Abstract
Background & Aims:
MASLD is defined by abnormalities in cardiometabolic risk factors (CMRFs). Characterizing the contribution of individual CMRFs to clinical outcomes may guide prioritization of interventions. We evaluated the association of individual CMRFs with all-cause mortality in US adults with metabolic dysfunction-associated steatotic liver disease (MASLD).
Methods:
Participants in the National Health and Nutrition Examination Survey (NHANES) III (1988-1994) and continuous NHANES (1999-2018) were linked to mortality data through 2019. Adults ≥ 20 years of age were included if Fatty Liver Index (FLI) > 60 and at least one CMRF (overweight/obesity, glucose intolerance, high blood pressure, triglycerides, or low high-density lipoprotein [HDL]). Cox regression analyzed risk of all-cause mortality adjusted for age, sex, and FIB-4.
Results
Among 21,872 participants included (mean age 50 years, 53% male), mean BMI was 33.6 kg/m2 with a median of 3 CMRF (99.5% overweight/obese, 55% glucose intolerance, 58% high blood pressure, 67% triglycerides, 40% low HDL). In adjusted analysis of individual CMRF, high blood pressure (aHR 1.39, 95%CI 1.24-1.55, p<0.001), glucose intolerance (aHR 1.26, 95%CI 1.16-1.38, p<0.01), and low HDL (aHR 1.15, 95%CI 1.05-1.26, p=0.003) exerted significant risks for mortality. When stratifying the overweight/obesity CMRF by discrete BMI ranges, a significantly greater risk for mortality was seen with BMI ≥ 35 – < 40 kg/m2 (aHR 1.18, 95%CI 1.02-1.36, p=0.03), BMI ≥ 40 – < 45 kg/m2 (aHR 1.55, 95%CI 1.24-1.94, p<0.001), BMI ≥ 45 kg/m2 (aHR 1.64, 95%CI 1.27-2.14, p<0.001) compared to those with a BMI ≥ 25 – < 30 kg/m2. In age-adjusted analysis, number of CMRFs was associated with greater risk for mortality (aHR 1.15 per additional CMRF, 95%CI 1.10-1.20, p<0.001)
Conclusions:
The differential risk for mortality between individual CMRFs supports distinct clinical profiles in MASLD. This study found that high blood pressure and glucose intolerance exerted the greatest risk for all-cause mortality in those with MASLD, suggesting a role for prioritization of CMRF optimization.
Keywords: fatty liver, metabolic syndrome, non-alcoholic fatty liver disease, nutrition surveys, cirrhosis
Brief Summary:
In this study utilizing data from the NHANES survey between 1988 and 2018, each of the five cardiometabolic risk factors comprising the consensus definition of MASLD carried differential risks for all-cause mortality, supporting distinct disease phenotypes in adults with MASLD.
Introduction
Metabolic dysfunction-associated steatotic liver disease (MASLD) is defined by hepatic steatosis and at least one cardiometabolic risk factor (CMRF) abnormality, emphasizing the shared pathophysiology of metabolic dysfunction and insulin resistance between MASLD and the conditions defined by these CMRFs. These CMRFs include elevated body mass index (BMI) and waist circumference (WC), glucose intolerance, hypertension (HTN), elevated triglycerides, and low high-density lipoprotein (HDL).1 Of these CMRFs, obesity and diabetes are clearly connected with negative outcomes in MASLD.2–8 MASLD is a growing concern, with recent estimates suggesting a 38% global prevalence, and this burden of disease supports a need to characterize those at risk for poor outcomes.9 In MASLD, this is complicated by a heterogenous presentation and natural history, and defining distinct clinical phenotypes or profiles may aid in the identification and prioritization of treatment strategies. It is likely that both number of CMRFs and individual CMRFs have differential contributions to liver-related outcomes, such as development of advanced fibrosis and cirrhosis, and non-liver-related outcomes, such as all-cause mortality. Recent work has highlighted that these cumulative CMRFs increase the risk for advanced hepatic fibrosis and all-cause mortality, in both US-based and non-US-based populations, as well as find that advanced fibrosis prevalence varies within subgroup combinations of CMRFs.10–12 While these studies focus on cumulative burden of CMRFs, a gap in knowledge exists for understanding the relative risks between individual CMRFs. In this study, we utilized a large, nationally representative, cross-sectional survey of the United States population to characterize the differential risk profiles for all-cause mortality in adults with MASLD by the number and presence of individual CMRFs.
Methods
Study Population
Adult participants (age ≥ 20 years) from the National Health and Nutrition Examination Survey (NHANES) III (1988 – 1994) and continuous NHANES (1999 – 2018) who met the definition of MASLD, defined by at least one CMRF abnormality and the presence of hepatic steatosis defined by a Fatty Liver Index (FLI) > 60, were included in our analysis.1,13,14 CMRFs were defined according to the consensus definition of MASLD (Supplemental Text)1, using anthropometric measurements, laboratory values, or medication history as appropriate. The Fatty Liver Index (FLI) is a validated algorithm for detection of hepatic steatosis in population samples, with a score ≥ 60 predictive of hepatic steatosis, and is calculated using BMI, WC, triglycerides, and gamma-glutamyl transferase (GGT).13–15 Participants were excluded if they were pregnant, had a history of human immunodeficiency virus (HIV) or chronic viral hepatitis (hepatitis C virus [HCV] antibody or HCV RNA positive or hepatitis B virus [HBV] surface antigen positive), had a history of heart failure, had a history of heavy alcohol use (an average of ≥ 4 drinks per day for males or ≥ 3 drinks per day for females). Imaging-based assessment of hepatic steatosis was available for NHANES III population with abdominal ultrasound and NHANES 2017-2018 participants with vibration-controlled transient elastography (VCTE).
