10) and had CVD (n=1529) and cancer (n=100), resulting in a final representative analysis of 5916 patients with T2DM.

Assessment of PNI

Predictive nutritional indices evaluate the nutritional status of individuals based on clinical markers, calculated using the following formula10–12: PNI=10×serum ALB (g/dL)+5×lymphocyte count(10ˆ9/L). Lymphocyte count is primarily obtained from complete blood cell count (CBC) tests, which derive CBC parameters using the Beckman Coulter counting and sizing method. Serum ALB levels are typically used to assess nutritional status and are measured in the NHANES database using the bromocresol violet dye method.13 Low PNI scores indicate inadequate nutritional status and have been linked to increased mortality rates in prior studies. Our analysis divided participants into PNI quartiles: quartile 1 (Q1) having the lowest PNI score, signaling a relatively high malnutrition risk, while quartile 4 (Q4) had the highest PNI score, indicating a relatively low malnutrition risk.

Determination of mortality

The determination of all-cause mortality and CVD mortality was conducted through the examination of records from the National Death Index, specifically up until 31 December 2019. The identification of disease-specific mortality was carried out by using the International Classification of Diseases, Tenth Revision (ICD-10). CVD mortality, in particular, was defined based on the designated ICD-10 codes, namely I00–I09, I11, I13, I20–I51, or I60–I69.

Assessment of covariates

Demographic information regarding the participants, including socioeconomic factors, smoking habits, alcohol consumption, duration of diabetes, medication usage, and hypertension, was gathered using a standardized questionnaire. To classify the participants, non-smokers were defined as individuals who had never smoked in their lifetime, while current smokers were those who had smoked over 100 cigarettes in the past and had not yet quit. Former smokers, on the other hand, referred to individuals who had smoked in the past but had since quit. Drinking habits were categorized as non-drinker, low to moderate drinker (less than two drinks per day for males and less than one drink per day for females), or heavy drinker (two or more drinks per day for males and one or more drinks per day for females). In terms of race/ethnicity, participants were classified as either non-Hispanic white or belonging to other racial/ethnic groups. Educational attainment was divided into three categories: less than high school, high school or equivalent, and college or higher. The poverty scores for household income were defined as 0–1.0, 1.0–3.0, and greater than or equal to 3.0. To determine the participants’ body mass index (BMI), weight was divided by height squared (kg/m2). Additionally, glycated hemoglobin was compared as previously described.14

Statistical analysis

In order to analyze the data from the NHANES database, special attention was given to the complex sampling design. This involved taking into account sample weights, clustering, and stratified analysis. When examining normally distributed data, SDs were used to express the results. On the other hand, non-normally distributed data were presented using the median and IQR. Categorical variables were expressed as percentages and χ2 tests were conducted. Quartiles of serum PNI levels were determined based on the distribution observed within the study population. One-way analysis of variance tests (for continuous variables with normal distribution), Kruska-Wallis test (for continuous variables with non-normal distribution), and χ2 tests (for categorical variables) were used to compare differences between the four groups. Cox proportional risk regression was used to estimate risk ratios (HRs) and 95% CIs for CVD mortality associated with serum PNI. The follow-up period was determined by calculating the duration between the NHANES interview date and either the date of death or the concluding date of follow-up, which was 31 December 2019. In the event that either of these events occurred earlier, the follow-up was terminated at that point.

In order to assess the relationship between serum PNI levels and mortality from various causes, particularly CVD, we used restricted three-sample regression models. The multivariate model employed in our analysis took into account several important factors. In model 1, adjustments were made for age (in years), gender (male or female), and race/ethnicity (non-Hispanic white or other). To further refine our analysis, model 2 incorporated additional adjustments for variables such as BMI (categorized as 25.0, 25.0–29.9, or ≥30.0 kg/m2), education level (below high school, high school or equivalent, or college or above), family income-poverty ratio (grouped as 0–1.0, 1.0–3.0, or >3.0), smoking status (never, former, or current smoker), and drinking status (none, low to moderate, or heavy drinkers). Finally, model 3 introduced further adjustments for diabetes duration (in years), diabetes medication use (including none, oral medication only, insulin only, or both insulin and medication), glycated hemoglobin levels (<7.0% or ≥7.0%), and hypertensive disease (classified as yes or no).

