Effect of Body Mass Index on the Prognosis of Liver Cirrhosis

Objective: At present, the association of body mass index (BMI) with the prognosis of liver cirrhosis is controversial. Our retrospective study aimed to evaluate the impact of BMI on the outcome of liver cirrhosis. Methods: In the first part, long-term death was evaluated in 436 patients with cirrhosis and without malignancy from our prospectively established single-center database. In the second part, in-hospital death was evaluated in 379 patients with cirrhosis and with acute gastrointestinal bleeding (AGIB) from our retrospective multicenter study. BMI was calculated and categorized as underweight (BMI <18.5 kg/m2), normal weight (18.5 ≤ BMI < 23.0 kg/m2), and overweight/obese (BMI ≥ 23.0 kg/m2). Results: In the first part, Kaplan–Meier curve analyses demonstrated a significantly higher cumulative survival rate in the overweight/obese group than the normal weight group (p = 0.047). Cox regression analyses demonstrated that overweight/obesity was significantly associated with decreased long-term mortality compared with the normal weight group [hazard ratio (HR) = 0.635; 95% CI: 0.405–0.998; p = 0.049] but not an independent predictor after adjusting for age, gender, and Child–Pugh score (HR = 0.758; 95%CI: 0.479–1.199; p = 0.236). In the second part, Kaplan–Meier curve analyses demonstrated no significant difference in the cumulative survival rate between the overweight/obese and the normal weight groups (p = 0.094). Cox regression analyses also demonstrated that overweight/obesity was not significantly associated with in-hospital mortality compared with normal weight group (HR = 0.349; 95%CI: 0.096-1.269; p = 0.110). In both of the two parts, the Kaplan–Meier curve analyses demonstrated no significant difference in the cumulative survival rate between underweight and normal weight groups. Conclusion: Overweight/obesity is modestly associated with long-term survival in patients with cirrhosis but not an independent prognostic predictor. There is little effect of overweight/obesity on the short-term survival of patients with cirrhosis and with AGIB.


INTRODUCTION
Overweight/obesity, which is defined as excessive body fat accumulation, is a common public health problem (1). The global age-standardized prevalence of obesity defined by high body mass index (BMI) is increased from 3.2 to 10.8% in men and from 6.4 to 14.9% in women between 1975 and 2014 (2). Overweight/obesity is considered a risk factor for liver diseases (3,4). Increased adipose tissues lead to triglyceride deposition in the liver, produce various transduction signals that alter lipid and glucose metabolisms, and then cause insulin resistance and increased release of free fatty acids, which are the causes of hepatic steatosis (5). Hepatic steatosis can contribute to lipid peroxidation and hepatic stellate cell activation, which further induce cellular injury and inflammation and accelerate the progression of liver fibrosis and cirrhosis (6,7). However, the impact of BMI on outcomes in liver cirrhosis is still controversial. Some studies demonstrated that obesity was an independent risk factor for cirrhosis-related death or hospitalization (8). Other studies supported that the patients with cirrhosis and with obesity had a lower mortality than those without (9). Considering the controversy of the existing evidence, this study aimed to examine the effect of BMI on the prognosis of patients with liver cirrhosis.

Study Design
This retrospective study was carried out following the rules of the 1975 Declaration of Helsinki and approved by the Medical Ethical Committee of the General Hospital of Northern Theater Command with an approval number of Y (2021) 023. It was divided into two major parts. In both of the two parts, if the patients are lacking height and weight, they were excluded from the current study.
In the first part, we retrospectively selected patients with cirrhosis and without malignancy from our prospectively established database (10). Eligible patients should be consecutively admitted to our department and underwent an endoscopy and contrast-enhanced CT or MRI scans between December 2014 and December 2020. They were regularly followed via telephone or through outpatient visits and/or by reviewing medical records until February 2021. Death and the patients with liver transplantation during follow-up were recorded. Liver transplantation-free survival was the major endpoint of the first part. Patients who underwent liver transplantation were followed until the time point when the liver transplantation was performed.
In the second part, we retrospectively selected patients with cirrhosis and with acute gastrointestinal bleeding (AGIB) who received terlipressin and/or somatostatin/octreotide from our multicenter study (registration number: NCT03846180), which has been further updated after some publications (11)(12)(13)(14). Notably, in this part, the eligible patients were consecutively admitted to 13 centers from 8 provinces or municipalities in China between January 2010 and December 2018, and patients who underwent transjugular intrahepatic portosystemic shunt, splenectomy, surgical shunt, or liver transplantation were excluded from the study. In-hospital death was the major endpoint of the second part.

