Impact of sarcopenia and sarcopenic obesity on survival in patients with primary liver cancer: a systematic review and meta-analysis

Background Sarcopenia and sarcopenic obesity are associated with an increased possibility of adverse clinical outcomes; however, the effects of sarcopenia and sarcopenic obesity on patients with primary liver cancer remain controversial. Therefore, the present study aimed to determine the impact of sarcopenia and sarcopenic obesity on survival in patients with primary liver cancer. Methods We searched studies published in English in PubMed, Embase, Web of Science, and Cochrane Library databases up to 13 November 2022. Cohort studies that reported the association among sarcopenia, sarcopenic obesity, and patient survival were included. Results A total of 64 cohort studies with data on 11,970 patients with primary liver cancer were included in the meta-analysis. Sarcopenia was associated with poor overall survival in patients with primary liver cancer [adjusted hazard ratio (HR) 2.11, 95% confidence interval (CI): 1.89–2.36, P < 0.0001], with similar findings for sarcopenic obesity (adjusted HR: 2.87, 95% CI: 2.23–3.70, P < 0.0001). Sarcopenia was also associated with poor overall survival across the subgroups analyzed by ethnicity, type of liver cancer, treatment modalities, method used to define sarcopenia, and etiology of liver cancer. We also found a negative correlation among sarcopenia, sarcopenic obesity, and recurrence-free/disease-free survival (adjusted HR: 1.73, 95% CI: 1.50–1.99, P < 0.001; adjusted HR: 2.28, 95% CI: 1.54–3.35, P < 0.001, respectively). Conclusion Sarcopenia and sarcopenic obesity were significantly associated with poor overall survival and recurrence-free/disease-free survival in patients with primary liver cancer. Systematic review registration https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=378433, PROSPERO [42022378433].


. Introduction
Primary liver cancer is the sixth most frequently occurring cancer and ranks as the third leading cause of cancer-related mortality worldwide, accounting for 8.3% of total cancer deaths (1), thus resulting in a global medical and economic burden.Hepatocellular carcinoma (HCC) is the predominant type of primary liver cancer, comprising 75%−85% of the cases, and intrahepatic cholangiocarcinoma (ICC) follows as the subsequent type (1).There are significant gender and racial differences in morbidity and mortality due to primary liver cancer; both morbidity and mortality rates are two-fold to three-fold higher in men than in women in most regions, and the disease is more common among Asians due to a high prevalence of hepatitis B and C (1).It is therefore critical to identify patients with high mortality risk based on the patient's prognosis for determining individualized treatments and improving the survival rate of patients with primary liver cancer.In recent years, researchers have made several efforts to determine the factors that influence the clinical outcomes of patients with liver cancer.Thus far, the Barcelona Clinic Liver Cancer (BCLC), Model for End-Stage Liver Disease (MELD), and the albumin-bilirubin (ALBI) scores have been used clinically to evaluate the prognosis of patients with liver cancer; however, these prognostic tools cannot adequately capture the nutritional and functional status of these patients.
Sarcopenia, as a marker of malnutrition, has been defined by the European Working Group on Sarcopenia in Older People (EWGSOP2) in 2018 as the presence of both low muscle mass and low muscle function (strength or performance) (2).It is difficult to diagnose sarcopenia because of different measuring methods and cutoff values.Sarcopenia is usually evaluated based on grip strength, dual-energy X-ray absorptiometry, computed tomography (CT), and magnetic resonance imaging (MRI) (2).Sarcopenia increases the risk of worse clinical outcomes such as reduced quality of life, development of complications, higher hospitalization cost, and death (3)(4)(5)(6).Previous studies have shown that the hospitalization cost of older patients with sarcopenia on admission was five-fold more than those without sarcopenia (7).Sarcopenia is a common condition in patients with oncological and chronic diseases.In a systematic review and meta-analysis that included 38 studies on sarcopenia and solid cancer outcomes, sarcopenia was significantly associated with the poor overall survival of patients (8).Similarly, according to a recent umbrella review of meta-analyses, sarcopenia is associated with adverse clinical outcomes across 12 cancer types: gastric, hepatocellular, urothelial, head and neck, hematologic malignancy, pancreatic, breast, colorectal, lung, esophageal, hematologic malignancies, and ovarian (9).The existence of sarcopenia was found to be associated with a higher risk of death in patients with liver cirrhosis (10), which is likely to progress into liver cancer.
The rate of fat deposition tends to increase in sarcopenic patients, resulting in systemic inflammatory activation and insulin resistance, which subsequently leads to progressive muscle reduction and fat accumulation, especially in conditions such as aging and cachexia (11).This vicious cycle finally results in sarcopenic obesity (SO), which is defined as the co-existence of obesity and sarcopenia (12).More recently, SO has received increasing interest from oncologists because of its adverse outcomes in patients with cancer.SO is an independent prognostic factor affecting the risk of adverse outcomes in oncological patients (13)(14)(15).
The influence of sarcopenia on survival in patients with liver cancer remains controversial, and a comprehensive analysis based on evidence-based medicine is required.To the best of our knowledge, previous meta-analyses have focused only on HCC patients and did not include data on patients with ICC.Recently, an increasing number of studies have examined the prognostic factors of liver cancer patients.Hence, we analyzed and summarized the relationship among sarcopenia, SO, and survival in patients with primary liver cancer.Our study aimed to determine the impact of sarcopenia and SO on the survival of patients with primary liver cancer.

