Liver Stiffness, Not Steatosis, Predicts Mortality in MASLD Patients: An NHANES Analysis

Abstract

Background: Metabolic dysfunction-associated steatotic liver disease (MASLD) has surged as a major cause of liver transplants in the United States. Existing studies have presented conflicting findings regarding the association between liver characteristics (specifically steatosis and fibrosis) and mortality. This study investigates the relationship between the controlled attenuation parameter (CAP) and liver stiffness measurement (LSM) via vibration-controlled transient elastography (VCTE) and all-cause mortality in MASLD patients. Methods: Using the NHANES 2017-2018 database, 3821 individuals representing the United States population with MASLD underwent VCTE for liver stiffness measurement. Exclusion criteria were applied, eliminating ineligible cases, incomplete examinations, underage individuals, and those with hepatitis B or C, along with significant alcohol consumption history. Cox proportional hazard models assessed the hazard ratio (HR) for all-cause mortality in CAP and LSM. Cox regression analysis with interaction terms was employed for deeper exploration. Results: The study unveiled a strong, independent correlation between LSM and all-cause mortality. However, the CAP failed to demonstrate a significant association with mortality in both univariate and adjusted analyses, contrary to recent findings. The analysis underscores the importance of accurately measuring liver stiffness via VCTE in predicting adverse outcomes in MASLD patients, emphasizing the pivotal role of fibrosis in assessing mortality risk. Conclusion: This study reaffirms the robust link between liver fibrosis (measured through VCTE) and mortality among MASLD individuals. The absence of a significant association between steatosis (indicated by CAP) and mortality challenges recent research, urging further comprehensive investigations with larger cohorts to delineate steatosis’ precise impact on MASLD-related mortality.

1. Introduction

Metabolic dysfunction-associated steatotic liver disease (MASLD) is the most rapidly increasing indication for liver transplantation in the United States [1]. MASLD is a complex, heterogeneous condition characterized by a wide spectrum of clinical manifestations. The histopathological progression of the disease spans from isolated hepatic steatosis, referred to as Metabolic Dysfunction-Associated Steatotic Liver (MASL), to Metabolic Dysfunction-Associated Steatohepatitis (MASH), a more advanced stage involving inflammation and liver injury. Over time, MASLD can further progress, with the accumulation of hepatic fibrosis, potentially advancing to cirrhosis and increasing the risk of developing hepatocellular carcinoma (HCC). This continuum of disease stages underscores the diverse nature of MASLD and the challenge of predicting its progression in individual patients [2]. The risk factors for developing MASLD include obesity, type 2 diabetes, hypertension, and dyslipidemia, such as high triglyceride and low HDL-C levels [3]. Pathogenetically, insulin resistance is believed to be the key trigger for the development of MASLD, which leads to unregulated hepatic triglyceride synthesis, peripheral lipolysis, and free fatty acid uptake into the liver hepatic oxidative stress, inflammation—all of which promote hepatic fat deposition [4,5].

Nowadays, more than 25% global population is living with MASLD, the prevalence increasing from 22% to 37% from 1991 to 2019 [4,6,7]. Although only 3–5% of patients with MASLD will eventually progress to cirrhosis [8], the mortality rate is significantly increased in MASLD patients with a hazard ratio of 1.32 [9]. It has been reported that the absolute excess risk of overall mortality is 10.7% with simple steatosis, 25.6% with non-cirrhotic fibrosis, and 49.4% with cirrhosis, when compared to control populations [10]. Despite the high mortality associated with MASLD, the specific pathological changes—whether steatosis or fibrosis—that are most strongly correlated with poor prognosis remain unclear. This uncertainty complicates efforts to accurately predict disease outcomes.

Vibration-controlled transient elastography (VCTE) is a noninvasive ultrasound-based device gaining popularity in the diagnosis of MASLD, for its ability to provide important information in determining steatosis and fibrosis in the liver [11,12]. It is the most validated and commonly used elastography method used worldwide [13,14,15]. This method measures the velocity of a low-amplitude shear wave as it propagates through the liver in a prespecified area of interest and converts the wave velocity into liver stiffness [16]. A recent study showed that both the controlled attenuation parameter (CAP), indicating steatosis, and liver stiffness measurement (LSM), indicating fibrosis, independently correlated with overall mortality over a three-year follow-up [17]. But there have been controversial conclusions made based on predictive results of the CAP for estimating mortality rates [18].

