Association of pan-immune inflammation value and systemic immune inflammation index with mortality in MASLD: A cohort study from NHANES 1999 to 2018

Abstract

Identifying prognostic risk factors in metabolic dysfunction-associated steatotic liver disease (MASLD) remains critical. This study investigates the association of the pan-immune inflammation value (PIV) and systemic immune inflammation index (SII) with mortality in the MASLD population. This cohort study used data from the National Health and Nutrition Examination Survey 1999 to 2018, associated with the National Death Index records, including 6252 patients with MASLD. Natural logarithms of PIV and SII (lnPIV, lnSII) were analyzed due to right-skewed distributions. Kaplan–Meier survival curves and multivariate Cox proportional hazards models were employed to assess the relationships between lnPIV, lnSII, all-cause and cardiovascular disease (CVD) mortality. Restricted cubic splines were applied to explore nonlinear trends, and segmented Cox regression was used to analyze threshold effects. Model performance was evaluated using the concordance index, time-dependent receiver operating characteristic curves, and calibration plots. Subgroup analyses were conducted to assess the robustness of the findings across different populations. Over a median follow-up of 9.17 years, 1154 all-cause deaths and 373 CVD-related deaths were recorded. Compared with individuals in the lowest quartile (Q1), those in the highest quartile (Q4) of lnPIV and lnSII had multivariable-adjusted hazard ratios of 1.36 (95% confidence interval [CI], 1.08–1.70) and 1.28 (95% CI, 1.02–1.61) for all-cause mortality, and 1.59 (95% CI, 1.00–2.52) and 1.76 (95% CI, 1.18–2.62) for CVD mortality, respectively. Nonlinear associations were observed between lnPIV, lnSII, and all-cause mortality, with thresholds identified at lnPIV = 6.015 and lnSII = 6.342. No significant associations were detected (P > .05) below these thresholds, whereas a significant positive association was found above the thresholds (P < .001). In contrast, a linear relationship was observed between lnPIV, lnSII, and CVD mortality. Time-dependent receiver operating characteristic curves and calibration plots demonstrated good model discrimination and calibration. No significant interactions were found across most subgroups (P for interaction > .05). Elevated PIV and SII are associated with an increased risk of all-cause and CVD mortality in MASLD. A significantly higher hazard of all-cause mortality was observed when lnPIV and lnSII were above specific thresholds. They are potential tools that require external validation in patients with MASLD.

1. Introduction

Metabolic dysfunction-associated steatotic liver disease (MASLD), formerly known as nonalcoholic fatty liver disease (NAFLD), was redefined in the 2023 Delphi consensus process.^[1] It is characterized by hepatocellular steatosis with the absence of secondary causes, such as excessive alcohol consumption. MASLD is also associated with at least 1 of 5 cardiometabolic risk factors: overweight or obesity, hyperglycemia, hypertension, elevated triglycerides (TG), or low levels of high-density lipoprotein cholesterol (HDL). Globally, approximately 35% of adults are affected by MASLD, which can progress to metabolic dysfunction-associated steatohepatitis (MASH), liver fibrosis, or even hepatocellular carcinoma (HCC).^[2,3] Beyond liver-related complications, MASLD is recognized as an independent risk factor for cardiovascular disease (CVD), metabolic syndrome, and extrahepatic malignancies, significantly contributing to all-cause mortality.^[4] As MASLD becomes a major global health challenge and imposes a substantial economic burden,^[5] identifying its prognostic risk factors remains important.

The pathogenesis of MASLD is complex and closely related to obesity, insulin resistance, and the activation of pro-inflammatory factors.^[6] Chronic immune activation and systemic inflammation play key roles in MASLD, driving damage in both the liver and extrahepatic organs.^[7,8] Excessive fat accumulation in the liver promotes the release of pro-inflammatory cytokines (such as tumor necrosis factor-α and interleukin-6) through activation of Kupffer cells and hepatic stellate cells, leading to the progression of liver fibrosis. Systemic inflammation also creates a 2-way vicious cycle with CVD and metabolic syndrome by inducing endothelial dysfunction, insulin resistance, and atherosclerotic plaque formation.^[9] Meta-analyses have demonstrated a significantly increased risk of CVD events and all-cause mortality in patients with MASLD.^[10,11]

Multiple inflammatory biomarkers have been used to evaluate the risk and prognosis of MASLD. The pan-immune inflammation value (PIV) and the systemic immune inflammation index (SII) are comprehensive inflammatory biomarkers that integrate different peripheral blood cell counts to quantify systemic inflammation.^[12,13] While simpler ratios such as the neutrophil-to-lymphocyte ratio (NLR) have been extensively studied in various disease settings, SII combines platelets, neutrophils, and lymphocytes, and PIV further incorporates monocytes, thereby capturing a broader spectrum of the innate and adaptive immune interplay. SII is calculated as (neutrophil count × platelet count)/lymphocyte count, and PIV is calculated as SII × monocyte count. Previous studies have applied SII and PIV to evaluate the risk and prognosis of CVD, cancer, and immune-related diseases.^[14,15] Both indices have also been associated with the presence and severity of MASLD^[16,17]; however, their association with long-term mortality in the MASLD population – particularly for PIV – remains unexplored.

In this study, we used data from the National Health and Nutrition Examination Survey (NHANES), a large, nationally representative cohort of the US population, characterized by rigorous methodology, quality control, and long-term follow-up. We identified the eligible MASLD population and employed various statistical analyses to evaluate the potential of PIV and SII to predict all-cause and CVD mortality in MASLD populations. These findings may contribute to the development of targeted strategies for inflammation management and risk stratification in individuals with MASLD.

2. Methods

2.1. Data source

The NHANES database, conducted by the National Center for Health Statistics, is a nationally representative survey of the US population, usually on a 2-year cycle, using a multi-stage, stratified sampling method to ensure representation across all age groups. Participants completed questionnaires with the assistance of trained staff and underwent physical examinations and laboratory tests at mobile examination centers. The NHANES investigation protocol is approved by the National Center for Health Statistics Ethics Review Committee, allowing data analysis without requiring additional ethical review.^[18] All participants were completely anonymous, and researchers had no access to information that could identify individual participants during or after data collection. The study used data from NHANES for 10 cycles (1999–2018), with mortality data obtained by linking NHANES records to the National Death Index, providing follow-up data through December 31, 2019. A detailed description of the dataset is available on the NHANES official website (https://www.cdc.gov/nchs/nhanes/). The dataset was accessed on February 28, 2025.

2.2. Participant selection

In total, 101,316 participants were reviewed during the 10 cycles (1999–2018) selected. After excluding the participants aged below 20 years old and pregnant women, 53,540 participants remained. Next, we excluded the following people: missing data required to calculate the fatty liver index (FLI), PIV, SII; those with FLI <60; hepatitis B virus or hepatitis C virus infection; autoimmune liver disease; excessive alcohol consumption (≥30 g/d for males and ≥ 20 g/d for females) and other causes of steatotic liver disease (SLD); no cardiometabolic risk factors; people lost to follow-up. We can get a cohort of 8411 MASLD patients. Additional exclusions were made for missing covariate data, including fibrosis-4 score (FIB-4), fasting blood glucose (FBG), glycated hemoglobin A1c (HbA1c), education level, marital status, poverty-income ratio (PIR), smoking status, alcohol consumption, and cancer history. Finally, 6252 patients with MASLD were included in this study (Fig. 1). The original data of the patients included in the study are in “raw data.xlsx” in the Supplementary material section.

