Association Between Liver Transplant Wait-list Mortality and Frailty Based on Body Mass Index

Key Points

Question  Among liver transplant candidates, does the association of frailty and increased risk of wait-list mortality differ by body mass index?

Findings  In this cohort study, the prevalence of frailty among nonobese candidates and those with class 1 obesity and class 2 or greater obesity was similar. Frailty was associated with a 2-fold higher risk of mortality in nonobese candidates and those with class 1 obesity, and a 3-fold higher risk of mortality in candidates with class 2 or greater obesity.

Meaning  Frailty assessment at transplant evaluation may aid in identification of patients at higher risk of death, mainly obese patients for whom the clinician’s visual evaluation may be less reliable to assess muscle mass and nutritional status.

Importance  Among liver transplant candidates, obesity and frailty are associated with increased risk of death while they are on the wait-list. However, use of body mass index (BMI) may not detect candidates at a higher risk of death owing to the fact that ascites and muscle wasting are seen across transplant candidates of all BMI measurements.

Objective  To evaluate whether the association between wait-list mortality and frailty varied by BMI of liver transplant candidates.

Design, Setting, and Participants  A prospective cohort study was conducted at 9 liver transplant centers in the United States from March 1, 2012, to May 1, 2018, among 1108 adult liver transplant candidates without hepatocellular carcinoma.

Exposures  At outpatient evaluation, the Liver Frailty Index score was calculated (grip strength, chair stands, and balance), with frailty defined as a Liver Frailty Index score of 4.5 or more. Candidates’ BMI was categorized as nonobese (18.5-29.9), class 1 obesity (30.0-34.9), and class 2 or greater obesity (≥35.0).

Main Outcomes and Measures  The risk of wait-list mortality was quantified using competing risks regression by candidate frailty, adjusting for age, sex, race/ethnicity, Model for End-stage Liver Disease Sodium score, cause of liver disease, and ascites, including an interaction with candidate BMI.

Results  Of 1108 liver transplant candidates (474 women and 634 men; mean [SD] age, 55 [10] years), 290 (26.2%) were frail; 170 of 670 nonobese candidates (25.4%), 64 of 246 candidates with class 1 obesity (26.0%), and 56 of 192 candidates with class 2 or greater obesity (29.2%) were frail (P = .57). Frail nonobese candidates and frail candidates with class 1 obesity had a higher risk of wait-list mortality compared with their nonfrail counterparts (nonobese candidates: adjusted subhazard ratio, 1.54; 95% CI, 1.02-2.33; P = .04; and candidates with class 1 obesity: adjusted subhazard ratio, 1.72; 95% CI, 0.99-2.99; P = .06; P = .75 for interaction). However, frail candidates with class 2 or greater obesity had a 3.19-fold higher adjusted risk of wait-list mortality compared with nonfrail candidates with class 2 or greater obesity (95% CI, 1.75-5.82; P < .001; P = .047 for interaction).

Conclusions and Relevance  This study’s finding suggest that among nonobese liver transplant candidates and candidates with class 1 obesity, frailty was associated with a 2-fold higher risk of wait-list mortality. However, the mortality risk associated with frailty differed for candidates with class 2 or greater obesity, with frail candidates having a more than 3-fold higher risk of wait-list mortality compared with nonfrail patients. Frailty assessments may help to identify vulnerable patients, particularly those with a BMI of 35.0 or more, in whom a clinician’s visual evaluation may be less reliable to assess muscle mass and nutritional status.

Paralleling escalating national trends in obesity, the proportion of candidates with obesity on the liver transplant wait-list is increasing, with up to 23% of candidates with class 1 obesity (ie, with a body mass index [BMI] of 30.0-34.9 [calculated as weight in kilograms divided by height in meters squared]), 10% of candidates with class 2 obesity (BMI of 35.0-39.9), and 4% of candidates with class 3 obesity (BMI≥40.0).¹ For liver transplant candidates, obesity is associated with increased risk of wait-list mortality,² wait-list dropout,¹ and prolonged waiting times for transplantation,³ yet patients with obesity still derive a survival benefit from liver transplantation.² However, BMI may not accurately reflect true body fat or size, as ascites and volume overload can often confound body weight measurements.⁴ Thus, better metrics are needed to assess risk in liver transplant candidates with obesity.

