Comparison Between Sarcopenia and Frailty as Predictors of Postoperative Complications: 65 Years and Older Non-Cardiac Surgery Cohort Study

Sarcopenia and frailty are two different concepts with specic measures. Nonetheless, they share similarities. Whether one is a better predictor of postoperative adverse outcomes in older adults has not been well studied. Our main objective was to evaluate the psoas muscle index (PMI), used as a surrogate for sarcopenia, and the Clinical Frailty Scale (CFS) association with postoperative complications incidence. Our secondary objective was to evaluate the correlation between PMI and CFS in a cohort of non-cardiac surgery patients. We conducted a prospective observational cohort study. Patients 65 years and older undergoing elective non-cardiac surgery in a tertiary academic center were included. Right and left psoas areas were measured on a CT scan at the fourth lumbar vertebra body level, and the sum was normalized for squared height (PMI). Sarcopenia was dened as PMI lowest tertile. The National Surgical Quality Improvement Program dened postoperative complications. We performed a negative binomial regression analysis to assess the association with postoperative complications and compared the model t using the Akaike information criterion (AIC). Correlation between PMI and CFS was analyzed using the Spearman correlation.


Abstract
Background Sarcopenia and frailty are two different concepts with speci c measures. Nonetheless, they share similarities. Whether one is a better predictor of postoperative adverse outcomes in older adults has not been well studied. Our main objective was to evaluate the psoas muscle index (PMI), used as a surrogate for sarcopenia, and the Clinical Frailty Scale (CFS) association with postoperative complications incidence. Our secondary objective was to evaluate the correlation between PMI and CFS in a cohort of non-cardiac surgery patients.

Methods
We conducted a prospective observational cohort study. Patients 65 years and older undergoing elective non-cardiac surgery in a tertiary academic center were included. Right and left psoas areas were measured on a CT scan at the fourth lumbar vertebra body level, and the sum was normalized for squared height (PMI). Sarcopenia was de ned as PMI lowest tertile. The National Surgical Quality Improvement Program de ned postoperative complications. We performed a negative binomial regression analysis to assess the association with postoperative complications and compared the model t using the Akaike information criterion (AIC). Correlation between PMI and CFS was analyzed using the Spearman correlation.

Conclusions
In our cohort, PMI was a better predictor of postoperative complications than CFS. The correlation between PMI and CFS was weak. An objective measure of sarcopenia, compared to the broader concept of frailty, might be an easier way to identify patients with higher risks of complications. Frailty is conceptually de ned as a decline in physiologic reserve, leading to decreased ability to cope with acute stressors. (1) There is considerable evidence that frail older patients represent a subgroup at higher risk of adverse postoperative outcomes, including longer hospital lengths of stay and increased mortality. (2,3) In a meta-analysis on general surgery patients, frail patients had a 30-day postoperative mortality of 8%, compared to 1% in the non-frail group. Readmission rates and complications were also higher in the frail group.(4) Sarcopenia, de ned as loss of skeletal muscle and muscle strength impairing physical performance, contributes to age-related functional impairment. It is associated with walking pace and grip strength and can be considered as a component of frailty. (5) Like frailty, it is common in older patients (6)(7)(8) and is associated with poor outcomes in patients undergoing general surgery. (6)(7)(8)(9)(10) Frailty and sarcopenia have both a close relationship with the aging process. However, they can be seen as distinct entities with different therapeutic approaches(11) Preoperative evaluation of frailty and sarcopenia in older patients could potentially be useful to clarify goals of care, obtain informed consent for surgery, and propose prehabilitation interventions, when indicated, in order to reduce postoperative morbidity. (12) The geriatric literature has proposed various instruments to help clinicians assess a patient's frailty state and the presence of sarcopenia. Frailty assessments rely on questionnaire-based instruments and physical assessments such as 4-meter walk test or grip strength. (4) Sarcopenia assessments share similarities with those frailty assessments. (5) In the perioperative literature, sarcopenia assessment usually involves the measurement of psoas muscle area (PMA) on cross-sectional computed tomography (CT) scans. PMA, as a surrogate of sarcopenia, has been shown to predict outcomes after cardiac and general surgery procedures. (13)(14)(15)(16)(17) A study on patients undergoing surgical aortic valve replacement, low PMA was associated with higher 1-year mortality (31.9% vs 16.9%, p = 0.03) and higher risk of postoperative adverse outcomes including prolonged ventilation (OR = 1.23, IC 95%: 1.05-1.45, p = 0.012) and longer hospital length of stay (LOS) (loss of one PMA unit adds 0.46 day to post-op LOS , IC 95% : 0.27-0.65, p < 0.0001). (18) To our knowledge, the association between psoas muscle area (PMA) and frailty in patients undergoing non-cardiac surgery has not previously been investigated.

