Determination of the cut-off of the L3-SMD for myosteatosis diagnosis and the impact factors of the L3-SMD
Our previous study has established the diagnostic criteria of sarcopenia based on L3-SMI<44.77 cm2/m2 in male and <32.50 cm2/m2 in female, and demonstrated that age, sex, height, weight, biceps circumference, and triceps skinfold thickness significantly impact on L3-SMI [21]. The same cohort including 911 healthy volunteers were used to determine the factors that affected the L3-SMD. The basic characteristics of the group were reported in the previous study [21]. According to the univariate analysis, several factors were associated with the value of the L3-SMD, including age, sex, height, weight, waist circumference, biceps circumference, triceps skinfold thickness, subscapular skinfold thickness, and abdominal skinfold thickness. In the multivariate analysis, only age, sex, weight, waist circumference and biceps circumference had remarkable influence on the L3-SMD (Supplementary Table 1). Among all of these factors, the impact of age on the L3-SMD is the most obvious. The average value of the L3-SMD gradually decreased with age. Considering the critical influence of age to muscle mass and quality, only the adults younger than 60 years in cohort 1 were enrolled to determine the cut-off of the L3-SMD for the diagnosis of myosteatosis. The imaging parameter data of the 665 individuals (365 male, 296 female) are listed in Table 1. Based on the mean − 1.28×SD among the adults aged<60 years, the cut-off of the L3-SMD for myosteatosis diagnosis was 38.93 Hu in males and 32.82 Hu in females. The prevalence of myosteatosis was 49.4% in individuals aged 60 to 69 years and up to 80.0% in patients older than 70 years (Supplementary Table 2).
Table 1
Characteristics of the imaging parameters in healthy population younger than 60 years in Cohort 1
Group
|
Male (n = 365)
|
Female (n = 296)
|
P value
|
Area
|
|
|
|
L3-SMA (cm2)
|
158.87 ± 22.20
|
104.27 ± 12.76
|
<0.001
|
L3-IMATA (cm2)
|
9.20 ± 5.12
|
8.57 ± 5.61
|
0.135
|
L3-VATA (cm2)
|
123.52 ± 70.17
|
63.89 ± 42.84
|
<0.001
|
L3-SATA (cm2)
|
127.57 ± 63.82
|
137.26 ± 54.73
|
0.039
|
Indexes
|
|
|
|
L3-SMI (cm2/m2)
|
54.70 ± 7.76
|
39.62 ± 5.56
|
<0.001
|
L3-IMATI (cm2/m2)
|
3.17 ± 1.76
|
3.25 ± 2.15
|
0.589
|
L3-VATI (cm2/m2)
|
42.54 ± 24.08
|
24.36 ± 16.65
|
<0.001
|
L3-SATI (cm2/m2)
|
43.87 ± 21.58
|
52.19 ± 22.00
|
<0.001
|
Density
|
|
|
|
L3-SMD (Hu)
|
44.84 ± 4.62
|
39.72 ± 5.39
|
<0.001
|
L3-IMATD (Hu)
|
-64.82 ± 5.45
|
-63.31 ± 4.52
|
<0.001
|
L3-VATD (Hu)
|
-96.8 ± 7.44
|
-92.56 ± 7.31
|
<0.001
|
L3-SATD (Hu)
|
-101.21 ± 7.71
|
-103.61 ± 5.09
|
<0.001
|
Liver density (Hu)
|
58.45 ± 8.78
|
60.08 ± 8.34
|
0.015
|
Spleen density (Hu)
|
51.69 ± 5.21
|
49.70 ± 4.71
|
<0.001
|
Ratio
|
|
|
|
Liver density / Spleen density
|
1.14 ± 0.20
|
1.22 ± 0.20
|
<0.001
|
SMA: skeletal muscle area; IMATA: intermuscular adipose tissue area; VATA: visceral adipose tissue area; SATA: subcutaneous adipose tissue area; SMI: skeletal muscle index; IMATI: intermuscular adipose tissue index; VATI: visceral adipose tissue index; SATI: subcutaneous adipose tissue index; SMD: skeletal muscle density; IMATD: intermuscular adipose tissue density; VATD: visceral adipose tissue density; SATD: subcutaneous adipose tissue density. |
