Frailty is an emerging concept as a strong predictor of outcome in patients with cirrhosis,^1,2 particularly those undergoing liver transplantation (LT). Unlike classic definition of frailty in geriatrics,³ “frailty” in the fields of hepatology/LT mainly focuses on one component of frailty: physical frailty. It refers to the clinical signs of cirrhotic patients’ diminished muscle contractile function, which results in decreased physical performance and impairment.⁴ Although the impact of frailty on health outcomes has been recognized in LT field for decades using so-called eyeball test, implementation into clinical practice has been hampered by a lack of consensus on evaluation tools and their implications for transplant decision-making.

Recently, several quantitative measures of frailty including Fried frailty phenotype,^5,6 Liver Frailty Index⁷ or Short Physical Performance Battery (SPPB)⁶ were developed and demonstrated the association with mortality before and after LT. Most studies focused on outcomes prior to LT, such as delisting in patients awaiting LT. The effects of frailty on liver transplant outcomes, on the other hand, are not established yet. Furthermore, only a few studies have investigated the implications of frailty in living donor transplant recipients. It could be due to the lower prevalence of living donor LT (LDLT) in Western countries, as well as the fact that living donor transplant recipients were less likely to be frail.⁸

Therefore, this study aimed to investigate whether pretransplant frailty, as measured by SPPB, can predict postoperative morbidity and mortality in patients with cirrhosis and prespecified subset after LT.

1. Study design and patients

We retrospectively analyzed data of LT recipients enrolled in a prospectively established LT cohort at Samsung Medical Center, Seoul, Korea between September 2018 and July 2020. Among 282 eligible patients, patients younger than 18 years old (n=10), no apparent cirrhosis (n=17), and missing preoperative SPPB score (n=13) were excluded. This study was approved by Institutional Review Board at the Samsung Medical Center (IRB number: 2021-11-015). Informed consent was waived because the study was based on de-identified existing clinical data routinely collected during hospital visits.

2. Data collection

Patients’ demographic, medical, transplantation, and posttransplant outcome data were collected from the electronic health record. Demographic data included age, sex, and body mass index (BMI). Medical data included comorbidity such as hypertension and diabetes mellitus, etiology of liver disease, history of hepatocellular carcinoma (HCC),⁹ laboratory findings, Child-Pugh classification and Model for End-Stage Liver Disease (MELD) score at the time of LT. Renal dysfunction was defined as baseline serum creatinine >1.5 mg/dL. Information related to transplantation including ABO incompatibility and graft type was also collected. At last, the following posttransplant outcomes were reviewed: posttransplant mortality, length of hospital stays, number of days in the intensive care unit (ICU) after transplantation, postoperative complication within 30 days, acute cellular rejection rate, and graft failure rate. Acute cellular rejection was graded using the Banff rejection activity index on histological examination.¹⁰ Graft failure was diagnosed in cases of clinical or histological signs of reappearing cirrhosis, graft failure leading to re-transplantation, or allograft-related death.¹¹

3. Frailty assessment using the SPPB test

The SPPB test was performed when LT recipients were admitted for pre-LT assessment. Median time interval from the SPPB test to LT was 1.0 month (interquartile range, 0.7 to 1.6 months). It consists of three objective physical assessments of low extremity function: standing balance, walking speed, and repeated char stands. Each of components had a score ranging from 0 to 4 and total score ranged 0 to 12 with higher scores corresponding to better performance. Frailty was defined as a SPPB score of <10 using established cutoff of cirrhotic patients on LT waitlist.¹²

4. Statistical analysis

Continuous variables were presented as the median with interquartile range. The categorical variables were presented as counts with percentages. Differences in baseline characteristics by status of frailty according to SPPB score were compared using the Wilcoxon rank sum or Fisher exact test for continuous and categorical variables, respectively. The primary outcome in this study was overall survival (OS). Posttransplant survival rates were estimated using the Kaplan-Meier method and log-rank test was used to examine differences in survival probabilities between frail and non-frail groups. The Cox proportional hazard regression analysis was performed to find risk factors for short-term and long-term mortality. For this analysis, those variables that were significant at p≤0.1 by univariable analysis were included into a multivariable model.