Frailty, a state of decreased physiologic reserve and increased vulnerability to stressors, may represent a novel metric to better assess risk in patients with cirrhosis. Frailty was initially described in community-dwelling older adults,⁵ but the construct has expanded to patients with cirrhosis through the Liver Frailty Index (LFI; consisting of 3 performance-based tests: grip strength, chair stands, and balance), derived and validated specifically in patients with cirrhosis who are awaiting liver transplantation.^6-8 Up to 25% of liver transplant candidates are frail,^6,7 and frailty is associated with an increased risk of wait-list mortality,^7,8 increased hospitalizations,⁹ and depression¹⁰ in liver transplant candidates. The increased risk of wait-list mortality in frail candidates is independent of age,¹¹ hepatic encephalopathy, and ascites⁸; thus, frailty conceptually captures the physiological reserve of a patient. Therefore, frailty may improve selection of transplant candidates who potentially have a higher risk for adverse outcomes, but it is unknown if BMI modifies the association between frailty and increased mortality risk among candidates on the liver transplant wait-list.

To clarify and quantify the interaction of candidate BMI and frailty on mortality for liver transplant candidates, we sought to quantify the prevalence of frailty, compare individual elements of the LFI score, and quantify the association of frailty and wait-list mortality in liver transplant candidates by BMI.

This was a prospective, longitudinal cohort study of 1108 participants 18 years or older who were being evaluated in the outpatient setting for liver transplant at the following institutions: University of California San Francisco (n = 765) from March 1, 2012, to April 1, 2018; Johns Hopkins Hospital, Baltimore, Maryland (n = 111), from August 9, 2016, to May 1, 2018; Columbia University, New York, New York (n = 50), from February 1, 2016, to April 1, 2018; Baylor University Medical Center, Dallas, Texas (n = 49), from January 1, 2016, to April 1, 2018; Duke University, Durham, North Carolina (n = 40), from May 1, 2016, to April 1, 2018; Loma Linda University Health, Loma Linda, California (n = 32), from July 1, 2016, to April 1, 2018, University of Pittsburgh, Pittsburgh, Pennsylvania (n = 36), from February 1, 2017, to April 1, 2018; University of Arkansas, Little Rock (n = 19), from March 1, 2017, to April 1, 2018; and Northwestern University (n = 6), Chicago, Illinois, from October 1, 2016, to April 1, 2018. No participants were lost to follow-up during the study period. We excluded participants with hepatocellular carcinoma because their wait-list mortality was expected to differ substantially from participants with other causes of liver failure. The institutional review board from each site approved the study. Written informed consent was obtained from each participant prior to enrollment in the study.

We measured the LFI as described below. Additional participant characteristics were abstracted from the electronic medical record (age, sex, race/ethnicity [white, African American, Asian, Hispanic, or other], indication for liver transplant, BMI, Model for End-stage Liver Disease Sodium [MELDNa] score, type 1 or 2 diabetes, hypertension, coronary artery disease, history of stroke, ascites, and hepatic encephalopathy). Ascites was ascertained at the baseline study visit from the hepatologists’ recorded physical examination or the management plan associated with the clinic visit that occurred on the same day at the assessment of frailty.⁸ Ascites was categorized as absent if ascites was not present on physical examination or as present if ascites was present on examination and/or the patient was undergoing large-volume paracentesis. Hepatic encephalopathy was defined as time greater than 60 seconds to complete the Numbers Connection Test as previously used in liver candidate cohorts.^7,11,12 A previous study reported that alternate definitions of ascites or hepatic encephalopathy (eg, based on the medical history or current medication use) did not substantially change the association between LFI score and wait-list mortality.⁸ Physicians were not aware of the measured LFI scores at liver transplant evaluation.

We measured the LFI as previously defined in participants with end-stage liver disease.^6,7,11,13 The LFI is composed of 3 components (grip strength, balance testing, and chair stands) and takes fewer than 5 minutes to administer in the clinic. These objective measures were recorded at the time of clinic liver transplant evaluation using the following:

1. Grip strength: mean of 3 trials in the patient’s dominant hand using a hand dynamometer, measured in kilograms;

2. Chair stands: measured as the number of seconds it takes to stand from seated in a chair 5 times with the patient’s arms folded across the chest; and

3. Balance testing: measured as the number of seconds that a patient can balance in 3 positions (feet side to side, semitandem, and tandem) for a maximum of 10 seconds each.

The LFI was calculated (https://liverfrailtyindex.ucsf.edu/) as: (−0.330 × sex-adjusted grip strength) + (−2.529 × number of chair stands per second) + (−0.040 × balance time) + 6.

