29,314 participants from the NHANES III cohort (1988–1994) with available ultrasonography and laboratory results were included in the study. Of them, 12,284 were excluded for age <20, 1192 for missing data on AST or ALT, 1982 for lack of liver ultrasound data, and 164 for missing data on BIA, leaving 13,692 participants for analysis. The mean age of the population was 43.7 ± 15.97 years, 6473/13,692 (47.3%) were male and the mean BMI was 27.3 ± 5.90 kg/m2. MAFLD was diagnosed in 4207/13,692 (30.73%) participants and sarcopenia in 5804/13,692 (42.39%) among the overall population, resulting in the following distribution of reciprocally independent groups: MAFLD [-]/sarcopenia [-] (46.30%), MAFLD [+]/sarcopenia [-] (11.31%), MAFLD [-]/sarcopenia [+] (22.97%) and MAFLD [+]/sarcopenia [+] (19.42%). Detailed baseline information of these groups are shown in Table 1.

During the follow-up of 23.7 ± 7.62 years, there were 4883 deaths in total. Among them, there were 1328 cardiovascular-related mortality, 1178 cancer-related death and 209 diabetes-related mortality (n = 209). MAFLD (aHR 1.152, 95% CI 1.070–1.241) and sarcopenia (aHR 1.123, 95% CI 1.042–1.210) were independently and simultaneously associated with increased all-cause mortality in the fully adjusted models (Table 2). Furthermore, to clarify whether the existence of MAFLD or sarcopenia increased the hazard for mortality, we categorized the cohort into four groups: (MAFLD [-]/sarcopenia [-], MAFLD [+]/sarcopenia [-], MAFLD [-]/sarcopenia [+] and MAFLD [+]/sarcopenia [+]). Compared with patients without MAFLD and sarcopenia, patients with MAFLD and sarcopenia were with higher risk of all-cause mortality (aHR 1.247, 95% CI 1.132–1.373) in the fully adjusted models. However, in the MAFLD [+]/sarcopenia [-] and MAFLD [-]/sarcopenia [+] groups, we could not demonstrate an increased risk for all-cause mortality compared with the MAFLD [-]/sarcopenia [-] group (aHR 1.076, 95% CI 0.952–1.217 and aHR 1.055, 95% CI 0.959–1.161, respectively) (Table 3).

Concerning cause-specific mortality, cardiovascular disease-related mortality was comparable for the MAFLD or sarcopenia group independently (Table 4). However, in multivariable analysis, the MAFLD [+]/sarcopenia [+] group was with a 22.9% increase in cardiovascular disease-related mortality risk compared with the MAFLD [-]/sarcopenia [-] group (aHR: 1.229 95% CI: 1.019–1.482) (Table 5).

In addition, regarding diabetes-related mortality, individuals with MAFLD had a 158.9% higher diabetes-related mortality than those without MAFLD in the age and gender adjusted model (HR: 2.589, 95% CI: 1.960–3.421). In the fully adjusted model, MAFLD was significantly associated with an increased risk for diabetes-related mortality (aHR: 2.532, 95% CI: 1.759–3.645). Individuals with sarcopenia had increased risk of diabetes-related mortality (aHR: 1.657; 95% CI: 1.226–2.239) in the age and sex adjusted model. However, sarcopenia failed to maintain this association in the fully adjusted model (aHR: 1.444, 95% CI: 0.988–2.112) (Table 6). Furthermore, compared with individuals without MAFLD and sarcopenia, individuals with MAFLD were significantly associated with a higher risk of diabetes-related mortality both with and without sarcopenia in the fully adjusted model. However, for the MAFLD [+]/sarcopenia [+] group, this association was failed to be maintained in the fully adjusted model (Table 7). In addtion, as shown in Tables S1–S2, individuals with MAFLD or sarcopenia, or in any mutually exclusive group were not significantly related to a higher risk of cancer-related mortality.

Subjects in the MAFLD [+]/sarcopenia [+] group had the highest rate of advanced fibrosis (2.8% by FIB-4 and 11.4% by NFS). Although MAFLD [-]/sarcopenia [+] has a slightly lower rate of advanced fibrosis evaluated by FIB-4, the overall median and advanced fibrosis in MAFLD [-]/sarcopenia [+] is significantly higher than that of MAFLD [+]/sarcopenia [-]. In the plateau of NFS evaluated liver fibrosis, followed by the MAFLD [-]/sarcopenia [+] group (1.6% by FIB-4 and 7.8% by NFS) and the MAFLD [+]/sarcopenia [-] group (2.0% by FIB-4 and 4.3% by NFS), and the MAFLD [-]/sarcopenia [-] group had the lowest rate of advanced liver fibrosis (1.9% by FIB-4 and 2.6% by NFS) (P < 0.001) (Fig. 1).

When A fibrosis-4 index (FIB-4) and B non-alcoholic fatty liver disease fibrosis score (NFS) were stratified by risk degree, the degree of fibrotic burden significantly increased from population without MAFLD and sarcopenia, to MAFLD population without sarcopenia and sarcopenic population.

As there were limited number of subjects with FIB-4 > 2.67 in the dataset, we used FIB-4 > 1.3 and NFS > 0.676 to define significant liver fibrosis [9, 15]. After adjustments for possible confounders (age, sex, race, marital status, education, and current smoking status), subjects with sarcopenia showed significantly higher aORs for significant liver fibrosis, by both FIB-4 and NFS compared with subjects without sarcopenia (all P < 0.001) (Tables S3–S4). When compared with the MAFLD [-]/sarcopenia [-] group, the MAFLD [-]/sarcopenia [+] group (aOR = 1.600 by FIB-4; aOR = 1.566 by NFS) and MAFLD [+]/sarcopenia [+] group (aOR = 1.647 by FIB-4; aOR = 2.296 by NFS) were both with higer risk of significant liver fibrosis (all P < 0.001). Furthermore, for the MAFLD [+]/sarcopenia [-] group, subjects failed to show a significantly increased risk for fibrosis by both FIB-4 and NFS after adequate adjustment (Tables S5–S6).