Duke Activity Status Index and Liver Frailty Index predict mortality in ambulatory patients with advanced chronic liver disease: A prospective, observational study

There remains a lack of consensus on how to assess functional exercise capacity and physical frailty in patients with advanced chronic liver disease (CLD) being assessed for liver transplantation (LT). Aim To investigate prospectively the utility of the Duke Activity Status Index (DASI) and Liver Frailty Index (LFI) in ambulatory patients with CLD.


| INTRODUC TI ON
Patients with advanced chronic liver disease (CLD) and assessed for liver transplantation (LT) in the current era tend to be sicker, medically more complex and are more often described as 'frail'. 1 This is largely due to an ageing population, increased prevalence of metabolic-related liver disease (i.e., diabetes and obesity) and worsening degree of liver disease severity at the time of presentation.
Frailty is a multidimensional clinical state of decreased physiological reserve and increased vulnerability to health stressors. 2More specifically, physical frailty refers to the functional ability (i.e., functional performance, capacity and disability) of a patient, 1 It is highly prevalent in CLD and is an independent predictor of adverse clinical outcomes in the United States (US). 3 Despite this, objective and reproducible assessments of physical frailty are scarce in Europe, with many clinicians adopting the subjective 'eyeball test' for assessing frailty in LT listing candidates. 4Consequently, the prevalence of physical frailty remains unknown in non-US countries and still the majority of US states.
The Liver Frailty Index (LFI), by Lai and colleagues, 5 is the most studied tool for physical frailty to date, consisting of three performance-based measures of physical function and strength (hand grip strength, balance and chair stands).LFI is simple, quick (3-5 min), can be carried out in any clinical setting (including outpatient clinic) and is reproducible. 68][9] However, it has not been studied or validated outside of the United States.Despite the positive contribution of the LFI to physical frailty assessments, it does not incorporate all aspects of physical frailty, including functional 'exercise' capacity (also referred to as 'aerobic exercise capacity'-the ability to sustain physical activity or endure physiological stress).1][12] However, 6MWT is limited in accuracy, practicability and the 'learning effect'.Although more accurate, CPET requires costly equipment, specifically trained staff and can be uncomfortable for patients with CLD, especially those with ascites.4][15][16] Furthermore, the DASI was able to predict adverse outcomes (30-day mortality, myocardial infarction and one-year new disability) over and above that of CPET and serological tests in 1401 patients undergoing major non-cardiac surgery. 17In view of these results, the ease of assessment and the cost-savings of completion; the investigation of the validity of DASI, alongside the LFI, warrants investigation in patients with CLD.
Simple and accurate assessment of a patient's physiological reserve and ability to cope with the physical stressors (i.e., LT, radiological interventions) remain key in CLD.Therefore, the aim of this prospective, observational UK study was to determine the prevalence, severity and predictors of physical frailty in outpatients with CLD and assessed for LT, in addition to investigating the ability of the DASI (functional exercise capacity) and LFI (physical frailty) to predict all-cause mortality, pre-LT list mortality and intensive care unit (ICU) length of stay.

| Study procedures
In addition to the routine outpatient clinic visit procedures, study participants were asked to complete the DASI questionnaire and the LFI under the supervision of trained personnel (e.g., physiotherapist [F.W., A.F.] or an exercise physiologist [JQ]).The DASI and LFI results were concealed from the participants and clinicians in order to avoid study potential influence on intervention, donor organ allocation and/or LT waiting list status.Assistance in the form of reading the DASI questions and circling the answer, from either study personnel or the caregiver/translator, was given for those who were unable to independently complete the questionnaire (i.e., English was not their first language).Study personnel, patients and caregivers were encouraged to ensure that the answers were provided by the patient alone.The self-reported DASI questionnaire consists of 12 questions related to functional capacity (i.e., can you climb a flight of stairs?) and is scored from 0 to 58.2, with the latter representing the highest functional status (Figure S1).The DASI score was converted into an estimated VO 2 peak using the following equation: VO 2 peak (mL/kg) = 0.43 x DASI +9.6. 17Physical frailty was measured using the LFI, 5 whereby every patient was asked to complete the following three performance-based measures: 1. Gender-adjusted hand grip strength (HGS): The participant was asked to stand up straight with their dominant arm straight down by their side holding the hand dynamometer (Takei, 5401 GRIP-D).The participant was instructed to squeeze the dynamometer as hard as they could for 5 s.HGS was repeated three times (1-min rest between each test) and the average recorded.
The HGS was adjusted for gender, as per LFI recommendations.
2. Timed 5 × chair stands: Using the same chair and with the patient folding their arms across their chest, the number of seconds required to complete 5 chair stands was recorded.