Covariate assessment
Race and ethnicity were categorized as non-Hispanic white, non-Hispanic Black, Hispanic, and other race/ethnicity, with Asian race not reported in NHANES III. Participant-reported Mexican American race/ethnicity was pooled with those reporting Hispanic race/ethnicity. Anthropometric measurements, laboratory testing, and self-reported medical and medication history were obtained at time of study examination. Blood pressure readings were measured by study personnel at time of examination and the average of all valid measurements was utilized if multiple readings were obtained for a participant. Participants were considered as having a history of tobacco smoking if they reported smoking at least 100 cigarettes in their lifetime. The five individual CMRFs were dichotomized as present or absent. CMRF-overweight/obesity was categorized into discrete ranges of BMI: 1) BMI ≥ 25 - < 30 kg/m2, 2) BMI ≥ 30 - < 35 kg/m2, 3) BMI ≥ 35 - < 40 kg/m2, 4) BMI ≥ 40 - < 45 kg/m2, 5) BMI ≥ 45 kg/m2, 6) BMI < 25 kg/m2 with elevated WC. The overweight BMI ≥ 25 – < 30 kg/m2 range was used as the reference category as it is hypothesized to carry the lowest risk within the range of those categorized with this CMRF. The CMRF-overweight/obesity variable was stratified to better explore differences within this CMRF given its high prevalence, as well as to explore the non-linear association between BMI and mortality. A CMRF-hierarchy of cardiometabolic risk variables was created by type and number of CMRFs where only participants meeting abnormal CMRF-overweight/obesity were categorized into mutually exclusive risk factor groups: CMRF-overweight/obesity alone, CMRF-overweight + a specific second CMRF (if only met two CMRFs), or to the number of CMRFs in addition to overweight/obesity (CMRF-overweight/obesity + two CMRFs, for example) for participants with three or more total CMRFs.
Outcomes
Participants were matched to mortality status through December 31, 2019 using the NHANES Public Use Linked Mortality File. Participant follow-up time was calculated from date of NHANES exam until death or end of data availability. Participants who were not matched to a mortality event were considered alive and censored at end of follow-up. All-cause mortality was the main outcome of interest. Causes of death are reported using re-coded categories in the Public Use Linked Mortality File. The secondary outcome of interest was the presence of advanced fibrosis, defined by a Fibrosis-4 Index (FIB-4) > 2.67, a cutoff validated to predict “at risk” fibrosis (≥F2 stage). 16 The FIB-4 was calculated using participant age, platelet count, AST, and ALT at time of examination.16
Statistical Analysis
All statistical analyses were performed in accordance with NHANES guidelines for survey analysis weighting to account for differential sampling probabilities and nonresponse. This survey weighting can be utilized to generate US population representative estimates. Sample weighting from each survey cycle was used to construct a combined analytical weight according to analytic guidance, where the corresponding cycle weight was multiplied by the proportion of total analyzed years comprised by that survey cycle.17–19 Cox proportional hazards regression was used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for the risk of all cause-mortality. All Cox models were adjusted for age (entry time and exit times utilized age at examination and age at death or censoring). The full, multivariable all-cause mortality analysis was also adjusted for sex and FIB-4, to address confounding by the presence of hepatic fibrosis and focus our analysis on the effects of individual CMRFs.
Collinearity was detected between the individual CMRFs and number of CMRFs (variance inflation factor (VIF) > 10), and therefore, were not included in the multivariable model simultaneously (VIF < 2 for all reported models).
The Wald test was used to compare HR estimates from individual CMRFs.
CMRFs were evaluated in separate Cox regression models:
Using number of CMRFs or individual CMRF alone or covariates of interest, adjusted for age. This model serves as a reduced, bivariable model.
Using all CMRFs in a single model, except the number of CMRFs variable, adjusted for age, sex, race/ethnicity, smoking history and FIB-4. This model is thus adjusted for the presence of other CMRFs and serves as a full, multivariable model.
Using CMRF-hierarchy including only those participants who met CMRF-overweight/obesity, designed to assess the type-specific CMRF effects of having a second CMRF in addition to CMRF-overweight/obesity (given the high prevalence of this CMRF), adjusted for age, sex, and FIB-4.
Analysis of the secondary outcome of advanced fibrosis (FIB-4 > 2.67) was performed using logistic regression adjusted for sex. All analyses were conducted in Stata BE 18. This research was conducted in accordance with the principles outlined in both the Declaration of Helsinki and the Declaration of Istanbul. This study was deemed exempt from review by the Institutional Review Board of the University of Southern California.
Results
Of 134,515 surveyed individuals within the included NHANES program years, a total of 21,872 met the criteria for MASLD and inclusion in this study, corresponding to a survey-weighted prevalence in the US adult population of 37% (95%CI 36.0%-38.0%) (Supplemental Figure 1). Characteristics of these participants are listed in Table 1, categorized by number of CMRFs, with an overall mean age of 51 years, 54% male, and 49% non-Hispanic white with a mean BMI of 33.6 kg/m2. CMRF-overweight/obesity was found in 99.3% of the participants, and CMRF-HDL was the least prevalent, found in 41.6% of the participants. Only one CMRF was found in 7.7% of the participants, while 21.8% met all five CMRFs. CMRF-overweight/obesity was present in nearly all (98.2%) of those with only one CMRF, while zero participants had CMRF-glucose intolerance as their only CMRF. In the subset of participants with valid liver imaging (NHANES III population with abdominal ultrasound or NHANES 2017-2018 population with valid vibration controlled transient elastography [VCTE], n = 4,441) and included in this study with a FLI ≥ 60, 84% met a definition of hepatic steatosis by imaging. Of the 16% (n = 724) that did not meet the definition of hepatic steatosis by imaging, all were from the VCTE sample and had a mean controlled attenuation parameter (CAP) of 246 dB/m.
Table 1.