Other posttransplant outcomes including hospitalization length, postoperative ICU stay, postoperative complication within 30 days, acute cellular rejection and graft failure were compared between frail and non-frail groups using the Wilcoxon rank sum or Fisher exact tests. p<0.05 was defined as statistically significant. Subgroup analysis according to history of HCC and donor type was performed. Lastly, we conducted inverse probability of treatment weighting analysis based on the propensity score to balance the baseline characteristics between frail and non-frail groups. The propensity score for each patient was calculated using a logistic regression model with baseline covariates including gender, renal dysfunction, underlying liver disease, history of HCC, BMI, MELD, and preoperative ICU care. After weighting baseline characteristics, standardized mean difference was recalculated and adequate balance was declared if standardized mean difference <0.2. Posttransplant survival was compared between frail and non-frail groups in weighted study population. Statistical analyses were conducted using SPSS Statistics 27.0 (IBM Corp., Armonk, NY, USA) and SAS version 9.4 (SAS institute Inc., Cary, NC, USA).

1. Patient characteristics

A total of 242 patients were analyzed in this study (Fig. 1). The baseline characteristics are shown in Table 1. The median age was 57 years (range, 50 to 61 years), 31.8% were female and median BMI was 24.4 kg/m². Based on preoperative SPPB scores, 182 patients (75.2%) were classified as non-frail and 60 patients (24.8%) were classified as frail. The baseline characteristics were different in several ways regarding the status of frailty. Compared with non-frail patients, frail patients were more likely to be female (55.0% vs 24.2%) and have higher proportion of renal dysfunction (38.3% vs 13.2%) and alcohol-related cirrhosis as primary etiology of liver disease (53.3% vs 20.9%). Frail patients had higher median MELD score (35 vs 11) and higher incidence of moderate-to-large amount of ascites and hepatic encephalopathy (35.0% vs 14.8% and 35.0% vs 8.2%, respectively). Approximately half of frail patients required ICU admission prior to LT, compared to just 5% of non-frail patients. The majority of LTs (78.1%) were performed on living donors, while only 21.9% were performed on deceased donors. In frail patients, deceased donor LT was conducted more frequently compared to non-frail patients (56.7% vs 10.4%).

Figure 1. Patient flowchart.
SPPB, Short Physical Performance Battery.

Table 1

Baseline Characteristics.

Variable	Total (n=242)	Non-frail (n=182)	Frail (n=60)	p-value
Age, yr	57 (50–61)	57 (51–61)	56 (47–61)	0.22
Sex				<0.001
Female	77 (31.8)	44 (24.2)	33 (55.0)
Male	165 (68.2)	138 (75.8)	27 (45.0)
Body mass index, kg/m2	24.4 (22.5–26.8)	24.7 (23.0–26.8)	23.4 (21.7–26.8)	0.05
Comorbidity
Hypertension	47 (19.4)	36 (19.8)	11 (18.3)	1.00
Diabetes	59 (24.4)	46 (25.3)	13 (21.7)	0.61
Etiology of liver disease				<0.001
Viral hepatitis	145 (59.9)	126 (69.2)	19 (31.7)
Alcohol	70 (28.9)	38 (20.9)	32 (53.3)
Others	27 (11.2)	18 (9.9)	9 (15.0)
History of HCC	138 (57.0)	126 (69.2)	12 (20.0)	<0.001
Laboratory findings
Total bilirubin, mg/dL	2.0 (0.8–14.0)	1.3 (0.7–3.6)	2.5 (1.6–3.2)	<0.001
PT, INR	1.36 (1.10–2.07)	1.24 (1.08–1.61)	2.47 (1.60–3.15)	<0.001
Albumin, g/dL	3.3 (2.8–4.0)	3.4 (2.9–4.1)	3.0 (2.7–3.4)	<0.001
Creatinine, mg/dL	0.8 (0.7–1.2)	0.8 (0.6–1.1)	1.2 (0.7–2.3)	<0.001
Renal dysfunction	47 (19.4)	24 (13.2)	23 (38.3)	<0.001
MELD score	14 (8–30)	11 (7–18)	35 (23–40)	<0.001
Ascites				0.001
Absent	130 (53.7)	108 (59.3)	22 (36.7)
Mild	64 (26.4)	47 (25.8)	17 (28.3)
Moderate-severe	48 (19.8)	27 (14.8)	21 (35.0)
Hepatic encephalopathy	36 (14.9)	15 (8.2)	21 (35.0)	<0.001
Child-Pugh class				<0.001
A	91 (37.6)	87 (47.8)	4 (6.7)
B	60 (24.8)	46 (25.3)	14 (23.3)
C	91 (37.6)	49 (26.9)	42 (70.0)
Preoperative ICU stay, day				<0.001
0	207 (85.5)	174 (95.6)	33 (55.0)
1–3	15 (6.2)	5 (2.7)	10 (16.7)
4–9	10 (4.1)	3 (1.6)	7 (11.6)
≥10	10 (4.1)	0	10 (16.7)
Donor type				<0.001
Living donor	189 (78.1)	163 (89.6)	26 (43.3)
Deceased donor	53 (21.9)	19 (10.4)	34 (56.7)