. . Search strategy
We searched studies relevant to the association of sarcopenia, SO, and survival of patients with liver cancer in PubMed, Embase, Web of Science, and Cochrane Library databases up to 13 November 2022.The search keywords included sarcopenia, muscle depletion, muscle weakness, liver cancer, and liver neoplasm.The detailed search strategies are presented in Supplementary Table S1.We restricted the studies to those published in English and conducted on humans.We also retrieved potential studies by reading through the relevant systematic reviews and metaanalyses.The present meta-analysis adhered to the PRISMA guidelines (21), and its protocol was registered on PROSPERO (CRD 42022378433).

. . Criteria for selection
Studies that met the following criteria were included: (1) participants: patients with liver cancer confirmed by clinical/imaging or liver biopsy criteria (may include patients evaluated or already listed for liver transplantation), including HCC and ICC; (2) exposures: pretreatment for sarcopenia and/or SO; (3) outcomes: the impact of sarcopenia and/or SO on patient survival; and (4) study design: prospective or retrospective cohort study.
Studies that met the following criteria were excluded: (1) studies lacking the criteria for diagnosing sarcopenia or SO; (2) studies lacking the statistical data on the impact of sarcopenia and/or SO on survival [hazard ratio (HR) and 95% confidence interval (CI)]; and (3) reviews, case reports, editorials, letters, posters, and/or conference abstracts.
The authors XL and XH independently screened the title/abstract of all the identified citations for eligibility by using the abovementioned inclusion/exclusion criteria.Next, they retrieved and rescreened the full texts of relevant articles.For studies with overlapping cohorts, studies having the latest data and/or a larger sample size and/or more data available for subgroup analysis were used.Disagreements were resolved by consensus or discussion.

. . Data acquisition and quality assessment
XL and XH extracted the following data independently from each included study: first author's name, first author's country, published year, ethnicity, study type, type of liver cancer, treatment modalities, etiology of liver cancer, enrolled numbers, patient demographics (including age and sex), duration of follow-up, definitions of sarcopenia and SO, cutoff values of sarcopenia, and number of sarcopenia or SO patients.The quality of the enrolled studies was independently evaluated by the two authors according to the Newcastle-Ottawa Scale (NOS).Discrepancies between both investigators were resolved by consensus and discussion.

. . Statistical analysis
The outcomes for the association between sarcopenia and overall survival (OS), recurrence-free survival (RFS), or diseasefree survival (DFS) were reported as crude and adjusted HR with corresponding 95% CI values.HR and 95% CI values were directly extracted from univariate and multivariate Cox regression analyses.The impact of sarcopenia and SO on the OS of primary liver cancer patients was assessed by the pooled unadjusted HR or adjusted HR and 95% CI by using a random-effects model (DerSimonian-Laird method) (22).A subgroup analysis for the adjusted HRs was conducted according to ethnicity, type of liver cancer, treatment modalities, method used to define sarcopenia, and etiology of liver cancer.Cochran's Q statistic and I 2 were used to evaluate heterogeneity, with a P-value of <0.1 and I 2 > 50% considered to show meaningful heterogeneity (23).We assessed publication bias through the utilization of funnel plots in meta-analysis and quantified it using Egger's regression test.We also conducted a meta-analysis of single proportions to determine the prevalence of sarcopenia.All analyses were performed with STATA software v15.0 (StataCorp, College Station, TX, USA), and a P-value of <0.05 was considered to be statistically significant.

