Continued muscle loss increases mortality in cirrhosis- impact of etiology of liver disease

Abstract

Background and Aims:

Sarcopenia or skeletal muscle loss adversely affects outcomes in cirrhosis. The impact of etiology of liver disease on the severity or the rate of muscle loss is not known.

Methods.

Consecutive, well-characterized adult patients with cirrhosis due to viral hepatitis (VH), alcoholic liver disease (ALD), or nonalcoholic fatty liver disease (NAFLD) and non-diseased controls with at least 2 temporally distinct abdominal CT (computed tomography) scans were evaluated. Psoas, paraspinal, and abdominal wall muscle areas at the L3 vertebra level were quantified on the CT scans. Standardized rate of change in muscle area was expressed as change in area/100 days. Univariate and multivariable analyses were performed to identify contributors to rate of muscle loss and survival.

Results:

Among 83 cirrhotics (NAFLD n=26, ALD n=39, VH n=18) there were 20(24.1%) deaths over 62.7±41.3 months. The mean percentage change in psoas area was −0.03±0.05/100d in controls and −3.52±0.45/100d in cirrhosis(p<0.001). The mean percentage change in psoas area was −1.72±0.27/100d in NAFLD, −5.28±0.86/100d in ALD and −2.29±0.28/100d in VH. Among cirrhotics, patients with ALD had the lowest initial muscle area and most rapid rate of reduction in muscle area. Etiology of liver disease, model for end-stage liver disease (MELD) and the rate of loss of muscle area were independent risk factors for survival.

Conclusions:

Etiology of liver disease is an independent risk factor for sarcopenia with the greatest rate of muscle loss noted in ALD. Survival in cirrhosis was dependent on initial muscle mass, rate of muscle loss and MELD score.

Lay summary.

Sarcopenia or skeletal muscle loss is one of the most frequent complications in cirrhosis that adversely affects clinical outcomes including survival, quality of life and post liver transplant outcomes. We show that the cause of liver disease, especially alcohol use related cirrhosis, is associated with the lowest muscle mass and most rapid rate of muscle loss. We also show that the rate of muscle loss as well as the initial muscle mass are independently associated with death in cirrhosis.

Introduction.

Despite significant advances in the management of patients with cirrhosis, sarcopenia remains one of the most frequent clinical complications, reported in 30–70% of patients, that adversely affects outcomes including survival^1,2. Weight loss is a poor indicator of disease stage or severity especially because of the confounding effects of fluid retention, ascites, and edema^3–5. As such, muscle loss and alterations in adipose tissue mass in different compartments are difficult to identify on clinical evaluation alone. Critical evaluation of published data on nutritional status and changes in body composition show that sarcopenia is the principal component of malnutrition in cirrhosis that influences clinical outcomes⁶. Nutritional indices that primarily measure muscle mass or function in cirrhosis are worse in patients with more advanced disease². However, unlike in aging-related sarcopenia, whether sarcopenia in cirrhosis is progressive has not been established. This is relevant because cirrhosis occurs in patients who also develop age-related sarcopenia^7,8, and dissecting age-related sarcopenia from muscle loss due to liver disease has prognostic and therapeutic implications. There is conflicting data that the etiology of liver disease affects the severity of muscle loss especially in patients with alcoholic liver disease (ALD)⁹. It is also not known if the severity of liver disease affects the rate of muscle loss independent of aging-related muscle loss. This is pertinent because, of the various causes of cirrhosis, ALD and non-alcoholic fatty liver disease (NAFLD) are likely to be the major causes of liver diseases in the next decade based on modeling and liver transplant waiting list registration data^10,11. The underlying cause of liver disease can also alter skeletal muscle metabolic and signaling pathways and consequently affect muscle mass. For instance, ethanol can directly and indirectly impair muscle protein homeostasis by sensitizing the muscle to the effects of increased ammonia due to impaired ureagenesis¹². In NAFLD, in addition to the adverse effects of elevated fatty acids and insulin resistance, impaired hepatic ammonia disposal with consequent hyperammonemia also contributes to sarcopenia^13,14. These data suggest that the underlying etiologies of liver disease can also cause muscle loss, but the impact of etiology on progression of muscle loss is not known. There is also limited data on reversal of sarcopenia in cirrhosis¹⁵ and unlike other complications that reverse, sarcopenia worsens after transplantation^16,17. Therefore, establishing the rate of muscle loss will help develop interventions that delay the progression of, and potentially reverse sarcopenia in cirrhosis.

Challenges with published data on nutritional assessment, including muscle mass, in cirrhosis include the heterogeneity in quantification methods, concerns with reproducibility, terminology used and contribution of different components of malnutrition to clinical outcomes^4–6. Recent use of imaging methods to quantify skeletal muscles, specifically using cross-sectional analyses of tissue on abdominopelvic computed tomograms (CT) in cirrhosis at prespecified levels of the lumbar vertebrae, provides precise and reproducible quantification of muscle area^6,18,19. Despite the lack of protocol CT scans in the management of cirrhosis in non-research settings, temporally distinct scans performed as part of standard clinical care have been used to determine the rate of muscle loss using standardized time intervals^17,20. The present study was performed to determine if the etiology of liver disease affects the rate of muscle loss and to determine if the rate of muscle loss is an independent risk factor for survival and clinical outcomes in patients with cirrhosis.