Characteristics of Participants with MASLD
| Variable | Total | 1 CMRF | 2 CMRFs | 3 CMRFs | 4 CMRFs | 5 CMRFs |
|---|---|---|---|---|---|---|
| Number | 21,872 | 1,691 (7.7%) | 4,544 (20.8%) | 5,899 (27.0%) | 4,971 (22.7%) | 4,767 (21.8%) |
| Age (years) | 51.0 (16.5) | 36.2 (12.5) | 42.4 (14.9) | 49.1 (15.4) | 55.4 (15.0) | 62.4 (12.3) |
| Male | 11,695 (53.5%) | 744 (44.0%) | 2,405 (52.9%) | 3,282 (55.6%) | 2,765 (55.6%) | 2,499 (52.4%) |
| Race/ethnicity | ||||||
| Mexican American | 4,077 (18.6%) | 339 (20.0%) | 986 (21.7%) | 1,138 (19.3%) | 871 (17.5%) | 743 (15.6%) |
| Non-Hispanic White | 10,699 (48.9%) | 705 (41.7%) | 1,996 (43.9%) | 2,790 (47.3%) | 2,709 (54.5%) | 2,499 (52.4%) |
| Non-Hispanic Black | 4,384 (20.0%) | 438 (25.9%) | 948 (20.9%) | 1,236 (21.0%) | 802 (16.1%) | 960 (20.1%) |
| Other | 2,712 (12.4%) | 209 (12.4%) | 614 (13.5%) | 735 (12.5%) | 589 (11.8%) | 565 (11.9%) |
| Smoking history | 10,494 (48.0%) | 676 (40.0%) | 1,909 (42.0%) | 2,771 (47.0%) | 2,583 (52.0%) | 2,555 (53.6%) |
| BMI (kg/m2) | 33.6 (6.0) | 34.7 (5.7) | 33.8 (6.0) | 33.6 (6.2) | 32.9 (5.8) | 33.6 (5.9) |
| BMI categories | ||||||
| BMI < 25 kg/m2 | 370 (1.7%) | 37 (2.2%) | 75 (1.7%) | 100 (1.7%) | 100 (2.0%) | 58 (1.2%) |
| BMI ≥ 25 – < 30 kg/m2 | 6,102 (27.9%) | 252 (14.9%) | 1,172 (25.8%) | 1,700 (28.8%) | 1,656 (33.3%) | 1,322 (27.7%) |
| BMI ≥ 30 – < 35 kg/m2 | 8,304 (38.0%) | 708 (41.9%) | 1,787 (39.3%) | 2,192 (37.2%) | 1,835 (36.9%) | 1,782 (37.4%) |
| BMI ≥ 35 – < 40 kg/m2 | 4,233 (19.4%) | 441 (26.1%) | 898 (19.8%) | 1,110 (18.8%) | 809 (16.3%) | 975 (20.5%) |
| BMI ≥ 40 – < 45 kg/m2 | 1,757 (8.0%) | 160 (9.5%) | 375 (8.3%) | 479 (8.1%) | 346 (7.0%) | 397 (8.3%) |
| BMI ≥ 45 kg/m | 1,106 (5.1%) | 93 (5.5%) | 237 (5.2%) | 318 (5.4%) | 225 (4.5%) | 233 (4.9%) |
| WC (cm) | 110.5 (12.5) | 110.2 (11.3) | 109.5 (12.3) | 110.0 (12.8) | 109.8 (12.3) | 112.9 (12.6) |
| A1c (%) | 6.0 (1.3) | 5.2 (0.3) | 5.4 (0.7) | 5.8 (1.1) | 6.1 (1.4) | 6.8 (1.6) |
| Fasting glucose (mg/dL) | 115.1 (44.4) | 91.7 (5.7) | 99.1 (21.0) | 110.6 (36.6) | 117.5 (48.9) | 137.2 (56.2) |
| Systolic BP (mmHg) | 128.3 (18.3) | 114.1 (8.4) | 121.2 (15.3) | 128.7 (17.6) | 132.6 (18.4) | 134.7 (19.6) |
| Diastolic BP (mmHg) | 74.0 (12.8) | 69.6 (9.1) | 72.8 (11.0) | 76.0 (13.3) | 75.9 (13.3) | 72.3 (14.3) |
| Triglycerides (mg/dL) | 203.8 (160.7) | 106.2 (44.5) | 168.1 (121.5) | 202.9 (146.6) | 249.2 (201.1) | 226.2 (166.2) |
| HDL (mg/dL) | 44.8 (12.6) | 54.8 (11.1) | 51.1 (13.1) | 46.2 (12.6) | 41.7 (11.6) | 39.5 (9.5) |
| LDL (mg/dL) | 122.6 (37.4) | 120.9 (38.2) | 123.1 (31.8) | 126.9 (35.3) | 128.9 (38.6) | 112.5 (40.7) |
| ALT (U/L) | 28.2 (19.9) | 27.7 (19.8) | 28.8 (20.2) | 29.8 (22.2) | 27.8 (18.7) | 26.2 (17.4) |
| AST (U/L) | 25.9 (14.2) | 25.1 (15.7) | 25.7 (14.9) | 26.4 (14.7) | 25.9 (13.1) | 25.9 (14.2) |
| Platelets (109/L) | 261.4 (70.0) | 270.1 (66.7) | 266.5 (67.8) | 263.9 (70.4) | 260.7 (70.4) | 251.0 (71.0) |
| Albumin (g/dL) | 4.2 (0.3) | 4.2 (0.3) | 4.2 (0.3) | 4.2 (0.3) | 4.2 (0.3) | 4.1 (0.3) |
| FLI | 83.6 (11.9) | 78.9 (11.9) | 81.4 (12.0) | 83.8 (11.7) | 84.6 (11.7) | 85.9 (11.3) |
| FIB-4 | 1.10 (0.86) | 0.72 (0.69) | 0.87 (0.66) | 1.04 (0.72) | 1.19 (0.68) | 1.42 (1.22) |
| FIB-4 > 2.67 (Advanced Fibrosis) | 607 (2.8%) | 11 (0.7%) | 71 (1.6%) | 141 (2.4%) | 148 (3.0%) | 236 (5.0%) |
| CMRF – overweight/obesity | 21,727 (99.3%) | 1,660 (98.2%) | 4,500 (99.0%) | 5,852 (99.2%) | 4,948 (99.5%) | 4,767 (100%) |
| CMRF – glucose intolerance | 12,884 (58.9%) | 0 (0%) | 1,191 (26.2%) | 3,474 (58.9%) | 3,452 (69.4%) | 4,767 (100%) |
| CMRF – HTN | 13,819 (63.2%) | 3 (0.2%) | 1,347 (29.6%) | 3,678 (62.3%) | 4,024 (80.9%) | 4,767 (100%) |
| CMRF – triglycerides | 14,667 (67.1%) | 27 (1.6%) | 1,745 (38.4%) | 3,510 (59.5%) | 4.618 (92.9%) | 4,767 (100%) |
| CMRF – low HDL | 9,098 (41.6%) | 1 (0.1%) | 305 (6.7%) | 1,183 (20.1%) | 2,842 (57.2%) | 4,767 (100%) |
Continuous variables shown as mean (standard deviation)
Count variables shown as count (proportion of total)
ALT: alanine aminotransferase; AST: aspartate aminotransferase; BMI: body mass index; BP: blood pressure; CMRF: cardiometabolic risk factor; FIB-4: Fibrosis-4 Index; FLI: Fatty Liver Index; HDL: high-density lipoprotein; HTN: hypertension; LDL: low-density lipoprotein; WC: waist circumference