Balance testing:
The participant was asked to adopt three balance positions (feet together, semi tandem and tandem), and the time that each three positions were held was recorded, up to a maximum of 10 s for each position.
The results of each test were inputted into the online LFI calculator available at http:// liver frail tyind ex.ucsf.edu, where a continuous score was provided and the patient categorised as robust (score < 3.2), pre-frail (score = 3.2-4.5)or frail (score >4.5).The LFI scores for all participants were plotted against the scores provided by the US cohort 5 for comparison of levels of physical frailty between the continents.

| Data collection
Demographic data were prospectively collected from the patient's electronic health records and laboratory blood sampling (full blood count, urea and electrolytes, liver function tests, international normalised ratio [INR]) on the same day of their clinic visit and completion of the DASI and LFI.Disease aetiology, severity (Model for End-stage Liver Disease [MELD], UK Model for End-stage Liver Disease [UKELD], history of variceal bleed, hepatic encephalopathy, ascites) and key medical comorbidities (i.e., ischaemic heart disease, atrial fibrillation, type 2 diabetes, hypertension and smoking history) were recorded.Body mass index (BMI) was calculated based on the participants estimated dry total body weight, which was corrected for the presence of ascites (minus 5% for mild, 10% for moderate, and 15% for severe ascites) and peripheral oedema (minus 5% for bilateral oedema). 18Participants were prospectively followed up until the censor date of the study on 31 May 2020, with regard to overall mortality, pre-LT mortality and post-LT ICU length.

| Statistical analysis
Participant demographics were presented as mean (SD), median (IQR) and number (%) depending on the variable.Single and multiple regression analysis were run between DASI and LFI with other patient variables (UKELD, MELD, age, sex, BMI, diabetes, variceal bleed, ascites, hepatic encephalopathy, sodium, creatinine, bilirubin, INR, white blood cells and neutrophil-to-lymphocyte ratio).Note UKELD and MELD were omitted from the multiple regression analysis due to a 0.8 correlation between those variables and the involvement of identical variables already in the regression model.Due to skewness of the DASI for the regression analysis, the log of DASI (+1 to account for 0 scores) was used for the outcome variable.Coefficients were then back-transformed for the regression models.Regression analysis was also used to compare LFI (and its individual components) between patients with and without cirrhosis and sex (male vs. female).Of note, balance (one component of LFI) was excluded from the comparison of the above groups due to minimum variability in that measure (89% had a perfect score of 30/30).
Pre-LT mortality was defined as the outcome of 'death' whilst on the waiting list for LT.Follow-up time for those who did not die or receive a LT was censored on 31 May 2020.Survival analysis for those listed for LT was calculated using Cox Proportional Hazards Model.
Kaplan Meier curves for the Cox Survival models were looked at for any proportional hazards assumption violations.ICU length of stay was defined as the time (days) from admission to ICU to the time of discharge to the ward for those who underwent a LT.Cox Survival analysis was used to calculate the relationship between LFI and ICU length of stay, as well as DASI and ICU length of stay.There was no need to adjust for competing risks in this model as there were no deaths during an ICU stay.The level of significance for all the tests stated above was set at p < 0.05.

| Patient recruitment
A total of 307 patients with CLD were recruited from the LT outpatient assessment clinic at QEUHB over the 12-month study period.

| Predictors of overall and pre-LT mortality
The overall mortality for the study population was 12.4% (38/307).

TA B L E 2
Unadjusted and adjusted regressions of LFI in patients assessed for a LT.

