Introduction

Chronic liver disease (CLD) imposes a substantial and increasing global health burden, affecting millions of individuals and accounting for significant morbidity and mortality worldwide. Estimates from the Global Burden of Disease Study indicate that in 2021, there were more than 58 million incident cases of cirrhosis and other CLDs, with approximately 1.43 million deaths attributable to these conditions globally, highlighting the rising absolute burden of liver pathology despite some improvements in age-standardized rates [1]. Data from multiple epidemiologic analyses further show that liver diseases collectively account for nearly two million deaths annually, representing about 4% of all global mortality and making them a leading cause of death across regions [2].

The progression of CLD from compensated stages to clinical decompensation represents a pivotal inflection point in the natural history of the disease and is accompanied by marked increases in morbidity, mortality, and healthcare utilization. Accurate stratification of disease severity and early identification of patients at higher risk of advanced or decompensated liver disease are central to effective management and prognostication. Alterations in lipid metabolism are well recognized in CLD due to impaired hepatocellular synthesis and disrupted lipoprotein processing, leading to characteristic reductions in circulating total cholesterol, low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C), with declining lipid levels consistently observed as liver disease severity increases across Child-Turcotte-Pugh (CTP) classes and Model for End-Stage Liver Disease (MELD) strata [3–9]. Among these alterations, reductions in HDL-C and its associated apolipoproteins appear particularly pronounced and have been documented across diverse etiologies of CLD, including viral, alcohol-related, and metabolic liver diseases [10–14]. In contrast, changes in triglyceride levels have been more heterogeneous, underscoring the complex and multifactorial relationship between hepatic injury, systemic inflammation, and lipid homeostasis [7, 15].

Beyond descriptive associations, emerging evidence suggests that lipid fractions and lipid-derived indices may carry prognostic relevance in CLD. Several studies have demonstrated that lower HDL-C levels are independently associated with adverse clinical outcomes, including mortality and hepatic decompensation, even after adjustment for established prognostic scores such as MELD and CTP [16, 17]. These observations imply that lipid parameters may capture dimensions of hepatic dysfunction and systemic derangement not fully reflected by conventional scoring systems.

More recently, attention has shifted toward composite lipid indices that integrate information from both atherogenic and protective lipoprotein fractions. The non-HDL/HDL-C ratio (NHHR) represents one such metric, incorporating all apoB-containing, potentially atherogenic lipoproteins relative to HDL-C, which exerts anti-inflammatory and immunomodulatory effects. In cardiovascular and metabolic research, NHHR has demonstrated superior discriminatory ability compared with isolated lipid measures [18–20]. Emerging data also link elevated NHHR to metabolic liver disease and hepatic fibrosis; however, its relationship with clinical severity and decompensation in established CLD remains insufficiently characterized [21–24]. In this context, the present study examined the association between NHHR and clinical decompensation in patients with CLD and evaluated whether the inclusion of NHHR enhances discrimination of clinical status beyond conventional prognostic indicators.

Materials and methods

Results

Baseline characteristics

A total of 220 patients with CLD were included in the study, with a mean age of 54.5 ± 11.9 years, of whom 138 (63%) were male. Figure 1 demonstrates the age distribution of the study population. The majority of patients were in the 50–59 year age group (33.6%), followed by 40–49 years (25.5%) and 60–69 years (21.8%). Younger patients aged 18–39 years accounted for 10%, while 9.1% were aged ≥ 70 years. Overall, 96 patients (43.6%) had decompensated liver disease. Participants were categorized into tertiles according to baseline NHHR. Baseline characteristics according to NHHR tertiles are presented in Table 1. Mean NHHR increased significantly across tertiles (P < 0.001), while age and sex distribution were comparable among groups. HDL-C levels decreased progressively, and non-HDL-C levels increased across increasing NHHR tertiles (both P < 0.001). Higher NHHR was also associated with greater liver disease severity, reflected by a stepwise increase in MELD-3.0 scores (P = 0.028) and a corresponding decrease in serum albumin levels (P < 0.001). In contrast, international normalized ratio (INR), platelet counts, and serum bilirubin did not differ significantly across tertiles. The distribution of CLD etiology was similar among NHHR categories. Although a trend toward a more advanced Child-Pugh class was observed with increasing NHHR, this did not reach statistical significance. Importantly, the proportion of patients with decompensated cirrhosis increased significantly across NHHR tertiles, rising from 31.1% in tertile 1 to 53.4% in tertile 3 (P < 0.001).