Myosteatosis is associated with poor liver function and portal hypertension
To determine the impact of myosteatosis on liver function, the data of 480 cirrhotic patients in Cohort 2 were subsequently analyzed. As shown in Table 2, based on the cut off of 38.93 Hu in males and 32.82 Hu in females, 147 (30.63%) patients were diagnosed with myosteatosis. The prevalence of myosteatosis in the male and female patients with liver cirrhosis was 30.41% and 30.98%, respectively. Compared with the nonmyosteatotic individuals, the patients with myosteatosis were much older and had higher BMI values (24.28 ± 3.90 vs. 23.14 ± 3.29 kg/m2, P = 0.003), lower red blood cell (RBC) counts (3.36 ± 0.76 vs. 3.55 ± 0.77 ×1012/L, P = 0.004), albumin levels (30.87 ± 6.00 vs. 33.40 ± 5.80 g/L, P<0.001), longer PT (16.46 ± 3.67 vs. 15.60 ± 2.89 s, P = 0.006) and higher total bilirubin (TBil) (40.57 ± 47.36 vs. 31.80 ± 34.23 µmol/L, P = 0.023) and direct bilirubin (DBil) concentrations (21.65 ± 38.50 vs. 11.82 ± 21.01 µmol/L, P<0.001) and INR (1.40 ± 0.33 vs. 1.31 ± 0.29, P = 0.005). In addition, Child–Pugh score (8.18 ± 2.27 vs. 7.18 ± 1.91, P<0.001) and MELD score (12.29 ± 4.49 vs. 11.25 ± 3.75, P = 0.009) were worse in the myosteatosis group. Among the 182 cirrhotic patients who underwent HVPG determination, 168 patients who met the enrolment criteria were included to evaluate the role of muscle alterations in portal hypertension (Supplementary Table 3). Based on our cut-off value, 19.05% (32/168) of the patients (26 males and 6 females) were diagnosed with sarcopenia, and 16.07% (27/168) of them were diagnosed with myosteatosis, including 16 males and 11 females. Among all of the investigated imaging-based nutritional indicators, L3-SMD was the only variable that was negatively related to the HVPG (r=-0.266, P<0.001, Supplementary Table 4). No correlation between L3-SMI and HVPG was observed. The average HVPG in the patients with myosteatosis was 21.57 ± 8.20 mmHg, which was much higher than that in the nonmyosteatosis group (16.13 ± 6.89 mmHg, P<0.001). Subsequently, we compared the CT-based nutritional indicators between the patients with different HVPGs (Supplementary Tables 5–6). Surprisingly, none of the patients with an HVPG < 10 mmHg were diagnosed with myosteatosis, but 27 of the 138 patients with an HVPG ≥ 10 mmHg had myosteatosis (P < 0.001). The mean L3-SMD was remarkedly decreased in patients with CSPH compared with those without CSPH (42.28 ± 7.47 Hu vs. 45.99 ± 5.97 Hu, P = 0.012). Similarly, the prevalence of myosteatosis (20.16% vs. 2.56%, P = 0.006) was much higher, and the average L3-SMD (42.09 ± 7.48 Hu vs. 45.74 ± 6.18 Hu, P = 0.006) was lower in patients with an HVPG ≥ 12 mmHg than in those with an HVPG < 12 mmHg (Supplementary Table 6). However, other CT-based nutritional indicators, including L3-SMI, seem to be irrelevant to the HVPG. Thus, we believe that, myosteatosis, not sarcopenia has obvious effect on portal hypertension.
Table 2
Baseline characteristics is compared between myosteatosis group and non-myosteatosis group according to the Cohort 2.