Standard cutoffs were used to define robust (LFI score <3.2), prefrail (LFI score of 3.2 to <4.5), and frail (LFI score ≥4.5).⁶

Among participants, the risk of wait-list mortality was estimated at 6 months, 1 year, and 3 years using a competing risk framework by candidate BMI (nonobese [18.5-29.9], class 1 obesity [30.0-34.9], and class 2 or greater obesity [≥35.0]), as defined by World Health Organization classification,¹⁴ and frailty status. Cumulative incidence curves for wait-list mortality were constructed using frailty and candidate BMI. Transplantation was considered a competing risk, and date the candidate was placed on the wait-list was the time origin. Subhazard ratios (SHRs) of wait-list mortality by candidate BMI were obtained using the Fine and Gray¹⁵ method for competing risks. Multivariable models included candidate age, sex, race/ethnicity, indication for transplant, MELDNa score, and ascites. To test whether wait-list mortality varied by candidate BMI, an interaction between candidate BMI and frailty was explored using a Wald test.

We then performed a sensitivity analysis excluding patients with refractory ascites, defined as repeated need for paracentesis, to evaluate the stability of our findings in patients without significant volume overload. We estimated the SHR of wait-list mortality by candidate BMI and, in a separate model, included an interaction between candidate BMI and frailty. Models were adjusted for candidate age, MELDNa score, and ascites.

Candidate characteristics were compared with χ² test for categorical variables and t tests or analysis of variance for continuous variables. All P values were from 2-sided tests, and results were deemed statistically significant at P < .05. All analyses were performed using Stata, version 14.2/MP (StataCorp).

Among the 1108 liver transplant candidates, the mean (SD) age was 55 (10) years, 474 (42.8%) were women, and 774 (69.9%) were white. With respect to BMI, 246 candidates (22.2%) had class 1 obesity and 192 (17.3%) had class 2 or greater obesity. The mean (SD) MELDNa score was 18 (6), and 290 candidates (26.2%) were frail.

Candidates with class 1 and class 2 or greater obesity were more likely to be older than nonobese candidates (mean [SD]: class 1 obesity, 57.3 [8.1] years; class 2 or greater obesity, 55.8 [8.7] years; and nonobese, 54.6 [10.9] years; P = .001) (Table 1). Candidates with class 1 and class 2 or greater obesity were more likely than nonobese patients to have nonalcoholic steatohepatitis (class 1 obesity, 58 of 246 [23.6%]; class 2 or greater obesity, 56 of 192 [29.2%]; and nonobese, 80 of 670 [11.9%]; P < .001) and were less likely to have alcoholic cirrhosis (class 1 obesity, 63 of 246 [25.6%]; class 2 or greater obesity, 41 of 192 [21.4%]; and nonobese, 178 of 670 [26.6%]; P < .001). Nonobese patients, candidates with class 1 obesity, and candidates with class 2 or greater obesity had similar mean (SD) MELDNa scores (nonobese, 18.2 [5.8]; class 1 obesity, 17.7 [5.8]; and class 2 or greater obesity, 18.8 [6.1]; P = .14), prevalence of ascites (nonobese, 248 of 670 [37.0%]; class 1 obesity, 92 of 246 [37.4%]; and class 2 or greater obesity, 72 of 192 [37.5%]; P = .99), and prevalence of hepatic encephalopathy (nonobese, 131 of 670 [19.5%]; class 1 obesity, 53 of 246 [21.5%]; and class 2 or greater obesity, 44 of 192 [21.3%]; P = .82).

Candidates with class 1 obesity and class 2 or greater obesity were more likely than nonobese candidates to have diabetes (class 1 obesity, 99 of 246 [40.2%]; class 2 or greater obesity, 70 of 192 [36.5%]; and nonobese, 172 of 670 [25.9%]; P < .001) and hypertension (class 1 obesity, 108 of 246 [43.9%]; class 2 or greater obesity, 84 of 192 [43.8%]; and nonobese, 237 of 670 [35.6%]; P = .02) than nonobese candidates (Table 1). Nonobese candidates, those with class 1 obesity, and those with class 2 or greater obesity had a similar prevalence of stroke (nonobese, 9 of 670 [1.4%]; class 1 obesity, 7 of 246 [2.8%]; and class 2 or greater obesity, 4 of 192 [2.1%]; P = .31) and coronary artery disease (nonobese, 33 of 670 [5.0%]; class 1 obesity, 22 of 246 [8.9%]; and class 2 or greater obesity, 12 of 192 [6.3%]; P = .08).