Diagnosis and Definitions
Liver cirrhosis was diagnosed based on clinical manifestations, laboratory tests, radiological examinations, and/or histological data. AGIB was defined as hematemesis, melena, and/or hematochezia within 5 days before admission (15).
BMI was calculated by dividing weight in kilograms by the square of height in meters (16). According to the WHO classification for Asian populations, all patients were categorized as underweight (BMI < 18.5 kg/m 2 ), normal weight (18.5 ≤ BMI < 23.0 kg/m 2 ), and overweight/obese (BMI ≥ 23.0 kg/m 2 ) (17).

Statistical Analyses
First, continuous variables were described as mean ± SD and median (range), and categorical variables were described as frequency (percentage). The difference was compared using Mann-Whitney U test and Chi-squared test or Fisher's exact test. Second, survival probability curves were calculated by the Kaplan-Meier curve analyses and compared by the logrank test. Third, univariate Cox regression analyses were performed to explore the association of BMI with mortality, and multivariable Cox regression analyses were performed by adjusting for age, gender, and Child-Pugh scores to identify whether BMI was an independent predictor of death. Hazard ratios (HRs) with 95% CIs were calculated. Fourth, timedependent receiver operating characteristic (T-ROC) curve analyses were used to evaluate the performance of BMI for predicting death, and area under the curve (AUC) and concordance index (C-index) were calculated. A two-tailed p < 0.05 was considered statistically significant. All statistical analyses were performed by using SPSS version 26.0 (IBM Corp, Armonk, New York, USA) and R version 4.0.3 with packages

First Part: Long-Term Outcomes of Patients With Cirrhosis
Overall, 436 of 527 patients with cirrhosis registered in our prospective database had BMI data at their admissions and were included in the present study. Among them, 30 (Figure 1). The Kaplan-Meier curve analyses demonstrated that overweight/obese group had a significantly higher cumulative survival rate than normal weight group (p = 0.047) (Figure 2A). Univariate  Cox regression analyses demonstrated that overweight/obesity was significantly associated with decreased long-term mortality (HR = 0.635; 95%CI: 0.405-0.998; p = 0.049). After adjusting for age, gender, and Child-Pugh score, overweight/obesity was not an independent predictor of decreased long-term mortality (HR = 0.758; 95%CI: 0.479-1.199; p = 0.236) (Supplementary Table 1). Time-dependent receiver operating characteristic analyses of BMI for predicting long-term mortality of patients with cirrhosis are shown in Figure 3A.  in overweight/obese group. During a median hospitalization period of 13 (range: 4-48) days, 14 patients died. Among them 11 (78.6%) patients died of liver diseases and 3 (21.4%) died of non-liver diseases. In-hospital mortality was 2.9% (1/34) in underweight group, 5.4% (10/185) in normal weight group, and 1.9% (3/160) in overweight/obese group (Figure 1).
Time-dependent receiver operating characteristic analyses of BMI for predicting in-hospital mortality of patients with cirrhosis and with AGIB are shown in Figure 3B.