Association of CMRFs and All-Cause Mortality
A total of 4,236 deaths occurred over a median follow-up time of 122 months (IQR 61-191 months), over a total analyzed follow-up time of 248,379 person-years. The proportion of deaths increased with increasing number of CMRFs (4.7% of the group with one CMRF to 30.7% of the group with five CMRFs). Causes of death are shown in Supplemental Table 1. The association of individual variables with all-cause mortality is reported in Supplemental Table 2.
In age-adjusted Cox regression analysis (Model 1) (Table 2), increasing number of CMRFs was associated with a greater risk for mortality (HR 1.15 per additional CMRF, 95%CI 1.10-1.20, p<0.001) (Supplemental Figure 2). Among covariables of interest, female sex (HR 0.72, 95%CI 0.68-0.78, p<0.001) and Hispanic ethnicity (vs. White non-Hispanic, HR 0.81, 95%CI 0.70-0.93, p=0.004) were protective against all-cause mortality. Increasing FIB-4 (per 1 unit increase, HR 1.09, 95%CI 1.06-1.12, p<0.001), Black race (vs. White race, HR 1.17, 95%CI 1.05-1.29, p=0.004), and smoking history (HR 1.71, 95%CI 1.54-1.88, p<0.001) were associated with greater risk for all-cause mortality. Statin therapy was not associated with risk for all-cause mortality (HR 0.93, 95%CI 0.85-1.03, p=0.16).
Table 2.
Analysis of CMRFs Associated with All-Cause Mortality (Model 1 and Model 2)
| CMRF | Model 1* | Model 2† | ||
|---|---|---|---|---|
| HR, 95% CI | P value | aHR, 95% CI | P value | |
| Female sex | 0.72, 0.68 - 0.78 | <0.001 | 0.76, 0.69 - 0.84 | <0.001 |
| FIB-4 | 1.09, 1.06 - 1.12 | <0.001 | 1.09, 1.06 - 1.12 | <0.001 |
| Race/ethnicity | ||||
| White, non-Hispanic | (reference) | (reference) | (reference) | (reference) |
| Hispanic | 0.81, 0.70 - 0.93 | 0.004 | 0.85, 0.74 - 0.98 | 0.02 |
| Non-Hispanic Black | 1.17, 1.05 - 1.29 | 0.004 | 1.14, 1.01 - 1.27 | 0.03 |
| Other | 1.17, 0.91 - 1.50 | 0.22 | 1.15, 0.88 - 1.51 | 0.30 |
| Smoking history | 1.71, 1.54 - 1.88 | <0.001 | 1.62, 1.46 - 1.80 | <0.001 |
| Statin therapy | 0.93, 0.85 - 1.03 | 0.16 | N/A | N/A |
| BMI categories | ||||
| BMI ≥ 25 - < 30 kg/m2 | (reference) | (reference) | (reference) | (reference) |
| BMI ≥ 30 - < 35 kg/m2 | 0.97, 0.88 – 1.06 | 0.47 | 1.01, 0.92 – 1.11 | 0.82 |
| BMI ≥ 35 - < 40 kg/m2 | 1.08, 0.94 – 1.24 | 0.28 | 1.17, 1.02 – 1.34 | 0.02 |
| BMI ≥ 40 - < 45 kg/m2 | 1.43, 1.16 – 1.76 | 0.001 | 1.55, 1.25 – 1.93 | <0.001 |
| BMI ≥ 45 kg/m2 | 1.50, 1.17 – 1.92 | 0.001 | 1.62, 1.25 – 2.11 | <0.001 |
| BMI < 25 kg/m2 with elevated WC | 1.69, 1.28-2.23 | <0.001 | 1.67, 1.27 – 2.21 | <0.001 |
| Lean (BMI < 25 kg/m2 with normal WC) | 1.83, 1.41 – 2.37 | <0.001 | 1.94, 0.94 – 4.00 | 0.07 |
| Glucose intolerance | 1.33, 1.23 – 1.45 | <0.001 | 1.26, 1.16 – 1.38 | <0.001 |
| Hypertension | 1.46, 1.31 – 1.62 | <0.001 | 1.38, 1.23 – 1.56 | <0.001 |
| Elevated Triglycerides | 1.05, 0.95 – 1.16 | 0.35 | 0.95, 0.85 – 1.06 | 0.37 |
| Low HDL | 1.17, 1.07 – 1.28 | 0.001 | 1.14, 1.05 – 1.25 | 0.003 |
| Number of CMRF (per additional CMRF) | 1.15, 1.10 – 1.20 | <0.001 | N/A | N/A |
Cox regression adjusted for age
Cox regression adjusted for age, sex, and FIB-4 score
Bold denotes statistically significant at α = 0.05 level
BMI: body mass index; BP: blood pressure; CMRF: cardiometabolic risk factor; HDL: high-density lipoprotein; LDL: low-density lipoprotein; WC: waist circumference
When assessing the individual CMRF components in separate age-adjusted models, compared to those with a BMI of ≥ 25 – < 30 kg/m2, those who did not meet the CMRF-overweight/obesity (i.e., those classified as lean, or with a normal BMI and normal WC) had a greater risk for mortality (HR 1.83, 95%CI 1.41-2.37, p<0.001) as did those with BMI ≥ 40 – < 45 kg/m2 (HR 1.43, 95%CI 1.16-1.76, p=0.001) and BMI ≥ 45 kg/m2 (HR 1.50, 95%CI 1.17-1.92, p=0.001). In contrast, those with a BMI ≥ 30 – < 35 kg/m2 (HR 0.97, 95%CI 0.88-1.06, p=0.47) or BMI ≥ 35 – < 40 kg/m2 (HR 1.08, 95%CI 0.94-1.24, p=0.28) did not have a significantly higher risk for all-cause mortality. CMRF-glucose intolerance (HR 1.33, 95%CI 1.23-1.45, p<0.001), CMRF-HTN (HR 1.45, 95%CI 1.31-1.62, p<0.001), and CMRF-low HDL (HR 1.17, 95%CI 1.07-1.28, p=0.001) were significantly associated with greater risk for all-cause mortality, while CMRF-triglycerides was not (HR 1.05, 95%CI 0.95-1.16, p=0.35).