The prevalence of frailty was similar across candidates who were nonobese, had class 1 obesity, and had class 2 or greater obesity (nonobese, 170 of 670 [25.4%]; class 1 obesity, 64 of 246 [26.0%]; and class 2 or greater obesity, 56 of 192 [29.2%]; P = .57) (Table 1) as was the prevalence of robustness (nonobese, 103 of 670 [15.4%]; class 1 obesity, 27 of 246 [11.0%]; and class 2 or greater obesity, 22 of 192 [11.5%]; P = .14) (Figure 1). The mean (SD) LFI score was similar across all BMI categories: 4.0 (0.9) for nonobese candidates, 4.1 (0.8) for those with class 1 obesity, and 4.2 (0.9) for those with class 2 or greater obesity (P = .08) (Table 2). The median grip strength and median chair stand components of the LFI were similar across all BMI categories (grip strength, men: nonobese, 33 kg [interquartile range (IQR), 26-39 kg]; class 1 obesity, 32 kg [IQR, 27-41 kg]; and class 2 or greater obesity, 33 kg [IQR, 27-39 kg]; P = .10; grip strength, women: nonobese, 20 kg [IQR, 16-24 kg]; class 1 obesity, 19 kg [IQR, 16-24 kg]; and class 2 or greater obesity, 22 kg [IQR, 17-27 kg]; P = .93; chair stand: nonobese, 15 seconds [IQR, 13-17 seconds]; class 1 obesity, 17 seconds [IQR, 13-20 seconds]; and class 2 or greater obesity, 14 seconds [IQR, 11-18 seconds]; P = .46) (Table 2). However, candidates with class 2 or greater obesity had poorer median performance for balance testing scores compared with nonobese candidates and those with class 1 obesity (nonobese, 30 seconds [IQR, 30-30 seconds]; class 1 obesity, 30 seconds [IQR, 30-30 seconds]; and class 2 or greater obesity, 30 seconds [IQR, 29-30 seconds]; P = .04) (Table 2). In addition, the percentage of candidates in each BMI category who could balance for the full 30 seconds decreased as BMI increased: 80.4% (539 of 670) of nonobese candidates, 77.2% (190 of 246) of candidates with class 1 obesity, and 72.9% (140 of 192) of candidates with class 2 or greater obesity.

The cumulative incidence of wait-list mortality at 6 months after being placed on the wait-list was 4.7% (31 of 670) in nonobese candidates, 5.2% (13 of 246) in those with class 1 obesity, and 5.6% (11 of 192) in those with class 2 or greater obesity; at 1 year was 10.0% (67 of 670) in nonobese candidates, 10.7% (26 of 246) in those with class 1 obesity, and 11.6% (22 of 192) in those with class 2 or greater obesity; and at 3 years was 19.8% (133 of 670) in nonobese candidates, 21.3% (52 of 246) in those with class 1 obesity, and 22.9% (44 of 192) in those with class 2 or greater obesity. After adjustment, candidates with class 1 obesity and class 2 or greater obesity had a similar risk of wait-list mortality compared with nonobese candidates (class 1 obesity: adjusted SHR [aSHR], 1.12; 95% CI, 0.81-1.54; P = .46; and class 2 or greater obesity: aSHR, 1.16; 95% CI, 0.80-1.68; P = .35).

The cumulative incidence of wait-list mortality in frail vs nonfrail candidates at 6 months was 8.0% (23 of 290) vs 4.0% (33 of 818), at 1 year was 16.5% (48 of 290) vs 8.4% (68 of 818), and at 3 years was 32.0% (93 of 290) vs 17.0% (139 of 818). After adjustment, frailty was associated with a nearly 2-fold higher risk of wait-list mortality (aSHR, 1.85; 95% CI, 1.36-2.50; P < .001).

The association of frailty with mortality increased as candidate BMI increased. For frail nonobese candidates, the risk of mortality was 1.5-fold higher than nonfrail nonobese candidates (aSHR, 1.54; 95% CI, 1.02-2.33; P = .04); for frail candidates with class 1 obesity, the risk of mortality was 1.7-fold higher than nonfrail candidates with class 1 obesity (aSHR, 1.72; 95% CI, 0.99-2.99; P = .06); and for frail candidates with type 2 or greater obesity, the risk of mortality was 3.2-fold higher than nonfrail candidates with class 2 or greater obesity (aSHR, 3.19; 95% CI, 1.75-5.82; P < .001) (Table 3). The association between increased wait-list mortality and frailty was not different by candidate BMI among nonobese candidates and those with class 1 obesity (P = .75 for interaction). However, the association between wait-list mortality and frailty varied and was stronger among candidates with class 2 or greater obesity (P = .047 for interaction) (Table 3 and Figure 2).