Display full size

Figure 1. Age distribution of study participants (n = 220).

Table 1. Baseline characteristics of study participants stratified by NHHR tertiles (n = 220).

Characteristic	Tertile 1 (NHHR ≤ 2.0) (n = 74)	Tertile 2 (NHHR 2.1–3.0) (n = 73)	Tertile 3 (NHHR > 3.0) (n = 73)	P value
NHHR, mean ± SD	1.79 ± 0.13	2.57 ± 0.27	3.80 ± 0.35	< 0.001
Age, years, mean ± SD	52.0 ± 10.5	55.1 ± 12.3	56.4 ± 11.7	0.178
Male sex, n (%)	48 (65)	42 (58)	48 (66)	0.423
HDL-C, mg/dL, mean ± SD	47.2 ± 10.1	38.7 ± 8.8	31.5 ± 7.1	< 0.001
Non-HDL-C, mg/dL, mean ± SD	111.0 ± 16.0	123.6 ± 18.7	138.2 ± 19.8	< 0.001
MELD-3.0 score, mean ± SD	12.3 ± 4.2	13.1 ± 4.5	15.0 ± 5.1	0.028
Serum albumin, g/dL, mean ± SD	3.4 ± 0.7	3.1 ± 0.8	2.7 ± 0.8	< 0.001
Total bilirubin, mg/dL, mean ± SD	2.3 ± 1.7	2.8 ± 2.0	3.1 ± 2.1	0.092
INR, mean ± SD	1.42 ± 0.31	1.45 ± 0.34	1.50 ± 0.38	0.166
Platelet count, × 109/L, mean ± SD	155 ± 60	148 ± 58	142 ± 55	0.212
Etiology of CLD, n (%)				0.357
Alcohol-related	33 (45)	35 (48)	38 (52)
Metabolic (MASLD)	21 (28)	19 (26)	16 (22)
Viral (HBV/HCV)	16 (22)	15 (21)	15 (21)
Other*	4 (5)	4 (5)	4 (5)
Child-Pugh class (A/B/C), n	40/26/8	34/28/11	30/32/11	0.052
Decompensated disease, n (%)	23 (31.1)	34 (46.6)	39 (53.4)	< 0.001

Display full size

*: Includes autoimmune, Wilson’s disease, cryptogenic, and other less common etiologies. CLD: chronic liver disease; HDL-C: high-density lipoprotein cholesterol; INR: international normalized ratio; MASLD: metabolic dysfunction-associated steatotic liver disease; MELD: Model for End-Stage Liver Disease; NHHR: non-high-density lipoprotein/HDL-C ratio; SD: standard deviation.

Discrimination and reclassification

Discrimination and reclassification performance of the predictive models for decompensation are summarized in Table 5. In the overall cohort, the base model incorporating age, sex, and MELD-3.0 score demonstrated moderate discrimination for the presence of decompensation, with an AUC of 0.73 (95% CI 0.67–0.79). The addition of NHHR to the base model resulted in a significant improvement in discrimination, increasing the AUC to 0.79 (95% CI 0.73–0.85; P < 0.001), along with significant gains in reclassification as reflected by positive NRI = 0.20 and IDI = 0.041. In analyses stratified by NHHR tertiles, the combined model consistently outperformed the base model across all groups. The improvement in discrimination was modest and not statistically significant in tertile 1 (P = 0.138), but was significant in tertile 2 (P = 0.045) and greatest in tertile 3, where the AUC increased from 0.75 to 0.81 (P < 0.001), accompanied by substantial improvements in reclassification (NRI = 0.27; IDI = 0.053). Overall, these findings indicate that incorporation of NHHR enhances the ability of clinical models to distinguish between decompensated and compensated CLD patients, particularly among those with higher NHHR levels. These results are also demonstrated in Figure 2 through the ROC curves, which show the improved discriminatory performance of the model after the addition of NHHR.