. . Study search and characteristics
The search in PubMed, Embase, Web of Science, and Cochrane Library databases yielded 2,622 relevant citations, of which 574 duplicates and 1,880 unavailable titles/abstracts were excluded.After reviewing the full text of the remaining 168 publications and previous reviews, we included 64 eligible cohort studies with 11,970 patients (Figure 1).
The demographics and characteristics of the included studies are shown in Table 1.In general, 47 of the 64 included studies were conducted in Asia, predominantly in Japan (n = 31), and 17 studies were from non-Asian regions.Four studies were prospective, and 60 studies were retrospective.Of the 11,970 patients, 8,919 were men (74.51%) and 3,051 were women (25.49%), with a median (or mean) age of 51.4-77 years.The median or mean age was not reported in three studies (19,63,76).Fifty-six of the 64 studies involved patients with HCC.Seventeen studies enrolled patients with resectable HCC; the remaining treatment regimens included transarterial chemoembolization (TACE), liver transplantation (LT), radiofrequency ablation (RFA), and administration of kinase inhibitors.The curative treatment included liver resection, RFA, and LT, whereas the palliative treatment included intra-arterial chemoembolization, administration of kinase inhibitors, and systemic chemotherapy.Viral hepatitis was the most common etiology of liver cancer, with HCV as the primary cause of viral hepatitis in most included studies (n = 36).The other causes of liver cancer include alcohol consumption, non-alcoholic steatohepatitis (NASH), and others.Six studies did not report the etiology of liver cancer.Thirty of the 64 studies reported the duration of follow-up (median or mean), which ranged from 8.3 months to 7.9 years.

. . Quality assessment
The quality of the included studies was determined by NOS.Patient selection, comparability, and outcomes were used to evaluate the methodological quality of each study.Supplementary Table S2 shows the evaluation of the quality of the included studies.Based on the NOS score of ≥7, all the included studies were considered to have high quality.

. . Definition of sarcopenia and SO
The approaches to identifying sarcopenia and SO in patients were different.In most studies (n = 61), the areas of visceral/subcutaneous fat and skeletal muscle were determined by a transverse analysis at the level of the third lumbar vertebra (L3).In 45 studies, sarcopenia was defined by the sex-specific cutoff values of the L3-SMI (skeletal muscle index, cm 2 /m 2 ), which showed a slight variation in those studies (shown in Table 1).Seven studies used different values to define sarcopenia depending on body mass index (BMI) (24, 37, 42, 51, 52, 70, 80).The majority of studies (44 studies) used cutoff values between 40 and 55 cm 2 /m 2 in men.Nine studies evaluated sarcopenia based on the L3-PMI (psoas muscle index, cm 2 /m 2 ), and one study used the total skeletal muscle mass (TSM) index.The other two studies defined sarcopenia as low fat-free muscle area (FFMA) based on MRI evaluation.According to these sex-specific cutoff values, 3,957 patients were diagnosed with sarcopenia in 62 studies, yielding a pooled prevalence of 43.2% (95% CI: 37.8%-48.5%).Among the 50 studies reporting the prevalence according to gender, a slight difference in prevalence was observed between female and male patients, with a higher pooled prevalence of 45% among men compared to 42.2% among women.
Four of the 64 studies reported the impact of SO on the survival of liver cancer patients.Regarding the definition of SO, Itoh et al.  68) used the co-existence of sarcopenia and obesity to define SO.However, variations were noted in the definition of obesity.Obesity was defined as the area of visceral adipose tissue at the level of the third lumbar vertebra ≥100 cm 2 in both men and women by Kobayashi et al., while the patients were considered obese if their BMI was ≥25 kg/m 2 in Liao's study; in Kroh's study, obesity was defined by categorizing individuals within the highest two quintiles body fat percentage for men and women.
Frontiers in Nutrition frontiersin.org

FIGURE
Forest plot of the pooled adjusted hazard ratios for the association between sarcopenia and overall survival in patients with primary liver cancer.

. . OS in patients with SO
Only four studies reported statistical data regarding the influence of SO on patient survival (14, 33,51,68).The prevalence of SO varied greatly among the studies, ranging from 6.67% to 30.00% (characteristics shown in Table 1).We conducted respective analyses of studies reporting unadjusted and adjusted HRs for OS.Patients with SO had a higher risk of death than those without SO (pooled unadjusted HR: 2.08.95% CI 1.67-2.60,P < 0.0001; pooled adjusted HR: 2.87; 95% CI: 2.23-3.70,P < 0.0001), as shown in

