ABSTRACT

PURPOSE

Computed tomography (CT)-based body composition parameters and the hepatic venous pressure gradient (HVPG) are key characteristics in patients with liver cirrhosis. The present study aims to explore the correlation between CT-based body composition parameters and HVPG, as well as the difference in HVPG between patients with and patients without sarcopenia.

METHODS

A literature search for studies reporting the correlation between HVPG and CT-based body composition parameters published in English up to August 2023 in four databases, Embase, MEDLINE (via PubMed), Web of Science, and Cochrane Library, was conducted. The correlation coefficient between HVPG and CT-based body composition parameters was the primary outcome, and the difference in the HVPG value between the sarcopenia and non-sarcopenia groups was the secondary outcome. A meta-analysis was conducted using a random-effects models. The methodologic quality was assessed using the Quality Assessment of Diagnostic Studies instrument.

RESULTS

A total of 652 articles were identified, of which nine studies (n = 1,569) met the eligibility criteria. Among them, seven studies reported the primary outcome via the muscle index, five via the skeletal muscle index (SMI), two via the psoas-muscle-related index (PRI), and three via two adipose tissue indexes. A total of five studies reported the secondary outcome: four via SMI and one via PRI. No evidence of a significant correlation was determined between the various body composition parameters and the HVPG value, either in the muscle index or the adipose tissue index. Higher HVPG values were observed in patients with sarcopenia than in patients without sarcopenia [pooled standardized mean difference (SMD): 0.628 (−0.350, 1.606), P < 0.001; I² = 92.8%; P < 0.001] when an Asian sarcopenia definition was adopted. In contrast, when a Western cut-off value was applied, the HVPG value was higher in patients without sarcopenia than in patients with sarcopenia [pooled SMD: −0.201 (−0.366, −0.037), P = 0.016; I² = 0.00%; P = 0.785].

CONCLUSION

No sufficient evidence regarding a correlation between the CT-based body composition and HVPG value was discovered. The difference in the HVPG value between the sarcopenia and non-sarcopenia groups was likely dependent on the sarcopenic cut-off value.

Sarcopenia, a disease entity representing a progressive and generalized skeletal muscle disorder, is a prevalent morbidity of liver cirrhosis (LC).¹ Due to the concomitant altered catabolic state, insulin resistance, chronic systemic inflammation and physical inactivity, sarcopenia exists in different LC stages and is closely related with decompensation risk and postoperative complications, as well as mortality independent of commonly used tools, such as Child–Pugh score or the model for end-stage liver disease (MELD) score.^2-4 Furthermore, the role of adipose quantity or distribution as a precipitating event for poor prognosis in patients with LC has also been proposed.^{5, 6} Importantly, as two body phenotypes, the muscle and adipose quantity may interact with each other instead of acting as two independent pathophysiological conditions.⁷

Computed tomography (CT) is considered the gold standard for assessing muscle or adipose quantity, and CT-based muscle quantity is recommended for defining sarcopenia.^{8, 9} In patients with LC, CT is routinely performed with the aim of monitoring portal-systemic collaterals and tumor development or recurrence; thus, CT-based body composition parameters are accessible and reproducible. In addition, the hepatic venous pressure gradient (HVPG) is recognized as the gold standard for evaluating portal hypertension (PH).¹⁰ To stratify the risk of decompensation with intent for early intervention, HVPG measurement has also been encouraged in patients with LC in real-life practice.¹¹