Data are presented as median (range) or number (%). p-values estimated by chi-square test or Fisher exact test for categorical variables and Mann-Whitney test for continuous variables..

HCC, hepatocellular carcinoma; PT, prothrombin time; INR, international normalized ratio; MELD, Model for End-Stage Liver Disease; ICU, intensive care unit..

2. Frailty and posttransplant outcome

Mortality (17%) was observed in 40 patients with a median follow-up of 340 days (range, 2 to 841 days). The vast majority of deaths (87.5%) were reported during the first 180 days after LT. The OS was 94.2% at 1 month, 85.2% at 6 months, and 79.6% at 24 months. OS rates were different according to the status of frailty. Frail patients showed poor survival probabilities compared to non-frail patients (98.3% vs 81.7% at 1 month, 91.4% vs 68.2% at 6 months, and 87.1% vs 58.8% at 24 months, respectively) (p<0.001) (Fig. 2). The median hospitalization length and postoperative ICU stay was 22 days and 4 days, respectively, in non-frail group while those were longer in frail group, with 24 days and 6 days, respectively (Supplementary Table 1). With respect to postoperative complications, 155 patients (64.0%) experienced complications within 30 days after LT. Among them, 98 out of 155 patients (63.2%) experienced major complications requiring interventions, defined as grade 3 or higher of the Clavien-Dindo classification. Anastomotic biliary complications including leakage or stenosis (9.1%), postoperative bleeding (8.3%), and anastomotic vascular complications including portal vein thrombosis, hepatic artery thrombosis or stenosis (7.4%) were observed with the highest frequency as a major complication. Patients in pretransplant frail group were more likely to experience complications within 30 days after LT (78.3% vs 59.3%, p=0.008) (Table 2). Of note, the frequency of major complications was higher in frail patients compared to non-frail patients (58% vs 35%, p=0.001) (Supplementary Table 2). Regarding acute allograft rejection, 25 patients experienced acute cellular rejection and two patients experienced graft versus host disease. There was no significant difference in rejection episodes between patients in frail and non-frail groups. In contrast, graft failure was more common in frail patients compared to non-frail patients (21.7% vs 4.9%, p<0.001) (Table 2).

Figure 2. Kaplan-Meier estimates of overall survival according to the frailty status.

Table 2

Posttransplant Outcomes.

	Total (n=242)	Non-frail (n=182)	Frail (n=60)	p-value
Posttransplant complications within 30 days after liver transplantation	155 (64.0)	108 (59.3)	47 (78.3)	0.008
Acute rejection	25 (10.3)	16 (8.8)	9 (15.0)	0.22
Graft failure	21 (9.1)	9 (4.9)	13 (21.7)	<0.001

Data are presented as number (%)..

3. Risk factors for posttransplant mortality

To identify independent risk factors for posttransplant mortality, we performed a Cox regression analysis. In the univariable analysis, male, normal or overweight BMI, and history of HCC were associated with favorable posttransplant outcome while high MELD score (≥30), preoperative ICU care, and status of frailty were associated with poor posttransplant outcome (Table 2). Frailty was an independent risk factor for posttransplant mortality (hazard ratio, 2.38; 95% confidence interval, 1.02 to 5.57) after adjustment of sex, renal dysfunction, BMI, MELD score, preoperative ICU care, and history of HCC (Table 3).