Subjects and Methods

A retrospective study to quantify the rate of muscle loss was performed including patients with cirrhosis due to viral hepatitis (VH), ALD who were abstinent for at least 6 months (6–1054 months, median 13 months), and NAFLD who were followed in the outpatient liver clinic between January 2007 and December 2017. Cirrhosis was diagnosed by clinical, biochemical and imaging criteria and/or liver biopsy. The etiology of liver disease was established by clinical and laboratory tests. All subjects included in this study had at least 2 temporally distinct CT scans of the abdomen at least 2 months apart. The time interval of 2 months was chosen based on the minimum time interval over which muscle loss can be determined²¹. Both CT scans were of sufficient quality to perform accurate measurements of tissue area and sufficient clinical information could be retrieved from the subjects’ electronic medical records. Survival was determined from the date of the first CT to evaluate the impact of initial muscle area as well as the rate of muscle loss. Clinical characteristics were retrieved from the electronic medical records. Diabetes mellitus was diagnosed by fasting glucose and/or treatment with anti-diabetic medications. Laboratory values on the initial and final CT scan dates, number of hospitalizations between the CT scans, and the date of death and liver transplantation after the last CT scan were also documented. Child-Pugh score and model for end-stage liver disease (MELD) score were calculated on the initial and final CT scan dates using standard formulae. In those patients for whom the data were not available on the date of the CT scans, data were obtained as close as possible but no more than 8 weeks from the date of the scans. Time interval between CT scans and duration of survival from the initial scan were used for calculating rate of change of muscle/fat area and overall survival, respectively. Exclusion criteria were: documented hepatocellular carcinoma or other malignancy, ingestion of medications that affect skeletal muscle mass, advanced organ dysfunction other than cirrhosis, and prior bariatric surgery. A simultaneous cohort of age matched subjects with no documented chronic diseases who had repeated CT scans for recurrent chronic abdominal pain were included as a control group to measure the rate of muscle loss. The primary outcomes were the standardized rate of muscle loss and survival.

The studies were approved by the Institutional Review Board of the Cleveland Clinic (#14–1595) and were performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. The requirement for obtaining an informed consent was waived by the Institutional Review Board.

Measurement of muscle area

All scans were obtained as part of clinical evaluation of these patients or control subjects. For each scan, a single axial slice at the level of the upper border of the L3 vertebra was selected and analyzed using the NIH ImageJ program as previously reported by us²². Cross-sectional areas for psoas, paraspinal, abdominal wall skeletal muscle and subcutaneous and visceral adipose tissue were measured using standard Hounsfield unit ranges (adipose tissue −190 to −30 and skeletal muscle −29 to + 150). A single assessor (SD) trained and experienced in the anatomy of the tissues of interest evaluated all scans and the surface areas were quantified in square centimeters based on the demarcations made by ImageJ. The validity of these measurements using inter- and intra-observer variability of the investigator has previously been reported²². The assessor was not aware of the outcomes of the patients or the temporal distinction of the scans. Total muscle area was quantified from the sum of the psoas, paraspinal and abdominal wall muscle areas on both sides. Similarly, total fat area was calculated as the sum of the visceral and subcutaneous adipose tissue area. Sarcopenia was diagnosed by age and gender based cutoff values reported by us in our population¹⁷. Change in muscle and fat area on the temporally distinct CT scans were expressed both as absolute values as well as a percentage change, normalized per 100 days²⁰, and referred to as the standardized change in muscle area, an approach that standardized the time interval between scans to allow for comparisons across patients. We determined if the standardized change in muscle area increased or decreased in the different groups stratified by etiology or severity of liver disease and survival. There was no difference in the interpretation of the data whether the standardized absolute change or standardized percentage change in muscle area was used.

Statistical analyses

Continuous variables are presented as mean±SD unless specified. Baseline qualitative variables were compared using a chi square test and for quantitative and rating variables, the Student’s t-test or analysis of variance were used. The Spearman correlation coefficient was used to determine the association of quantitative variables. Group responses as well as subgroup analyses for different etiologies of cirrhosis were calculated. A receiver operating characteristic curve was generated to determine the optimum cutoff based on sensitivity and specificity in this cohort for the standardized rate of muscle loss with death as the outcome variable. Kaplan-Meier survival analyses with log-rank tests were performed to determine the median survival in cirrhotics stratified by Child-Pugh score, sarcopenia defined based on initial psoas muscle area, and by the standardized percentage rate of muscle loss. Statistical tests for all comparative analyses were set at 5% and Bonferroni multiple testing correction was used. Univariate analyses were performed to identify the risk factors for death and the rate of muscle loss. Significant factors were then input into a linear regression model for independent risk factors for the rate of muscle loss and a Cox proportional hazard model for risk factors for mortality. All analyses were done using SPSS v. 25 (IBM, Armonk, NY).

Results

During the study period, there were 704 patients with cirrhosis who had at least 2 CT scans and were alive at the time of follow up. Final analysis was done in 83 cirrhotic patients who satisfied the inclusion criteria and in 16 control subjects (Supplementary Fig. 1). Their clinical and demographic data are shown in Table 1. Age and body surface area at inclusion were similar in controls and patients with cirrhosis. However, among the cirrhotic patients, those with NAFLD had significantly higher body mass index (BMI) than the other groups (p<0.05 or lower) while the BMI in non-NAFLD cirrhosis was not different from that in controls. Diabetes mellitus was significantly more frequent (p<0.001) in patients with NAFLD cirrhosis (65.4%) compared to either controls (6.3%) or other causes of cirrhosis (17.5%). There was no significant difference in the presence of diabetes between non-NAFLD cirrhosis and controls. The number of hospitalizations was similar among patients with cirrhosis of different etiologies. Number of patients with ascites, encephalopathy, varices, gastrointestinal bleeding, and infections and their severity at inclusion were also similar in patients with different etiologies of cirrhosis (Table 1). There was no significant difference in the ascites score at the time of the initial CT vs. final CT, but the encephalopathy score was significantly different (p<0.001) at these times (1.3±0.5 vs.1.6±0.7 respectively). Child-Pugh and MELD scores for all cirrhotics at the time of the initial CT vs. final CT were 7.5±2.2 vs. 8.3±2.6 and 12.0±4.6 vs. 14.7±7.3, respectively, and were significantly different (p<0.01). However, the change in Child-Pugh and MELD scores and the proportion of patients who survived were similar in the 3 groups of cirrhotics. Using cutoff values from age and gender matched control subjects, the prevalence of sarcopenia was consistently higher in non-NAFLD than NAFLD subjects using any of the individual muscles evaluated or the total muscle area (Supplementary Table 1). The laboratory findings in the subjects at inclusion and change over time are shown in Table 2. Expectedly, patients with cirrhosis had lower initial serum albumin, higher INR, and higher aspartate (AST) and alanine aminotransferases (ALT) compared to controls. Patients with cirrhosis due to an underlying viral etiology had significantly higher AST compared to controls and NAFLD at inclusion. There were 16 patients with HCV and 2 patients with HBV. Of the 16 HCV patients, only 10 received treatment and 2 had viral clearance; both patients with HBV were on reverse transcriptase inhibitors and did not clear the virus. Compared to the initial values, the values for INR, serum creatinine, and AST at the final scan were significantly different only in patients with cirrhosis (p<0.05) but not controls. 5 patients had liver transplant prior to the last CT scan (close to the last CT except in 2 patients) and their clinical and muscle areas did not differ from those who were not transplanted.