Body composition, especially muscle quantity, and HVPG have been characterized as important characteristics in patients with LC. With the progress of LC, clinically significant PH is concomitant. Muscle depletion and fat accumulation or redistribution also likely occur in this course.^{1, 12} Specifically, the metabolism changes of such a population are characterized by insulin resistance, dysregulated muscle protein turnover, and altered lipid redistribution.¹³ Furthermore, some clinical events, such as loss of appetite, fluid retention, and sedentary behavior, contribute to alterations of the body phenotype. A large sample cross-sectional study revealed that muscle mass depletion was independently associated with the liver fibrosis stage.¹⁴ In addition, a preclinical study showed that ammonia-lowering therapy could result in an increase of skeletal muscle mass.¹⁵ Nevertheless, the evidence on the correlation between HVPG and body composition is still weak. The number of existing studies is too limited to provide relevant data. Discrepant results were yielded among these studies. The study by Matsui et al.¹⁶ showed that the HVPG value was inversely correlated with the skeletal muscle index (SMI). In contrast, other published data showed a null association.^{5, 17-19} Similarly inconsistent results have also been observed regarding the adipose tissue index and HVPG. Rodrigues et al.⁵ concluded that there was a significant negative correlation between the subcutaneous adipose tissue index (SATI) and the HVPG value, but Cho et al.¹⁸ and Zeng et al.¹⁹ did not.

Whether the HVPG value is correlated with a certain body composition parameter, and to what extent the HVPG value differs between patients with sarcopenia and patients without sarcopenia remains unknown. Knowledge of the impact of PH on muscle or adipose tissue is highly desirable, guiding nutrition support and tailoring individualized therapy. The additional value of HVPG, known as a validated index mirroring PH, would be detected for association with body tissue alternations in patients with LC. Hence, a meta-analysis was conducted to overview the current evidence and address this issue.

Methods

Protocol registration

The present review was performed following the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.²⁰ The PRISMA checklist is shown in Online Resource 1. This study was registered prospectively in the International Prospective Register of Systematic Reviews in 2023 (registration number: CRD42023392942). The requirement for informed consent and ethical approval from the Institutional Review Board were waived because the study quantified all existing publicly available data instead of involving specific patients.

Eligibility criteria

Population, interventions, comparisons, outcomes: the population of interest was patients with LC. The interventions of interest included CT scanning and HVPG within an acceptable interval. The outcomes of interest included: (1) the correlation analysis between various body composition parameters and HVPG; and (2) the HVPG value reported in patients with or without sarcopenia. The comparison and study of interest were not applicable or limited.

The abstract of a conference poster containing relevant information was also eligible. The authors contacted the corresponding author for detailed information. References cited in the text of selected articles were also further searched to minimize publication bias.

Search strategy

Peer-reviewed articles written in English and published up to August 2023 were searched in Embase, MEDLINE (via PubMed), Web of Science, and Cochrane Library. The retrieval protocol combined medical subject headings and text, which were mostly derived from entry terms in the PubMed and Embase databases. The search strategy is available in Online Resource 2.

Study selection

The exclusion criteria were as follows: (1) duplicate and irrelevant articles; (2) cell-line studies; (3) review articles; (4) case reports; (5) letters; (6) comments and editorials; (7) subjects from pediatric and non-human sources; and (8) cadavers.

The further exclusion criteria in a full-text assessment were as follows: patients with (1) LC with non-intrahepatic causes; (2) presence of evident intrahepatic vessel communication in measuring HVPG; and (3) a history of transjugular intrahepatic portosystemic shunt.

The HVPG value and body composition parameter on a continuous scale were eligible for analysis.

The correlation analysis should be performed using Pearson’s (r) or Spearman’s rho analysis according to the normality of the raw data. Presently, the impact of tumors not involving an intra- or extra-hepatic great vessel on the HVPG value remains unclear. Measurements of HVPG were performed in selected patients with hepatocellular carcinoma (HCC) and LC in real-life practice; thus, patients with HCC with a Barcelona Clinic Liver Cancer stage of 0, A, or B would not have been excluded in this meta-analysis. In addition, this potential effect could be further eliminated in the subgroup analysis.

Definitions

Transversal-psoas muscle thickness and psoas muscle thickness by height are the same measurement with different names, referring to the transversal diameter of the psoas muscle perpendicular to the largest axial psoas muscle diameter at the L3 plane normalized by height. Therefore, these two indexes were replaced with the psoas-muscle-related index (PRI) for analysis. All muscle and adipose indexes are defined and illustrated in Supplementary Figure 1.

Outcomes

The primary outcome was the correlation coefficient between various body composition parameters and HVPG. The difference in HVPG value between the sarcopenia group and the non-sarcopenia group was the secondary outcome. Due to a lack of a validated cut-off value to define adipopenia, the secondary outcome analysis was not performed in adipose indexes.

