Sarcopenia negatively impacts short-term outcomes in patients undergoing hepatic resection for colorectal liver metastasis

Abstract

Background

As indications for liver resection expand, objective measures to assess the risk of peri-operative morbidity are needed. The impact of sarcopenia on patients undergoing liver resection for colorectal liver metastasis (CRLM) was investigated.

Methods

Sarcopenia was assessed in 259 patients undergoing liver resection for CRLM by measuring total psoas area (TPA) on computed tomography (CT). The impact of sarcopenia was assessed after controlling for clinicopathological factors using multivariate modelling.

Results

Median patient age was 58 years and most patients (60%) were male. Forty-one (16%) patients had sarcopenia (TPA ≤ 500 mm²/m²). Post-operatively, 60 patients had a complication for an overall morbidity of 23%; 26 patients (10%) had a major complication (Clavien grade ≥3). The presence of sarcopenia was strongly associated with an increased risk of major post-operative complications [odds ratio (OR) 3.33; P = 0.008]. Patients with sarcopenia had longer hospital stays (6.6 vs. 5.4 days; P = 0.03) and a higher chance of an extended intensive care unit (ICU) stay (>2 days; P = 0.004). On multivariate analysis, sarcopenia remained independently associated with an increased risk of post-operative complications (OR 3.12; P = 0.02). Sarcopenia was not significantly associated with recurrence-free [hazard ratio (HR) = 1.07] or overall (HR = 1.05) survival (both P > 0.05).

Conclusions

Sarcopenia impacts short-, but not long-term outcomes after resection of CRLM. While patients with sarcopenia are at an increased risk of post-operative morbidity and longer hospital stay, long-term survival is not impacted by the presence of sarcopenia.

Introduction

Advances in multimodality treatment of metastatic colorectal cancer has increased survival with complete resection of colorectal liver metastasis (CRLM) now being associated with a 5-year survival of 50–55%.^1–4 With the introduction of more effective chemotherapy regimens and widening criteria for surgical therapy, the number of patients with CRLM eligible for surgery has also expanded.⁵ Advances in operative techniques and peri-operative care have minimised mortality with contemporary series reporting a mortality of less than 5%.^4,6–9 Post-operative morbidity, however, remains a concern with some studies reporting complication rates of up to 20–50%.^4,7,10–13 Post-operative morbidity can lead to prolonged hospital stays, increased need for resource expenditure, greater medical costs and possibly even worse long-term survival.^14,15 Identification of those patients at greatest risk for developing peri-operative complications is therefore critical.

Unfortunately, data to stratify patients with regard to the risk of peri-operative morbidity remain ill-defined. The American Society of Anesthesiologist (ASA) classification, Eastern Cooperative Oncology Group (ECOG) performance status and body mass index (BMI) – among others – have been utilised with varying degrees of success, but often fail to identify correctly those patients at highest risk of peri-operative complications. Frailty, a measure thought to estimate the patient's physiologic reserves, has recently been proposed as a more robust predictor of post-operative complications.¹⁶ Assessments of frailty have been criticised, however, for relying on scales that depend on subjective evaluations of weakness, exhaustion and physical activity. As such, some investigators have proposed using sarcopenia – depletion of muscle mass assessed by computed tomography (CT) – as a potentially better predictor of outcome. In general, sarcopenia is assessed by measuring the surface area of skeletal muscle at the level of the third lumbar vertebrae by measuring the total area of skeletal muscle in this particular cross-section or by measuring the surface area of the psoas muscle. Previous data have suggested that sarcopenia was associated with worse outcomes among patients being treated with chemotherapy for breast, pancreatic, prostate and renal cell carcinoma.^17–21 In addition, Englesbe et al. reported that sarcopenia was predictive of mortality among patients undergoing liver transplantation.²²

We herein sought to investigate the incidence of sarcopenia and sarcopenic obesity in patients undergoing liver resection for CRLM. Moreover, we investigated the impact of sarcopenia on both short- and long-term outcome among patients undergoing hepatic resection for CRLM. We hypothesise that sarcopenia and sarcopenic obesity may be associated with increased morbidity in patients undergoing hepatic resection for CRLM.

Methods

Patients and data collection

Between January 2000 and December 2009, 389 patients who underwent curative intent surgery for CRLM were identified from the Johns Hopkins Hospital liver database. Peri-operative abdominal CT images (i.e. within 30 days of surgery) were available for re-review for 259 patients who, in turn, represent the study cohort. Clinical and pathological data were collected including information on demographics, primary tumour stage, extent of metastatic disease, operative details, as well as length of intensive care unit (ICU) and overall hospital stay. Data on peri-operative mortality and morbidity were also collected. Specifically, complications were scored using the Clavien–Dindo classification with major complications being defined as Clavien grade ≥ 3.²³ The study was approved by the Johns Hopkins Institutional Review Board.

Image analysis

Sarcopenia was assessed by measuring the cross-sectional area of the right and left psoas muscles (total psoas muscle area = TPA) by three trained observers (I.K., P.P. and M.V.). TPA was measured at the level of L3 on the first image where both vertebral spinae were clearly visible. Measurements were performed in a semi-automated fashion by manually outlining the borders of both psoas muscles and setting the threshold between −30 and 110 Hounsfield Units (HU). This allowed for automatic calculation of the total surface area of skeletal muscle in the outlined area that excluded both vasculature and, more importantly, areas with fatty infiltration based on the respective HU. All images were analysed using the Ultravisual software package (Merge Emageon, Birmingham, AL, USA). The measured psoas area was then normalised for height, as is conventional for other body composition measures (TPA mm/m²).^24–26

Statistical analyses

Data are provided as mean and standard deviation (SD) for continuous variables and proportions for binary variables. The impact of sarcopenia was evaluated both as a continuous and a categorical variable. To obtain the categorical cut-off value for sarcopenia, optimum stratification was assessed through a series of sensitivity analyses that defined 500 mm²/m² as the most relevant cut-off value. As such, sarcopenia was defined as a TPA of ≤500 mm²/m². Sarcopenic obesity was defined as the presence of sarcopenia in patients with a BMI ≥ 30.