Figure 1.

Flowchart depicting the process of selecting the study population. FBG = fasting blood glucose, FLI = fatty liver index, HbA1c = glycated hemoglobin A1c, MASLD = metabolic dysfunction-associated steatotic liver disease, NHANES = National Health and Nutrition Examination Survey, PIR = poverty-income ratio, PIV = pan-immune inflammation value, SII = systemic immune inflammation index, SLD = steatotic liver disease.

2.3. Assessment of MASLD

We identified hepatic steatosis using the FLI,^[19] a reliable tool for assessing steatosis liver disease (SLD), due to the absence of ultrasound assessments of hepatic steatosis in most study cycles. As a surrogate marker for hepatic steatosis, FLI may misclassify steatosis status compared with direct imaging modalities or elastography; this limitation should be considered when interpreting our findings. FLI is calculated by the following formula:

$$\begin{array}{c}{\displaystyle }\mathrm{F}\mathrm{L}\mathrm{I}=\frac{\left(e^{0.953\times ln\left(TG\right)+0.139\times BMI+0.718\times ln\left(GGT\right)+0.053\times \left(WC\right)-15.745}\right)}{(1+e^{0.953\times ln\left(TG\right)+0.139\times BMI+0.718\times ln\left(GGT\right)+0.053\times \left(WC\right)-15.745})}\times 100\end{array}$$

People with FLI ≥60 were selected, who were considered to have a high probability of hepatic steatosis.^[19] It is also needed to rule out viral hepatitis, autoimmune liver disease, alcoholic liver disease (≥30 g/d for males and ≥20 g/d for females^[1]), and other factors that contribute to SLD. Daily alcohol consumption was obtained using the United States Department of Agriculture’s automated multiple-pass method based on 24-hour dietary recall.^[20] Finally, MASLD diagnosis required at least one of the following 5 cardiometabolic risk factors^[1]:

1. Body mass index (BMI) ≥25 kg/m² or waist circumference (WC) >94 cm for males and >80 cm for females.

2. Fasting serum glucose ≥100 mg/dL or 2-hour post-load glucose levels ≥140 mg/dL or HbA1c ≥5.7% or diabetes mellitus (DM) or treatment for DM.

3. Blood pressure ≥130/85 mm Hg or antihypertensive drug treatment.

4. Plasma TG ≥150 mg/dL or lipid-lowering treatment.

5. Plasma HDL ≤40 mg/dL for males and ≤50 mg/dL for females or lipid-lowering treatment.

2.4. Measurement of PIV and SII

At mobile examination centers, blood samples were collected from NHANES participants to measure complete blood cell counts, expressed as ×10³ cells/µL. The formulas for calculating PIV and SII are as follows^[21,22]:

$$\begin{array}{c}{\displaystyle }\mathrm{P}\mathrm{I}\mathrm{V}=\frac{\mathrm{p}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{l}\mathrm{e}\mathrm{t}\mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}\times \mathrm{n}\mathrm{e}\mathrm{u}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{p}\mathrm{h}\mathrm{i}\mathrm{l}\mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}\times \mathrm{m}\mathrm{o}\mathrm{n}\mathrm{o}\mathrm{c}\mathrm{y}\mathrm{t}\mathrm{e}\mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}}{\mathrm{l}\mathrm{y}\mathrm{m}\mathrm{p}\mathrm{h}\mathrm{o}\mathrm{c}\mathrm{y}\mathrm{t}\mathrm{e}\mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}}\end{array}$$

$$\begin{array}{c}{\displaystyle }\mathrm{S}\mathrm{I}\mathrm{I}=\frac{\mathrm{p}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{l}\mathrm{e}\mathrm{t}\mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}\times \mathrm{n}\mathrm{e}\mathrm{u}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{p}\mathrm{h}\mathrm{i}\mathrm{l}\mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}}{\mathrm{l}\mathrm{y}\mathrm{m}\mathrm{p}\mathrm{h}\mathrm{o}\mathrm{c}\mathrm{y}\mathrm{t}\mathrm{e}\mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}}\end{array}$$

As PIV and SII were positively skewed distributions in the included population, their natural logarithmic forms, lnPIV and lnSII, were used in data analysis. MASLD participants were divided into 4 groups based on lnPIV and lnSII quartiles, with quartile 1 set as the reference group.

2.5. Clinical characteristics and covariates

The demographic characteristics were collected, including age, gender (male or female), race (Hispanic, non-Hispanic White, non-Hispanic Black, or other races), marital status (single or not single), educational level (less than high school, high school or equivalent, college or above), and PIR (<1.3, 1.3–3.5, or > 3.5^[23]). Lifestyle habits, including smoking (never, former, and current^[24]) and alcohol use (never, former, and current^[25]), were collected through questionnaires.

Medical history includes hypertension, DM, CVD, cancer, and hyperlipidemia (all classified as yes or no). Hypertension was determined by self-reported physician diagnosis, the use of antihypertensive medication, or a measured blood pressure ≥140/90 mm Hg during a physical examination.^[26] DM was determined by self-reported DM diagnosis, using hypoglycemic drugs or insulin, or by FBG ≥126 mg/dL or HbA1c ≥6.5% on an examination.^[27] CVD was identified based on questionnaire responses regarding congestive heart failure, coronary heart disease, angina/angina pectoris, heart attack, or stroke. Cancer was diagnosed through self-reported questionnaires. Hyperlipidemia was determined by a self-reported medical history, taking lipid-lowering drugs, or meeting the following diagnostic criteria: LDL ≥ 130 mg/dL, or total cholesterol (TC) ≥ 200 mg/dL, or TG ≥ 150 mg/dL, or HDL < 40 for males and HDL < 50 for females.

Physical examination data included BMI (kg/m²) and WC (cm). Laboratory test data included alanine aminotransferase (U/L), aspartate aminotransferase (U/L), albumin (Alb, g/L), FBG (mg/dL), HbA1c (%), TC (mg/dL), and HDL (mg/dL). The FIB-4, a reliable indicator of liver fibrosis, is calculated as FIB-4=$\mathrm{F}\mathrm{I}\mathrm{B}-4=(\mathrm{a}\mathrm{g}\mathrm{e}\times \mathrm{A}\mathrm{S}\mathrm{T})/(\mathrm{p}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{l}\mathrm{e}\mathrm{t}\mathrm{c}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{t}\times \sqrt{\mathrm{A}\mathrm{L}\mathrm{T}})$, and FIB-4 ≥ 1.3 is considered advanced fibrosis.^[28]

2.6. Outcome measurement

We obtained all-cause and CVD mortality information from the National Death Index database associated with NHANES for participants with follow-up available through December 31, 2019 (https://www.cdc.gov/nchs/linked-data/mortality-files/?CDC_AAref_Val=https://www.cdc.gov/nchs/data-linkage/mortality-public.htm). Causes of death were classified using the International Classification of Diseases, Tenth Revision. Death from any cause was defined as all-cause mortality. CVD mortality refers to deaths due to major cardiovascular and cerebrovascular diseases (International Classification of Diseases codes: I00–I09, I11, I13, I20–I51, and I60–I69).