Data extraction

Two review authors (S.Y. and Q.C.) blindly and independently extracted the following items from each article: the first co-author, year of publication, country, study design, sample size, body mass index (BMI), sex, cause of liver disease, albumin, decompensation proportion, Child–Pugh score, MELD score, the interval between CT scan and HVPG measurement, sarcopenia definition, sarcopenia cut-off value, sarcopenia proportion, HVPG value in the sarcopenia and non-sarcopenia groups, correlation coefficient between body composition parameters and the HVPG value, and details of the HVPG measurement technique.

All data were respectfully recorded by two review authors using Microsoft Excel. Any inconsistency was resolved by reviewing the original article to achieve a consensus.

Risk of bias and certainty of evidence assessment

Two review authors independently assessed the methodological quality with regard to risk of bias and applicability concern using the Quality Assessment of Diagnostic Studies instrument. The Grading of Recommendations Assessment, Development and Evaluation (GRADE) system and online tool (GRADE Pro GDT, https://gdt.gradepro.org/) were used to rate the outcome if possible. The certainty of evidence was classified into four levels based on the five domains (https://training.cochrane.org/resource/grade-handbook) high, moderate, low, and very low.

Statistical analysis

The HVPG values in the sarcopenia and non-sarcopenia groups presenting as mean ± standard deviation were summarized. Values presenting as the median (interquartile range) would have been converted using an established fashion if necessary.²¹

The difference in the HVPG values was compared using the standardized mean difference (SMD) with a 95% confidence interval (CI). The Pearson correlation coefficient was collected and converted to the Fisher-Z value according to the following equation: Z = 0.5 [ln (1 + r) – ln (1 − r)]; the corresponding standard error was calculated according to the following equation: SEz = 1/√^--n-3, and summary r was recovered using the following equation: r = (e^2Z − 1) / (e^2Z + 1).²²

A Fisher transformation was used to convert the Spearman coefficient into an approximately normal distribution and further calculate the 95% CI. Subsequently, the same summary process was conducted as a Pearson analysis. Fisher’s Z value was used in the meta-analysis and shown in the plots, and the correlation coefficient derived from the inverse Fisher’s transformation was presented as the summary result. The heterogeneity was identified using Cochran’s Q test and further quantified using the I² statistic among the studies. When the P value was <0.05 or the I² value was >50%, the heterogeneity was considered high, and the source of bias was explored. Publication bias was assessed if the number of included studies was >10.²³ In the prespecified sensitivity analysis, pooled correlation coefficient estimates were further stratified as per presence of HCC and different sarcopenic cut-off values.

A P value of <0.05 was indicative of a significant difference. Considering the heterogeneity and sample size, a random effects model was selected to calculate the pooled effect size. The Stata MP (version 16.0, Stata Corp, College Station, USA) package was used for meta-analysis, and Review Manager (version 5.3) was used to evaluate the methodological quality.

Results

Study characteristics

Of the 652 studies screened initially, nine involving a total of 1,569 patients with LC were included for meta-analysis.^{5,16-19,24-27} A corresponding flow diagram is shown in Figure 1.

One poster including relevant data was excluded because it had not been published officially, and the request for raw data or effect size had not been answered.²⁸ The characteristics of the included studies are shown in Table 1.

Regarding the characteristics of the included patients, the sarcopenia proportion ranged from 34.7% to 71% across the eligible studies. The most common cause of liver disease was alcohol in six studies, followed by virus in the remaining three studies. There were 13 participants with HCC in the context of LC included in one study.¹⁷

A total of seven studies reported the primary outcome. Of these, five comprised 718 patients reported via SMI^{5, 16-19} (one reported the correlation coefficient separately in the sarcopenia and non-sarcopenia subgroups)¹⁷ and two comprised 268 patients reported in PRI.^{25, 26} A total of three studies provided the primary outcome in SATI and the visceral adipose tissue index (VATI).^{5, 18, 19}

In addition, five studies reported the secondary outcome: four reported via SMI and one reported via PRI.²⁵ Among the four studies reporting via SMI, the cut-off value was 42 cm²/m² for men and 38 cm²/m² for women in two studies^{16, 17} and 52.4 cm²/m² in men and 38.5 cm²/m² in women in the other two studies.^{24, 27} Considering that SMI was recommended for defining sarcopenia by most societies, the study reporting via PRI was not included for the secondary outcome.