Table 1.

Clinical and demographic data

	Control	Cirrhotic	Cirrhotics
			NAFLD	Alcoholic	Viral
Number	16	83	26	39	18
Male (%)	5 (31.3)	59 (71.1)**	14 (53.8)	29 (74.4)†	16 (88.9)
Age (years)	50.1 ± 11.5	52.7 ± 9.5	57.3 ± 8.6e	50.3 ± 10.2b	51.2 ± 7.1
BMI (kg/m2)	30.2 ± 6.7	31.6 ± 8.3	36.8 ± 8.9fh	29.5 ± 7.9c	28.7 ± 3.9b
Body Surface Area (m2)	2.0±0.3	2.1 ± 0.4	2.2 ± 0.3	2.1 ± 0.4	2.1 ± 0.3
Sarcopeniaš (initial)	0 (0)	70 (84.3)***	13 (50)‡	39 (100)	18 (100)
Dead	1(6.3)	20(24.1)	8(30.8)	7(26.9)	5(27.8)
MELD (initial)	n/a	12.0±4.6	11.0±4.0	12.5±5.6	12.2±2.4
Child Pugh score (initial)	n/a	7.5±2.2	7.1±2.3	7.7±2.3	7.4±1.9
Co-morbidities
Diabetes (%)	1(6.3)	27(32.5)*	17(65.4)‡	7(17.9)	3(16.7)
Insulin		14 (51.9)	7 (41.2)	5 (71.4)	2 (66.7)
Oral hypoglycemics		3 (11.1)	3 (17.6)	0	0
ESRD on HD (%)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)
Hospitalizations	4.3 ± 5.1	7.3 ± 7.3	8.1 ± 9.7	6.5 ± 6.0	7.9 ± 6.2
Duration of follow up (interval between CT in months)	108.3 ± 34.6	62.7 ± 41.3***	61.7 ± 41.3	62.4 ± 40.2	64.7 ± 45.8
Duration of cirrhosis (years)		2.1 ± 1.9	2.0 ± 1.9	1.8 ± 2.3	2.7 ± 3.1
Complications of cirrhosis (initial)
Ascites (%)
Absent	n/a	46 (55.4)	17 (65.4)	18 (46.2)	11 (61.1)
Slight	n/a	17 (81.6)	3 (23.1)	11 (28.2)	3 (16.7)
Moderate	n/a	20 (24.1)	6 (23.1)	10 (25.6)	4 (22.2)
Hepatic encephalopathy (%) No	n/a	45 (54.2)	21 (80.8)	28 (71.8)	12 (66.7)
Grade 1–2	n/a	19 (22.9)	5 (19.2)	8 (20.5)	6 (33.3)
Grade 3–4	n/a	2 (2.4)	0 (0)	3 (7.7)	0 (0)
Varices grade initial	n/a	1.1 ± 1.1	1.1 ± 1.1	1.1 ± 1.2	1.3 ± 1.1
Varices grade final		0.9 ± 1.0	0.9 ± 1.1	0.9 ± 0.9	1 ± 1
Number of GI bleeds		0.5 ± 0.8	0.4 ± 0.7	0.7 ± 1.0)	0.4 ± 0.6
GI bleed (number)			8 (30.8)	16 (41.0)	6 (33.3)
Infection: SBP		15 (46.9)	2 (18.2)	7 (53.8)	6 (75)
Infection: sepsis		5 (15.6)	3 (27.3)	1 (7.7)	1 (12.5)
Infection: UTI		4 (12.5)	2 (18.2)	2 (15.4)	0
Infection: other		8 (25)	4 (36.4)	3 (23.1)	1 (12.5)

^šsarcopenia defined by initial psoas muscle area less than 5^th percentile of age and gender matched control values. NAFLD: Non-alcoholic fatty liver disease; BMI: Body Mass Index; MELD: Model for End-stage Liver Disease; ESRD on HD: End-stage renal disease on hemodialysis

Figures in parentheses are %

Cirrhotic vs. control:

^*p<0.05,

^**p<0.010,

^***p<0.001

vs other cirrhotic groups:

^†:p<0.05,

^‡:p<0.001

vs. NAFLD:

^a:p<0.05;

^b:p<0.01;

^c:p<0.001

vs. Alcoholic:

^d:p<0.05;

^e:p<0.01;

^f:p<0.001

vs. Viral:

^g:p<0.05;

^h:p<0.01;

^i:p<0.001

Table 2.