Table 5. Classification performance of base and combined models for identifying decompensated vs. compensated patients by NHHR tertiles.

Group	Model	AUC (95% CI)	ΔAUC vs. Base	P (DeLong)	NRI	IDI
Overall cohort (n = 220)	Base Model	0.73 (0.67–0.79)	-	-	-	-
	Base + NHHR	0.79 (0.73–0.85)	+ 0.06	< 0.001	0.20	0.041
NHHR tertile 1 (≤ 2.0) (n = 74)	Base Model	0.70 (0.58–0.82)	-	-	-	-
	Base + NHHR	0.74 (0.62–0.85)	+ 0.04	0.138	0.08	0.010
NHHR tertile 2 (2.1–3.0) (n = 73)	Base Model	0.74 (0.63–0.85)	-	-	-	-
	Base + NHHR	0.78 (0.68–0.88)	+ 0.04	0.045	0.17	0.026
NHHR tertile 3 (> 3.0) (n = 73)	Base Model	0.75 (0.65–0.85)	-	-	-	-
	Base + NHHR	0.81 (0.72–0.89)	+ 0.06	< 0.001	0.27	0.053

Display full size

Base Model comprised age, sex, and MELD-3.0 score. AUC: area under the receiver operating characteristic curve; CI: confidence interval; NRI: net reclassification improvement; IDI: integrated discrimination improvement; MELD: Model for End-Stage Liver Disease; NHHR: non-high-density lipoprotein/high-density lipoprotein cholesterol ratio.

Display full size

Figure 2. ROC curves of the base model and base model plus NHHR across NHHR tertiles. ROC curves comparing the discriminatory performance of the base predictive model (incorporating age, sex, and MELD-3.0 score) and the model augmented with NHHR across chronic liver disease cohorts: (A) overall cohort, (B) NHHR tertile 1 (≤ 2.0), (C) NHHR tertile 2 (2.1–3.0), and (D) NHHR tertile 3 (> 3.0). For each panel, the true positive rate (sensitivity) is plotted against the false positive rate (1 – specificity). The AUC for each model is indicated in the legend. The diagonal dashed line represents the reference line for random classifier performance. AUC: area under the receiver operating characteristic curve; MELD: Model for End-Stage Liver Disease; NHHR: non-high-density lipoprotein/high-density lipoprotein cholesterol ratio; ROC: receiver operating characteristic.

Discussion

In this cross-sectional study of patients with CLD, we found that an elevated NHHR was associated with markers of advanced liver disease and with the presence of clinical decompensation. Patients with higher NHHR tertiles had progressively greater incidence of ascites, hepatic encephalopathy, and overall decompensation. A higher NHHR correlated moderately with MELD-3.0 and total bilirubin and inversely with serum albumin and platelet count. NHHR remained independently associated with decompensation after adjustment for key confounders. Furthermore, NHHR significantly enhanced discrimination and reclassification for decompensation when added to established clinical predictors such as the MELD-3.0 score. These findings highlight the potential utility of NHHR as a simple, accessible biomarker to complement conventional risk stratification in CLD.

A substantial body of literature demonstrates profound disturbances in lipid metabolism in CLD. Hepatic synthesis and secretion of lipoproteins are impaired in cirrhotic patients due to reduced apolipoprotein production and defective lipoprotein assembly, leading to progressive decreases in total cholesterol, HDL-C, and LDL-C with advancing liver dysfunction. Several observational studies have confirmed that lipid parameters, particularly HDL-C, decline with worsening Child-Pugh and MELD scores, reflecting impaired hepatic synthetic capacity and more severe disease states. Reduced HDL-C has been linked to poorer outcomes in cirrhosis. Previous research demonstrated that HDL-related biomarkers correlate inversely with disease severity and are independent predictors of survival in CLD, suggesting that HDL-C may serve as a prognostic indicator. Trieb et al. [17] demonstrated that HDL-related biomarkers reflect deterioration in liver function and reliably predict disease progression, development of complications, and survival in patients with cirrhosis. In particular, HDL-C and apoA-I decrease with increasing severity of the disease regardless of the etiology of cirrhosis. The authors further emphasized that HDL-C and apoA-I show excellent diagnostic accuracies for 90-day and 12-month mortality, which are very similar to those of composite scores such as Child-Pugh and MELD [17]. Similarly, Habib et al. [27] established HDL-C as a liver function test and prognostic factor in non-cholestatic cirrhosis. In patients hospitalized with spontaneous bacterial peritonitis, Llovet et al. [28] found that HDL-C was one of the parameters independently associated with short-term prognosis. In line with these observations, we found that patients in the highest NHHR tertile, where HDL-C levels were lowest, had significantly higher decompensation rates.