After exclusion of 78 participants with refractory ascites, frail participants were still at a 1.77-fold increased risk of wait-list mortality compared with nonfrail participants (95% CI, 1.31-2.39; P < .001). The association between increased risk of wait-list mortality and frailty did not differ by candidate BMI among nonobese candidates and those with class 1 obesity (P = .84 for interaction). Specifically, frail nonobese candidates had a higher risk of wait-list mortality compared with nonfrail nonobese candidates (aSHR, 1.55; 95% CI, 1.01-2.35; P = .04), as well as a trend toward higher wait-list mortality in frail candidates with class 1 obesity compared with nonfrail candidates with class 1 obesity (aSHR, 1.67; 95% CI, 0.95-2.95; P = .08). Similarly, the association between wait-list mortality and frailty was not different for candidates with class 2 or greater obesity (P = .18 for interaction); frail candidates with class 2 or greater obesity had a 2.57-fold higher risk of wait-list mortality compared with nonfrail candidates with class 2 or greater obesity (95% CI, 1.40-4.71; P = .002).

In this 9-center prospective cohort study of frailty in 1108 liver transplant candidates, we report 2 novel findings. First, we observed that the prevalence of frailty, as measured by the composite LFI score, was similar across the spectrum of BMIs (ie, nonobese, class 1 obesity, and class 2 or greater obesity), with more than 1 in 4 candidates meeting the criteria for frailty. Next, frailty was associated with a 2-fold higher risk of wait-list mortality and as candidate BMI increased, the risk of wait-list mortality increased. In addition, the mortality risk associated with frailty differed for candidates with class 2 or greater obesity, with frail candidates having more than 3 times higher risk of wait-list mortality vs nonfrail candidates.

We originally sought to explore the association between frailty and BMI owing to concerns that performance-based testing of grip strength and chair stands might be altered by BMI alone. In the original Fried Frailty Phenotype, scoring of grip strength to calculate the frailty index was adjusted by BMI.⁵ In addition, there is a theoretical concern that obesity may be associated with the time it takes an individual with obesity to perform 5 chair stands owing to the additional body weight. If LFI scores varied significantly by obesity status, this finding might suggest that the LFI score captured obesity itself in patients with cirrhosis, rather than the true construct of decreased physiologic reserve and increased vulnerability to health stressors as originally described in community-dwelling older adults.⁵ Instead, we observed that LFI score and rates of frailty were similar across obesity categories, demonstrating the independent association of frailty with mortality and underscoring the clinical significance of assessing frailty in patients with cirrhosis.

Participants with class 2 or greater obesity displayed poorer balance than nonobese candidates and those with class 1 obesity. This finding is consistent with studies evaluating balance in other populations showing that obesity is associated with poor balance in both younger and older adults.^16,17 Given the relatively small contribution of balance to the LFI score, as well as the relatively small difference in balance time by obesity status, this statistical difference in balance did not result in a clinically (or statistically) meaningful difference in LFI score by obesity status.

We believe that our analyses demonstrating the interaction between class 2 or greater obesity and frailty on the outcome of wait-list mortality is of particular significance to the hepatology and liver transplant communities. Obesity is rising in prevalence among patients with end-stage liver disease, especially those seeking liver transplantation. Alongside this rise is the increasing recognition of the association of sarcopenia with transplant-related outcomes, which is a dominant component of the frailty phenotype.^18,19 Although muscle wasting can often be identified relatively easily in nonobese patients, subcutaneous adipose tissue may mask underlying muscle wasting, making it nearly undetectable until the most advanced stages, if at all. For these patients, the LFI score may enhance risk prediction. Furthermore, identification of frail patients may guide pretransplant interventions such as prehabilitation, which have been studied in candidates for kidney transplant.²⁰ Prehabilitation appears to be a feasible option in candidates for liver transplant to improve frailty status prior to transplant and is currently being studied at several centers.²¹