Multivariable Models of CMRF and All-Cause Mortality
In multivariable Cox regression analysis adjusted for all CMRF variables including categorical CMRF-overweight/obesity (Model 2) (Table 2) (Figure 1), adjusted for age, sex, FIB-4, race/ethnicity, and smoking history, similar effects of individual CMRFs on all-cause mortality were seen. Compared to those with a BMI of ≥ 25 – < 30 kg/m2, greater risk for all-cause mortality was seen in those with BMI ≥ 35 – < 40 kg/m2 (aHR 1.17, 95%CI 1.02-1.34, p=0.02), those with BMI ≥ 40 – < 45 kg/m2 (aHR 1.55, 95%CI 1.25-1.93, p<0.001), those with BMI ≥ 45 kg/m2 (aHR 1.62, 95%CI 1.25-2.11, p<0.001), and those with normal BMI and elevated WC (aHR 1.67, 95%CI 1.27-2.21, p<0.001). Mortality risk was similar in those with BMI ≥ 30 – < 35 kg/m2 compared to those with BMI ≥ 25 – < 30 kg/m2 (aHR 1.01, 95%CI 0.92-1.11, p=0.82). CMRF-glucose intolerance (aHR 1.26, 95%CI 1.16-1.38, p<0.001), CMRF-HTN (aHR 1.38, 95%CI 1.23-1.56, p<0.001), and CMRF-low HDL (aHR 1.14, 95%CI 1.05-1.25, p=0.003) were associated with greater risk for all-cause mortality, while CMRF-triglycerides was not (aHR 0.95, 95%CI 0.85-1.06, p=0.37). The aHR for CMRF-HTN did not significantly differ from the aHR for CMRF-glucose intolerance (Wald test p=0.20), while it did differ significantly from CMRF-low HDL (Wald test p=0.01) and CMRF-elevated triglycerides (Wald test p=0.001). The aHR for CMRF-glucose intolerance did not significantly differ from the aHR for CMRF-low HDL (p=0.14), but did differ from the aHR for CMRF-elevated triglycerides (p=0.002). The aHR for CMRF-low HDL differed significantly from the aHR for CMRF-elevated triglycerides (p=0.02).
Figure 1. Age, Sex, FIB-4, and CMRF Adjusted Hazard Ratios for CMRF-overweight/obesity Categories for Risk of All-Cause Mortality.
This figure shows hazard ratios (and 95% confidence intervals) for all-cause mortality, adjusted for age, sex, and FIB-4 score, and presence of individual CMRF, for the categorical CMRF-overweight/obesity variable of discrete BMI categories
Hierarchy Analysis of CMRF with All-Cause Mortality
The CMRF-hierarchy model (Model 3) analyzed only the population who met CMRF-overweight/obesity (n=21,727) and evaluated the addition of other CMRFs and the number of CMRFs (Table 3, Figure 2, and Figure 3). Compared to participants who were CMRF-overweight/obesity, those who met two CMRF (CMRF-overweight/obesity plus additional CMRF) did not have a greater risk for mortality if the additional CMRF was CMRF-triglycerides ( aHR 0.78, 95%CI 0.53-1.16, p=0.22) or CMRF-low HDL (aHR 1.34, 95%CI 0.82-2.20, p=0.24), but the risk was greater with CMRF-glucose intolerance (aHR 1.56, 95%CI 1.04-2.35, p=0.03) or CMRF-HTN (aHR 1.65, 95%CI 1.23-2.20, p=0.001). In addition, compared to CMRF-overweight/obesity, risk of mortality was higher in those with three CMRFs (aHR 1.66, 95%CI 1.24-2.23, p=0.001), four CMRFs (aHR 1.80, 95%CI 1.37-2.37, p<0.001), and five CMRFs (aHR 2.18, 95%CI 1.63-2.92, p<0.001). These results support a rank of CMRF aHR of CMRF-HTN > CMRF-glucose intolerance > CMRF-low HDL > CMRF-triglycerides.
Table 3.