2.7. Statistical analysis

R software (version 4.4.2; R Foundation for Statistical Computing) was used for statistical analyses. Following the NHANES analysis guidelines (https://wwwn.cdc.gov/nchs/nhanes/tutorials/weighting.aspx) (accessed on March 1, 2025), sample weights, clustering, and stratification were considered in data analysis to ensure the reliability and representation of the analysis results in the US MASLD population. P < .05 (two-sided) is considered statistically significant. Missing covariate data were handled by complete-case analysis, excluding participants with any missing values. The proportional hazards assumption was formally tested using Schoenfeld residuals. Both global and variable-specific tests were performed; a nonsignificant P value (>.05) indicates that the assumption holds.

The Kolmogorov-Smirnov test was used to check the assumed distribution of normality for each variable, and all continuous variables in this study were not normally distributed. Consequently, continuous variables were expressed as median (interquartile range, IQR) and compared using the rank-sum test for 2 independent samples. The classified data were represented by numbers (weighted percentage, %) using the chi-square test or the Fisher exact probability method.

To assess the association between lnPIV, lnSII, and all-cause and CVD mortality, multivariate Cox proportional hazard models were used, with results expressed as hazard ratios (HR) and 95% confidence intervals (CI). Specifically, model 1 serves as an unadjusted analysis. In model 2, we adjust for age, gender, and race. In model 3, we adjust for age, gender, race, education level, marital status, PIR, smoking, alcohol use, hypertension, DM, CVD, cancer, BMI, serum TC, HDL, Alb, and FIB-4. We take age, BMI, TC, HDL, and Alb as continuous variables, and gender, race, education level, marital status, smoking, alcohol use, hypertension, DM, CVD, and cancer as categorical variables. PIR and FIB-4 are converted to categorical variables according to thresholds. Kaplan–Meier survival curves were generated to illustrate differences in all-cause and CVD mortality across quartiles of lnPIV and lnSII, with comparisons made using the log-rank test. Restricted cubic splines (RCS) were applied to investigate potential nonlinear relationships between lnPIV, lnSII, and mortality. Segmented Cox regression models were used to identify threshold effects and breakpoints, with likelihood ratio tests employed to compare segmented models with standard Cox models. To assess the discriminatory performance of the Cox proportional hazards models, the concordance index (C-index) was calculated, and time-dependent receiver operating characteristic (ROC) curves were generated at 12, 36, and 60 months of follow-up. For calibration assessment, calibration plots were constructed comparing the predicted survival probabilities from the models with the observed Kaplan–Meier estimates at the same time points.

Subgroup analyses were conducted to examine the association between lnPIV, lnSII, and mortality across different populations, stratified by age (age < 60, age ≥ 60), gender, race (combining Other Hispanic with Other Race), smoking status, alcohol consumption, BMI (BMI < 30, BMI ≥ 30), DM, CVD, and FIB-4. Interaction terms were included to evaluate effect modifications, and significant interactions were further explored to provide subgroup-specific insights.

3. Results

3.1. Baseline characteristics of participants

Table 1 describes the weighted baseline characteristics of 6252 participants grouped according to all-cause and CVD mortality, with 1154 (18.46%) all-cause deaths and 373 (5.97%) CVD deaths. The median age of participants was 52 years, and most were male (n = 3380, 55.15%), non-Hispanic White (n = 3027, 73.36%), with a college degree or above (n = 2872, 54.63%), non-single (n = 4589, 77.83%), and had a high income (n = 1840, 41.22%). There were 3603 (53.83%) patients with hypertension, 1832 (23.35%) patients with diabetes, 1005 (13.19%) patients with CVD, 5496 (88.05%) patients with hyperlipidemia, and 711 (11.30%) patients with tumors. A total of 2108 (27.96%) participants were considered to have advanced fibrosis. The median values for inflammatory indices were as follows: PIV (270.13), lnPIV (5.60), SII (504.70), and lnSII (6.22). Compared with survivors, non-survivors were characterized by older age, non-Hispanic White ethnicity, lower BMI, lower education level, being single, concurrent with comorbidities, tended to smoke, lower income level, and higher FBG, HbA1c, FIB-4, lnPIV, and lnSII.

Table 1.

Baseline characteristics of participants with MASLD in the present study.