Biochemical test results

		Cirrhotic		Cirrhotics
	Control	Initial	Final	NAFLD	Alcoholic	Viral
Number	16	83	83	26	39	18 (initial)
Sodium (mmol/L)	138.4 ± 2.3	136.4 ± 4.9	136.3 ± 5.4	137.9±3.5	135.6 ± 6.1	136.1 ± 3.3
S. creatinine (mg/dL)	0.9 ± 0.2	0.9 ± 0.3	1.3 ± 1.0###	0.9±0.4	0.8 ± 0.3	1.0 ± 0.4
Albumin (g/dL)	4.1 ± 0.5	3.4 ± 0.7***	3.3 ± 0.9	3.3±0.8	3.4 ± 1.0	3.0 ± 0.7
T. bilirubin (mg/dL)	0.6 ± 0.5	2.3 ± 2.8	1.8 ± 0.9	1.7±1.4	2.9 ± 3.9	1.9 ± 0.8
AST (u/L)	21.4 ± 7.8	76.0 ± 49.3***	58.3 ± 48.6##	56.1±21.2h	76.6 ± 48.2	103.5 ± 66.7b
ALT (u/L)	20.3 ± 10.6	51.3 ± 39.9***	43.6 ± 35.2	48.5±26.1	44.3 ± 33.8	70.5 ± 60.1
INR	1.0 ± 0.1	1.3 ± 0.3*	1.4 ± 0.4#	1.3±0.2	1.3 ± 0.4	1.3 ± 0.2
Δ Sodium	1.0 ± 4.4	0.3 ± 3.6		0.6 ± 1.9	1.0 ± 4.8	−0.1 ± 1.5
Δ S. creatinine	−0.1 ± 0.2	0.1 ±0.2		0.0 ± 0.1	0.1 ± 0.3	0.0 ± 0.1
Δ Albumin	−0.1 ± 0.2	0 ± 0.4		0.0 ± 0.4	0.1 ± 0.5	0.0 ± 0.2
Δ T. bilirubin	−0.1 ± 0.3	0.4 ±1.9		0.4 ± 1.2	0.5 ± 2.4	0.5 ± 1.4
Δ AST	−2.4 ± 6.7	−9.2 ± 21.2		−3.4 ± 6.5	−12.2 ± 25.3	−11.1 ± 24.2
Δ ALT	−2.0 ± 7.4	7.1 ± 30.5		3.6 ± 17.1	12.5 ± 40.3	0.6 ± 17.2
Δ INR	0.0 ± 0.3	0.0 ± 0.3		0.0 ± 0.2	0.0 ± 0.3	0.0 ± 0.3
Δ Platelets (x1000)		−2.8 ± 21.2		1.3 ± 11.5	−6.6 ± 28.8	−0.6 ± 7.6

NAFLD: Non-alcoholic fatty liver disease; S. creatinine: serum creatinine; T. bilirubin: total bilirubin; AST: Aspartate aminotransferase; ALT: Alanine aminotransferase; INR: International normalized ratio

Δ values represent change over 100 days to normalize for different time intervals between initial and final values.

Cirrhotic vs. control:

^*p<0.05,

^**p<0.010,

^***p<0.001

Cirrhotic initial vs. cirrhotic final:

^#p<0.05,

^##p<0.010,

^###p<0.001

vs. NAFLD:

^a:p<0.05;

^b:p<0.01;

^c:p<0.001

vs. Alcoholic:

^d:p<0.05;

^e:p<0.01;

^f:p<0.001

vs. Viral:

^g:p<0.05;

^h:p< 0.01;

^i:p<0.001

The initial and final muscle areas and fat areas in the different groups are shown in Table 3. The mean interval between CTs was similar in the different groups (control, cirrhosis of different etiologies). The initial psoas, paraspinal, abdominal wall, and total muscle areas were significantly lower in cirrhosis compared to controls (p<0.001). The measured values correlated highly with the areas normalized to height² (r²=0.625–0.978, p<0.001 for all correlations) and the differences between groups were similar whether the area or normalized area was used (Supplementary Table 2). No differences were noted in the initial areas between males and females. Muscle area at the end of follow-up was also significantly lower in cirrhosis than controls (p<0.001). Patients with alcoholic cirrhosis had the lowest (p<0.001) muscle areas on the initial and on the final scan compared to those in patients with viral hepatitis or NAFLD cirrhosis. Visceral, subcutaneous and total abdominal fat areas on the initial and final CT scans were significantly higher in cirrhosis compared to controls. Amongst patients with cirrhosis, those with NAFLD had significantly higher (p<0.001) fat areas (visceral, subcutaneous and total) on the initial and final scans compared to those with non-NAFLD etiologies. No differences were noted among patients who received treatment vs. those who did not receive treatment for VH.

Table 3.

Initial and final muscle and fat mass

	Control	Cirrhotic	Cirrhotics
			NAFLD	Alcoholic	Viral
Number	16	83	26	39	18
Interval between initial and final CT (months)	36.67 ± 34.48 (2–109; 31.5)	36.95 ± 30.43 (6–123; 31.3)	39.19 ± 29.77 (2.4–114; 29.7)	35.18 ± 28.51 (2–123; 32.6)	37.56 ± 36.46 (2–123; 28.7)
Initial psoas area	44.78 ± 4.99	27.56 ± 4.15***	32.02 ± 3.42fi	24.45 ± 2.24ci	27.85 ± 1.47cf
Final psoas area	44.72 ± 5.59	20.77 ± 7.13***	27.43 ± 5.35fi	16.04 ± 4.41ci	21.40 ± 6.30cf
Initial paraspinal area	61.72 ± 6.04	44.52 ± 6.05***	50.58 ± 5.09fi	39.81 ± 2.91ci	45.98 ± 3.12cf
Final paraspinal area	61.39 ± 6.84	37.41 ± 9.06***	45.72 ± 6.7fh	30.95 ± 5.87ci	39.40 ± 6.89bf
Initial abdominal wall area	74.64 ± 6.75	49.21 ± 8.84***	57.55 ± 6.21ij	41.88 ± 4.27ci	53.06 ± 5.83af
Final abdominal wall area	74.05 ± 6.77	42.20 ± 11.62***	52.82 ± 8.07ij	33.16 ± 6.01ci	46.45 ± 9.78af
Initial total muscle area	181.14 ± 11.99	121.29 ± 18.03***	140.15 ± 12.55fi	106.14 ± 8.07ci	126.88 ± 9.32cf
Final total muscle area	180.16 ± 14.01	100.38 ± 26.96***	125.97 ± 18.53fh	80.14 ± 15.22ci	107.24 ± 22.18bf
Initial visceral fat area	182.27 ± 24.76	223.77 ± 57.97***	303.32 ± 24.73fi	187.77 ± 15.07c	186.85 ± 26.56c
Final visceral fat area	188.04 ± 22.78	216.73 ± 55.42**	293.61 ± 21.58fi	180.71 ± 13.16c	183.72 ± 23.9c
Initial subcutaneous fat area	194.50 ± 17.58	215.62 ± 28.73***	241.65 ± 23.7fi	206.24 ± 23.45c	198.33 ± 19.23c
Final subcutaneous fat area	197.55 ± 17.44	213.70 ± 37.68**	246.52 ± 24.79fi	197.41 ± 36.76c	201.62 ± 22.65c
Initial total fat area	376.77 ± 33.61	439.13 ± 76.84***	544.16 ± 28.79fi	394.01 ± 26.40c	385.18 ± 32.85c
Final total fat area	385.59 ± 30.31	430.69 ± 81.48***	540.93 ± 27.60fi	378.12 ± 34.70c	385.34 ± 33.27c