Hierarchy of CMRF by Type and Number Associated with All-Cause Mortality in Those Meeting CMRF-overweight/obesity* (Model 3)
| CMRF | aHR, 95%CI | P value |
|---|---|---|
| Overweight/obesity alone | (reference) | (reference) |
| Overweight/obesity + elevated triglycerides | 0.78, 0.53 – 1.16 | 0.22 |
| Overweight/obesity + low HDL | 1.34, 0.82 – 2.20 | 0.24 |
| Overweight/obesity + glucose intolerance | 1.57, 1.04 – 2.35 | 0.03 |
| Overweight/obesity + hypertension | 1.65, 1.23 – 2.20 | 0.001 |
| Overweight/obesity + 2 CMRF | 1.66, 1.24 – 2.23 | 0.001 |
| Overweight/obesity + 3 CMRF | 1.80, 1.37 – 2.37 | <0.001 |
| Overweight/obesity + 4 CMRF | 2.18, 1.63 – 2.92 | <0.001 |
Adjusted for age, sex, and FIB-4
Bold denotes statistically significant at α = 0.05 level
BP: blood pressure; CMRF: cardiometabolic risk factor; HDL: high density lipoprotein; HR: hazard ratio
Figure 2. Overall Survival Stratified by Hierarchy Model of Additional CMRFs.
The figure shows overall survival, by participant age at death, stratified by the hierarchy model of cardiometabolic risk factors (CMRFs) present in addition to CMRF-overweight/obesity.
Figure 3. Hazard Ratios of Hierarchy Model of CMRF for All-Cause Mortality.
This figure shows hazard ratios (and 95% confidence intervals) for all-cause mortality, adjusted for age, sex, and FIB-4 score, stratified by the hierarchy model of cardiometabolic risk factors (CMRFs) present in addition to CMRF-overweight/obesity.
Association of Hierarchy of CMRF with Hepatic Fibrosis
The CMRF-hierarchy was evaluated for association with the secondary outcome of hepatic fibrosis (FIB-4 > 2.67) (Supplemental Table 3). Compared to those with only CMRF-overweight/obesity, odds for advanced fibrosis were greater with a second CMRF of glucose intolerance (aOR 3.04, 95%CI 1.15-8.00, p=0.03), a second CMRF of HTN (aOR 7.90, 95%CI 3.65-17.1, p<0.001), three CMRFs (aOR 5.35, 95%CI 2.78-10.3, p<0.001), four CMRFs (aOR 6.41, 95%CI 3.27-12.6, p<0.001), or five CMRFs (aOR 14.4, 95%CI 7.26-28.8, p<0.001).
Discussion
MASLD is defined by hepatic steatosis and at least one abnormal CMRF, but the exact contribution of individual CMRFs to clinical outcomes is largely unknown. In our analysis of a large, representative U.S. population with MASLD defined by noninvasive algorithm, we found differential risks for all-cause mortality for individual CMRFs. Among this population, over 99% met criteria for CMRF-overweight/obesity, which was the sole CMRF in 98% of those with only one CMRF. Additionally, among those with MASLD and only one CMRF, CMRF-glucose intolerance was not seen as the sole CMRF, a surprising finding considering the clear pathophysiologic connection between insulin resistance and hepatic steatosis. We found that individual CMRFs exerted independent, differential effects on all-cause mortality with CMRF-blood pressure, CMRF-glucose intolerance, and CMRF-low HDL independently associated with a 40%, 25% and 15%, respectively, greater adjusted risk for all-cause mortality. When analyzing the risk for all-cause mortality across the range of BMI included in the CMRF definition (i.e., BMI ≥ 25 kg/m2), increased risk for mortality was not seen until BMI was greater than 35 kg/m2. These independent associations persisted despite adjustment for demographic factors such as sex and race/ethnicity, noninvasive measures of advanced liver fibrosis such as FIB-4 score, and other covariables associated with mortality such as tobacco smoking history. We explored the effect of adding each individual CMRF as a second CMRF (in addition to CMRF-overweight/obesity) to generate a relative hierarchy of CMRF effects. Within this model, among those with two CMRFs, CMRF-blood pressure exerted a greater risk for mortality than CMRF-glucose intolerance, while CMRF-low HDL and CMRF-triglycerides had a nonsignificant effect on risk for mortality. This hierarchy model also demonstrated point estimates for the risk for mortality that progressively increased with three or more CMRFs, which matches other findings.12 Additionally, this hierarchy model demonstrated increasing odds of presence of advanced hepatic fibrosis with increasing numbers of CMRFs. These findings suggest that not only does the number of CMRFs matter to clinically-relevant outcomes, but the individual CMRFs exert independent, distinct risks. This lays a practical foundation for characterizing distinct clinical profiles useful in prioritization of interventions as well as a framework toward understanding the heterogenous natural history of MASLD.
Our analysis found that stratification of the CMRF-overweight/obesity category into discrete ranges of BMI captured distinct risks for all-cause mortality. In particular, an increased risk for mortality was seen at the extremes of BMI and a BMI ≥ 45 carries a 62% greater risk for all-cause mortality, relative to those with a BMI in the ≥ 25 – < 30 kg/m2 range. In parallels with the findings in other studies, our study supports that there are different risk groups within this single CMRF.4–7,20, highlighting that the longitudinal shifting between these risk categories may translate into changes in mortality and would have significant implications for real-world patient counseling and treatment. The heterogeneity across the range of BMI also supports the need for improved methods of characterizing the risk of excess adiposity in those with MASLD. Our study included a relatively low prevalence of participants with ‘lean’ MASLD compared to other BMI categories. Future studies aiming to assess the specific cardiometabolic risks in this subgroup may require enriched enrollment to better characterize these associations
Our results also show that the CMRFs associated with lipids, CMRF-low HDL and CMRF-triglycerides, do not predict risk of mortality. The association of lipid CMRFs with unique disease phenotypes is supported in other MASLD work, as analyses of interventional trials found that histologic disease improvements was associated with improvements in HDL and triglycerides.21–23 Some of this diminished risk of lipid CMRFs may also be explained by the common use of statin pharmacotherapy. Future work is needed to identify the exact interplay of these lipid-related CMRFs with MASLD disease activity.