Variable	Total (n = 6252)	All-cause death			CVD death
		No (n = 5098, 81.54%)	Yes (n = 1154, 18.46%)	P	No (n = 5879, 94.3%)	Yes (n = 373, 5.97%)	P
Age, M (Q1, Q3), years	52.00 (40.00, 63.00)	50.00 (39.00, 60.00)	68.00 (58.00, 76.00)	<.001	51.00 (40.00, 62.00)	69.00 (59.00, 76.00)	<.001
BMI, M (Q1, Q3), kg/m2	33.00 (29.94, 37.20)	33.10 (30.10, 37.36)	31.96 (29.00, 36.08)	<.001	33.00 (29.96, 37.20)	32.52 (29.28, 37.21)	.224
WC, M (Q1, Q3), cm	110.50 (103.80, 119.60)	110.40 (103.70, 119.50)	111.50 (104.70, 120.10)	.047	110.40 (103.70, 119.30)	113.50 (106.00, 123.80)	<.001
ALT, M (Q1, Q3), U/L	25.00 (19.00, 33.00)	25.00 (19.00, 34.00)	21.00 (17.00, 28.00)	<.001	25.00 (19.00, 33.00)	21.00 (17.00, 27.00)	<.001
AST, M (Q1, Q3), U/L	23.00 (19.00, 28.00)	23.00 (20.00, 28.00)	23.00 (19.00, 28.00)	.083	23.00 (19.00, 28.00)	23.00 (19.00, 28.00)	.121
GGT, M (Q1, Q3), U/L	26.00 (19.00, 39.00)	26.00 (19.00, 38.00)	27.00 (19.00, 43.00)	.03	26.00 (19.00, 39.00)	28.00 (19.00, 46.00)	.218
Platelet count, M (Q1, Q3), 103/μL	247.00 (208.00, 294.00)	249.00 (210.00, 294.00)	237.00 (197.00, 289.00)	<.001	248.00 (209.00, 294.00)	229.00 (191.00, 278.00)	<.001
Neutrophil count, M (Q1, Q3), 103/μL	4.10 (3.30, 5.10)	4.00 (3.20, 5.00)	4.20 (3.50, 5.40)	<.001	4.00 (3.30, 5.10)	4.50 (3.50, 5.50)	<.001
Monocyte count, M (Q1, Q3), 103/μL	0.50 (0.40, 0.70)	0.50 (0.40, 0.60)	0.60 (0.50, 0.70)	<.001	0.50 (0.40, 0.70)	0.60 (0.50, 0.70)	.001
Lymphocyte count, M (Q1, Q3), 103/μL	2.00 (1.60, 2.40)	2.00 (1.70, 2.40)	1.80 (1.50, 2.30)	<.001	2.00 (1.70, 2.40)	1.80 (1.40, 2.20)	<.001
Alb, M (Q1, Q3), g/L	42.00 (40.00, 44.00)	42.00 (40.00, 44.00)	41.00 (39.00, 43.00)	<.001	42.00 (40.00, 44.00)	41.00 (38.00, 43.00)	<.001
FBG, M (Q1, Q3), mg/dL	104.00 (97.00, 117.00)	104.00 (96.40, 115.00)	109.00 (99.00, 130.40)	<.001	104.00 (97.00, 116.00)	111.80 (100.00, 139.00)	<.001
HbA1c, M (Q1, Q3), %	5.60 (5.30, 6.00)	5.60 (5.30, 5.90)	5.80 (5.40, 6.50)	<.001	5.60 (5.30, 6.00)	5.90 (5.50, 6.70)	<.001
HDL, M (Q1, Q3), mg/dL	45.00 (38.00, 53.00)	45.00 (38.00, 53.00)	45.00 (38.00, 54.00)	.971	45.00 (38.00, 53.00)	45.00 (36.00, 54.00)	.756
TC, M (Q1, Q3), mg/dL	198.00 (171.00, 226.00)	199.00 (172.00, 226.00)	194.00 (167.00, 228.00)	.199	198.00 (171.00, 226.00)	192.00 (164.00, 230.00)	.293
FLI, M (Q1, Q3)	85.77 (73.73, 94.84)	85.87 (73.61, 94.87)	85.39 (74.86, 94.67)	.597	85.72 (73.59, 94.73)	88.50 (78.19, 95.80)	.014
PIV, M (Q1, Q3)	270.13 (182.25, 405.70)	264.60 (179.96, 391.26)	318.26 (206.40, 509.47)	<.001	267.73 (181.60, 401.10)	340.96 (214.40, 515.82)	<.001
SII, M (Q1, Q3)	504.70 (370.00, 701.50)	495.65 (368.68, 688.00)	561.05 (381.43, 790.40)	<.001	501.67 (369.60, 697.67)	571.65 (417.59, 786.17)	<.001
LnPIV, M (Q1, Q3)	5.60 (5.21, 6.01)	5.58 (5.19, 5.97)	5.76 (5.33, 6.23)	<.001	5.59 (5.20, 5.99)	5.83 (5.37, 6.25)	<.001
LnSII, M (Q1, Q3)	6.22 (5.91, 6.55)	6.21 (5.91, 6.53)	6.33 (5.94, 6.67)	<.001	6.22 (5.91, 6.55)	6.35 (6.03, 6.67)	<.001
FIB-4, M (Q1, Q3)	0.94 (0.66, 1.36)	0.89 (0.63, 1.27)	1.41 (0.95, 1.96)	<.001	0.93 (0.65, 1.33)	1.45 (1.00, 2.01)	<.001
Gender, n (%)				.421			.011
Male	3380 (55.15)	2699 (54.91)	681 (56.65)		3151 (54.82)	229 (62.63)
Female	2872 (44.85)	2399 (45.09)	473 (43.35)		2728 (45.18)	144 (37.37)
Race, n (%)				<.001			.001
Mexican American	1344 (9.11)	1159 (9.93)	185 (4.05)		1292 (9.33)	52 (4.11)
Other Hispanic	512 (5.07)	459 (5.31)	53 (3.56)		491 (5.11)	21 (4.27)
Non-Hispanic White	3027 (73.36)	2323 (72.18)	704 (80.69)		2805 (73.14)	222 (78.27)
Non-Hispanic Black	1061 (8.26)	874 (8.21)	187 (8.54)		987 (8.11)	74 (11.65)
Other Race	308 (4.20)	283 (4.36)	25 (3.16)		304 (4.31)	4 (1.69)
PIR, n (%)				<.001			<.001
PIR < 1.3	1892 (20.05)	1490 (18.97)	402 (26.83)		1770 (19.85)	122 (24.59)
PIR1.3–3.5	2520 (38.73)	1997 (37.48)	523 (46.56)		2340 (38.33)	180 (47.88)
PIR > 3.5	1840 (41.22)	1611 (43.55)	229 (26.61)		1769 (41.82)	71 (27.53)
Education level, n (%)				<.001			<.001
Less than high school	1861 (18.88)	1397 (17.03)	464 (30.45)		1711 (18.30)	150 (32.16)
High school or equivalent	1519 (26.49)	1230 (26.16)	289 (28.49)		1427 (26.41)	92 (28.11)
College or above	2872 (54.63)	2471 (56.80)	401 (41.06)		2741 (55.28)	131 (39.74)
Marital status, n (%)				<.001			<.001
Not single	4589 (77.83)	3868 (79.44)	721 (67.79)		4369 (78.45)	220 (63.80)
Single	1663 (22.17)	1230 (20.56)	433 (32.21)		1510 (21.55)	153 (36.20)
Alcohol use, n (%)				<.001			<.001
Never	866 (11.40)	683 (11.02)	183 (13.78)		808 (11.35)	58 (12.56)
Former	1620 (23.06)	1176 (20.72)	444 (37.69)		1479 (22.36)	141 (38.91)
Current	3766 (65.54)	3239 (68.26)	527 (48.54)		3592 (66.28)	174 (48.53)
Smoking, n (%)				<.001			<.001
Never	3174 (51.47)	2736 (53.91)	438 (36.26)		3038 (52.24)	136 (34.00)
Former	2040 (31.85)	1511 (29.90)	529 (44.07)		1851 (31.07)	189 (49.78)
Current	1038 (16.67)	851 (16.19)	187 (19.68)		990 (16.69)	48 (16.22)
Hypertension, n (%)				<.001			<.001
Yes	3603 (53.83)	2723 (50.75)	880 (73.09)		3302 (52.76)	301 (78.29)
No	2649 (46.17)	2375 (49.25)	274 (26.91)		2577 (47.24)	72 (21.71)
Diabetes, n (%)				<.001			<.001
Yes	1832 (23.35)	1335 (20.90)	497 (38.62)		1667 (22.51)	165 (42.31)
No	4420 (76.65)	3763 (79.10)	657 (61.38)		4212 (77.49)	208 (57.69)
CVD, n (%)				<.001			<.001
Yes	1005 (13.19)	597 (9.73)	408 (34.81)		849 (11.90)	156 (42.53)
No	5247 (86.81)	4501 (90.27)	746 (65.19)		5030 (88.10)	217 (57.47)
Hyperlipidemia, n (%)				.022			.022
Yes	5496 (88.05)	4465 (87.65)	1031 (90.56)		4465 (87.65)	1031 (90.56)
No	756 (11.95)	633 (12.35)	123 (9.44)		633 (12.35)	123 (9.44)
Cancer, n (%)				<.001			.004
Yes	711 (11.30)	475 (9.70)	236 (21.29)		654 (11.08)	57 (16.29)
No	5541 (88.70)	4623 (90.30)	918 (78.71)		5225 (88.92)	316 (83.71)
FIB-4, n (%)				<.001			<.001
≤1.3	4144 (72.04)	3713 (76.48)	431 (44.31)		4025 (73.47)	119 (39.28)
>1.3	2108 (27.96)	1385 (23.52)	723 (55.69)		1854 (26.53)	254 (60.72)

Non-normally distributed variables are displayed as median with 1st and 3rd quartile (M, Q1, and Q3). Categorical variables are displayed as numbers with weighted percentages (n, %).

ALT = alanine aminotransferase, AST = aspartate aminotransferase, BMI = body mass index, CVD = cardiovascular disease, FBG = fasting blood glucose, FIB-4 = fibrosis-4 score, FLI = fatty liver index, GGT = gamma-glutamyl transferase, HbA1c = glycated hemoglobin A1c, HDL = high-density lipoprotein cholesterol, HR = hazard ratio, M = median, PIR = poverty-income ratio, Q = quartile, TC = total cholesterol, WC = waist circumference.