In stratified analyses, we stratified by age (≤60 or >60 years), sex (male or female), race/ethnicity (non-Hispanic white or other), smoking status (current or never/past), BMI (<30 or ≥30 kg/m2), and glycated hemoglobin (<7% or ≥7%). The significance of the interaction was estimated using the p value of the product term of serum PNI values and stratified variables. All analyses were performed using IBM SPSS Statistics V.25.0. A two-sided p value <0.05 was considered statistically significant. Several sensitivity analyses were also performed to test the stability of the results. First, to reduce potential reverse causality bias, we excluded participants who died 2 years prior to follow-up (n=680). Second, the main analysis was repeated based on quintiles of serum PNI levels. Third, to investigate the potential role of inflammation, lipid levels, and liver and kidney indices with any observed associations, we further adjusted for C-reactive protein (CRP) levels, lipid profiles (including triglycerides, low-density lipoprotein (LDL) cholesterol, and high-density lipoprotein (HDL) cholesterol), indicators of kidney function (estimation of glomerular filtration rate and uric acid levels), and indicators of liver function (aspartate aminotransferase and alanine aminotransferase levels). Fourth, because serum PNI levels were not normally distributed, an unadjusted Spearman correlation coefficient was used for correlation analysis at baseline. Finally, we performed a restricted three-sample analysis in which we added relevant hematological indicators to the baseline data for further sensitivity analysis.

Results

Among the 5916 patients with diabetes (with a mean age of 59.01 years and 51.7% male), the median serum PNI was 53.0 (IQR: 49.0–58.0). The baseline characteristics of serum PNI quartiles are presented in table 1. Participants in lower serum PNI quartiles were generally older, less educated, obese, had lower household income, and had a longer duration of diabetes (table 1).

Table 1

Baseline characteristics of patients with T2DM in PNI and NHANES 1999–2018

During an average follow-up duration of 8.17 years, a total of 1248 deaths occurred due to various causes, out of which 370 were directly linked to CVD. Through a comprehensive analysis that accounted for several factors like age, gender, race, BMI, duration of diabetes, use of diabetes medication, and hypertensive disease, it was observed that individuals with lower serum PNI had a significantly higher risk of mortality from both all causes and CVD (refer to table 2 for detailed findings). The estimated HRs and their corresponding 95% CIs for all-cause mortality, from lowest to highest serum PNI category (<49.00, 49.00–52.99, 53.00–57.99, and ≥58.00), were 1.00 (reference), 0.76 (0.66, 0.88), 0.62 (0.53, 0.73), and 0.72 (0.61, 0.85), respectively (p<0.001). The estimated HRs for CVD mortality, from lowest to highest serum PNI category, were 1.00, 0.70 (0.53, 0.93), 0.73 (0.55, 0.97), and 0.74 (0.55, 1.00), respectively (p=0.033). Figure 1 depicts the non-linear dose–response relationship between serum PNI (range 8.39–270) and all-cause and CVD mortality after multivariate adjustment (p for non-linearity <0.001, p for non-linearity=0.047).

Table 2

HR (95% CIs) for all-cause mortality and cardiovascular mortality associated with PNI in patients with diabetes mellitus in the NHANES study, 1999–2018

Figure 1

Association between prognostic nutritional index (PNI) and all-cause mortality (A) and cardiovascular disease (CVD) mortality (B) in patients with diabetes in the National Health and Nutrition Examination Survey (NHANES) study from 1999 to 2018. HRs (solid lines) and 95% CI (shaded areas) according to age (continuous), sex (male or female), race and ethnicity (non-Hispanic white or other), body mass index (BMI; <25.0, 25.0–29.9, or ≥30.0), education level (below high school, high school or equivalent, or above high school), household income level (lower, normal, or higher), smoking status (never, past, or current), alcohol use (none, low to moderate, or heavy), duration of diabetes (<3 years, 3–10 years, >10 years), medication use (no insulin or pills, only diabetes pills, only insulin, diabetes pills and insulin) and HbA1c (<7.0%, ≥7.0%) to adjust. P<0.001 for all-cause mortality and p=0.096 for cancer mortality.