| D ISCUSS I ON
Our prospective, single-centre UK study highlights that both poor functional exercise capacity and physical frailty, as determined by simple easy-to-use tools of the DASI and LFI, are common in ambulatory patients with CLD; with only 19% of patients defined as 'robust' in a tertiary liver outpatient unit.Furthermore, both DASI and LFI predicted both pre-LT and overall 'all-cause' mortality.Both female sex and hyponatraemia were independent predictors of both poor functional capacity (low DASI) and physical frailty (high LFI).In addition, older age and hepatic encephalopathy predicted physical frailty, whilst high BMI predicted poor functional capacity.Understanding and identifying those patients with CLD who are at higher risk of poor functional capacity, physical frailty and subsequent mortality, will aid with targeting and tailoring future prehabilitation programmes (nutrition, exercise, psychology).
In outpatient liver departments, in which time and space can be limited, evaluation of functional exercise capacity has remained a challenge in patients with CLD.Our study is the first to investigate the utility of the DASI questionnaire in this patient population.Not only is the DASI questionnaire user-friendly, cost-effective, time-efficient (<2 min), but also it provides a simpler alternative to either the 6MWT or CPET, in predicting all-cause and pre-LT mortality.4][15][16] Similar to our findings in patients with CLD, Wijeysundera et al. 17 highlighted in 1401 patients undergoing major non-cardiac surgery (excluding LT surgery) that the DASI was able to predict 30-day and 1-year survival.Similarly, Ney and colleagues (2016) performed a meta-analysis of CPET in 1107 patients and highlighted that functional capacity (i.e., weighted mean VO 2 peak) was below the threshold required for independent living in CLD and was associated with pre-and post-LT survival. 12Despite these significant findings, the use of CPET in CLD and the LT setting is not uniform throughout Europe and the United States, largely as a result of cost, specialist equipment, workforce requirement and perception that the logistical burden of CPET outweighs the additional information provided to guide patient care. 4Based on our findings, the DASI may be utilised as a quick, cheap screening tool in liver outpatients to determine which patients with Note: Due to the inclusion of identical variables, MELD and UKELD were not included in the multivariate analysis.Also, since DASI was log transformed and the coefficients above have been transformed back these represent a per cent increase rather than a point increase.
Abbreviations: BMI, body mass index; DASI, Duke Activity Status Index; INR, international normalised ratio; MELD, model for end-stage liver disease; NLR, neutrophil-to-lymphocyte ratio; UKELD, UK model for end-stage liver disease; WBC, white blood cell count.
Shaded area represent statistical significance.a Significant variable.
TA B L E 3 Unadjusted and adjusted regression of DASI in patients assessed for a LT.
CLD need more intricate analysis and individualised prehabilitation prior to radiological procedures and/or LT.In the era of virtual 'clinic' monitoring, the DASI could be completed by the patient at home and reported back to the specialist hospital team.
The prevalence of physical frailty within our UK-based study and its ability to predict mortality is similar to that reported by Lai and colleagues in the United States (Figure 2), 5,7 thereby further validating the use of LFI in ambulatory patients with CLD.Most notably, in our study, female sex was a predictor of both poor functional capacity and physical frailty.In particular, females performed significantly worse on the gender-adjusted hand grip strength and chair stand components of the LFI.This finding is supported by a multicentre cohort US study (2020) of 1405 patients with cirrhosis waiting for LT, in which females presented with worse physical frailty scores despite similar liver disease severity.Moreover in the US study, physical frailty accounted for 13% of the known gender gap in pre-LT mortality. 20Socioeconomic status and/or sociocultural experiences may contribute to the gender variations seen in physical frailty, in addition to the more widely recognised physical differences, such as biological or genetic factors. 21These findings are important, because unlike factors such as liver disease severity and age, physical frailty is a potentially modifiable contributor of pre-LT mortality. 22In addition to female sex and age, key clinical determinants of the severity of liver failure (including hyponatraemia, hepatic encephalopathy, ascites and UKELD) were all significant predictors of increased physical frailty in our cohort.In addition, patients with cirrhosis performed  significantly worse in the physical frailty subscale chair stands than those with non-cirrhotic disease aetiologies, such as severe recurrent cholangitis (e.g., PSC).These findings may be explained by the multiple mechanisms driving physical frailty in cirrhosis (i.e., chronic inflammation, 'accelerated starvation' state/malnutrition and hyperammonaemia), 23 which ultimately result in disruption of the maintenance of muscle health.
Hepatic encephalopathy and ascites are the two most common debilitating complications of CLD, 24,25 with both being strongly associated with poor functional capacity and frailty in our study.Due to reduced hepatic function and/or portal systemic shunting those with hepatic encephalopathy have higher levels of circulating ammonia, 19 which directly upregulates myostatin (i.e., increases muscle protein breakdown) 26,27 and increases mitochondrial dysfunction. 28tients with ascites, as highlighted by our study, are particular susceptible to physical frailty due to reduced appetite, early satiety, delayed gut motility 29 and subsequent decreased calorie intake; all of which exacerbate the state of 'accelerated starvation' (impaired hepatic glycogen stores) found in cirrhosis. 23Both hepatic encephalopathy and ascites should therefore be optimised in the ambulatory setting (i.e., medications and easy-to-access paracentesis), in parallel to prehabilitation programmes (nutrition/exercise), in order to minimise physical frailty and functional decline in patients with CLD. or post-LT.Throughout the study, patients continued to receive the standard of care guidance for nutrition and physical activity 18 ; however, the study was unable to control for potential varying degrees of healthcare intervention.If anything, those who were subjectively perceived to be the frailest or functionally dependent would likely have received the most healthcare intervention, yet still the baseline DASI and LFI predicted poor clinical outcomes.