3.2. Relationship of lnPIV and lnSII to mortality

Compared with the low-quartile subgroup, MASLD participants with higher lnPIV and lnSII exhibited a significantly increased risk of all-cause and CVD mortality (P < .01, log-rank test; Fig. 2). In Cox proportional hazard models (Table 2), when lnPIV and lnSII were analyzed as continuous variables, each unit increase was significantly and positively associated with higher all-cause and CVD mortality across all 3 models (HR > 1, P < .05). Compared with quartile 1 (reference group), participants in quartile 4 of lnPIV and lnSII had a significantly elevated risk of both all-cause and CVD mortality in all 3 models (HR > 1, P ≤ .05). Quartile 2 of lnSII showed a lower risk of all-cause mortality in model 1 compared with quartile 1 (HR [95% CI]: 0.74 [0.58–0.96], P = .021). No significant differences were observed for the remaining quartiles in relation to all-cause and CVD mortality. Further trend analysis demonstrated a statistically significant association between increasing quartiles of lnPIV and lnSII and elevated all-cause and CVD mortality (P for trend < .05), suggesting that higher lnPIV and lnSII were strongly associated with increased mortality risk in the MASLD population. A sensitivity analysis using multiple imputation yielded consistent results (Table S1, Supplemental Digital Content). The proportional hazards assumption was met for all models (all Schoenfeld residual P > .05; Table S2, Supplemental Digital Content).

Figure 2.

Kaplan–Meier curves show the survival patterns of MASLD adults. (A, B) All-cause mortality of MASLD adults with different quartile levels of lnPIV and lnSII. (C, D) CVD mortality of MASLD adults with different quartile levels of lnPIV and lnSII. CVD = cardiovascular disease, MASLD = metabolic dysfunction-associated steatotic liver disease, PIV = pan-immune inflammation value, SII = systemic immune inflammation index.

Table 2.

Hazard ratios of lnPIV and lnSII for all-cause and CVD mortality among participants with MASLD.

Models	Model 1		Model 2		Model 3
	HR (95% CI)	P	HR (95% CI)	P	HR (95% CI)	P
All-cause mortality
lnPIV	1.52 (1.29–1.78)	<.001	1.38 (1.20–1.60)	<.001	1.34 (1.17–1.54)	<.001
Q1	1.00 (Reference)		1.00 (Reference)		1.00 (Reference)
Q2	0.85 (0.64–1.13)	.254	0.90 (0.69–1.18)	.451	0.96 (0.75–1.22)	.719
Q3	1.06 (0.81–1.37)	.68	1.06 (0.84–1.34)	.633	1.13 (0.88–1.44)	.332
Q4	1.57 (1.21–2.03)	<.001	1.43 (1.14–1.79)	.002	1.36 (1.08–1.70)	.008
P for trend		<.001		<.001		<.001
lnSII	1.31 (1.07–1.60)	.008	1.29 (1.09–1.53)	.003	1.27 (1.08–1.50)	.004
Q1	1.00 (Reference)		1.00 (Reference)		1.00 (Reference)
Q2	0.74 (0.58–0.96)	.021	0.80 (0.64–1.01)	.066	0.85 (0.67–1.06)	.146
Q3	0.86 (0.67–1.10)	.229	0.87 (0.69–1.11)	.264	0.92 (0.72–1.16)	.467
Q4	1.27 (1.01–1.60)	.047	1.30 (1.05–1.62)	.016	1.28 (1.02–1.61)	.031
P for trend		.002		<.001		.005
CVD mortality
lnPIV	1.65 (1.23–2.20)	<.001	1.50 (1.14–1.99)	.004	1.47 (1.09–2.00)	.013
Q1	1.00 (Reference)		1.00 (Reference)		1.00 (Reference)
Q2	0.87 (0.58–1.33)	.529	0.96 (0.65–1.42)	.844	1.02 (0.68–1.53)	.915
Q3	1.19 (0.73–1.94)	.483	1.24 (0.78–1.96)	.368	1.31 (0.81–2.11)	.267
Q4	1.79 (1.16–2.78)	.009	1.68 (1.10–2.57)	.017	1.59 (1.00–2.52)	.05
P for trend		<.001		.004		.02
lnSII	1.44 (1.06–1.97)	.02	1.47 (1.09–1.98)	.011	1.48 (1.06–2.06)	.022
Q1	1.00 (Reference)		1.00 (Reference)		1.00 (Reference)
Q2	0.82 (0.57–1.18)	.291	0.94 (0.66–1.35)	.745	1.01 (0.69–1.48)	.954
Q3	1.12 (0.76–1.63)	.572	1.23 (0.84–1.79)	.289	1.33 (0.90–1.96)	.146
Q4	1.56 (1.12–2.18)	.009	1.73 (1.22–2.44)	.002	1.76 (1.18–2.62)	.005
P for trend		.002		<.001		.003

Model 1: unadjusted analysis. Model 2: adjusted for age, gender, and race. Model 3: adjusted for age, gender, race, education level, marital status, PIR, smoking, alcohol use, hypertension, DM, CVD, cancer, BMI, TC, HDL, Alb, and FIB-4.

Alb = albumin; FIB-4 = fibrosis-4 score, BMI = body mass index, CVD = cardiovascular disease, DM = diabetes mellitus, HDL = high-density lipoprotein cholesterol, HR = hazard ratio, PIR = poverty-income ratio, PIV = pan-immune inflammation value, SII = systemic immune inflammation index, TC = total cholesterol.

3.3. The detection of nonlinear relationships

A multifactor-adjusted RCS analysis was conducted to evaluate potential nonlinear associations between lnPIV, lnSII, and mortality (Fig. 3). The results indicated a nonlinear correlation between lnPIV, lnSII, and all-cause mortality (P for nonlinearity < .001, P for overall < .001), with the RCS curve displaying an inflection point. In contrast, lnPIV and lnSII demonstrated linear relationships with CVD mortality (P for nonlinearity > .05, P for overall < .05). Given the presence of a nonlinear association between lnPIV, lnSII, and all-cause mortality, a segmented Cox regression model was used to explore threshold effects (Table 3). The analysis revealed a significant threshold effect (P for likelihood test < .001) with breakpoints at lnPIV = 6.015 and lnSII = 6.342. Overall, lnPIV and lnSII were positively associated with all-cause mortality (HR > 1, P < .001), with a significant threshold effect observed for both variables. Below these thresholds (lnPIV = 6.015 and lnSII = 6.342), no significant association was found (P > .05), whereas above the thresholds, a substantially higher hazard of mortality was observed (HR = 1.86, 95% CI: 1.44–2.40 for lnPIV; HR = 1.98, 95% CI: 1.52–2.58 for lnSII).

Figure 3.

Restricted cubic spline fitting of the relationship between lnPIV and lnSII and mortality. Correlation of lnPIV and lnSII levels with all-cause (A, C) and CVD mortality (B, D). Hazard ratios were adjusted for age, gender, race, education level, marital status, PIR, smoking, alcohol use, hypertension, DM, CVD, cancer, BMI, TC, HDL, Alb, and FIB-4. Alb = albumin, BMI = body mass index, CVD = cardiovascular disease, DM = diabetes mellitus, FIB-4 = fibrosis-4 score, HDL = high-density lipoprotein cholesterol, PIR = poverty-income ratio, PIV = pan-immune inflammation value, SII = systemic immune inflammation index, TC = total cholesterol.