There is significant overlap between the CMRFs of MASLD and the components of metabolic syndrome. Previous studies analyzing those with metabolic syndrome have supported the differential relationships of these individual components with all-cause mortality, with similar overall findings to this current study.24,25 Despite this, understanding these differential risks is uniquely relevant to MASLD as treatments, diagnostics, and expert guidance for prognostication and management characterize MASLD as a homogenous disease entity. This homogeneity is contradicted by the difficulties that arise with its heterogenous disease presentation, progression, and outcomes, suggesting the existence of distinct clinical phenotypes. The findings of our study suggest that individual CMRFs provide distinct, differential risks in those with MASLD, supporting a framework for further stratification of this disease beyond one umbrella diagnosis. In particular, our hierarchy model supported a ranked order of CMRFs for risk of all-cause mortality, with high blood pressure exerting the greatest risk, followed by glucose intolerance, then low HDL, and finally high triglycerides. Furthermore, our findings suggest that it is not only number of CMRFs, but the individual CMRFs, that matter in MASLD. Our findings aligns with several recent publications that find that cumulative number of CMRFs are associated with greater risk for all-cause mortality, cardiovascular mortality, and advanced hepatic fibrosis, with those with the greatest number of CMRFs also having the most significant abnormalities in these CMRF parameters.10–12 Additionally, while we found differences in mortality risk in stratification of discrete ranges of BMI, these same principles may be applicable to the other CMRFs and could be explored in more focused studies, particularly taking into account whether pharmacotherapy is used within each stratified category.
This study has some limitations. One imitation is posed by the use of cross-sectional survey data, particularly as CMRFs may change over time. A second limitation is defining hepatic steatosis using FLI rather than direct measurement of hepatic steatosis. Using FLI to define hepatic steatosis allowed a much larger sample with long-term survival data in NHANES database. FLI has been validated to predict hepatic steatosis in numerous studies, including NHANES III participants with VCTE measurement of hepatic steatosis, where FLI demonstrated an optimal cut point for prediction of hepatic steatosis at a value of 59.5 and in a study comparing FLI to magnetic resonance imaging-derived fat fraction with a specificity of greater than 80%, negative predictive value of greater than 90%, and an AUROC of greater than 0.85 for prediction of hepatic steatosis.26,14 Additionally, we recognize that there is possibility of overlap in the classifications of the CMRFs of triglycerides and low HDL, due to shared treatments such as statins. We are unable to generate racially adjusted BMI definitions for CMRF-overweight/obesity due to reporting limitations of race/ethnicity in NHANES. Liver-related mortalities are not reported in publicly available NHANES-linked mortality files after 2006, therefore we are unable to analyze this outcome.
In conclusion, we show the presence of individual CMRFs exert differential effects on all-cause mortality in those with MASLD. These differential effects suggest that there may be distinct clinical phenotypes among MASLD driven by unique CMRF, and these phenotypes can likely be refined further by including additional MASLD risk factors such as genetic background, dietary habits, and alcohol consumption (particularly with consumption in the spectrum of metALD).
A deeper understanding of clinical phenotypes of persons with MASLD will aid in identifying those most at risk for disease progression and mortality, assisting health systems in prioritizing resources, and enhancing outcomes with current and future emerging classes of therapeutics.
Supplementary Material
Financial support statement:
Matthew Dukewich is supported by an NIH Translational Research in Hepatology grant 5T32DK127977
Abbreviations:
- aHR
adjusted hazard ratio
- ALT
alanine aminotransferase
- AST
aspartate aminotransferase
- aOR
adjust odds ratio
- AUROC
area under the receiver operating characteristic
- BMI
body mass index
- BP
blood pressure
- CI
confidence interval
- CMRF
cardiometabolic risk factor
- FIB4
Fibrosis-4 Index
- FLI
Fatty Liver Index
- GGT
gamma-glutamyl transferase
- HBV
hepatitis B virus
- HCV
hepatitis C virus
- HDL
high-density lipoprotein
- HR
hazard ratio
- LDL
low-density lipoprotein
- MASLD
metabolic dysfunction-associated steatotic liver disease
- NHANES
National Health and Nutrition Examination Survey
- VIF
variance inflation factor
- WC
waist circumference
Footnotes
Conflict of interest statement: There are no conflicts of interest for any of the authors relevant to this work
References
- 1.Rinella ME, Lazarus JV, Ratziu V, et al. A multisociety Delphi consensus statement on new fatty liver disease nomenclature. Hepatology. Dec 1 2023;78(6):1966–1986. doi: 10.1097/HEP.0000000000000520 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Wang T, Xi Y, Raji A, et al. Overall and subgroup prevalence of non-alcoholic fatty liver disease and prevalence of advanced fibrosis in the United States: An updated national estimate in National Health and Nutrition Examination Survey (NHANES) 2011-2018. Ann Hepatol. 2024;29(1):101154. doi: 10.1016/j.aohep.2023.101154 [DOI] [PubMed] [Google Scholar]
