The purpose of this study is to examine the association of sarcopenia as measured by psoas muscle area and outcomes in patients undergoing left ventricular assist device (LVAD) implantation. We retrospectively examined 333 consecutive patients who underwent implantation of a HeartMate II LVAD at our institution from June 2008 to August 2013. Patients were included if they had a perioperative computed tomography that spanned the L3-L4 vertebrae. Sarcopenia was defined as having the lowest tertile psoas muscle area by gender. The primary end point was the composite of inpatient death or prolonged length of stay of >30 days. One hundred patients met inclusion criteria. The psoas muscle area cut-off values for the lowest tertiles were 12.0 cm 2 for men and 6.5 cm 2 for women, resulting in 32 sarcopenic patients (32%). The primary outcome of inpatient death or prolonged length of stay occurred in 81% of patients in the sarcopenic versus 60% in the nonsarcopenic group (p = 0.043). There was a trend toward prolonged length of stay in sarcopenic patients but no difference in overall mortality. This demonstrates that sarcopenia as measured by psoas muscle area is associated with increased composite length of stay and mortality after LVAD implantation and may serve as correlate for frailty.
The assessment of psoas muscle size has recently emerged as a marker for frailty. Less prone to temporary fluctuations in mass, the psoas muscle is a relatively lean core muscle the size and density of which can be relatively easily measured on computed tomography (CT) scans of the abdomen. This assessment has the advantage of being objective and widely reproducible. Studies have shown it to have significant prognostic implications for patients undergoing liver transplantation, abdominal aortic aneurysm repair, resection of pancreatic adenocarcinoma, and esophagectomy for cancer. No studies, however, have yet been conducted examining its prognostic value in patients being evaluated for left ventricular assist device (LVAD). This study examines the association of sarcopenia as measured by psoas muscle size and outcomes in patients with end-stage heart failure undergoing LVAD implantation.
Methods
This study was approved by the Washington University Institutional Review Board. We retrospectively analyzed the records of 333 consecutive patients who underwent implantation of a HeartMate II LVAD at our institution from June 1, 2008 to August 31, 2013. Patients were included if they had a perioperative CT that spanned the L3-L4 vertebrae and occurred from 6 months before to 30 days after LVAD implantation. All CT scans of the chest, abdomen, or abdomen or pelvis were screened, and those spanning L3-L4 were included. Exclusion criteria were age <18 years, loss to follow-up, or LVAD exchange.
Psoas muscle area and maximum diameter were measured on a single transaxial CT image at the L3-L4 interspace by readers who were blinded to patient outcomes. Films were reviewed independently by a novice reader (internal medicine resident) and an American Board of Radiology–certified radiologist who specializes in body CT. The primary end point was the composite of inpatient death or prolonged length of stay (LOS) of >30 days. Other outcomes of interest included time to discharge, survival to discharge, and time to death.
Definitions of sarcopenia vary widely in the medical reports with investigators using parameters such as >2 SD below the mean for young healthy controls, derived values–based receiver operating characteristic (ROC) curves or Cox models, and the lowest tertile or quartile for a study cohort. Given that this last method seemed to be the most common in similarly designed studies, patients were classified as sarcopenic if their psoas area fell into the lowest, gender-based tertile (i.e., psoas tertiles were determined by gender, and patients in the lowest female and male tertiles were categorized as sarcopenic). Patient characteristics were compared between sarcopenic and nonsarcopenic patients using the Student 2-sample t test and Fisher’s exact test for continuous and categorical variables, respectively. Non-normal and ordinal variables were summarized by their median (first quartile, third quartile) and compared using the Kruskal–Wallis test. Patients who died while inpatient or remained in hospital for at least 30 days were defined as having met the primary end point, whereas those who met neither of the criteria were said to not have met the end point.