In this study, through analyzing the NHANES 2017–2018 database, we investigated the correlation between all-cause mortality in MASLD patients and the controlled attenuation parameter (CAP), liver stiffness measurement (LSM) via vibration-controlled transient elastography (VCTE). The result unveiled a strong, independent correlation between LSM and all-cause mortality. However, the CAP failed to demonstrate a significant association with mortality in both univariate and adjusted analyses.

2. Methods

2.1. Study Design and Population

This study is a retrospective analysis utilizing data from the National Health and Nutrition Examination Survey (NHANES) 2017–2018 cycle. NHANES is a cross-sectional survey designed to assess the health and nutritional status of adults and children in the United States through interviews, physical examinations, and laboratory tests. The 2017–2018 cycle included VCTE measurements, providing data on liver stiffness and steatosis in a nationally representative sample.

Participants aged 18 years and older who underwent VCTE examinations were considered for inclusion in this analysis. From the initial pool of participants, specific exclusion criteria were applied to focus on individuals with MASLD and to eliminate other causes of liver disease.

2.2. Inclusion and Exclusion Criteria

Participants were excluded if they were ineligible for VCTE based on NHANES protocols, which include (1) being unable to lie down on the exam table; (2) being pregnant (or unsure if pregnant) at the time of the exam, or if a urine could not be obtained to test for pregnancy; (3) having an implanted electronic medical device; and (4) currently wearing a bandage or having lesions to the right side of their abdomen. Individuals with incomplete or invalid VCTE examinations were also excluded; a valid VCTE examination required at least 10 successful measurements with an interquartile range to median ratio (IQR/M) of ≤30% for LSM.

Individuals under the age of 18 were excluded to limit the study to the adult population. Participants with evidence of hepatitis B virus (HBV) infection, indicated by positive hepatitis B surface antigen (HBsAg), and those with active hepatitis C virus (HCV) infection, indicated by positive HCV RNA, were excluded to remove other causes of chronic liver disease.

To isolate MASLD cases, participants with significant alcohol consumption were excluded. Significant alcohol consumption was defined as daily alcohol intake exceeding two drinks for women and three drinks for men in the year preceding study enrollment. Alcohol consumption data were obtained from the NHANES Alcohol Use Questionnaire, which assessed the average daily alcohol intake over the past 12 months.

After applying these exclusion criteria, the final study population comprised 3821 individuals representative of the U.S. adult population with MASLD.

2.3. Vibration-Controlled Transient Elastography Measurements

VCTE examinations were conducted using the FibroScan^® 502 V2 Touch device by trained technicians following standardized procedures. Participants were instructed to fast for at least three hours prior to the examination. The appropriate probe (M or XL) was selected based on the participant’s body mass index (BMI) and thoracic perimeter, as recommended by the manufacturer. LSM was obtained to assess hepatic fibrosis, reported in kilopascals (kPa). CAP was measured simultaneously to assess hepatic steatosis, reported in decibels per meter (dB/m). Hepatic steatosis was defined as a CAP value greater than 285 dB/m.

2.4. Outcome Assessment

The primary outcome of interest was all-cause mortality. Mortality data were obtained by linking NHANES participants to the National Death Index (NDI) records, with follow-up data available through 31 December 2019. The time to event was calculated from the date of the VCTE examination to the date of death or end of follow-up. Participants were followed from the date of their VCTE examination until the date of death or the end of the follow-up period, whichever occurred first. Individuals who were alive at the end of the follow-up period were censored at that date. Survival time was calculated accordingly, and all analyses accounted for the censoring of participants who did not experience the event by the end of the study period.

2.5. Statistical Analysis

Time-to-event analyses were conducted using Cox proportional hazards regression models to evaluate the association between CAP, LSM, and all-cause mortality. Sampling weights were applied in all analyses to account for the complex sampling design of NHANES and to produce nationally representative estimates.