We acknowledge the following limitations. First, we could not accurately classify dry weight because we did not have exact volume status assessment on patients. Although sensitivity analysis excluding participants with refractory ascites confirmed that frailty was associated with an increased risk of wait-list mortality for individuals in all BMI categories, we were underpowered to detect an interaction; however, the direction, magnitude, and dose response seen between wait-list mortality and frailty with increasing candidate BMI were consistent with the primary analysis. However, the proportion of participants with ascites did not differ by BMI category, and a previous study has shown that the association between frailty and wait-list mortality did not change based on different methods of ascites ascertainment (eg, medical history, current medication use, or current physical examination).⁸ Last, this study enrolled only outpatients; thus, our findings may not be generalizable to the inpatient population. Furthermore, frailty has been associated with poor surgical outcomes in other populations.^22,23 However, to our knowledge, the association between frailty and BMI has not been reported in other surgical populations prior to surgery, and thus may be a future area of study.

Despite these limitations, to our knowledge this is the largest study to date evaluating the association between frailty, obesity, and wait-list mortality in patients with end-stage liver disease, with important clinical implications. Our findings suggest that liver transplant candidates with obesity who are frail should be counseled on their potential risk of wait-list mortality in excess of their liver disease severity. More important, our findings highlight the importance of frailty in the risk assessment of patients with obesity with cirrhosis and that frailty, not necessarily obesity, matters. There appears to be a need to determine whether a similar interaction between frailty and obesity exists with respect to the outcome of death after liver transplantation is an area of future research for the Multi-Center Functional Assessment in Liver Transplantation (FrAILT) Study. In the interim, our data strongly support the incorporation of systematic frailty assessment with the LFI in all patients with end-stage liver disease awaiting liver transplantation, which is particularly critical for individuals with stage 2 or greater obesity in whom frailty is associated with the highest risk of mortality.

Accepted for Publication: June 14, 2019.

Published Online: September 11, 2019. doi:10.1001/jamasurg.2019.2845

Correction: This article was corrected on October 23, 2019, to fix an error in the byline.

Corresponding Author: Jennifer C. Lai, MD, MBA, Department of Medicine, University of California San Francisco, 513 Parnassus Ave, PO Box 0538, San Francisco, CA 94143 (jennifer.lai@ucsf.edu).

Author Contributions: Drs Haugen and Lai had full access to the all the data in the study and take responsibility for the integrity of the data and accuracy of the data analysis.

Concept and design: Haugen, McAdams-DeMarco, Verna, Rahimi, Kappus, Dunn, Volk, Ganger, Ladner, Segev, Lai.

Acquisition, analysis, or interpretation of data: Haugen, McAdams-DeMarco, Verna, Rahimi, Kappus, Dunn, Volk, Gurakar, Duarte-Rojo, Ganger, O’Leary, Ladner, Garonzik-Wang, Lai.

Drafting of the manuscript: Haugen, McAdams-DeMarco, Kappus, Ganger, Lai.

Critical revision of the manuscript for important intellectual content: Haugen, McAdams-DeMarco, Verna, Rahimi, Dunn, Volk, Gurakar, Duarte-Rojo, Ganger, O’Leary, Ladner, Garonzik-Wang, Segev, Lai.

Statistical analysis: Haugen, McAdams-DeMarco.

Obtained funding: Haugen, Volk, Lai.

Administrative, technical, or material support: Verna, Rahimi, Kappus, Dunn, Volk, Gurakar, O’Leary, Ladner, Lai.

Supervision: Dunn, Garonzik-Wang, Segev, Lai.

Conflict of Interest Disclosures: Dr Haugen reported receiving grants from the National Institutes of Health during the conduct of the study. Dr McAdams-DeMarco reported receiving grants from the National Institute on Aging during the conduct of the study. Dr Verna reported receiving grants from Salix and personal fees from Gilead outside the submitted work. Dr Dunn reported receiving personal fees from Axcella Health Inc outside the submitted work. Dr Ganger reported receiving personal fees from Gilead outside the submitted work. Dr Ladner reported receiving grants from the National Institutes of Health during the conduct of the study. No other disclosures were reported.

Funding/Support: Funding for this study was provided by the National Institute of Diabetes and Digestive and Kidney Disease and the National Institute of Aging: grant numbers F32AG053025 (principal investigator [PI], Dr Haugen), R01AG055781 (PI, Dr McAdams-DeMarco), K24DK101828 (PI, Dr Segev), K23AG048337 (PI, Dr Lai), and R01AG059183 (PI, Dr Lai).

Role of the Funder/Sponsor: The National Institutes of Health had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.