For the publication nation, one study was conducted in Australia,²⁵ one in Switzerland,⁵ and the remaining seven in Asian countries, including China,¹⁹ Japan,^{16, 17} and the Republic of Korea.^{18, 24, 26, 27}

All included studies were retrospective studies published in the last 5 years.

Quality assessment and risk of bias

All included studies were considered to be of low or moderate risk of bias, as illustrated in Figure 2. The detailed scales are shown in Online Resource 3. The GRADE summary of findings for the outcome is provided in Supplementary Table S1.

Primary outcome

Muscle index

Only Matsui et al.¹⁶ reported a significantly negative correlation between SMI and HVPG in 202 patients; the remaining studies reported a null correlation.

The pooled correlation coefficient, regardless of muscle index, was −0.08 (−0.25, 0.09; P = 0.368), with significant heterogeneity observed (overall: I²= 85.3%; P < 0.001); similar results were observed in the SMI and PRI subgroups [SMI: r = −0.09 (−0.31, 0.14); P = 0.442; I²= 88.4%; P < 0.001; PRI: r = −0.01 (−0.15, 0.12); P = 0.852; I²= 16.1%; P = 0.275] (Figure 3).

Adipose tissue index

No significant correlation was pooled [r = −0.03 (−0.12, 0.05), I²= 34.5%, P = 0.177] in either of the adipose index subgroups [SATI: r = −0.06 (−0.24, 0.13), P = 0.545; VATI: r = −0.03 (−0.12, 0.07), P = 0.586]. The high heterogeneity was detected in the SATI subgroup (I²= 71.1%, P = 0.032) but not in the VATI subgroup (I²= 0.0%, P = 0.695). The corresponding forest plot is shown in Figure 4.

Secondary outcome

The summary difference of the HVPG value between the sarcopenia and non-sarcopenia groups indicated statistical significance, with unstable results due to different sarcopenia definitions. When using the cut-off value from the Japan Society of Hepatology guidelines for sarcopenia (SMI <42 cm²/m² for men or <38 cm²/m² for women), higher HVPG values were observed in patients with sarcopenia than in patients without sarcopenia [pooled SMD: 0.628 (−0.350, 1.606), P < 0.001; I²= 92.8%; P < 0.001]. When a commonly used cut-off value in the Western population was applied (50 cm²/m² for men and 39 cm²/m² for women), the HVPG value was higher in patients without sarcopenia than in patients with sarcopenia [pooled SMD: −0.201 (−0.366, −0.037), P = 0.016; I²= 0.00%; P = 0.785] (Figure 5).

Sensitivity analysis

After exclusion of the study including 13 patients with HCC, the correlation between either PRI or SMI and HVPG was not significant [overall: r = −0.10 (−0.30, 0.11), P = 0.341; I²= 89.1%, P < 0.001; SMI: r = −0.13 (−0.40, 0.17), P = 0.401; I²= 92.6%; P < 0.001]. The corresponding forest plot is shown in Figure 6.

Discussion

In the present review, a meta-analysis was performed to identify and quantify the current evidence regarding the correlation between body composition parameters and HVPG. The pooled results indicated that there was no significant correlation between muscle or adipose quantity and the HVPG value, regardless of muscle index. The results of the secondary outcome were unstable due to different sarcopenia definitions. With consideration of the statistical significance and ethnicity-specific cut-off value of sarcopenia, the result appears to reveal that patients with lower muscle mass may have a higher HVPG value.

Body composition and HVPG are of paramount importance for patients with LC. Nevertheless, a knowledge gap remains in the correlation between them. To the best of the present authors’ knowledge, this meta-analysis is the first to quantitatively combine current data to assess the correlation between body composition parameters and HVPG.