Table 3.

Threshold effect analysis of the lnPIV and lnSII on all-cause mortality in patients with MASLD.

Outcome	Effect	P
lnPIV and all-cause mortality
Fitting model by standard COX regression	1.30 (1.18–1.44)	<.001
Fitting model by 2-piecewise COX regression
Inflection point	6.015
<6.015	1.10 (0.94–1.28)	0.252
≥6.015	1.86 (1.44–2.40)	<.001
P for likelihood test		<.001
lnSII and all-cause mortality
Model 1 Fitting model by standard COX regression	1.30 (1.15–1.47)	<.001
Model 2 Fitting model by 2-piecewise COX regression
Inflection point	6.342
<6.342	0.93 (0.76–1.13)	0.464
≥6.342	1.98 (1.52–2.58)	<.001
P for likelihood test		<.001

Cox regression models were adjusted for age, gender, race, education level, marital status, PIR, smoking, alcohol use, hypertension, DM, CVD, cancer, BMI, TC, HDL, Alb, and FIB-4. The log-likelihood test was used to compare the standard and 2-piecewise COX regression, indicating the improved fit of the segmented model.

Alb = albumin, FIB-4 = fibrosis-4 score, BMI = body mass index, CVD = cardiovascular disease, DM = diabetes mellitus, HDL = high-density lipoprotein cholesterol, PIR = poverty-income ratio, PIV = pan-immune inflammation value, SII = systemic immune inflammation index, TC = total cholesterol.

3.4. Assessment of model performance

The C-index and time-dependent ROC curves were employed to assess the predictive accuracy of both lnPIV and lnSII for mortality outcomes among participants with MASLD. The C-index in predicting all-cause mortality of MASLD was 0.839 for lnPIV and 0.940 for lnSII. Similarly, the C-index for indices in predicting CVD mortality of MASLD was 0.864 for lnPIV and 0.865 for lnSII (Table S3, Supplemental Digital Content). Figure 4 displays the time-dependent AUC values of lnPIV and lnSII for predicting all-cause and CVD mortality at 12, 36, and 60 months among patients with MASLD, based on the fully adjusted model 3. Both biomarkers demonstrated good predictive accuracy across all time points, with all AUC values exceeding 0.8. Notably, lnPIV achieved the highest predictive performance for 12-month CVD mortality (AUC = 0.887), whereas the lowest AUC values were observed for 36-month all-cause mortality prediction by both lnPIV and lnSII (AUC = 0.846). Overall, the AUCs for 12-month mortality were higher than those for 36-and 60-month mortality, indicating better short-term predictive performance. Figures S1 and S2, Supplemental Digital Content, present the time-dependent ROC curves for model 1 (unadjusted) and model 2 (adjusted for age, gender, and race), respectively, both of which demonstrated lower AUC values than the fully adjusted Model 3 (Fig. 4) across all follow-up time points. Figure 5 presents the calibration curves for model 3 (fully adjusted), assessing the agreement between predicted and observed mortality risks for lnPIV and lnSII. Both biomarkers demonstrated good calibration for all-cause mortality, with the calibration plots closely approximating the 45° diagonal line. In contrast, the calibration for predicting CVD mortality was suboptimal, indicating a degree of miscalibration for this outcome. Calibration curves for model 1 (unadjusted) and model 2 (adjusted for age, gender, and race) are presented in Figures S3 and S4, Supplemental Digital Content, respectively.

Figure 4.

Time-dependent ROC curves show the predictive performance of the lnPIV and lnSII for mortality. The ROC curve and time-dependent AUC for predicting all-cause and CVD mortality using the lnPIV (A, B) and lnSII (C, D). Hazard ratios were adjusted for age, gender, race, education level, marital status, PIR, smoking, alcohol use, hypertension, DM, CVD, cancer, BMI, TC, HDL, Alb, and FIB-4. Alb = albumin, AUC = area under the curve, BMI = body mass index, CVD = cardiovascular disease, DM = diabetes mellitus, FIB-4 = fibrosis-4 score, HDL = high-density lipoprotein cholesterol, PIR = poverty-income ratio, PIV = pan-immune inflammation value, ROC = receiver operating characteristic, SII = systemic immune inflammation index, TC = total cholesterol.

Figure 5.

The calibration curves demonstrate the degree of calibration of lnPIV and lnSII in predicting mortality. The calibration curves of lnPIV (A, B) and lnSII (C, D) for predicting all-cause and CVD mortality. Hazard ratios were adjusted for age, gender, race, education level, marital status, PIR, smoking, alcohol use, hypertension, DM, CVD, cancer, BMI, TC, HDL, Alb, and FIB-4. Alb = albumin, BMI = body mass index, CVD = cardiovascular disease, DM = diabetes mellitus, FIB-4 = fibrosis-4 score, HDL = high-density lipoprotein cholesterol, PIR = poverty-income ratio, PIV = pan-immune inflammation value, SII = systemic immune inflammation index, TC = total cholesterol.

3.5. Subgroup analysis

We conducted subgroup analyses based on age (age < 60, age ≥ 60), gender, race (combining Other Hispanic with Other Race), smoking status, alcohol consumption, BMI (BMI < 30, BMI ≥ 30), DM, CVD, and FIB-4 (Table 4). The association between lnPIV, lnSII, and mortality remained consistent across most subgroups, with no significant interactions observed (P for interaction > .05). However, a significant interaction was detected between lnPIV and all-cause mortality in the gender subgroup (P for interaction = .036) and between lnSII and CVD mortality in the race subgroup (P for interaction = .011). Despite these interactions, both lnPIV and lnSII remained positively associated with mortality risk in these subgroups (HR > 1).

Table 4.

Associations of lnPIV and lnSII with mortality in different subgroups among participants with MASLD.