All areas are in cm². Total muscle area was calculated as the sum of the psoas, paraspinal, and abdominal wall muscle areas on both sides. NAFLD: Non-alcoholic fatty liver disease; CT: Computed tomography scan

Figures in parenthesis are range;median

Cirrhotic vs. control:

^*p<0.05,

^**p<0.010,

^***p<0.001

vs. NAFLD:

^a:p<0.05;

^b:p<0.01;

^c:p<0.001

vs. Alcoholic:

^d:p<0.05;

^e:p<0.01;

^f:p<0.001

vs. Viral:

^g:p<0.05;

^h:p<0.01;

^i:p<0.001

As mentioned, to adjust for the variability in interval between the CT scans, we compared the standardized absolute and percentage change in muscle and fat areas over a 100-day period (Fig. 1, Supplementary Table 3) and both the absolute and percentage changes showed similar trends in each of the groups evaluated. The standardized rate of muscle loss was greater (p<0.001) in patients in all 3 groups of cirrhosis compared with that in controls. Amongst patients with cirrhosis, those with ALD had the highest standardized rate of muscle loss compared to that in NAFLD and VH related cirrhosis. There was significantly greater reduction in standardized change in psoas muscle area in patients with ALD compared to those with VH (p<0.01) and NAFLD (p<0.05). There was no difference in the standardized rate of change in any of the muscle areas between VH and NAFLD cirrhosis. The standardized rate of change for the different muscle groups (psoas, paraspinal or abdominal wall muscle area) was similar within each of the disease categories and in controls with no advantage to using one or more muscle groups for our analyses. Similar to the reduction in muscle area, patients with cirrhosis had reduction in standardized change in visceral, subcutaneous and total fat areas compared to controls (p<0.05 or less). Only the standardized subcutaneous fat area was increased over time in NAFLD compared to the other causes of cirrhosis.

Figure 1. Rate of muscle loss was greater in cirrhosis.

Panel A. Representative scans of a patient with alcoholic cirrhosis on initial and final scan. Panel B. Standardized percentage change in psoas muscle area. Patients with cirrhosis had a greater standardized rate of psoas muscle area compared to age matched controls. Amongst cirrhotics, patients with alcoholic cirrhosis had a greater standardized rate of change in psoas muscle area. Panel C. Standardized percentage change in total muscle area in cirrhosis was greater than that in controls. Amongst cirrhotics, patients with alcoholic cirrhosis had a greater standardized rate of change in total muscle area. *** p<0.001; ** p<0.01; * p<0.05. A alcoholic cirrhosis, CIR cirrhosis, CTL control, N cirrhosis due to non-alcoholic steatohepatitis, V viral hepatitis related cirrhosis. All data mean±SEM.

Univariate analyses of differences between patients with cirrhosis who survived or those who died over the follow up period are shown in Table 4. Patients with cirrhosis who had diabetes mellitus had a significantly higher mortality compared to those without (p<0.05) but there was no difference in muscle area between patients with and without diabetes mellitus or those who were treated with insulin or oral hypoglycemic agents. Initial serum sodium and albumin were lower and bilirubin higher in patients with cirrhosis who died compared to those who were alive at follow up (p<0.05). Consistently, Child-Pugh at the time of initial (p<0.05) and final scans (p<0.01) were significantly higher in patients who died, but the change in Child-Pugh score was not different between survivors and non-survivors. Interestingly, ascites and hepatic encephalopathy prior to the first scan were not different between cirrhotics who were alive or died. Expectedly, patients who died had a shorter duration of follow-up (p<0.001) compared with those who survived.

Table 4.

Survival by lab values and physical exam

	Cirrhotics
	Alive	Dead
Number	63	20
Age (years)	51.36 ± 9.6	56.87 ± 8.3*
Gender (M:F)	45:18	14:6
BMI (m2/kg)	31.75 ± 8.72	31.19 ± 6.77
Complication or comorbidity at initial CT
Diabetes	16	11*
Ascites	24	13
Encephalopathy	17	5
Initial laboratory values
Sodium (mmol/L)	137.1 ± 4.3	134.4 ± 6.2*
S. creatinine (mg/dL)	0.87 ± 0.35	0.91 ± 0.27
Albumin (g/dL)	3.47 ± 0.74	3.10 ± 0.44*
T. bilirubin (mg/dL)	1.93 ± 1.73	3.41 ± 4.79*
ALT (u/L)	48.2 ± 34.6	61.1 ± 53.1
AST (u/L)	73.2 ± 44.4	84.8 ± 62.7
INR	1.30 ± 0.30	1.32 ± 0.32
MELD (initial)	11.6 ± 4.1	13.1 ± 5.8
MELD (final)	13.2 ± 6.0	19.5 ± 9.0**
Δ MELD	1.5 ± 6.3	6.5 ± 7.9**
Δ MELD/100d	0.01 ± 1.45	2.25 ± 4.17***
Child-Pugh (initial)	7.2 ± 2.2	8.4 ± 1.9*
Child-Pugh (final)	7.8 ± 2.4	9.8 ± 2.8**
Δ Child-Pugh	0.65 ± 2.89	1.40 ± 1.88
Δ Child-Pugh/100d	0.06 ± 0.67	0.39 ± 1.55
Number of admissions	7.6 ± 7.6	6.3 ± 6.5
Duration of follow-up (months)	72.0 ± 41.0	33.4 ± 26.4***