- 3.Le MH, Devaki P, Ha NB, et al. Prevalence of non-alcoholic fatty liver disease and risk factors for advanced fibrosis and mortality in the United States. PLoS One. 2017;12(3):e0173499. doi: 10.1371/journal.pone.0173499 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Chung GE, Yu SJ, Yoo JJ, et al. Lean or diabetic subtypes predict increased all-cause and disease-specific mortality in metabolic-associated fatty liver disease. BMC Med. Jan 04 2023;21(1):4. doi: 10.1186/s12916-022-02716-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Wijarnpreecha K, Li F, Lundin SK, et al. Higher mortality among lean patients with non-alcoholic fatty liver disease despite fewer metabolic comorbidities. Aliment Pharmacol Ther. May 2023;57(9):1014–1027. doi: 10.1111/apt.17424 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Behari J, Wang R, Luu HN, McKenzie D, Molinari M, Yuan JM. Severe obesity is associated with worse outcomes than lean metabolic dysfunction-associated steatotic liver disease. Hepatol Commun. Jul 01 2024;8(7)doi: 10.1097/HC9.0000000000000471 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Flegal KM, Kit BK, Orpana H, Graubard BI. Association of all-cause mortality with overweight and obesity using standard body mass index categories: a systematic review and meta-analysis. JAMA. Jan 02 2013;309(1):71–82. doi: 10.1001/jama.2012.113905 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Sanyal AJ, Van Natta ML, Clark J, et al. Prospective Study of Outcomes in Adults with Nonalcoholic Fatty Liver Disease. N Engl J Med. Oct 21 2021;385(17):1559–1569. doi: 10.1056/NEJMoa2029349 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Younossi ZM, Wong G, Anstee QM, Henry L. The Global Burden of Liver Disease. Clin Gastroenterol Hepatol. Jul 2023;21(8):1978–1991. doi: 10.1016/j.cgh.2023.04.015 [DOI] [PubMed] [Google Scholar]
- 10.Yang T, Yin J, Li J, Wang Q. The influence of different combinations of cardiometabolic risk factors on the prevalence of MASLD and risk of advanced fibrosis deserves attention. J Hepatol. Feb 2024;80(2):e82–e85. doi: 10.1016/j.jhep.2023.09.030 [DOI] [PubMed] [Google Scholar]
- 11.Kim JH, Lee Y, Nam CM, Kwon YJ, Lee JW. Impact of cardiometabolic risk factors for metabolic dysfunction-associated steatotic liver disease on mortality. Nutr Metab Cardiovasc Dis. Jun 2025;35(6):103965. doi: 10.1016/j.numecd.2025.103965 [DOI] [PubMed] [Google Scholar]
- 12.Tamaki N, Kimura T, Wakabayashi SI, et al. Cardiometabolic criteria as predictors and treatment targets of liver-related events and cardiovascular events in metabolic dysfunction-associated steatotic liver disease. Aliment Pharmacol Ther. Aug 08 2024;doi: 10.1111/apt.18205 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ruhl CE, Everhart JE. Fatty liver indices in the multiethnic United States National Health and Nutrition Examination Survey. Aliment Pharmacol Ther. Jan 2015;41(1):65–76. doi: 10.1111/apt.13012 [DOI] [PubMed] [Google Scholar]
- 14.Ho B, Thompson A, Jorgensen AL, Pirmohamed M. Role of fatty liver index in risk-stratifying comorbid disease outcomes in non-alcoholic fatty liver disease. JHEP Rep. Dec 2023;5(12):100896. doi: 10.1016/j.jhepr.2023.100896 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Bedogni G, Bellentani S, Miglioli L, et al. The Fatty Liver Index: a simple and accurate predictor of hepatic steatosis in the general population. BMC Gastroenterol. Nov 02 2006;6:33. doi: 10.1186/1471-230X-6-33 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Shah AG, Lydecker A, Murray K, et al. Comparison of noninvasive markers of fibrosis in patients with nonalcoholic fatty liver disease. Clin Gastroenterol Hepatol. Oct 2009;7(10):1104–12. doi: 10.1016/j.cgh.2009.05.033 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Nguyen VK, Middleton LYM, Huang L, et al. Harmonized US National Health and Nutrition Examination Survey 1988-2018 for high throughput exposome-health discovery. medRxiv. Feb 08 2023;doi: 10.1101/2023.02.06.23284573 [DOI] [Google Scholar]
- 18.Akinbami LJ, Chen TC, Davy O, et al. National Health and Nutrition Examination Survey, 2017-March 2020 Prepandemic File: Sample Design, Estimation, and Analytic Guidelines. Vital Health Stat 1. May 2022;(190):1–36. [PubMed] [Google Scholar]
- 19.CDC. NHANES Survey Methods and Analytic Guidelines. Accessed September 22, 2024. https://wwwn.cdc.gov/nchs/nhanes/analyticguidelines.aspx
- 20.Nabi O, Lapidus N, Boursier J, et al. Lean individuals with NAFLD have more severe liver disease and poorer clinical outcomes (NASH-CO Study). Hepatology. Jul 01 2023;78(1):272–283. doi: 10.1097/HEP.0000000000000329 [DOI] [PubMed] [Google Scholar]
- 21.Brunt EM, Kleiner DE, Wilson LA, Sanyal AJ, Neuschwander-Tetri BA, Network NSCR. Improvements in Histologic Features and Diagnosis Associated With Improvement in Fibrosis in Nonalcoholic Steatohepatitis: Results From the Nonalcoholic Steatohepatitis Clinical Research Network Treatment Trials. Hepatology. Aug 2019;70(2):522–531. doi: 10.1002/hep.30418 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Hoofnagle JH, Van Natta ML, Kleiner DE, et al. Vitamin E and changes in serum alanine aminotransferase levels in patients with non-alcoholic steatohepatitis. Aliment Pharmacol Ther. Jul 2013;38(2):134–43. doi: 10.1111/apt.12352 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Corey KE, Vuppalanchi R, Wilson LA, Cummings OW, Chalasani N, CRN N. NASH resolution is associated with improvements in HDL and triglyceride levels but not improvement in LDL or non-HDL-C levels. Aliment Pharmacol Ther. Feb 2015;41(3):301–9. doi: 10.1111/apt.13035 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Yu WW, Randhawa AK, Blair SN, Sui X, Kuk JL. Age- and sex- specific all-cause mortality risk greatest in metabolic syndrome combinations with elevated blood pressure from 7 U.S. cohorts. PLoS One. 2019;14(6):e0218307. doi: 10.1371/journal.pone.0218307 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Kuk JL, Ardern CI. Age and sex differences in the clustering of metabolic syndrome factors: association with mortality risk. Diabetes Care. Nov 2010;33(11):2457–61. doi: 10.2337/dc10-0942 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Wu J, Tian S, Li H, et al. Population-specific cut-off points of fatty liver index: a study based on the National Health and Nutrition Examination Survey data. BMC Gastroenterol. May 27 2022;22(1):265. doi: 10.1186/s12876-022-02303-z [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.