Variables	n (%)	lnPIV and all-cause mortality		lnPIV and CVD mortality		lnSII and all-cause mortality		lnSII and CVD mortality
		HR (95% CI)	P-int	HR (95% CI)	P-int	HR (95% CI)	P-int	HR (95% CI)	P-int
All patients	6252 (100.00)	1.34 (1.17–1.54)*		1.47 (1.09–2.00)*		1.27 (1.08–1.50)*		1.48 (1.06–2.06)*
Age			0.753		0.968		0.617		0.541
<60	3541 (56.64)	1.25 (0.91–1.70)		1.38 (0.68–2.78)		1.12 (0.75–1.69)		1.86 (0.76–4.55)
≥60	2711 (43.36)	1.33 (1.17–1.52)*		1.49 (1.14–1.95)*		1.27 (1.06–1.53)*		1.39 (0.98–1.96)
Gender			0.036		0.634		0.064		0.799
Male	3380 (54.06)	1.16 (0.97–1.40)		1.51 (1.03–2.21)*		1.10 (0.89–1.35)		1.46 (0.91–2.32)
Female	2872 (45.94)	1.59 (1.30–1.95)*		1.45 (0.98–2.14)		1.53 (1.21–1.93)*		1.52 (1.03–2.26)*
BMI			0.854		0.307		0.84		0.13
<30	1671 (26.73)	1.29 (1.04–1.59)*		1.04 (0.64–1.69)		1.22 (0.92–1.62)		0.85 (0.48–1.52)
≥30	4581 (73.27)	1.31 (1.10–1.58)*		1.63 (1.13–2.37)*		1.24 (1.01–1.52)*		1.79 (1.22–2.64)*
Alcohol use			0.47		0.843		0.469		0.954
Never	866 (13.85)	1.30 (0.91–1.86)		1.31 (0.78–2.21)		1.46 (0.93–2.29)		1.50 (0.76–2.97)
Former	1620 (25.91)	1.51 (1.15–1.97)*		1.40 (0.90–2.19)		1.50 (1.10–2.03)*		1.44 (0.86–2.40)
Current	3766 (60.24)	1.21 (0.98–1.50)		1.45 (1.00–2.12)		1.08 (0.82–1.40)		1.38 (0.81–2.34)
Smoking			0.699		0.42		0.67		0.851
Never	3174 (50.77)	1.48 (1.17–1.87)*		1.47 (1.08–2.00)*		1.46 (1.12–1.91)*		1.69 (1.15–2.48)*
Former	2040 (32.63)	1.23 (1.03–1.47)*		1.64 (1.15–2.35)*		1.21 (0.96–1.53)		1.42 (0.91–2.22)
Current	1038 (16.60)	1.27 (0.94–1.70)		1.03 (0.49–2.18)		1.07 (0.73–1.59)		1.21 (0.48–3.02)
Hypertension			0.463		0.501		0.208		0.512
Yes	3603 (57.63)	1.43 (1.22–1.66)*		1.53 (1.11–2.10)*		1.40 (1.15–1.70)*		1.51 (1.09–2.08)*
No	2649 (42.37)	1.15 (0.83–1.57)		1.34 (0.61–2.97)		1.00 (0.70–1.41)		1.48 (0.54–4.06)
Diabetes			0.386		0.159		0.525		0.357
Yes	1832 (29.30)	1.24 (0.99–1.56)		1.22 (0.80–1.85)		1.15 (0.88–1.52)		1.30 (0.85–2.00)
No	4420 (70.70)	1.42 (1.20–1.68)*		1.68 (1.17–2.40)*		1.35 (1.10–1.64)*		1.63 (1.06–2.49)*
CVD			0.604		0.497		0.802		0.464
Yes	1005 (16.07)	1.19 (0.93–1.53)		1.29 (0.83–2.01)		1.20 (0.90–1.60)		1.23 (0.81–1.86)
No	5247 (83.93)	1.41 (1.18–1.69)*		1.64 (1.14–2.34)*		1.30 (1.03–1.66)*		1.77 (1.08–2.88)*
Race			0.437		0.238		0.224		0.011
Mexican American	1344 (21.50)	1.22 (0.77–1.95)		1.10 (0.48–2.55)		1.24 (0.68–2.27)		1.17 (0.43–3.20)
Non-Hispanic White	3027 (48.42)	1.37 (1.17–1.60)*		1.44 (0.99–2.10)		1.25 (1.04–1.52)*		1.37 (0.92–2.06)
Non-Hispanic Black	1061 (16.97)	1.09 (0.85–1.38)		1.33 (0.86–2.06)		1.04 (0.77–1.41)		1.27 (0.72–2.23)
Other Race	820 (13.12)	1.25 (0.63–2.48)		2.38 (1.06–5.35)*		1.45 (0.68–3.10)		5.10 (1.82–14.24)*
FIB-4			0.24		0.389		0.582		0.522
≤1.3	4144 (66.28)	1.53 (1.16–2.01)*		1.25 (0.71–2.20)		1.42 (1.01–2.00)*		1.33 (0.70–2.54)
>1.3	2108 (33.72)	1.19 (1.02–1.37)*		1.57 (1.15–2.14)*		1.16 (0.94–1.42)		1.55 (1.06–2.27)*

Categorical variables are displayed as numbers with weighted percentages (n, %). Hazard ratios were adjusted for age, gender, race, education level, marital status, PIR, smoking, alcohol use, hypertension, DM, CVD, cancer, BMI, TC, HDL, Alb, and FIB-4.

Alb = albumin, BMI = body mass index, CVD = cardiovascular disease, DM = diabetes mellitus, FIB-4 = fibrosis-4 score, HDL = high-density lipoprotein cholesterol, HR = hazard ratio, P-int = P for interaction, PIR = poverty-income ratio, PIV = pan-immune inflammation value, SII = systemic immune inflammation index, TC = total cholesterol.

^*P < .05.

4. Discussion

This study aimed to evaluate the association of the pan-immune inflammation value (PIV) and SII with all-cause and cardiovascular mortality in individuals with MASLD using data from a nationally representative cohort. We found that elevated levels of both lnPIV and lnSII were independently associated with increased mortality risk, with nonlinear relationships observed for all-cause mortality and linear relationships for CVD mortality. Threshold effects were identified at lnPIV = 6.015 and lnSII = 6.342, above which the hazard of all-cause mortality increased. Furthermore, both biomarkers demonstrated strong discriminative ability in fully adjusted models, with AUC values exceeding 0.8 for all-cause and CVD mortality. It was observed that their accuracy in predicting short-term mortality rates was higher than that in predicting long-term mortality rates. These findings suggest that PIV and SII, as inexpensive and readily available biomarkers derived from routine blood counts, may serve as useful adjunctive tools for identifying high-risk MASLD patients who could benefit from intensified surveillance and targeted interventions. However, the identified thresholds are exploratory and require external validation before clinical application.

Systemic inflammation drives MASLD progression to MASH and advanced fibrosis.^[29] MASH and systemic inflammation affect 10% to 30% of MASLD patients, contributing to liver damage and fibrosis. Inflammation and fibrosis are key prognostic indicators, driving both liver-related complications (cirrhosis and hepatocellular carcinoma) and extrahepatic outcomes such as CVD and malignancies.^[30] Consequently, systemic immune-inflammatory biomarkers have garnered increasing research attention in MASLD.

SII is a biomarker that reflects systemic inflammation and immune status by integrating neutrophil, lymphocyte, and platelet counts. Initially developed to assess the prognosis of hepatocellular carcinoma,^[13] SII has since been used to evaluate the prognostic value of various malignancies,^[31,32] cardiovascular events,^[33,34] infectious diseases,^[35] and other conditions, demonstrating strong predictive capabilities. Zhao et al^[17] found that log2-SII was associated with all-cause mortality in patients with NAFLD based on a large, nationally representative survey of US adults. After adjusting for multiple potential confounders, log2-SII exhibited a nonlinear relationship with all-cause mortality, with a threshold of 8.8. High log2-SII levels were linked to lower survival rates in NAFLD patients. Zhang et al^[36] examined the predictive ability of 4 systemic inflammatory markers for cardiovascular mortality risk in MASLD patients using the NHANES database. SIRI (AUC = 0.70) and NLR (AUC = 0.69) were found to be superior to SII (AUC = 0.60) and PLR (AUC = 0.52) in predicting the risk of cardiovascular death, indicating their value as effective tools for risk stratification in MASLD patients.