CAP and LSM were treated as continuous variables. An interaction term between CAP and LSM was created by multiplying the CAP and LSM variables to explore potential effect modification between hepatic steatosis and fibrosis. Univariate Cox models were first performed to assess the individual associations of CAP and LSM with all-cause mortality. Subsequent models incorporated the interaction term to evaluate whether the relationship between CAP or LSM and mortality was influenced by their interaction. Additional multivariate Cox regression analyses was applied by selecting potential confounders, including age, gender, HbA1c, body mass index, congestive heart failure, coronary artery disease, history of acute myocardial infarction, hypertension, and smoking status, using a stepwise selection method with entry criteria of p  <  0.20 and retention criteria of p  <  0.05 for the final model.

To visualize the combined effect of CAP and LSM on mortality risk, a three-dimensional contour plot was generated based on a Cox regression model without the interaction term. Predicted hazard ratios were calculated over a range of CAP values (220 to 400 dB/m, in increments of 20 dB/m) and LSM values (2 to 30 kPa, in increments of 2 kPa) as the X- and Y-axis with supplementing overlay of p values as the Z-axis.

All statistical analyses were performed using the Stata software package (version 17.0; StataCorp LLC, College Station, TX, USA) and Python 3.10. A two-sided p-value of less than 0.05 was considered statistically significant.

2.6. Ethical Considerations

The NHANES protocol was approved by the National Center for Health Statistics Research Ethics Review Board, and all participants provided written informed consent. This study utilized publicly available de-identified data and was exempt from additional institutional review board approval.

2.7. Data Availability

The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request at yang.liu@mayo.edu.

3. Results

We excluded 2853 patients ineligible for VCTE, 907 with incomplete examinations, 867 individuals who are less than 18 year old, 27 with positive HbsAg, 43 with positive HCV RNA, and 736 individuals with alcohol use history (Supplementary Materials Table S1). Consequently, our participant cohort comprised 3821 individuals (Figure 1).

Patient characteristics can be found in

Table 1. Baseline characteristics of patients stratified by survival status.

	Alive (N = 3725) †	Decreased (N = 71) †
Age, median (IQR), years	54 (37–66)	72 (62–80)
Female, n (%)	1947 (52.3)	24 (33.8)
History of AMI, n (%)	147 (4.0)	14 (19.7)
CHF, n (%)	90 (2.5)	12 (17.1)
CAD, n (%)	153 (4.1)	15 (21.1)
Hypertension, n (%)	1419 (38.1)	46 (64.8)
Smoking, n (%) *	1393 (37.4)	47 (66.2)
Hemoglobin A1c, median (IQR), %	5.6 (5.3–6.0)	6.0 (5.6–6.8)
CAP, median (IQR), 10 dB/m	26.3 (21.9–30.7)	26.2 (21.4–32.1)
LSM, median (IQR), kPa	4.9 (4.0–6.1)	5.1 (4.4–6.9)

Abbreviations: AMI, acute myocardial infarction; CHF, congestive heart failure; CAD, coronary artery disease; CAP, controlled attenuation parameter; LSM, liver stiffness measurement. * Defined by smoked more than 100 cigarettes in lifetime. ^† Unweighted numbers.

.

The Univariate Cox proportional hazard model (Figure 2) showed a nonsignificant hazard ratio (HR) for all-cause death in CAP (every 10 dB/m) of 1.03 (95% confidence interval, CI 0.98–1.08, p = 0.264). The overall results did not change when adjusted for the interaction term with LSM with an HR of 1.01 (95% CI 0.96–1.06, p = 0.662). For LSM (every 1 kPa), a univariate analysis showed an HR of 1.04 (95% CI 1.02–1.06, p < 0.001), which was adjusted to 1.01 (95% CI 0.93–1.09, p = 0.856) after accounting for the interaction term. In multivariate Cox regression models including age, gender, HbA1c, and history of acute myocardial infarction as covariates (

Table 2. Adjusted hazard ratios (aHRs) from multivariate cox regression.

	aHR	95% CI	p
Controlled Attenuation Parameter (CAP)
CAP	0.99	0.94–1.05	0.730
Age (per year increase)	1.06	1.02–1.11	0.002
Female	0.45	0.21–0.96	0.039
Hemoglobin A1c (per one percent increase)	1.31	1.08–1.60	0.007
History of AMI	0.42	0.18–1.00	0.050
Liver Stiffness Measurement (LSM)
LSM	1.03	1.00–1.05	0.043
Age (per year increase)	1.06	1.02–1.11	0.002
Female	0.46	0.21–1.02	0.057
Hemoglobin A1c (per one percent increase)	1.30	1.06–1.58	0.010
History of AMI	0.45	0.19–1.04	0.062

Abbreviations: AMI, acute myocardial infarction.