BMI: Body mass index(kg/m²); CT: Computed tomography scan; T. bilirubin: total bilirubin; S. creatinine: serum creatinine; INR: International Normalized Ratio; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; Δ: Change in; MELD: Model for End-stage Liver Disease;/100d: per 100 days

^*p<0.05,

^**p<0.01,

^***p<0.001

The different muscle and fat areas on the initial and final scan as well as the rate of change of muscle areas (p<0.001 for all groups) and fat areas (visceral fat p<0.001; total fat area p<0.01) in survivors and non-survivors among patients with cirrhosis are shown in Table 5. Muscle areas on the initial scan and final scan were significantly lower (p<0.05 or lower) in patients with cirrhosis who died compared to those who were alive at follow up. In contrast, visceral and subcutaneous fat areas on the initial and final scans were similar between survivors and non-survivors amongst patients with cirrhosis. The standardized rates of loss of muscle area and visceral fat area were greater (p<0.001) in cirrhotics who died compared to those who were alive at follow-up.

Table 5.

Survival by measures of body composition

	Cirrhotics
	Alive	Dead
Number	63	20
Interval between initial and final CT (months)	42.0 ± 31.8	21.1 ± 18.5**
Initial psoas area	28.24 ± 4.32	25.41 ± 2.70**
Final psoas area	21.67 ± 7.43	17.93 ± 5.29*
Δ psoas area/100d	−0.58 ± 0.28	−1.80 ± 1.33***
%Δ psoas area/100d	−2.2 ± 1.25	−7.66 ±6.71***
Initial paraspinal area	45.38 ± 6.15	41.83 ± 4.91*
Final paraspinal area	38.63 ± 9.20	33.57 ± 7.60*
Δ paraspinal area/100d	−0.59 ± 0.28	−1.98 ± 1.38***
%Δ paraspinal area/100d	−1.37 ± 0.75	−5.07 ± 4.11***
Initial abdominal wall area	50.43 ± 8.97	45.38 ± 7.36*
Final abdominal wall area	43.69 ± 12.04	37.51± 8.92*
Δ abdominal wall area/100d	−0.58 ± 0.32	−2.01 ± 1.54***
%Δ abdominal wall area/100d	−1.25 ± 0.81	−4.88 ± 4.46***
Initial total muscle area	124.05 ± 18.39	112.61 ± 13.97*
Final total muscle area	103.99 ± 27.89	89.00 ± 20.45*
Δ total muscle area/100d	−1.75 ± 0.86	−5.79 ± 4.19***
%Δ total muscle area/100d	−1.51 ± 0.87	−5.57 ± 4.78***
Initial visceral fat area	217.15 ± 53.57	244.61 ± 67.35
Final visceral fat area	212.19 ± 53.35	231.04 ± 60.68
Δ visceral fat area/100d	−0.42 ± 0.60	−4.09 ± 5.87***
%Δ visceral fat area/100d	−0.21 ± 0.32	−1.63 ± 2.39***
Initial subcutaneous fat area	213.69 ± 28.61	221.69 ± 28.97
Final subcutaneous fat area	210.51 ± 39.11	223.76 ± 31.56
Δ subcutaneous fat area/100d	−0.34 ± 2.10	−0.54 ± 6.10
%Δ subcutaneous fat area/100d	−0.18 ± 1.02	−0.29 ± 2.96
Initial total fat area	430.50 ± 72.59	466.30 ± 85.23
Final total fat area	423.03 ± 80.36	454.79 ± 82.31
Δ total fat area/100d	−0.73 ± 2.32	−4.63 ± 8.91**
% Δ total fat area/100d	−0.19 ± 0.56	−1.08 ± 2.18**

All areas are in cm². Total muscle area was calculated as the sum of the psoas, paraspinal, and abdominal wall muscle areas on both sides. Total fat area was calculated as the sum of the visceral and subcutaneous fat areas. CT: Computed tomography scan; Δ: Change in;/100d: per 100 days; MELD: Model for End-stage Liver Disease; INR: International Normalized Ratio; AST: Aspartate aminotransferase; ALT: Alanine aminotransferase

^*p<0.05,

^**p<0.01,

^***p<0.001

Univariate analyses of data between cirrhotic patients with and without sarcopenia defined by the initial psoas area as less than the 5^th percentile of controls are shown in Supplementary Table 4. Patients with sarcopenia were younger (p<0.01), had lower BMI (p<0.05) and had non-NAFLD etiology of cirrhosis (p<0.001). There were no differences in patients with hepatic encephalopathy or ascites amongst those with and without sarcopenia. The rate of muscle and fat loss (both the absolute and percentage standardized change) were greater (p<0.001) in patients who were diagnosed with sarcopenia on the initial scan (Supplementary Table 5). Clinical, laboratory and body composition characteristics that correlated with the initial psoas muscle area and standardized percentage change in muscle area are shown in Supplementary Table 6. Serum sodium, other groups of muscle areas, and visceral and total fat area were positively correlated with the initial psoas muscle area and the standardized percentage change in psoas area. An inverse correlation was observed between the initial and standardized percentage change in psoas muscle area and the ascites score, total bilirubin and initial MELD and Child-Pugh scores.