During the analysis, the data were stratified based on various factors such as age (≤60 or >60 years), sex (male or female), race/ethnicity (non-Hispanic white or other), BMI (<30 or ≥30 kg/m2), smoking status (current or never/past), and glycated hemoglobin (<7% or ≥7%). After conducting multiple tests, no significant interactions were found between PNI and the occurrence of all-cause mortality and CVD mortality. These results have been displayed in tables 3 and 4.

Table 3

Stratified analysis of the association between serum PNI and CVD mortality in patients with T2DM in NHANES 1999–2018

Table 4

Stratified analysis of the association between serum PNI and all-cause mortality in patients with T2DM in NHANES 1999–2018

In our sensitivity analyses, the negative correlations between serum PNI values and all-cause mortality and CVD mortality did not significantly change even after excluding participants who died within 2 years before follow-up (online supplemental table S1). Quintile analysis of serum PNI values revealed a p value <0.001 for all-cause mortality, and no statistically significant difference in mortality from CVD. However, the negative correlation remained (online supplemental table S2). In addition to that, we took into account the relevant serological indicators, which can be seen in online supplemental table S3. The analysis revealed that there were no significant changes in the overall mortality rates. This was demonstrated by the correlations observed after adjusting for CRP (model 2), lipid indicators such as triglycerides, total cholesterol, LDL, and HDL (model 3), liver indicators like alanine aminotransferase and aspartate aminotransferase (model 4), as well as nephrological indicators like the estimation of glomerular filtration rate and uric acid (model 5). Further analysis using Spearman correlation revealed significant associations between PNI levels and age, glycated hemoglobin, triglycerides, total cholesterol, LDL, HDL, alanine aminotransferase, aspartate aminotransferase, estimation of glomerular filtration rate, and uric acid (online supplemental table S4). Notably, when all values were included, there was no substantial change in the restricted three sample bars (online supplemental figure S1).

Discussion

In a comprehensive study involving a substantial number of American adults with diabetes, it has been observed that individuals with lower serum PNI values are significantly more likely to experience both all-cause mortality and CVD-related mortality. Remarkably, this association holds strong even after accounting for other well-established risk factors including lifestyle choices, BMI, duration of diabetes, and usage of diabetes medication. Further analysis conducted within specific subgroups further strengthens the validity and reliability of these findings. Previous studies have highlighted that malnutrition is linked to an increased risk of CVD and death.15–17 Initially used as a nutritional indicator to predict postoperative morbidity and mortality,18 PNI has been used to measure nutritional status. Due to the complexity of earlier PNIs in clinical practice, Onodera’s calculation method, based on peripheral blood lymphocyte counts and serum ALB levels, became simpler.19 Several studies have demonstrated a clear association between PNI and atherosclerotic disease.17 20 21 However, the majority of these studies have been performed in the general population or in those at high risk of CVD. In addition, studies conducted in patients with acute heart failure,22 or in the cohort study by Fan who used four indicators of nutritional status,23 indicate that PNI may provide more useful predictive values than other nutritional scores. Chien et al conducted a cohort study of obese Asians,24 concluding that obese individuals with poor nutritional status had the highest comorbidity burden and the most adverse cardiac outcomes. Taken together, from these studies we hypothesize that there is a correlation between malnutrition as reflected by PNI and mortality in T2DM. The mechanisms linking ALB levels to mortality from diabetes are unclear. However, several mechanisms may explain why serum ALB is associated with diabetes risk. First, increased oxidative stress may lead to insulin resistance, and the antioxidant properties of serum ALB, such as multiple ligand-binding capacity and free radical trapping properties, may play an important role in the development of diabetes.25 26 Second, low ALB levels are associated with low skeletal muscle mass, which is associated with insulin resistance and increased risk of diabetes.27 Third, inflammation is associated with pathogenesis of diabetes. Low serum ALB levels have been associated with inflammation, and inflammatory cytokines such as interleukin-8 (IL-8), IL-1β, IL-6, IL-10, and tumor necrosis factor-α are associated with reduced serum ALB levels.28