|
Non-myosteatosis (n = 333)
|
myosteatosis (n = 147)
|
P value
|
Age, years (mean ± SD)
|
52.10 ± 11.51
|
62.44 ± 10.09
|
<0.001
|
Gender (Male/Female)
|
206/127
|
90/57
|
0.895
|
Body mass index (kg/m2)
|
23.14 ± 3.29
|
24.28 ± 3.90
|
0.003
|
Child-Pugh score
|
7.18 ± 1.91
|
8.18 ± 2.27
|
<0.001
|
MELD
|
11.25 ± 3.75
|
12.29 ± 4.49
|
0.009
|
Serum index
|
|
|
|
RBC (1012/L)
|
3.55 ± 0.77
|
3.36 ± 0.76
|
0.011
|
WBC (109/L)
|
4.28 ± 3.32
|
4.58 ± 2.85
|
0.346
|
PLT (109/L)
|
120.24 ± 165.88
|
103.59 ± 76.15
|
0.245
|
HCT (%)
|
30.86 ± 7.63
|
30.40 ± 6.90
|
0.527
|
Hemoglobin (g/L)
|
102.28 ± 28.38
|
100.55 ± 25.48
|
0.526
|
TBil (µmol/L)
|
31.80 ± 34.23
|
40.57 ± 47.36
|
0.023
|
DBil (µmol/L)
|
11.82 ± 21.01
|
21.65 ± 38.50
|
<0.001
|
Albumin (g/L)
|
33.40 ± 5.80
|
30.87 ± 6.00
|
<0.001
|
Scr (mmol/L)
|
63.81 ± 41.01
|
75.10 ± 46.73
|
0.008
|
PT (seconds)
|
15.60 ± 2.89
|
16.46 ± 3.67
|
0.006
|
INR
|
1.31 ± 0.29
|
1.40 ± 0.33
|
0.005
|
Complications
|
|
|
|
Ascites (%)
|
212/333(64.0)
|
110/147(74.8)
|
0.016
|
Grade
|
|
|
<0.001
|
0
|
142
|
39
|
|
1
|
115
|
42
|
|
2
|
25
|
14
|
|
3
|
51
|
52
|
|
HE(%)
|
26/333(7.8)
|
20/147(13.6)
|
0.047
|
SBP(%)
|
9/333(3)
|
12/147(8.2)
|
0.007
|
UGIB(%)
|
208/333(62.5)
|
63/147(42.9)
|
<0.001
|
HRS(%)
|
2/333(0.6)
|
6/147(17.7)
|
0.006
|
PVT(%)
|
107/333(32.1)
|
25/147(17.0)
|
0.001
|
L3 body composition parameters
|
|
|
L3-SMI (cm2/m2)
|
46.84 ± 8.21
|
43.35 ± 9.11
|
<0.001
|
L3-IMAT (cm2/m2)
|
2.45 ± 1.42
|
5.08 ± 3.90
|
<0.001
|
L3-VAT (cm2/m2)
|
25.68 ± 17.78
|
39.28 ± 27.32
|
<0.001
|
L3-SAT (cm2/m2)
|
35.36 ± 21.35
|
42.67 ± 27.58
|
0.002
|
Sarcopenia%
|
55/333(16.5)
|
53/147(36.0)
|
<0.001
|
MELD: model for end-stage liver disease; RBC: red blood cell; WBC: white blood cell; PLT: blood platelet; HCT: hematocrit; TBil: total bilirubin; DBil: direct bilirubin; Scr: serum creatinine; PT: prothrombin time; INR: international normalized ratio; HE: hepatic encephalopathy; SBP: spontaneous bacterial peritonitis; UGIB: upper gastrointestinal bleeding; AKI: acute kidney injury; HRS: hepatorenal syndrome; PVT: portal venous thrombosis; SMI: skeletal muscle index; IMATI: intermuscular adipose tissue index; VATI: visceral adipose tissue index; SATI: subcutaneous adipose tissue index. |
Concurrence of sarcopenia and myosteatosis reflect severe cirrhosis and is an indicator for poor prognosis of cirrhosis
Among the 480 cirrhotic patients in cohort 2, 53 (11.04%) patients had both sarcopenia and myosteatosis, 55 (11.46%) had only sarcopenia, 94 (19.58%) had only myosteatosis, and 278 (57.92%) had neither sarcopenia nor myosteatosis. As shown in Table 3, compared with the individuals without any kind of muscle alterations or those who had only sarcopenia or myosteatosis, the patients with complex skeletal muscle alterations had lower RBC counts and albumin levels, longer PT, higher TBil and DBil concentration and INR. Moreover, the Child–Pugh score and MELD score, two important indicators of the extent of cirrhosis, were both worse in the complex skeletal muscle alterations group (Child–Pugh score, P<0.001; MELD score, P = 0.045). Therefore, all the observations suggested that concurrence of sarcopenia and myosteatosis has a strong correlation with poorer liver function and more serious liver cirrhosis. Furthermore, the mortality within the 2-year follow-up period was 39.62% (21/53) in the patients with complex skeletal muscle alterations, significantly higher than that in those with only sarcopenia (14.54%, 8/55) or myosteatosis (18.09%, 17/94), and without any skeletal muscle alterations (8.99%, 25/278). A total of 10 patients received liver transplantations, including 3 in complex skeletal muscle alterations group, 3 in patients with only myosteatosis and 4 in patients without skeletal muscle alterations. As shown in Fig. 2A, patients with complex skeletal muscle alterations had significantly elevated mortality. The concurrence of sarcopenia and myosteatosis evidently reduced the overall and liver transplantation-free survival of the cirrhotic patients (Fig. 2B and Supplementary Fig. 1, both P < 0.001). Moreover, the episodes of breakthrough of HE, spontaneous bacterial peritonitis (SBP), esophageal and gastric variceal bleeding (EGVB) and acute kidney injury (AKI) / hepatorenal syndrome (HRS) occurred in 22.6% (12/53), 13.2% (7/53), 67.9% (36/53), and 7.6% (4/53) of the patients with complex skeletal muscle alterations, respectively, which were all much higher than those in the other groups (HE, P = 0.002; SBP, P = 0.004; EGVB, P = 0.002; AKI/HRS, P = 0.003 ).