), LSM remained significantly associated with all-cause mortality (adjusted hazard ratio [aHR] 1.03 per 1 kPa increase, 95% CI 1.00–1.05, p  =  0.043), while the CAP was not associated with mortality (aHR 0.99 per 10 dB/m increase, 95% CI 0.94–1.05, p  =  0.726).

Figure 3 shows Cox regression of CAP and LSM, supplementing an overlay of p values as a Z-axis. The p-value and CI remained statistically significant with an increase in LSM, though not CAP.

4. Discussion

In concurrence with previous research [19,20], we established a strong and independent correlation between LSM and mortality. This link aligns with established medical knowledge that emphasizes the role of fibrosis as a robust predictor of adverse outcomes in individuals with liver diseases. The results of our univariate analysis, as well as the Cox regression analysis incorporating LSM, provided substantial evidence for the prognostic value of fibrosis in relation to all-cause mortality. These findings highlight the clinical significance of measuring liver stiffness using VCTE in assessing the risk of mortality among MASLD patients.

Our study did not reveal a significant association between CAP and mortality. This finding is noteworthy and contrasts with some recent studies that indicated a potential predictive relationship between steatosis and mortality [17]. Vilar-Gómez et al. reported that both the CAP and LSM were independent predictors of all-cause mortality in the U.S. adult population [17]. Notably, the reported hazard ratio for CAP was 1.01 (95% CI: 1.00–1.05), indicating only a marginal association. The discrepancy between our results and theirs may stem from subtle methodological differences, such as variations in sample selection criteria or statistical analysis methods. The near-borderline significance reported by Vilar-Gómez et al. suggests that the association between the CAP and mortality is weak, and small changes in the study design or analysis could render it non-significant. This is further evidenced by the observation that while the distribution of the strength of association (hazard ratio) between CAP/LSM levels and mortality is similar in both studies, we observe stronger statistical significance (lower p values) for LSM, but not so much for the CAP [4]. Our multivariate analysis further supports this distinction by demonstrating that the CAP was not significantly associated with mortality after adjusting for confounders, whereas LSM remained robustly predictive. Therefore, we agree with Vilar-Gómez et al. on the association between LSM and mortality, meanwhile underscoring the need for further large-scale studies to clarify the relationship between hepatic steatosis measured by the CAP and mortality.

There are several limitations to our study. First, there is a potential selection bias introduced by excluding participants who were ineligible for VCTE or had incomplete examinations. This exclusion may disproportionately affect individuals with higher BMI or obesity-related comorbidities who are at increased risk of mortality. As a result, the generalizability of our findings regarding the CAP and all-cause mortality may be impacted. Second, the relatively short follow-up period (median follow-up of approximately 1–2 years) may not be sufficient to observe long-term mortality outcomes associated with hepatic steatosis. This limited duration could contribute to the lack of a significant association between CAP and all-cause mortality. Third, our data lacked specific information on liver-related mortality, as the NHANES dataset only provides broad categories for causes of death. This limitation prevented a more detailed analysis of the relationship between the CAP, LSM, and liver-specific outcomes.

In conclusion, our study reaffirms the robust link between liver fibrosis, as measured by LSM via VCTE, and all-cause mortality among individuals with MASLD. However, we did not find a significant association between hepatic steatosis, indicated by CAP, and mortality. This lack of association may be attributed to the relatively short follow-up period, which limits the observation of long-term outcomes associated with steatosis. Additionally, selection bias resulting from the exclusion of participants unable to undergo VCTE or with incomplete examinations may have impacted our findings. It is also possible that hepatic steatosis alone may have a less immediate impact on mortality compared to fibrosis. Further comprehensive studies with larger cohorts and longer follow-up periods are needed to delineate the precise impact of steatosis on mortality in MASLD patients.