While HDL-C has been widely studied in CLD, composite lipid indices such as NHHR have been less extensively evaluated in this context. HDL-C has recognized anti-inflammatory and endothelial protective properties, whereas non-HDL fractions reflect atherogenic and proinflammatory lipoproteins; the NHHR therefore captures a balance between potentially protective and harmful lipid fractions. Studies in metabolic and cardiovascular populations show that ratios incorporating non-HDL and HDL cholesterol capture this balance more effectively than individual lipid measures [18–20]. In our cohort, NHHR demonstrated consistent associations with markers of hepatic dysfunction, correlating positively with MELD-3.0 score, total bilirubin, and INR, and inversely with serum albumin. These relationships indicate that higher NHHR aligns with more advanced liver dysfunction, reflecting impaired synthetic capacity and cholestatic injury. Importantly, our cohort comprised patients with CLD of diverse etiologies, including alcoholic liver disease, metabolic dysfunction-associated steatotic liver disease (MASLD), viral hepatitis-related liver disease, and other causes, thereby enhancing the generalizability of these findings across the etiologic spectrum of CLD. Notably, NHHR was independently associated with clinical decompensation in multivariable logistic regression models, underscoring its potential utility as an integrated indicator of disease severity. Our findings resonate with emerging evidence linking lipid and lipoprotein disturbances to hepatic outcomes. An et al. [29] reported that the NHHR was associated with worse liver function and independently predicted adverse outcomes in liver disease cohorts, suggesting that composite lipid ratios capture metabolic and lipoprotein derangements beyond conventional measures. Similarly, in non-alcoholic fatty liver disease populations, studies have demonstrated positive associations between NHHR and the severity of steatosis and fibrosis, reinforcing the concept that dysregulated lipoprotein balance parallels progressive hepatic injury [24, 30]. Kim et al. [20] further showed that NHHR outperformed the apolipoprotein B/apolipoprotein A1 ratio in identifying metabolic syndrome and insulin resistance. Population-based analyses from NHANES have also demonstrated significant associations between higher NHHR and the presence of MASLD and hepatic steatosis [21, 22]. Extending these observations to prognostic outcomes, Zhang et al. [23] identified elevated NHHR as a predictor of mortality in patients with MASLD, particularly cardiovascular mortality in those without advanced fibrosis, indicating prognostic implications beyond liver biochemistry alone. Shi et al. [13] demonstrated that low HDL-C independently predicted 90-day transplant-free mortality in patients with hepatitis B virus (HBV)-related acute-on-chronic liver failure, underscoring the critical role of HDL-related pathways in immune dysfunction and hepatic failure. Likewise, Mo et al. [14] developed a prognostic model based on serum apolipoprotein A-I that accurately stratified mortality risk in patients with HBV-related acute-on-chronic liver failure and acutely decompensated cirrhosis, highlighting the loss of protective lipoproteins as a marker of hepatic reserve and systemic inflammation. Although these studies primarily focused on metabolic and virus-related liver disease, their findings align closely with our observations in a broader CLD population. This convergence suggests that composite lipid measures such as NHHR may reflect a shared pathophysiologic axis linking impaired hepatic synthetic function, systemic inflammation, and metabolic stress across diverse etiologies and stages of liver disease. The positive associations between higher NHHR and adverse liver outcomes likely reflect multiple interrelated mechanisms. Specifically, an imbalance between non-HDL and HDL cholesterol may signify an excess of remnant and atherogenic lipoproteins relative to protective HDL particles, fostering systemic inflammation and endothelial dysfunction that can contribute to progressive hepatic injury and the development of portal hypertension. At the same time, a relative deficiency of functional HDL may compromise reverse cholesterol transport. Experimental and clinical studies have shown that CLD can alter HDL composition and reduce cholesterol efflux capacity, indicating impaired HDL functionality rather than merely decreased HDL concentrations [10, 17]. This functional impairment may diminish HDL-mediated reverse cholesterol transport and weaken its antioxidant and anti-inflammatory effects, thereby contributing to hepatocellular injury and systemic inflammation [31]. Although the precise mechanisms linking dysregulated lipoprotein balance to clinical decompensation remain incompletely understood, these alterations likely represent the combined effects of impaired hepatic synthesis and systemic metabolic stress inherent to advanced liver disease. Collectively, these data support the concept that NHHR integrates pro-inflammatory and protective lipoprotein signals in a manner that reflects hepatic reserve and the overall systemic metabolic burden across the spectrum of CLD.