A
single-centre, prospective observational cohort study was conducted at the LT Unit, Queen Elizabeth University Hospital Birmingham (QEUHB), UK.A service quality improvement audit code (ID: 15209) was obtained from QEUHB clinical governance and ethics department in 2018.Between 1 September 2018 and 1 September 2019, adult patients (≥18 years) with CLD were consecutively recruited from the liver outpatient assessment clinic at QEUHB.The term CLD incorporated all UK indications for LT, including decompensated cirrhosis (all aetiologies), >2 cm single and/ or multiple hepatocellular carcinoma (HCC) with cirrhosis/portal hypertension (Milan criteria), non-cirrhotic portal hypertension (i.e., refractory ascites, varices) and other 'variant' indications (refractory recurrent cholangitis, polycystic liver disease).Outpatients were excluded if they were unable to give written consent or unable to complete one or more of the tests, because they required urgent hospital admission for acute illness, severe hepatic encephalopathy (grade ≥3 or 4) or an acute musculoskeletal injury impeding completion of one or more elements of the tests.All patients with grade 1 or 2 hepatic encephalopathy underwent two independent capacity assessments (MA, NR) prior to written informed consent.Those who lacked capacity to consent were excluded from the study.

F I G U R E 2
Abbreviations: BMI, body mass index; CI, confidence interval; INR, international normalised ration; LFI, Liver Frailty Index; MELD, model for endstage liver disease; NLR, neutrophil-to-lymphocyte ratio; UKELD, UK model for end-stage liver disease; WBC, white blood cell count.

F I G U R E 3
Overall mortality (A) Overall mortality by Liver Frailty Index (LFI) quartiles.The lower the LFI the more 'robust' and the higher the LFI the more 'frail'.(B) Overall mortality by Duke Activity Status Index (DASI) quartile.The lower the DASI score the lower the functional capacity and the higher the DASI the greater the functional capacity.TA B L E 4 Cox survival models with LFI or DASI: Overall mortality, waiting list mortality and ICU length of stay.
Our study has several strengths and limitations.Primarily, this is the first study to investigate the predictive ability of the DASI questionnaire on overall and pre-LT mortality in patients with CLD and assessed for LT.Subsequently, the DASI provides clinicians with a time and cost-effective alternative to CPET to identify those most at risk and/or potentially require further in-depth investigation of their functional status.In addition, even though LFI has been assessed in several states in the United States, this is the first non-US study to validate its utility in European outpatient units.The findings highlight the pressing need for other liver centres to validate and consider incorporating these simple and cheap measures within outpatient clinics that manage CLD and assess for LT.Another strength is that in 2018-2020, both DASI and LFI were not part of the routine outpatient assessment or monitoring of CLD at our study centre.Therefore, by blinding the patient's clinician/multidisciplinary teams to the DASI and LFI findings, it mitigated any potential selection bias pre-LT.The limitations are largely due to the fact that the data collection ran in parallel to routine 'real-world' clinical practice.Firstly, our findings are only applicable to patients in the ambulatory setting and can't be extrapolated to hospitalised patients with CLD.Secondly, the functional/frailty assessments only represent a cross-sectional 'snapshot' of the patient, rather than serial measures overtime to enable observation of dynamic changes.Thirdly, our study recruitment likely underrepresents patients with HCC (6% of study cohort), as 19% of all UK registrations for liver transplantation had HCC during the same time period (NHSBT 2018/2019 annual report).Fourthly, patients with CLD were recruited from a tertiary care LT assessment unit and may not be a true reflection of patients with CLD in the community or non-specialist centres, as a result of referrers' selection bias.Interestingly, both DASI and LFI remained significant in predicting pre-LT waiting list mortality, despite the limitation that subjectively the 'frailest' may have been deemed ('eyeball test') too high risk at the time of LT assessment.Concerns regarding selection bias were mitigated with overall mortality, which included all patients who died after LT assessment, on the pre-LT waiting list, F I G U R E 4 Pre-LT mortality (A) Pre-LT mortality by Liver Frailty Index (LFI) quartiles.The lower the LFI the more 'robust' and the higher the LFI the 'frailer'.(B) Pre-LT mortality by Duke Activity Status Index (DASI) quartile.The lower the DASI score the lower the functional capacity.
In conclusion, poor functional exercise capacity and physical frailty are highly prevalent in UK ambulatory patients with CLD, assessed for LT.Both DASI and LFI predict pre-LT and overall mortality.Female sex and hyponatraemia, in particular, are significant predictors for both poor functional capacity and physical frailty.Both the DASI and LFI, which measure different aspects of physiological reserve, should be utilised in ambulatory patients with CLD, in order to target and individualise exercise/nutritional interventions.