Methods
Our objective was to compare the association between the Clinical Frailty Scale (CFS), a validated frailty assessment tool, and psoas muscle index (PMI), representing PMA normalized for height surrogate of sarcopenia, with adverse postoperative outcomes in order to assess whether one marker is a better predictor of postoperative outcomes. We also aimed to assess the correlation between low PMI and a frailty state with adverse postoperative outcomes.

Design and Settings
We conducted a retrospective analysis of a subgroup of patients from a prospective cohort (19) undergoing elective noncardiac surgery in a tertiary academic hospital (Hôpital Maisonneuve-Rosemont, Montréal, Canada) from January 2017 to January 2018.

Study participants
All patients aged 65 years or older scheduled to undergo elective non-cardiac surgery were eligible for inclusion (vascular and general surgery in our center). We excluded patients undergoing emergent surgery or those unable to provide consent. We contacted patients by phone or met them in person to obtain consent. Our institution's ethics review board approved the prospective study, and an amendment has been made to include this sub-study.
A total of 270 patients underwent major surgery in our center and 134 patients underwent vascular or general surgery (the other patients underwent orthopedic surgery). Data for the complete cohort was published in Canadian Journal of Anesthesia (ref). A total of 78 patients with abdominal CT performed within 6-months of surgery were included in the nal analysis. Figure 1 shows the number of patients meeting the inclusion and exclusion criteria. Patients with and without abdominal CT-scans did not differ signi cantly in baseline characteristics and frailty state (Additional le 1). Our nal cohort's surgical procedures were mostly intraperitoneal vascular bypass surgery, gastrointestinal tract surgery, hyperthermic intraperitoneal chemotherapy, and biliary tract surgery.
Of 134 potential patients, 78 met the inclusion criteria and had a CT scan performed within 6-months before surgery.

Frailty and Sarcopenia Assessment
Frailty was assessed prospectively using the CFS, an instrument developed and validated by Rockwood and colleagues. (20) Based on self-report of comorbidities and help with instrumental activities of daily living (IADLs) and activities of daily living (ADLs), (21) the CFS is scored on an ordinal scale from 1 to 9, where a score of 1 corresponds to being robust and a score of 9 being severely frail. Patients were then classi ed into three categories according to their score; 1-3 being considered robust, 4 being vulnerable, and 5-8 frail.
A trained research assistant with previous experience using CFS performed a semi-structured interview either by phone or in-person to determine the frailty level. We have used this method in the past in a cohort of orthopedic surgery patients, and CFS was predictive of hospital LOS. (22) CFS has been previously used in both the vascular and general surgery population and was associated with higher postoperative mortality, greater risk of 6-months readmission, and postoperative functional decline. (23)(24)(25) PMI was used as a surrogate of skeletal muscle loss. Preoperative CT scans were downloaded in digital DICOM format and imported into the web-based CoreSlicer software platform (www.coreslicer.com) and analysed retrospectively. After selecting the axial slice at the top of the L4 vertebral level, a research assistant unaware of frailty score and patient outcomes traced the right and left psoas muscles with a segmental brush tool as validated in previous studies. (15,17,26,27) The research assistant underwent training and inter-rater reliability testing with a co-investigator (LAM) who developed the CoreSlicer software. The psoas muscles' measured areas were summed and normalized for height, as is conventional for other body composition measures, yielding PMI in cm 2 /m 2 . As muscle volume is correlated to patient height, this normalization procedure reduces patient sex and morphology variations.
(28) PMI results were then strati ed by gender and divided into tertiles, with patients in the rst (lowest) tertile being most sarcopenic, and patients in the third-highest tertile, least sarcopenic. (15,17,29) Variables The surgical procedure and patient characteristics (age, gender, comorbidity, preoperative status, and body mass index) were collected through medical chart review. Comorbidity was de ned as the coexistence of at least two separate chronic illnesses. The burden of comorbidity was quanti ed by the

Statistical Analysis
Continuous variables were summarized with the sample median and interquartile range (IQR) and compared using the Spearman rank correlation test. Dichotomous variables were summarized with frequency tables and compared across PMI tertiles using the chi-square test. A multivariable negative binomial regression model was used to determine the association between PMI or CFS and the number of severe postoperative complications after adjusting for covariates (age, sex, comorbidity, and baseline hemoglobin level). Covariates were chosen because they are known predictors of postoperative complications and are also factors associated with frailty or sarcopenia. The risk was reported as incidence risk ratios (IRR) with 95% con dence intervals (95% CI). We chose to analyze the total number of complications since patients might have more than one complication during the hospital stay. We constructed two models using the same covariates and only interchanging PMI for CFS. We compared the Akaike's Information Criterion of both models. Collinearity between PMI and CFS and other covariates was assessed using the Variance In ation Factor (VIF). All statistical analyses were performed using SPSS 25.