Kaplan-Meier survival analysis (Fig 2,Panel A) showed that mean survival was significantly lower in sarcopenic patients compared to those without sarcopenia based on the 5^th percentile of the initial psoas muscle area (p=0.003). Similar observations were made when patients were stratified by either the 20th percentile of the initial psoas muscle area or the total muscle area (Supplementary Table 7). Using a receiver operating characteristic curve, the patients were stratified for standardized rate of change of psoas muscle area (Fig 2, Panel B). Kaplan-Meier survival analysis showed a significantly lower (p=0.001) survival in patients with greater than 3.66%/year loss of psoas muscle area (Fig 2, Panel C). Analyses for total muscle area showed similar observations (data not shown). Patients stratified by Child-Pugh score showed significantly better survival (p<0.005) in those in class A compared to those in classes B and C (Fig 2, Panel D). Using linear regression, variables that associated with the standardized change in psoas muscle area are shown in Table 6. A. Independent factors associated with the standardized change in psoas muscle area included etiology of cirrhosis, gender, MELD score at the time of the initial scan and the initial psoas area. The same variables were also significant when the standardized percentage change in psoas muscle area was used as the dependent variable. Use of standardized change (or percentage change) in total muscle area as the dependent variable yielded results similar to those noted with the psoas muscle area above. Cox regression analysis for survival showed that etiology of liver disease, initial MELD score, initial psoas muscle area and the standardized change in psoas muscle area were associated with mortality (Table 6. B). Similar observations were made if, instead of the change in psoas muscle area, the standardized change in total muscle area was used as the independent variable. Interestingly, both initial muscle area and the standardized change in muscle area (psoas or total muscle area) were fit into the model for predicting survival.

Figure 2. Survival analysis of patients with cirrhosis.

Panel A. Kaplan-Meier survival curve for cirrhotic patients stratified by Child-Pugh scores. Mean(SEM) survival (months) was 125.1(5.5) for Child-Pugh A, 105.7(11.5) for Child-Pugh B, and 89.1(18.0) for Child-Pugh C patients. Log rank test Child-Pugh A vs. B,C p=0.005.

Panel B. Receiver operating characteristic curve to determine the optimum cutoff value for standardized percentage change of psoas muscle area with death as outcome/independent categorical variable.

Panel C. Kaplan-Meier survival curve for cirrhotic patients stratified by sarcopenia using a cutoff of 5th percentile of the initial psoas muscle area. Mean(SEM) survival (months) was 172.6(4.3) for those without sarcopenia vs. 112.5(7.8) for those with sarcopenia. Log rank test p=0.003.

Panel D. Kaplan-Meier survival curve for cirrhotic patients stratified by standardized percentage change of psoas muscle area. Mean (SEM) survival (months) was 31.22(10.55) vs. 132.57(6.14) for patients with >3.66%/y vs. those with < 3.66%/y; log rank test p<0.001.

Table 6. A.

Independent predictors of rate of muscle loss

Model	Unstandardized beta	Coefficients standard error	Standardized coefficients beta	t	Sig
Constant	−3.942	0.774		−5.227	0.000
Etiology of cirrhosis	0.311	0.120	0.261	2.593	0.011
Psoas muscle area (initial)	0.123	0.021	0.589	5.959	0.000
Gender	−0.420	0.171	−0.222	−2.454	0.016
MELD (initial)	−0.050	0.017	−0.266	−3.002	0.004
Dependent variable is change in psoas area/100 days.

Table 6. B.

Cox proportional hazard for mortality

Model	Beta	SE	Wald	df	Sig	Exp (beta)
Etiology of cirrhosis
NALFD vs. Alcoholic			27.266	2	0.000
NAFLD vs. Viral	1.687	0.617	7.474	1	0.006	5.404
Viral vs. Alcoholic	−4.33	1.081	16.143	1	0.000	0.013
Psoas area (initial)	−0.526	0.147	12.808	1	0.000	0.591
Δ psoas area/100d	−0.973	0.246	15.596	1	0.000	0.378
MELD (Initial)	0.131	0.063	4.312	1	0.038	1.140

All areas are in cm². Δ: Change in;/100d: per 100 days; MELD: Model for End-stage Liver Disease; NAFLD: Non-alcoholic fatty liver disease

Discussion

Low muscle area on imaging, a measure of whole-body muscle mass, has been consistently shown to adversely predict survival in cirrhotics^1,23,24. Using image analysis of CT scans, we noted that nearly all patients with cirrhosis had sarcopenia using age and gender-based criteria reported by us¹⁷. Patients with cirrhosis had a greater rate of loss of muscle mass compared to age matched controls. The rate of muscle loss was associated with shorter survival in cirrhosis and, of the major causes of cirrhosis, patients with alcoholic cirrhosis had the lowest muscle area for comparable severity of liver disease at entry into the study and also the greatest standardized rate of muscle loss. Etiology of liver disease, initial muscle mass, gender and MELD score were independent risk factors for the rate of muscle loss. Mortality in cirrhosis was higher in patients with lower initial muscle mass, greater rate of muscle loss and higher initial MELD score.

Muscle and fat area measured on single abdominal cross-sectional images have been used to quantify whole body skeletal muscle and adipose tissue mass²⁵. We and others have reported that low muscle area in cirrhosis is associated with increased mortality, but the impact of the etiology of liver disease on muscle loss is not well characterized^1,9,19. There are also conflicting views about reporting absolute muscle area or those normalized using different denominators²⁶. However, we observed that there were no differences in interpretation of our data when absolute muscle area or that normalized to height was used. Anthropometric measures of lean body mass or muscle area have shown that patients with alcoholic and cholestatic diseases have the most severe muscle loss for similar severity of liver disease^6,27. There have been conflicting reports that patients with ALD have more severe muscle loss than patients with other etiologies of liver disease⁹. In the present study, patients with ALD had significantly lower muscle mass on the initial evaluation and a greater rate of muscle loss compared to non-ALD related liver disease. Potential reasons for the greater muscle loss in ALD could include continued alcohol consumption; however, this is unlikely in the present study because these patients were self-reportedly abstinent and the change in laboratory measures of the severity of liver disease was not different in the three groups of cirrhotics, as a more rapid deterioration in laboratory tests would be expected in patients who continued to drink. Other potential explanations for our observations of greater severity of muscle loss in ALD could be due to ethanol-induced sensitization of the skeletal muscle to hyperammonemia, as reported in preclinical models, or epigenetic changes, but these need to be evaluated in human patients¹². Even though patients with NAFLD have greater whole-body weight, their muscle mass has been reported to be lower than that of controls but whether the severity of muscle loss is similar to that in other liver diseases has not been systematically evaluated²². Using cutoff values for sarcopenia in a population similar to that in the present study¹⁷, the prevalence of sarcopenia was lower in NAFLD than in non-NAFLD cirrhosis. This may be due to the relative preservation of muscle mass in these patients or relatively less severe underlying disease. However, in this study, measures of the severity of liver disease were similar to those in non-NAFLD patients suggesting relative preservation of muscle mass in NAFLD cirrhosis compared to other causes of cirrhosis. We also noted that unlike that in ALD, the rate of muscle loss was similar in patients with VH and NAFLD cirrhosis, suggesting that the reason for lower prevalence of sarcopenia in NAFLD needs to be evaluated further.