Table 3

Univariable and Multivariable Cox Regression Analysis for Postoperative Mortality.

Factor	Univariable analysis			Multivariable analysis
	HR (95% CI)	p-value		HR (95% CI)	p-value
Age	1.00 (0.97–1.04)	0.81
Male (vs female)	0.54 (0.29–1.00)	0.05		0.86 (0.43–1.71)	0.67
Diabetes	1.21 (0.60–2.42)	0.59
Renal dysfunction	1.86 (0.95–3.66)	0.07		0.52 (0.20–1.34)	0.18
Body mass index, kg/m2
<18.5	Reference			Reference
18.5 to <25	0.18 (0.07–0.47)	<0.001		0.20 (0.07–0.58)	0.003
≥25	0.15 (0.05–0.42)	<0.001		0.22 (0.07–0.70)	0.01
MELD ≥30	3.67 (1.97–6.83)	<0.001		2.75 (0.86–8.82)	0.09
Preoperative ICU care	4.41 (2.32–8.37)	<0.001		1.43 (0.54–3.81)	0.47
Etiology of liver disease
Alcohol (vs non-alcohol)	1.84 (0.98–3.44)	0.06		1.29 (0.61–2.74)	0.50
History of HCC	0.47 (0.25–0.89)	0.02		1.48 (0.60–3.64)	0.40
Frail (vs non-frail)	4.00 (2.15–7.44)	<0.001		2.38 (1.02–5.57)	0.04

Donor type was excluded from the multivariable Cox regression model due to high intercorrelations between donor type and Model for End-Stage Liver Disease (MELD) ≥30..

HR, hazard ratio; CI, confidence interval; ICU, intensive care unit; HCC, hepatocellular carcinoma..

4. Frailty and posttransplant outcome in prespecified subgroup

We further evaluated the association between frailty and posttransplant mortality in prespecified subgroups defined by history of HCC and donor type. Among 138 patients (57.0%) who had history of HCC, only 12 patients (8.7%) were classified as frail group. Nevertheless, frail patients showed notably lower posttransplant survival rates compared to non-frail patients, with the survival disparity being apparent 1 month after transplant (79.2% vs 96.4% at 1 month, 68.6% vs 88.9% at 6 months, p=0.004 for non-HCC patients; 75.0% vs 98.4 at 1 month, 65.6% vs 91.7% 6 months, p=0.001 for HCC patients) (Fig. 3A and 3B). In terms of donor type, 189 patients (78.1%) received LT from living donors, and 26 patients (13.8%) were classified as frail group. In comparison to non-frail patients, frail patients had worse posttransplant survival rates (84.6% vs 98.8% at 1 month, 80.6% vs 92.3% at 6 months, p=0.025) (Fig. 3C). Similarly, the discrepancy in survival was most pronounced 1 month following transplantation. The survival difference between frail and non-frail patients was also observed in patients who underwent deceased donor LT, albeit statistical significance was not achieved due to the small number of enrolled patients (73.5% vs 89.5% at 1 month, 58.8% vs 78.9% at 6 months, p=0.099) (Fig. 3D).

Figure 3. Kaplan-Meier estimates of 6-month posttransplant survival in each subgroup of patients. (A) Non-hepatocellular carcinoma (HCC), (B) HCC, (C) living donor liver transplantation, (D) deceased donor liver transplantation.

5. Sensitivity analysis

After balancing the baseline characteristics by inverse-probability-of-treatment-weighting, there was no longer any significant difference between the frail and non-frail groups (Supplementary Table 3). The Kaplan-Meier survival analysis in weighted study population also revealed that frail patients had a significantly lower posttransplant survival compared to non-frail patients (63.2% vs 89.8% at 6 months, p=0.002) (Fig. 4).

Figure 4. Kaplan-Meier estimates of overall survival according to the frailty status after inverse probability of treatment weighting.