PIV is an immune biomarker derived from SII by incorporating monocyte counts. Initially developed to assess the prognosis of metastatic colorectal cancer,^[22] PIV demonstrated strong predictive power for survival outcomes, outperforming other immune-inflammatory biomarkers. In addition, PIV has shown promising efficacy in prognostic evaluations for other malignancies,^[37] cardiovascular postoperative complications,^[38] sepsis,^[39] and more. Coste et al^[40] conducted a cohort study involving 80 MASLD participants and found that PIV levels (336.4 vs 228.63, P = .0107) and SII levels (438.47 vs 585.39, P = .0238) were significantly lower in the medium-high risk group compared with the low-risk group, highlighting the potential of PIV and SII as biomarkers for distinguishing between moderately high-risk and low-risk MASLD patients with liver fibrosis. Similarly, Jiang et al^[16] conducted a cross-sectional study of 5026 participants, revealing that higher PIV – but not SII – was associated with an increased likelihood of NAFLD and liver fibrosis, suggesting that PIV may serve as a more valuable inflammatory marker for evaluating NAFLD and liver fibrosis.

Blood cell subtypes play distinct roles in MASLD progression. In our study (Table 1), deceased patients had lower platelet and lymphocyte counts but higher neutrophil and monocyte counts. Platelet activation contributes to thrombosis and inflammation^[41]; thrombocytopenia in advanced liver disease reflects impaired liver function and splenomegaly. Neutrophils promote inflammation via reactive oxygen species, cytokine release, and neutrophil extracellular trap formation^[42–44]; elevated NLRs predict poor prognosis in cirrhosis.^[45] Lymphocytes coordinate inflammatory processes driving MASLD: Th1/Th17 cells produce pro-inflammatory cytokines,^[46] Treg cells are susceptible to oxidative stress-induced apoptosis,^[47] CD8+ T cells exert direct hepatocyte cytotoxicity, and γδ T cells respond to early metabolic stress.^[48] B cells infiltrate the liver in NASH, secreting interleukin-6 and tumor necrosis factor-α.^[49] Monocytes differentiate into macrophages upon entering tissues; Kupffer cells and monocyte-derived macrophages critically regulate MASLD progression to MASH and fibrosis.^[50] Crosstalk between peripheral and hepatic macrophages underscores the systemic nature of metabolic disease.^[51–53]

Numerous studies have investigated prognostic indicators for MASLD, with FIB-4 being one of the most extensively validated noninvasive markers for predicting liver-related outcomes and mortality. In large cohort studies, FIB-4 has demonstrated Area Under the Receiver Operating Characteristic curve values ranging from 0.72 to 0.78 for predicting all-cause mortality.^[54] Longitudinal studies have further shown that dynamic changes in FIB-4 are associated with increased risks of all-cause mortality, cardiovascular events, and liver-related complications, with adjusted HR ranging from 1.5 to 3.7 depending on the outcome and population studied.^[55] Regarding insulin resistance markers, the triglyceride-glucose (TyG) index and its derivatives (TyG-WC, TyG-Waist-to-Height Ratio) have emerged as significant predictors of mortality in MASLD. Gao et al^[56] developed a prognostic model based on insulin resistance-related indicators, achieving an AUC of 0.866 for all-cause mortality and 0.902 for CVD mortality. In comparison, our fully adjusted models demonstrated C-index values of 0.84 for lnPIV and 0.94 for lnSII in predicting all-cause mortality, and 0.86 for both biomarkers in predicting CVD mortality. These values compare favorably with the prognostic performance reported in previous studies, suggesting that PIV and SII may offer complementary value for mortality prediction in MASLD. However, it is important to note that our findings are derived from a single cohort and have not yet been externally validated; therefore, these results should be interpreted with caution. Notably, the collective evidence indicates that the prognosis of MASLD is influenced by multiple factors, including liver fibrosis, systemic inflammation, and insulin resistance. A comprehensive, multifactorial approach that integrates these diverse dimensions may represent a more scientifically robust and clinically reliable strategy for risk stratification in patients with MASLD.

A major strength of this study is the use of the NHANES database, a nationally representative dataset of the US population with a large sample size and long-term follow-up, ensuring the reliability and generality of the findings. In addition, the study design employed a variety of statistical analysis methods, including Kaplan–Meier survival curves, Cox proportional hazards models, and restricted cubic splines, to comprehensively examine the relationship between PIV, SII, and mortality in the MASLD population, with findings effectively visualized using figures and tables. The robustness of the results was further validated through subgroup analyses. Despite its strengths, this study has several limitations that should be addressed in future research. First, although we have rigorously defined the parameters, potential deviations from real-world conditions remain. For example, hepatic steatosis was defined using FLI, which, despite its good diagnostic performance, carries a risk of false positives and false negatives. More accurate diagnostic methods, such as the controlled attenuation parameter or liver biopsy, were unavailable due to database limitations. Second, our reliance on single-time measurements limits the ability to capture dynamic changes in PIV and SII over time. Future studies incorporating repeated biomarker assessments and longitudinal inflammatory trajectories are warranted to validate and extend our findings. Third, despite extensive covariate adjustment, residual confounding due to unmeasured factors cannot be entirely excluded. Notably, NHANES lacks information on anti-inflammatory medication use (e.g., NSAIDs, statins) and undiagnosed inflammatory conditions, which may influence both PIV/SII levels and mortality risk. In addition, while the NHANES database is representative of the multiracial population in the United States, it still has regional limitations. To enhance the generality of these findings, future research should incorporate large, multicenter cohort studies across diverse geographic and demographic populations. It is important to emphasize that the thresholds identified in this study are exploratory and require external validation in independent cohorts before any clinical application.

5. Conclusion

This study suggests a significant association between elevated PIV and SII and higher all-cause and CVD mortality in the US MASLD population. After adjusting for covariates, lnPIV and lnSII exhibited a nonlinear relationship with all-cause mortality and a linear relationship with CVD mortality. Segmented Cox regression analysis indicated that when lnPIV and lnSII exceeded the thresholds (lnPIV = 6.015 and lnSII = 6.342), a substantially higher hazard of all-cause mortality was observed. Time-dependent ROC curves and calibration plots indicated strong model discrimination and calibration. Subgroup analysis further confirmed the robustness of these findings across diverse populations. Given the observational design of this study, these findings should be interpreted as associative rather than causal, and external validation in independent cohorts is warranted before any clinical application.

Acknowledgments

We extend our sincere appreciation to the Centers for Disease Control and Prevention (CDC) and the National Center for Health Statistics (NCHS) for designing, coordinating, and openly sharing this comprehensive dataset.

Author contributions

Conceptualization: Jun Yu, Jingwen Zhou, Fang Ye.

Data curation: Jun Yu, Meng Sun, Jie Wan, Xiaoying Chen, Xiaoyun Dou.

Formal analysis: Jun Yu, Meng Sun, Xiaoying Chen.

Investigation: Jun Yu, Meng Sun, Jie Wan.

Methodology: Jun Yu, Meng Sun, Jie Wan, Xiaoyun Dou.

Software: Jun Yu, Meng Sun, Jie Wan, Xiaoying Chen, Xiaoyun Dou.

Supervision: Jun Yu, Jingwen Zhou, Xiaoyun Dou, Fang Ye.

Funding acquisition: Jingwen Zhou, Fang Ye.

Visualization: Jingwen Zhou.

Resources: Xiaoying Chen.

Writing – original draft: Jun Yu, Meng Sun.

Writing – review & editing: Jingwen Zhou, Fang Ye.