Discussion
In the present study, low PMI was independently associated with postoperative complications in older adults undergoing vascular and general surgical procedures, while CFS was not.
Despite having a small cohort, PMI values were similar to those reported in the literature. (32)(33)(34)(35) In a study of 96 patients undergoing hepatic resection or liver transplantation, the mean PMI was 6.39 cm 2 /m 2 for women and 8.18 cm 2 /m 2 for men, while ours were 6.23 cm 2 /m 2 for women and 7.77 cm 2 /m 2 for men. (33) Similarly, low PMI has been associated with postoperative complications in past studies. In a cohort of 259 patients undergoing liver resection for colorectal liver metastases, patients with a PMI below 5 cm 2 /m 2 were exposed to over a three-fold increased risk (OR = 3.12, p = 0.02) of developing major postoperative complications, after adjusting for sex, age, BMI and extent of liver resection. (36) Although they only retained Grade 3 complications on the Clavien-Dindo classi cation, the effect size is still similar to our data.
In our cohort, PMI was poorly correlated to frailty, as measured by the CFS. The small number of patients in our cohort and the fact that few were frail could have affected the correlation. Nevertheless, other studies have shown a weak correlation between sarcopenia and frailty. (7,8) Using the European Working Group on Sarcopenia in Older People Criteria, Davies and al. also observed a weak correlation coe cient (0.16) between frailty and sarcopenia in a population of 1611 community-dwelling older adults. (7) Frailty and sarcopenia are two concepts that are related, but they are not interchangeable. In our cohort, patients with low PMI were mostly non-frail (19 out of 23). PMI re ects muscle mass, and frailty is a more complex concept that encompasses multifactorial aspects of health such as fatigue, cognitive impairment, and social isolation. (1,6,11). Frailty includes but also goes beyond physical reserves. Independence in IADLs and ADLs necessitate physical capacities and depend on cognitive capacities and environmental resources. Sarcopenia and frailty concepts overlap in terms of physical function impairment. Sarcopenia has been noted to have low sensitivity and high speci city to detect clinical frailty by other authors. (7) Moreover, CFS, an estimation of diminished physiological reserves, re ecting functional dependence (37), might not capture the physical frailty as much as other frailty measures. For example, the Fried phenotype includes exhaustion, weakness, slow gait, and low physical activity, which could have a higher correlation with sarcopenia. (38) PMI was a better predictor of severe postoperative complications than CFS scores. Since sarcopenia represents reduce skeletal muscle mass, strength, and power, sarcopenic patients are more likely to remain bedridden and for a longer time after surgery, further worsening muscle wasting. (39) This immobilization could predispose patients to develop thromboembolic, pulmonary, and infectious postoperative complications. (40)(41)(42)(43)(44) In our cohort, a signi cant portion of complications was infectious, which lends credence to the association observed with PMI. On the other hand, frailty measured by CFS has been associated with long-term morbidity and mortality. (2,45) Thus, frailty could have a more substantial impact on long-term outcomes, not measured in this study. (6,46) Therefore, both of these geriatric markers can be useful at predicting patient outcomes in the surgical population, but at different time points.
Our study has many strengths. To our knowledge, this is the rst study that compared PMI and frailty as predictors of postoperative complications in the non-cardiac surgical population. We included patients from different surgical specialties, making the results more generalizable. Besides, as a measurement tool for preoperative assessment, PMI has the advantage of being readily available from CT scans performed as part of a routine preoperative workup for several surgical procedures without requiring additional measurements.
Our results must be interpreted with the following limitations. First, our study was based on a singlecentered cohort, which limits its generalizability. Second, we have a relatively small cohort of patients compared to other studies, which might have affected the association between CFS and postoperative complications. Also, we limited our outcome ascertainment to short-term perioperative complications and did not assess mid-and long-term outcomes, including functional recovery.
In conclusion, PMI and CFS represent two different but interrelated processes. PMI was associated with higher postoperative complications in our study. Further studies would be needed to examine the complementarity effect of integrating both PMI and frailty assessment to determine older surgical patients' care trajectory. Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
Not applicable Authors' contributions ABS was a major contributor in writing the manuscript. ABS, MB and LB contributed to data collection and analysis. JA, LAM, MFF, RG and HTW substantively revised the project and helped with the interpretation of data. All authors read and approved the nal manuscript.