In this study, we have shown a lower posttransplant survival in frail patients (SPPB score <10) when compared with non-frail patients (SPPB score ≥10). Notably, frailty was an independent predictor of posttransplant survival after adjusting for confounders including MELD score. Moreover, frail patients were more likely to experience both major posttransplant complications within 30 days and graft failure along with longer hospital length including days of ICU stay. In predefined subgroups who had history of HCC or graft from living donor, non-frail patients had a greater survival rate than those who were frail, although the survival benefit was reduced. The survival discrepancy was most pronounced within 30 days following LT.

Although the underlying pathophysiology of frailty in liver disease is not completely understood, frailty may result from multiple organ system dysfunction including neuromuscular, endocrine, immunological, and skeletal muscle dysfunction.¹³ One of the major predisposing factors that contributes to frailty is cirrhosis. Increased protein catabolism accelerates muscle breakdown¹⁴ as well as impaired ammonia clearance results in increased systemic ammonia concentration with myotoxic effects.^15,16 Complications of cirrhosis such as encephalopathy¹⁵ and ascites¹⁷ also contribute to frailty by decreased appetite, reduced physical activity and frequent hospitalization. Hence, the severity of liver disease has a direct impact on the prevalence of frailty. In this study, patients in frail group showed more than 3-fold higher median MELD score than the patients in non-frail group (Table 1).

The etiology of liver disease has been known to be associated with the development of sarcopenia. Compared to other etiologies, alcohol exposure has demonstrated a rapid reduction of muscle mass through increasing muscle autophagy,¹⁸ inhibiting proteasome activity,¹⁹ and decreasing the anabolic hormone including insulin-like growth factor 1.²⁰ Consistent with previous studies, alcohol-related cirrhosis was more prevalent in frail patients in the current study (Table 1). Nevertheless, frailty was found as an independent factor for posttransplant mortality after adjusting severity of liver disease as determined by the MELD score and etiology of liver disease. It is consistent to previous findings of multicenter study that the frailty was more frequently observed in patients with ascites or hepatic encephalopathy, but frailty was independently associated with waitlist mortality in ascites- and hepatic encephalopathy-adjusted models.²¹ These findings suggest that, regardless of liver disease-related parameters, frailty is a distinct assessment tool for predicting mortality in LT recipients.

Despite the fact that many previous studies have demonstrated strong and consistent associations between frailty and adverse pre- and post-LT outcomes in cirrhotic patients, how pretransplant frailty impacts outcomes after LT are still inconclusive. In this study, mortality in frail patients were mostly observed within 90 days after LT (n=18/21, 85.7%). Infection was the leading cause of mortality (n=9/18, 50.0%), followed by primary graft non-function (n=6/18, 33.3%), bleeding (n=2/18, 11.1%), and stroke (n=1/18, 5.6%). In addition, frail patients had a higher incidence of postoperative complications related to anastomotic biliary or vascular complications and graft rejection. Moreover, frail patients had a higher rate of early mortality than non-fail patients, regardless of history of HCC or donor type. These findings imply that patients with pretransplant frailty may be vulnerable to postoperative complications such as infection, bleeding, and graft failure, resulting in higher posttransplant mortality.

Several lines of evidence suggest that the pro-inflammatory state in frail patients are linked to increased rates of acute cellular rejection and graft loss in kidney, lung and liver transplant recipients.^22-25 In this study, the rate of acute cellular rejection was higher in frail patients than in non-frail patients, albeit statistical significance was not found (15.0% vs 8.8%, p=0.22). In contrast, the rate of graft failure during the study period was significantly higher in frail patients than in non-frail patients (21.7% vs 4.9%, p<0.001). However, this result needs careful interpretation since we did not collect information regarding immunosuppressive agents, and multivariable analysis was not performed due to the small number of patients. Further prospective studies are warranted to investigate this association in a larger cohort of LT recipients.