DISCUSSION
The first objective of the present work was to explore the relationship of BMI with the long-term prognosis of patients with cirrhosis. We found that overweight/obesity was inversely associated with long-term mortality of patients with liver cirrhosis. This finding supported the "obesity paradox" that overweight/obese patients could have superior survival. It was first proposed by Fleischmann et al. (18) in patients undergoing hemodialysis (18) and further validated in subjects with chronic diseases, such as cardiovascular diseases, hypertension, and diabetes (19)(20)(21)(22).
The pathophysiology of the "obesity paradox" remains to be elucidated, and there are some underlying explanations (Supplementary Figure 1). First, fat storage in overweight/obese patients may protect the balance of muscle protein catabolism in chronic wasting diseases (23). Body protein is crucial for survival because it can maintain cell function and support cell architecture (24). Muscle protein metabolism is preserved in patients with obesity and with chronic cardiac failure, indicating better outcomes but increased in patients without obesity (25). Similarly, sarcopenia, which is mainly caused by increased muscle protein metabolism (26,27), is associated with lower BMI in patients with liver cirrhosis (28), and further leads to higher mortality (29). Second, adipose tissue, which has been recognized as an endocrine organ, can secrete diverse adipokines (30,31). Adiponectin, an anti-inflammatory adipokine, can inhibit the proliferation and activation of hepatic stellate cells, which produce extracellular matrix proteins in the case of liver injury and promote the occurrence of liver fibrosis (32). Leptin, another adipokine, can prevent ectopic lipid accumulation in non-adipose tissues, augment immune response, and improve bacterial clearance and survival in animal models (33,34). Both of which are increased in overweight/obese patients with liver cirrhosis, probably improving the outcomes of the patients (35). Third, patients with cirrhosis and with hepatic edema have systemic vasodilation and underfilled arteries, decreasing effective circulatory blood volume and activating cardiac sympathetic nervous system (SNS) and reninangiotensin-aldosterone system (RAAS) which can stimulate sodium and water retention. Prolonged sodium and water retention will cause hyponatremia and pulmonary edema and increase cardiac afterload (36,37). Overweight/obesity can alleviate the activities of cardiac SNS and RAAS, thereby inhibiting hyperdynamic circulation and conferring survival benefits (38). Fourth, overweight/obese patients are more likely to receive medical interventions, including antihypertensive drugs for decreasing systolic blood pressure, which can make shortterm hemodynamic status more stable (39,40), and statins for treating hyperlipidemia (41), which can have a favorable impact on outcomes of cirrhosis and portal hypertension (42). Age, gender, and body function may interact with the relationship between BMI and prognosis (43,44). Accordingly, this study adjusted some potential confounding factors, including age, gender, and Child-Pugh score, in multivariable Cox regression analysis. By comparison, the previous study by Karagozian et al. (9) selected patients with cirrhosis from the National Inpatient Sample database, which was lacking laboratory data, such as hepatic function (9). Thus, these results should be more reliable. This study found that BMI was not an independent risk factor of decreased long-term mortality, indicating that the prognostic impact of BMI might not be as strong as Child-Pugh score. This finding can be explained by the hypothesis of "reverse causation" that overweight/obesity may not be a cause for a better outcome, but a consequence (45)(46)(47). Another explanation is that BMI is convenient but unable to comprehensively measure body composition (2,48), such as muscle and subcutaneous and visceral adipose tissue and their specific distributions in the body (49). Besides, body weight may be masked by fluid retention resulting from ascites in patients with cirrhosis, inaccurately or falsely evaluating the prognostic impact of BMI.
This study demonstrated that BMI was not significantly associated with in-hospital outcomes of patients with cirrhosis and with AGIB, which is consistent with the previous study regarding the association of obesity with in-hospital mortality of patients with non-variceal gastrointestinal bleeding (50). This may be explained by the complexity of evaluating the outcomes of acute injuries, which should not be attributed to the effect of body weight alone. Fatal injuries brought by decompensated events are far beyond the potential benefits of overweight/obesity. There were some limitations in this study. First, we did not obtain the dynamic changes of BMI during follow-up. Second, we did not have an external validation cohort to verify the present findings. Third, BMI data were missing in a proportion of our AGIB patients, probably producing a selection bias. Fourth, the interventions used during the hospitalization and follow-up, which might be beneficial for the outcomes, were not available. Fifth, we did not evaluate the specific values of muscle mass or the distribution of adipose tissue due to the absence of dualenergy X-ray absorptiometry, waist circumference, and waist to hip ratio.
In conclusion, there is a modest association of overweight/obesity with decreased long-term mortality of patients with cirrhosis. However, BMI cannot act as an independent prognostic predictor of liver cirrhosis. In the future, prospective large-scale studies should be attempted to combine BMI with other indicators involved in measuring muscle mass and adipose tissue to more precisely predict the clinical outcomes of liver cirrhosis.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Medical Ethical Committee of the General Hospital of Northern Theater Command. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.