The association between sarcopenia and the composite end point was evaluated using a logistic regression model. Separate multivariate models were created that adjusted for age, gender, preoperative LOS, body mass index (BMI), and HeartMate II risk score (HMRS). The odds ratio and 95% confidence interval were obtained from the models and describe the likelihood of the composite event occurring in sarcopenic patients relative to nonsarcopenic patients. Given the lack of a consensus definition of sarcopenia in the previous reports but the apparent correlation of psoas area with outcomes in a number of studies, a similar evaluation was conducted using psoas area as a continuous variable. Statistical models were developed that adjusted for (1) gender and age, (2) gender and HMRS, (3) gender and preoperative LOS, and (4) gender and BMI. The models were created to show the increased likelihood of an event occurring for 1 cm 2 decrease in area. To normalize for differences in body habitus, psoas area was then indexed to body surface area (BSA) and evaluated as mentioned previously. To determine if the association between psoas area and inpatient mortality or prolonged LOS was different between genders, a logistic model was developed that included the interaction between gender and psoas area.
Kaplan–Meier curves were created for long-term mortality by sarcopenia status where start time was the implant date and were compared using the log-rank test. Long-term mortality was further evaluated through a Cox proportional hazards model. Multivariate models that adjusted for age and HMRS separately were also created. The hazard ratio (HR) represents the increase in hazard (likelihood of death occurring) in sarcopenic versus nonsarcopenic patients. Cox models that examined psoas area as a continuous variable were also created, with separate models built to adjust for (1) gender and age and (2) gender and HMRS. To determine if the association between psoas area and long-term mortality was different between genders, a Cox model was developed that included the interaction between gender and psoas area.
In an attempt to delineate the ideal cut point for sarcopenia to predict inpatient mortality or prolonged LOS, the psoas area was evaluated by the area under the ROC curve. To account for variances in average psoas area by gender, the area was indexed to BSA. The area ranges from 0.5 to 1.0, where 0.5 indicates that the measure is no better than chance and 1.0 is perfect discriminative ability. The optimal cut point was determined by the Youden index, which equally weights sensitivity and specificity and maximizes the number of correctly classified patients.
To determine the accuracy of a novice reader, psoas area was first measured by a resident physician in internal medicine and then subsequently by an American Board of Radiology–certified radiologist. The accuracy was assessed using the interclass correlation coefficient (ICC). The ICC ranges from 0 to 1 with 1 representing perfect agreement.
All analysis conducted in SAS version 9.4 (SAS Institute Inc., Cary, North Carolina).
Results
A total of 333 patients received a continuous-flow LVAD from June 1, 2007, until August 31, 2013. Twenty-eight patients were excluded from analysis: 24 were LVAD exchanges, 2 were pediatric patients, and 2 were followed up at other institutions. Of the remaining 305 patients, 100 were found with a CT scan with adequate psoas muscles views within 6 months before LVAD implantation to within 30 days postoperatively. The mean follow-up of the cohort was 1.24 ± 1.21 years.
Most patients were white, men, and had a median Interagency Registry for Mechanical Circulatory Support (INTERMACS) profile of 2 ( Table 1 ). The psoas muscle area cut-off values for the lowest tertiles were 12.0 cm 2 for men and 6.5 cm 2 for women, resulting in 32 (32%) sarcopenic versus 68 (68%) non-sarcopenic patients. Similarly, the cutoff for psoas area indexed to BSA was 5.72. There was a trend toward older age, lower BMI, greater percentage of patients with BMI <25, and greater HMRS in the sarcopenic group. There were no significant between-cohort differences in INTERMACS profile, history of stroke, chronic obstructive pulmonary disease, creatinine, hemoglobin level, percentage with BMI <20, or ischemic cause of heart failure.