Given that literatures for implication of frailty in patients with cirrhosis accumulated, experts in American Association for the Study of Liver disease and European Association for the Study of the Liver recently consented assessment for frailty with a standardized tool at baseline and longitudinally in all patients with cirrhosis.^26,27 However, there are insufficient data to recommend the use of one frailty tool over another. The Karnofsky Performance Status is one of the most validated tools for predicting mortality not only in general oncology practice but also in patients with end-stage liver disease. However, Karnofsky Performance Status score is vulnerable to bias and is less reproducible due to its subjective nature. In contrast, SPPB score is a simple, objective, and reproducible tool for assessing frailty that does not require specialized equipment or trained personnel compared with other objective metrics such as Fried frailty phenotype, Liver Frailty Index, and cardiopulmonary exercising testing. A study of 294 listed LT patients demonstrated that SPPB score was associated with risk of waitlist mortality (hazard ratio, 1.19; 95% confidence interval, 1.07 to 1.32 with each 1-unit decrease).⁶ Another study of 300 outpatients with cirrhosis reported that SPPB <10 increased rates of unplanned hospitalization or death by 2.5 times.¹² This study found that pretransplant SPPB score <10 was associated with posttransplant outcomes in LT patients, including posttransplant survival, complications, ICU stays, and allograft rejection.

In this study, 78.1% of patients received LDLT, which is consistent with the previous report from Korean transplantation registry.²⁸ The prevalence of frailty (SPPB <10) was 13.8% among LDLT patients, which was lower than the previously reported prevalence of frailty in patients with advanced liver disease, which ranged from 17% to 43%.¹³ This phenomenon could be attributed to a selection bias where patients who are “too frail to transplant” are more likely to be delisted if they are scheduled to receive grafts from living donors. Nevertheless, frail patients showed significantly lower 6-month posttransplant survival compared to non-frail patients (80.6% vs 92.9%, p=0.025). Consistent findings were reported in the previous Japanese study of 124 patients undergoing LDLT. They found that low skeletal muscle mass was associated with worse outcomes and perioperative nutritional therapy significantly increased OS.²⁹ This finding implicated the importance of “prehabilitation,”⁴ which refers to multidisciplinary training to enhance physical strength and nutritional status prior to LT. Although clinical studies focusing on impact of prehabilitation was limited to LT patients, recent meta-analysis of 15 randomized clinical trials reported prehabilitation before major abdominal surgery significantly reduced OS and pulmonary morbidity.³⁰

There are several limitations in this study. First, SPPB score was evaluated only at baseline in this study. According to previous studies utilizing the Karnofsky Performance Status and Liver Frailty Index, patients who did not improve their performance status following LT had worse outcomes.^31,32 The impact of change in SPPB score after LT cannot be determined from the current study. Second, although SPPB score has not been validated in an inpatient setting, adjunctive performance scales including Fried frailty phenotype or markers of frailty such as skeletal muscle index were not evaluated in this study. Lastly, the retrospective nature of the study and the small number of patients enrolled from a single site lead to selection bias, overfitting in multivariable regression model, and limited subgroup analyses. To reduce selection bias, we conducted inverse-probability-of-treatment-weighting analysis and obtained comparable results. Additionally, we compared the regression coefficient with the regression model using stepwise variable selection to detect overfitting, and a similar result was observed (hazard ratio, 2.68; 95% confidence interval, 1.26 to 4.71; p=0.01) (Supplementary Table 4).

Despite these caveats, our findings may help predict posttransplant outcomes in LT recipients, including complications within 30 days, graft failure and OS.

In conclusion, assessment of pretransplant frailty using SPPB provides crucial predictive information for clinical outcomes in cirrhotic patients undergoing LT. Measuring SPPB score and providing appropriate prehabilitation before LT might be beneficial to optimize care in LT recipients.

Supplementary materials can be accessed at https://doi.org/10.5009/gnl220337.

This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (NRF-2016R1C1B2015463 and NRF-2019R1C1C1007729). This research was also supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (HR20C0025).

No potential conflict of interest relevant to this article was reported.

No potential conflict of interest relevant to this article was reported.

Study concept and design: W.K., J.M.K., W.H.C. Data acquisition: J.K. Data analysis and interpretation: M.J.G., D.H.S., G.Y.G. Drafting of the manuscript: M.J.G., W.H.C. Critical revision of the manuscript for important intellectual content: W.K., J.M.K., Y.H.P. Statistical analysis: M.J.G., J.K. Obtained funding: W.K. Administrative, technical, or material support: M.S.C., J.H.L. Study supervision: K.C.K., S.W.P. Approval of final manuscript: all authors.