In cross-sectional studies in cirrhotic patients, muscle area has been correlated to survival¹. However, there are very limited data evaluating the temporal course of muscle area in cirrhosis. Our use of a contemporaneous group of age-matched subjects without documented disorders that alter muscle mass showed that the standardized rate of muscle loss (either absolute or percentage) in cirrhosis of any etiology is significantly greater than that due to age alone. Unlike earlier reports²⁸, in the present study we observed that all 3 muscle groups studied had similar predictive value for outcomes. We therefore believe that the psoas muscle area alone can be used as an indicator of muscle mass due to the ease of measurement. In addition to the severity of the underlying liver disease and low muscle mass, we also observed that continued muscle loss is an independent risk factor for mortality, suggesting that it may be possible to improve clinical outcomes even in sarcopenic patients by targeting ongoing muscle loss. Others have reported that malnutrition, diagnosed primarily using measures of muscle mass, is associated with a greater risk of complications of cirrhosis²⁹ but we did not find an increased prevalence of ascites or encephalopathy in sarcopenic cirrhotic patients. Also, similar to that reported in longitudinal studies of aging-related sarcopenia in which women have a lower rate of muscle loss⁷ and in studies of cirrhotics, a greater muscle loss in males³⁰, on multivariate analysis, men had a greater rate of muscle loss even though the initial muscle mass was similar between the genders. We also noted that cirrhotics with sarcopenia had a greater rate of muscle loss which may have resulted in the lower muscle mass on the initial scan. Continued muscle loss despite becoming sarcopenic may suggest a failure of adaptive mechanisms to slow the rate of muscle loss and the lack of any interventions to reverse muscle loss. Hence, targeting muscle loss in patients with established sarcopenia may be beneficial.

A multivariable analysis of the risk factors for outcome showed that survival in cirrhotics was independently associated with the underlying etiology of liver disease, the initial muscle area, the rate of muscle loss (measured as the standardized change in muscle area), and the severity of underlying liver disease. These data suggest that adding both static as well as dynamic measures of muscle area are independently and robustly associated with mortality in cirrhosis and may be considered for prognostic models³¹.

Limitations of the present studies include the retrospective nature, biases in inclusion, and lack of evaluation of the muscle quality. As mentioned earlier, since protocol CT scans are generally not performed for evaluation of muscle mass in cirrhosis, and other methods for estimation of body composition are not as robust as imaging to differentiate muscle from non-muscle components¹⁸, the present studies provide evidence that a dynamic measurement of muscle mass may be an important component of clinical evaluation in patients with cirrhosis and potentially other chronic diseases. However, given the radiation exposure and costs of CT scans, alternative reliable methods for quantification of muscle mass are required to determine the rate of muscle mass as an indicator of outcomes in cirrhotics. Another concern is the modeling used for quantifying the standardized rate of muscle loss depended only on two points in time. Change in muscle mass is a dynamic process given that both protein synthesis and proteolysis contribute to the overall muscle mass⁶, however, we chose to use the two most distant temporal scans because integrating multiple data points at varying intervals in different patients may yield disparate data unless the sample size is very large. Even though the study duration was long, the number of patients included is modest because we excluded a number of subjects to generate a contemporaneous cohort with similar standard of care and image quality while avoiding confounding factors that could affect muscle mass independent of the severity and etiology of liver disease. Furthermore, only patients with the most frequent causes of cirrhosis were included and it is possible that patients with other causes including metabolic and cholestatic diseases may have a different pattern of muscle loss²⁷. The relatively lower proportion of patients with VH in the study is also a limitation and may have been related to differences in follow-up but we are unable to identify the specific reason for the distribution patent of patients in each category in this retrospective study. In addition to muscle mass, quality of muscle is increasingly recognized as a contributor to muscle function. There have been reports that muscle density on imaging may reflect muscle quality¹⁵ but the validity and clinical relevance of this is yet to be established and will be part of future studies. Finally, despite the necessarily observational nature of these studies, our data provide potential mechanistic insights thereby laying the foundation for future studies dissecting the cause(s) of etiological differences in muscle loss in cirrhosis.

In conclusion, we show that patients with alcoholic cirrhosis have a more rapid rate of muscle loss compared to other etiologies of liver disease and a dynamic measure of muscle loss is of clinical relevance while planning interventions to prevent or treat sarcopenia. We also observed that the rate of muscle loss is independently and robustly associated with mortality and outperformed any single measure of muscle area. These observations suggest that even if muscle mass does not increase in response to interventions, approaches to reduce the rate of muscle loss have the potential to retain muscle mass in cirrhosis and potentially other chronic diseases.

List of Abbreviations:

ALD

Alcoholic liver disease

NAFLD

Non-alcoholic fatty liver disease

VH

viral hepatitis

CT

Computed tomogram or computed tomogram scan

MELD

Model for End-stage Liver Disease