Research on the relationship between lymphocytes and all-cause mortality and cardiovascular mortality in T2DM is relatively limited. Some studies have reported that lymphocytes regulate the production of inflammatory mediators by macrophages and are also essential for obesity-associated inflammation.29 Lymphocytes are increased in adipose tissue in individuals with diet-induced obesity, insulin resistance, and diabetes.30 31 It has been suggested that lymphocytes appear to be the major leucocyte subpopulation associated with the development of diabetes. The authors also attributed this association to obesity and impaired insulin.32 Zhang conducted a large cohort study (n=8991) and showed that elevated levels of leucocyte counts, neutrophils, and lymphocytes were all predictors of the incidence of type 2 diabetes (T2D).33

In addition, we found a non-linear relationship between PNI and both all-cause mortality and cardiovascular mortality in patients with T2D. The results indicated that the risk of mortality due to any cause and cardiovascular mortality in patients with diabetes was elevated when PNI levels were below 53 or above 80, compared with the appropriate range of PNI (53–80). Similar studies were retrieved, such as the NHANES cohort study conducted by Zhang, which also examined the relationship between PNI and all-cause mortality in patients with T2DM. The study concluded that PNI, a marker of immunonutrition, was an independent predictor of all-cause mortality in patients with T2DM.33 Notably, this study also investigated the relationship between PNI and the incidence of diabetic nephropathy, an aspect that was not addressed in our own study. By exploring both all-cause mortality and cardiovascular mortality, our study aims to contribute to the groundwork for future large-scale prospective studies.

In our large prospective study, we used a nationally representative sample of US adults with diabetes to define PNI, based on previous studies. After multifactorial adjustment, we found a non-linear relationship between serum PNI levels (range 8.39–270) and all-cause and CVD mortality. To account for the variation in PNI levels across individuals, we stratified our analysis by race/ethnicity (white or non-white) and observed similar results in each stratum. Similarly, we observed consistent results when we further stratified the analysis by smoking, obesity status, and duration of diabetes mellitus.

The current investigation is an extensive, long-term study with a representative sample, making it highly reliable. We took great care to account for various potential factors, such as lifestyle choices, dietary habits, duration of diabetes, glycemic control, and lipid levels, in order to ensure the applicability of our findings. Nonetheless, it is crucial to acknowledge certain limitations. In light of the study’s observational design, it is not possible to establish a causal relationship. Additionally, the fact that we only had a single baseline measurement of serum PNI prevented us from assessing PNI values at the time of the endpoint event. Although we attempted to account for factors such as diabetes duration, medication use, and glycated hemoglobin levels, it should be noted that detailed information on the severity of diabetes was not available. Moreover, since our results were derived from a cohort of US adults with diabetes, their generalizability to other populations may be somewhat limited. The subgroup analysis lacked sufficient statistical power and should thus be interpreted with caution. Finally, the presence of unknown confounding variables cannot be entirely ruled out.

Conclusion

Lower serum PNI levels were significantly associated with higher all-cause and CVD mortality. These findings suggest that maintaining an appropriate range of serum PNI status may reduce the risk of death in patients with diabetes.

Data availability statement

No data are available.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

This study involves human participants and the National Center for Health Statistics Research Ethics Review Board approved all NHANES protocols (protocol numbers: 98-12, 2005-06, 2011-17, 2018-01). Participants gave informed consent to participate in the study before taking part.

Acknowledgments

We thank all the blood donors who agreed to this scientific study.