FIGURE
Forest plot of the pooled adjusted hazard ratios for the association between sarcopenic obesity and overall survival in patients with primary liver cancer.
Supplementary Figure S5 and Figure 3.The subgroup analysis was not conducted because of the limited number of studies reporting SO data.
Moderate heterogeneity was observed in univariate analysis, while mild heterogeneity was noted in multivariate analysis (P = 0.027 and I 2 = 44.8%for the unadjusted HRs; P = 0.377 and I 2 = 7.0% for the adjusted HRs).Publication bias was not found in the univariate analysis (Egger's test, P = 0.587), whereas the multivariate analysis showed the existence of publication bias (Supplementary Figure S7).We further conducted a subgroup analysis on RFS/DFS for the adjusted HRs according to ethnicity, type of liver cancer, and sarcopenia definitions.However, we did not perform a subgroup analysis on RFS/DFS for the adjusted HRs based on treatment modalities because all subjects were treated by hepatectomy.The results are shown in Supplementary Table S4.
The association between SO existence and its influence on RFS/DFS in HCC was analyzed in the same manner.The summary of crude and adjusted HRs were 1.80 (95% CI: 1.33-2.44,P < 0.001; Supplementary Figure S8) and 2.28 (95% CI: 1.54-3.35,P < 0.001; Figure 5), respectively; this indicated that SO was associated with a lower RFS/DFS.

. Discussion
In the present meta-analysis, we analyzed 64 studies comprising 11,970 participants diagnosed with primary liver cancer.Sarcopenia is a common disorder in this population, with a pooled prevalence of 43.2%, and it is more prevalent in men.The results revealed a robust association between sarcopenia and/or SO and patient survival.A strong relationship between sarcopenia and adverse clinical outcomes in cancer patients has been reported in prior studies, including depression (83), risk of fall (84), higher risk of complications (85), and cancer recurrence and mortality (9,13,15,86).In 2022, a meta-analysis comprising 280 publications involving 81,814 patients with solid tumors demonstrated that sarcopenia is a prevalent condition in oncological patients with a prevalence of 35.3%; moreover, it is particularly higher in patients with specific cancer types such as esophageal cancer, urothelial cancer, cholangiocarcinoma, prostate cancer, and thyroid cancer (87).A recent study reported that SO affected 20% of cancer patients, demonstrating a significant association with various poor Frontiers in Nutrition frontiersin.org

FIGURE
Forest plot of the pooled adjusted hazard ratios for the association between sarcopenia and recurrence-free survival or disease-free survival in patients with primary liver cancer.
outcomes in cancer patients, such as low OS, RFS, and longer length of hospital stay, particularly in patients with oropharyngeal cancer and liver cancer (88).A previous systematic review of three articles involving 1,515 liver transplant recipients demonstrated a two-fold increase in mortality rates linked to pre-transplant SO (89).The underlying mechanisms of sarcopenia and SO are poorly understood.According to prior studies, sarcopenia and SO are multifactorial conditions.The key mechanisms of sarcopenia include aging, inflammation, hormonal changes, inactivity, and low-protein intake.The available evidence indicates that the elderly population, especially individuals aged 65 years and above, is susceptible to anabolic resistance due to decreased availability of post-prandial amino acid, diminished muscle perfusion, and decreased digestive ability caused by the sequestration of amino acids in the splanchnic region (90).Body fat increases with age until 70 years.This accumulation of body fat activates macrophages, mast cells, and T lymphocytes, resulting in the secretion of inflammatory factors such as tumor necrosis factor (TNF), leptin, IL-6, and growth hormone (GH), which induces an array of inflammatory responses (91).Inflammatory factors such as TNFα and IL-6 facilitate skeletal muscle wasting; the former directly catabolizes skeletal muscle, leading to increased gluconeogenesis, proteolysis, and upregulation of uncoupling proteins (UCPs) 2 and 3 in cachectic skeletal muscle, while the latter suppresses protein synthesis in muscle cells by the Janus kinase signaling pathway (92).Testosterone not only modulates inflammation in skeletal muscle by activating satellite cells to promote muscle regeneration but also increases the utilization of amino acids and androgen receptor expression in skeletal muscle to promote muscle protein synthesis; however, the levels of testosterone decline with age, which is likely to have a negative effect on muscle mass (93).Inactivity can affect muscle metabolism, further exacerbating the catabolic response and decreasing muscle protein synthesis (11).Prior studies have shown that exercise can improve muscle strength and mass, and both resistance training and aerobic exercise are beneficial to sarcopenia (94, 95).A previous review elaborated on the mechanisms of SO, which included lipotoxicity, adipose tissue inflammation, adipose tissue dysfunction, insulin resistance, and systemic chronic sterile low-grade inflammation.The authors proposed intricate interactions between adipose tissue and skeletal muscle, leading to the establishment of a detrimental vicious circle as individual's age, resulting in chronic low-grade local inflammation and systemic inflammation (91).Currently, there is a lack of specific medicines to treat SO.Lifestyle intervention is the most important method to treat SO, including calorie restriction; aerobic exercise; resistance exercise; and supplementation of protein, calcium, and vitamin D (93).Exercise intervention can positively change body composition and improve body weight, BMI, fat mass, body fat percentage, grip strength, and walking speed in the SO population; nutritional intervention can decrease fat mass with no improvement in grip strength (96).