In fact, limited LC-related studies have reported both composition parameters and HVPG values at the same time, seldom exploring the association between them. Specifically, CT-based quantitative analysis and invasive operation hamper the acquisition of data in clinical practice. Despite the fact that the limited evidence grade leads to a cautious interpretation of the results, the findings of this meta-analysis could help explore the impact of PH on body composition parameters and might be instrumental in refining a comprehensive evaluation algorithm of patients with LC.

In this meta-analysis, several points merit attention. First, the HVPG value was used to evaluate the PH instead of the portosystemic pressure gradient, largely because the portosystemic pressure gradient was commonly collected in the transjugular intrahepatic portosystemic shunt procedure with a limited clinical application prospect. Second, to reduce the bias derived from different global cut-off values of sarcopenia, only the muscle or adipose tissue quantity as the continuous variable normalized to height or height² was extracted and comparable. In addition, other statistics would have been summarized if they could have been converted to the correlation coefficient using a validated statistical method, including the contingency coefficient and standardized beta value; however, such a study was not found in the study screening. Third, SMI is recognized as the gold standard for measuring muscle quantity in defining sarcopenia, and psoas-muscle-related parameters have been shown to be less strongly correlated with the total body protein or mortality risk compared with SMI.^{29, 30} Therefore, of the five studies reporting the secondary outcome, one study reporting via PRI was not included in the meta-analysis. Last, all included studies were published in the past 5 years, thereby enabling a standard care for patients with LC.

Negative results of the primary outcome are partly explainable because of a considerable interindividual variation of the liver function reserve among the included patients. In the included studies, decompensated cirrhosis or clinical signs of PH, such as ascites, gastro-esophageal varices, and hepatic encephalopathy, were deemed indications of HVPG measurement. Among all the evaluable patients, the mean values of the MELD score were 9–13, the decompensation proportions were 54.8%–100%, and the baseline HVPG values were 14–19 mmHg. In fact, sarcopenia is relatively frequently found in advanced liver disease or the decompensated stage.^{31, 32} Furthermore, some characteristics of patients with LC, including the cause of liver disease, decompensated cirrhosis, or oral beta‐blocker administration should have been used in the subgroup analyses with the aim of ruling out confounding factors and further identifying a potential association between the muscle quantity and the HVPG value in a certain subgroup of patients with LC. Likewise, adipose tissue change and re-distribution could be affected by BMI and sex.³³ Therefore, for the primary outcome of the adipose tissue, the non-significant summary result may indicate the likelihood of the correlation between adipose tissue indexes and HVPG depending on the baseline characteristics of the included patients.

In addition, the result of the secondary outcome was not robust. It is speculated that a lower cut-off value (42 cm²/m² for men or <38 cm²/m² for women) could identify more individuals with a low muscle quantity and further re-classify a proportion of patients as having sarcopenia; that is, a lower cut-off value of sarcopenia has more statistic power to differentiate patients with different PH stratifications. It is noted that all included studies on the secondary outcome were from Asian countries (Japan and the Republic of Korea). The Asian sarcopenia definition (42 cm²/m² for men or <38 cm²/m² for women) thus allows for better interpretability and practical applicability.⁹

As the present study is a pilot meta-analysis exploring the unknown relationship between two important characteristics of patients with LC, some limitations exist. First, a considerable interindividual variation of baseline characteristics among included patients, especially liver function status, leads to a cautious interpretation of the results. Second, some included studies only presented the effect size instead of analyzing it in the subgroups. The evidence grade is limited by the number of included studies and the data blank. Most importantly, the number of available studies that fulfilled the present study’s inclusion criteria is low, precluding meta-regression to further identify the potential confounding factors. Hence, a prospective study dedicated to recording relevant information is required in the future.

In conclusion, overall, this meta-analysis showed a non-significant correlation between body composition parameters, including muscle and adipose tissue quantity, and the HVPG value. However, its current clinical usefulness is uncertain due to a lack of universal definition and limited research.

Reporting checklist

This review was performed following the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.