Table 3
Comparison of baseline characteristics among the cirrhotic patients with different skeletal muscle alterations according Cohort 2
|
Neither sarcopenia nor myosteatosis (n = 278)
|
Only Sarcopenia (n = 55)
|
Only myosteatosis (n = 94)
|
Both sarcopenia and myosteatosis (n = 53)
|
P value
|
Age, years (mean ± SD)a
|
50.99 ± 11.19
|
51.91 ± 12.68
|
62.40 ± 9.90
|
60.89 ± 9.54
|
<0.001
|
Gender (Male/Female)
|
118/160
|
9/46
|
44/50
|
13/40
|
<0.001
|
Body mass index (kg/m2) a
|
23.74 ± 3.16
|
20.15 ± 2.12
|
25.30 ± 3.87
|
22.59 ± 3.36
|
<0.001
|
Child-Pugh score a
|
7.13 ± 1.94
|
7.42 ± 1.82
|
7.88 ± 2.26
|
8.70 ± 2.21
|
<0.001
|
MELD
|
11.50 ± 3.85
|
11.40 ± 3.65
|
11.90 ± 4.10
|
13.33 ± 5.06
|
0.043
|
Serum index
|
|
|
|
|
|
RBC (1012/L) a
|
3.59 ± 0.77
|
3.37 ± 0.76
|
3.43 ± 0.71
|
3.23 ± 0.82
|
0.006
|
WBC (109/L) a
|
4.15 ± 3.32
|
4.95 ± 3.29
|
5.50 ± 9.81
|
4.65 ± 2.80
|
0.006
|
PLT (109/L)
|
113.76 ± 133.15
|
152.96 ± 277.36
|
105.47 ± 68.44
|
100.26 ± 88.81
|
0.176
|
HCT (%)
|
31.18 ± 7.60
|
29.27 ± 7.67
|
30.92 ± 6.72
|
29.48 ± 7.18
|
0.195
|
Hemoglobin (g/L)
|
103.41 ± 28.07
|
96.60 ± 29.47
|
101.72 ± 25.64
|
98.47 ± 25.31
|
0.465
|
TBil (µmol/L) a
|
31.54 ± 32.65
|
33.16 ± 41.74
|
35.44 ± 27.67
|
49.67 ± 69.25
|
0.072
|
DBil (µmol/L) a
|
11.26 ± 18.34
|
14.67 ± 31.42
|
18.45 ± 29.29
|
27.44 ± 50.95
|
<0.001
|
Albumin (g/L)
|
33.52 ± 5.77
|
32.77 ± 5.99
|
31.20 ± 6.19
|
30.28 ± 5.67
|
<0.001
|
Scr (mmol/L) a
|
64.08 ± 44.27
|
62.51 ± 17.32
|
71.05 ± 45.17
|
82.21 ± 49.03
|
0.003
|
PT (seconds) a
|
15.56 ± 2.94
|
15.80 ± 2.64
|
16.15 ± 3.83
|
17.00 ± 3.34
|
0.008
|
INR a
|
1.31 ± 0.30
|
1.33 ± 0.25
|
1.37 ± 0.35
|
1.43 ± 0.30
|
0.005
|
Complications
|
|
|
|
|
|
Ascites (%)
|
169/278(60.8)
|
43/55(78.2)
|
62/94(66.0)
|
48/53(90.6)
|
<0.001
|
Grade
|
|
|
|
|
<0.001
|
0
|
126
|
16
|
34
|
5
|
|
1
|
92
|
23
|
24
|
18
|
|
2
|
21
|
4
|
9
|
5
|
|
3
|
29
|
12
|
27
|
25
|
|
HE(%)
|
25/278(9.0)
|
1/55(1.8)
|
8/94(8.5)
|
12/53(22.6)
|
0.002
|
SBP(%)
|
6/278(2.2)
|
3/55(5.5)
|
5/94(5.3)
|
7/53(13.2)
|
0.004
|
EGVB(%)
|
122/278(43.9)
|
23/55(41.8)
|
42/94(44.7)
|
36/53(67.9)
|
0.002
|
AKI/HRS(%)
|
1/278(0.36)
|
1/55(1.8)
|
2/94(2.1)
|
4/53(7.6)
|
0.003
|
L3 body composition parameters
|
L3-SMI (cm2/m2) a
|
48.48 ± 7.80
|
38.57 ± 4.35
|
46.95 ± 8.58
|
36.97 ± 6.01
|
<0.001
|
L3-IMAT (cm2/m2) a
|
2.58 ± 1.44
|
1.79 ± 1.09
|
6.09 ± 4.27
|
3.31 ± 2.24
|
<0.001
|
L3-VAT (cm2/m2) a
|
27.45 ± 18.09
|
16.76 ± 12.95
|
45.28 ± 28.14
|
28.63 ± 22.34
|
<0.001
|
L3-SAT (cm2/m2) a
|
38.50 ± 21.33
|
19.46 ± 12.59
|
49.78 ± 27.73
|
30.06 ± 22.51
|
<0.001
|