FIGURE
Forest plot of the pooled adjusted hazard ratios for the association between sarcopenic obesity and recurrence-free survival or disease-free survival in patients with primary liver cancer.
To the best of our knowledge, this is the first meta-analysis to demonstrate a significant relationship between sarcopenia, SO, and survival in patients with primary liver cancer in a large sample.The study included patients who underwent either curative or palliative treatment.A strength of this meta-analysis is that the study population included patients with HCC and ICC for the first time.Although a prior meta-analysis of studies published before 2017 examined the association between sarcopenia and mortality in HCC patients, one of the studies included not only patients with HCC but also metastatic liver cancer (97).Another meta-analysis included cohorts that overlapped (98), while one study focused only on the prognosis of sarcopenia in HCC patients treated with sorafenib or lenvatinib (99).
Additionally, in the present meta-analysis, a comprehensive search was performed, and studies with overlapping cohorts were excluded.Consequently, our meta-analysis added 40 additional studies that were not analyzed in previous meta-analyses, thus contributing to 64 included studies.We found a correlation between SO and decreased OS.There is, however, a lack of extensive research on the effect of SO on survival in patients diagnosed with liver cancer.
Our study has several limitations.First, this study was constrained by insufficient data from each of the included studies, which is inherent to the nature of meta-analysis.Additionally, not all studies reported an adjusted HR, which could potentially restrict our ability to determine the precise magnitude of the mortality risk between sarcopenic and non-sarcopenic patients.Second, the selection of adjusted variables for the multivariate Cox regression models varied among the studies.Third, the significant results of Egger's tests indicated the presence of publication bias.
Sarcopenia was defined as the presence of both low muscle mass and impaired muscle function, according to EWGSOP2 (2).However, the diagnosis of sarcopenia remains controversial.On the one hand, there is a lack of standardized and feasible methods to measure muscle function or physical performance.Importantly, this limitation applies not only to the current meta-analysis but also to all existing studies that have investigated the impact of sarcopenia and/or SO on patients with malignant carcinomas.On the other hand, only prospective cohort studies are likely to document muscle function.Therefore, all the included studies can only partially define sarcopenia by measuring muscle mass, depending on CT/MRI images at the L3 level, which can be easily obtained from medical records.Hence, the retrospective nature of the included studies is recognized as a limitation of the current study.Given that the majority of the included studies were retrospective cohort studies, it is probable that the results were influenced by selection bias, as only patients who underwent CT scans were included.Furthermore, several methods were used to measure muscle mass, and the studies evaluating SMI and PMI at the L3 level based on CT imaging were included in the present meta-analysis; this may partially result in heterogeneity.We observed slight variations in the actual sex-specific cutoff values used in different studies.In some studies, the authors predefined the thresholds, whereas other studies derived these thresholds from their own study population to calculate the cutoff values.The thresholds were higher in European and American populations than in Asian populations, which might be explained by the fact that previous western studies were deemed unsuitable for Asian patients.Additionally, different cutoff values might change the magnitude of the association between sarcopenia and survival.Hence, we performed a subgroup analysis to overcome these limitations.The subgroup analysis showed a significant association between sarcopenia and poor OS and RFS/DFS across all the analyzed subgroups.Further prospective studies evaluating both muscle mass and muscle function are necessary to accurately and timely identify sarcopenic patients and to better clarify the relationship between muscle loss and survival.
Unlike previous studies examining the association between sarcopenia and survival in liver cancer patients, there is limited research on the association between SO and survival in liver cancer patients.Of the five studies searched, one study was excluded because it involved patients with other diseases, such as liver cirrhosis and cholestatic diseases (100).Furthermore, the four studies defined SO by using different approaches.Hence, further studies are required to confirm the association between SO and survival.

. Conclusion
Sarcopenia and SO exhibited a significant association with reduced OS and RFS/DFS in patients with liver cancer.Additionally, the evaluation of sarcopenia and SO needs a consensus regarding their definitions and the utilization of appropriate cutoff values.We suggest that patients with liver cancer should undergo initial evaluation for sarcopenia and SO and receive regular monitoring because of poor prognosis.Further prospective studies are required to integrate sarcopenia and SO into an established prognostic scale specifically tailored for patients with liver cancer.

FIGURE
FIGUREPRISMA flow diagram of the search strategy.
a Values are median unless indicated otherwise.b Values are mean.