We also demonstrated that incorporation of NHHR into established prognostic models resulted in a significant improvement in risk discrimination and patient reclassification, as reflected by higher AUC values and favorable NRI and IDI indices compared with a base clinical model comprising age, sex, and MELD-3.0 score. Notably, the gain in predictive performance was most pronounced within the highest NHHR tertile, suggesting that its added value is greatest among patients with substantial lipoprotein imbalance and more advanced metabolic derangement. From a clinical perspective, these results underscore the potential utility of NHHR as a pragmatic risk stratification tool beyond established clinical severity scores. Because NHHR can be readily derived from routinely available lipid panels, it represents a low-cost and accessible biomarker that may help identify patients who warrant closer surveillance and timely clinical intervention. This simplicity and broad availability may be particularly advantageous in resource-limited settings, where access to advanced imaging modalities or repeated comprehensive scoring assessments is often constrained.

This study has certain limitations that merit consideration. First, the cross-sectional design precludes causal inference. Consequently, although higher NHHR was associated with the presence of decompensation, it remains unclear whether elevated NHHR contributes to the development of decompensation or whether advanced liver dysfunction and metabolic alterations accompanying decompensation lead to changes in lipid profiles and consequently a higher NHHR. This possibility of reverse causality should be considered when interpreting the findings, and prospective longitudinal studies are required to clarify the temporal relationship between NHHR and hepatic decompensation. Second, this was a single-center study conducted at a tertiary-care hospital, which may limit the generalizability of the findings to other populations or healthcare settings. Although the cohort included patients with diverse etiologies of CLD, including alcohol-related liver disease, MASLD, viral hepatitis, and other causes, the study population may be more representative of patients with moderate to advanced disease typically seen in tertiary centers. Therefore, multicenter studies involving more diverse and community-based populations are needed to confirm the broader applicability of these findings. Third, detailed lipid characterization was not available. Specifically, data on apolipoproteins, LDL and HDL subclasses, and direct measures of HDL functionality were not assessed. These parameters may provide additional insights into lipid metabolism and inflammatory processes in CLD and could further clarify the biological mechanisms underlying the observed association between NHHR and decompensation. In addition, although key clinical confounders were adjusted for in multivariable analyses, residual confounding from unmeasured factors cannot be entirely excluded. NHHR should therefore be interpreted in the appropriate clinical context, as lipid levels may be influenced by factors such as acute infection, systemic inflammation, nutritional status, and metabolic comorbidities. Nevertheless, the relatively large sample size, systematic assessment of liver disease severity, and integration of discrimination and reclassification analyses strengthen the validity of the findings.

In conclusion, this study provides robust evidence that an elevated NHHR is independently associated with the presence of clinical decompensation in patients with CLD. The consistent correlations observed between NHHR and established liver function markers, including albumin, INR, total bilirubin, and MELD-3.0, reinforce the relevance of systemic metabolic imbalance in advanced hepatic dysfunction. The incremental discriminatory value conferred by NHHR beyond traditional clinical and biochemical parameters supports its consideration as a simple and accessible biomarker for risk stratification and targeted clinical evaluation in CLD. Integration of NHHR into routine clinical assessment may help identify patients with decompensation in CLD and support more comprehensive clinical evaluation and management.