| Variable | Sarcopenic | p -Value | |
|---|---|---|---|
| No ( N = 68) | Yes ( N = 32) | ||
| Age (years) | 53 ± 14 | 57 ± 14 | 0.21 |
| Male | 52 (76%) | 25 (78%) | 1.00 |
| White | 48 (71%) | 25 (78%) | 0.48 |
| Body mass index (kg/m 2 ) | 31.4 ± 5.9 | 28.4 ± 6.2 | 0.02 |
| BMI <20 | 0 | 1 (3%) | 0.32 |
| BMI <25 | 9 (13%) | 10 (31%) | 0.05 |
| Atrial fibrillation | 37 (54%) | 20 (63%) | 0.52 |
| Smoking | 41 (60%) | 16 (50%) | 0.39 |
| Coronary artery disease | 29 (43%) | 20 (63%) | 0.09 |
| Diabetes mellitus | 35 (51%) | 13 (41%) | 0.39 |
| INTERMACS profile | 2.0 (1.0, 2.0) | 2.0 (1.0, 2.0) | 0.76 |
| Chronic obstructive pulmonary disease | 12 (18%) | 7 (22%) | 0.60 |
| Stroke | 4 (6%) | 6 (19%) | 0.07 |
| Bridge to transplant | 44 (65%) | 17 (53%) | 0.21 |
| Ischemic etiology of heart failure | 28 (41%) | 17 (53%) | 0.29 |
| Labs on admission | |||
| Blood urea nitrogen (mg/dL) | 38 ± 21 | 41 ± 24 | 0.60 |
| Creatinine (mg/dL) | 1.4 (1.1, 1.9) | 1.5 (0.9, 2.0) | 0.96 |
| Albumin (mg/dL) | 3.6 (3.2, 4.1) | 3.3 (2.9, 3.7) | 0.01 |
| Hemoglobin (g/dL) | 11.3 ± 1.8 | 10.5 ± 1.8 | 0.06 |
| Platelets (g/dL) | 202 ± 86 | 193 ± 74 | 0.62 |
| International normalized ratio | 1.5 ± 0.4 | 1.6 ± 0.5 | 0.22 |
| HeartMate II Risk Score | 1.65 ± 0.97 | 2.15 ± 0.86 | 0.01 |
| Pre-op length of stay | 7.5 (4.0, 12.0) | 9.0 (5.5, 15.0) | 0.25 |
The primary outcome of inpatient death or prolonged LOS occurred in 81% of patients in the sarcopenic versus 60% of the nonsarcopenic groups (p = 0.043; Table 2 ). There were higher rates of inpatient mortality, prolonged LOS, and overall death in the sarcopenic group, although these individually did not rise to the level of statistical significance. In time to all-cause mortality, there was no significant difference in the sarcopenic versus nonsarcopenic cohorts (p = 0.50) ( Figure 1 ).
| Variable | Sarcopenic | p -Value | |
|---|---|---|---|
| No ( N = 68) | Yes ( N = 32) | ||
| Inpatient mortality/prolonged LOS | 41 (60%) | 26 (81%) | 0.04 |
| Inpatient mortality | 13 (19%) | 9 (28%) | 0.31 |
| Prolonged LOS | 38 (56%) | 23 (75%) | 0.08 |
| Overall mortality | 32 (47%) | 20 (63%) | 0.20 |
When evaluated as a continuous variable, psoas area was significantly associated with inpatient mortality or prolonged LOS after adjusting for age and gender (odds ratio = 1.434 per 1 cm 2 decrease, p <0.001), after adjusting for gender and HMRS, and after adjusting for gender and preoperative LOS ( Table 3 ). Similar results were found when psoas area was indexed to BSA where it was significantly associated with inpatient mortality or prolonged LOS after adjusting for age and gender, after adjusting for gender and HMRS, and after adjusting for gender and preoperative LOS ( Table 4 ). It should be noted that the interaction between psoas area index and gender was found to be nonsignificant (p = 0.11) suggesting that the association of psoas area index and the composite of inpatient mortality and prolonged LOS was not different between genders.
| Variable | OR for Adjustment #1 | OR for Adjustment #2 |
|---|---|---|
| Psoas area (per 1 cm 2 decrease) | 1.351 (p=0.001) | 1.362 (p=0.001) |
| Gender (male vs. female) | 4.591 (p=0.019) | 7.310 (p=0.004) |
| HeartMate II risk score | 1.664 (p=0.06) | – |
| Pre-op length of stay (per 1 day increase) | – | 1.152 (p=0.006) |
| Variable | OR for Adjustment #1 | OR for Adjustment #2 |
|---|---|---|
| Psoas area (per 1 cm 2 decrease) | 2.249 (p<0.001) | 2.355 (p<0.001) |
| Gender (male vs. female) | 3.815 (p=0.030) | 6.192 (p=0.005) |
| HeartMate II risk score | 1.638 (p=0.070) | – |
| Pre-op length of stay (per 1 day increase) | – | 1.160 (p=0.005) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