MELD: model for end-stage liver disease; RBC: red blood cell; WBC: white blood cell; PLT: blood platelet; HCT: hematocrit; TBil: total bilirubin; DBil: direct bilirubin; Scr: serum creatinine; PT: prothrombin time; INR: international normalized ratio; HE: hepatic encephalopathy; SBP: spontaneous bacterial peritonitis; UGIB: upper gastrointestinal bleeding; AKI: acute kidney injury; HRS: hepatorenal syndrome; SMI: skeletal muscle index; IMATI: intermuscular adipose tissue index; VATI: visceral adipose tissue index; SATI: subcutaneous adipose tissue index. |
a: compared with Meld Kruskal–Wallis test. The other continuous variables were determined by one-way analysis of variance (ANOVA). Categorical variables were analyzed with two-tailed χ2 tests |
Establishment of a new prognostic prediction model for liver cirrhosis and performance of the prognostic nomograms
We incorporated sarcopenia, myosteatosis and nine variables commonly used in previous non-invasive diagnostic models to select the major factors associated with the overall survival of liver cirrhosis, including sex, PLT, TBil, albumin, ALT, AST, INR, Scr, ascites grade, and so on. Univariate analysis found that TBil, albumin, INR, history of HE, ascite-grade, sarcopenia, myosteatosis had remarkable effect on the outcomes of cirrhosis (Supplementary Table 7). Of these initial 7 variables, 6 predictors were integrated into the new model for the prognostic prediction of liver cirrhosis according to the stepwise Cox regression hazard model analysis, inclulding TBil, albumin, history of HE, ascite-grade, sarcopenia and myosteatosis (Fig. 3A). The model is presented as nomograms in Fig. 3B. Six-month survival, 1-year survival and 2-year survival probability can be estimated with the nomograms. The calibration curve showed that the survival probabilities predicted by the nomogram agreed well with the actual survival probabilities (Fig. 3C-E). The area under the receiver operating characteristic curve (AUC) of the nomograms for predicting the prognosis in cirrhotic patients is 0.874(95% CI, 0.800-0.949) for 6-month survival, 0.831 (95% CI, 0.764–0.898) for 1-year survival and 0.813 (95% CI, 0.756–0.871) for 2-year survival (Fig. 3F). The AUC of Child-Pugh and MELD score for prognostic prediction were 0.857 (95% CI, 0.783–0.932), 0.767 (95% CI, 0.677–0.857) for 6-month survival, 0.820 (95% CI, 0.754–0.887), 0.718 (95% CI, 0.639–0.797) for 1-year survival, 0.802 (95% CI, 0.748–0.855), 0.726 (95% CI, 0.665–0.786) for 2-year survival. The time-ROC analysis and DCA determination both revealed that the accuracy of the new nomograms for predicting prognosis of liver cirrhosis was superior to Child-Pugh and MELD score (Fig. 4G-I).