The impact of sarcopenia and sarcopenic obesity on postoperative morbidity and mortality was examined using univariate and multivariate analyses. Multivariate logistic regression models included other potential predictors of outcome such as gender, BMI and extent of liver resection. Recurrence-free and overall survival were evaluated using the non-parametric Kaplan–Meier method. A P-value < 0.05 was considered statistically significant. All statistical analyses were performed using Stata 10.0 (StataCorp, College station, TX, USA).

Results

Clinical, operative and sarcopenia characteristics

The clinical and pathological characteristics of the 259 patients included in the study are outlined in Table 1. Most patients had metastatic disease from a primary colon tumour (n = 191; 73%), while 68 (27%) patients had a primary rectal lesion. The majority of patients had T3/T4 primary colorectal tumors (n = 218; 84%) and associated lymph node metastasis (n = 176; 70%). Regarding the extent of CRLM, the median number of treated hepatic metastasis per patient was 2 (range, 1 to 15) and the median size of the largest lesion was 2.6 cm (range, 0.3 to 10.0 cm). At the time of surgery, surgical treatment was resection only (n = 198; 76%) or resection plus radiofrequency ablation (RFA) (n = 61; 24%). No patient underwent RFA alone. A major hepatic resection (>3 segments) was undertaken in 121 patients (47%). Specifically, the extent of hepatic resection was less than a hemihepatectomy in 7 (6%) patients, a hemihepatectomy in 84 (69%) patients and an extended hemihepatectomy in 30 (25%) patients.

Table 1.

Clinicopathological characteristics of 259 patients included in the study

Characteristic	Number (%)
Demographics
Age, mean ± SD	58 ± 12
Gender, n (%)
Male	155 (60%)
Female	104 (40%)
Anthropomorphic characteristics
BMI (kg/m2)
<30	191 (74%)
≥30	68 (26%)
TPA mm2/m2
≤500 mm2/m2 (sarcopenic)	41 (17%)
>500 mm2/m2	217 (83%)
Primary tumour characteristics
Node status, n (%)
Positive	176 (84%)
Negative	74 (16%)
Missing	9
Primary location, n (%)
Colon	191 (74%)
Rectum	68 (27%)
CRLM characteristics
Size, n (%)
<3 cm	132 (54%)
≥3 cm	113 (46%)
Number, n (%)
≤3	196 (87%)
>3	62 (13%)
Interval, n (%)
Synchronous	155 (60%)
Metachronous	104 (40%)
Hepatic surgery operative details
Extent of liver resection
Minor resection (<3 segments)	138 (53.3)
Major resection (≥3 segments)	121 (46.7)
RFA
Resection only	198 (76%)
Resection + RFA	61 (24%)

SD, standard deviation; BMI, body mass index; TPA, total psoas area; CRLM, colorectal liver metastasis; RFA, radiofrequency ablation.

Among the 259 patients evaluated, the mean TPA was 2070 ± 690 mm²; after normalising for patient height, the mean TPA/m² was 700 ± 200 mm²/m². Overall, 41 patients (17%) had sarcopenia defined as a TPA/m²≤ 500 mm²/m². Sarcopenia was more prevalent among women (32%) compared with men (5%) (Fig. 1). While sarcopenia was observed across a wide range of BMIs, sarcopenia was less frequently observed among obese patients with a BMI ≥ 30 kg/m² (Fig. 2). Of the 68 patients who had a BMI ≥ 30 kg/m² only 5 (2%) also had sarcopenia and therefore were characterised as having sarcopenic obesity. Other demographic and clinicopathological factors were similar among sarcopenic patients vs. non-sarcopenic patients (Table 2).

Figure 1.

Boxplot showing the distribution of total psoas area (TPA)/m² stratified by gender

Figure 2.

Scatterplot showing the distribution of total psoas area (TPA)/m² according to body mass index (BMI)

Table 2.

Comparison of demographic, clinical and surgical characteristics of sarcopenic vs. non-sarcopenic patients

Characteristic	Sarcopenia	No sarcopenia	P-value
	n = 41 (15.8%)	n = 218 (83.2%)
Gender
Male	8 (19.5%)	147 (67.4%)	<0.001
Female	33 (80.5%)	33 (32.6%)
Mean age (SD)	59 (1.4)	58 (1.2)	0.68
BMI
<30	36 (87.8)	155 (71.1)	0.026
≥30	5 (12.2)	63 (28.9)
Primary tumour nodal status
Positive	28 (68.3)	148 (67.9)	0.85
Negative	11 (26.8)	63 (28.9)
No data	2 (4.9)	7 (3.3)
Timing of CRLM diagnosis
Synchronous	23 (56.1)	132 (60.6)	0.61
Metachronous	18 (43.9)	86 (39.4)
Mean size of largest CRLM, cm (SD)	2.8 (1.5)	3.1 (1.9)	0.26
Mean number of CRLM treated (SD)	2.7 (2.4)	2.8 (2.5)	0.75
Extent of liver resection
Minor	22 (53.7)	116 (53.2)	0.99
Major	19 (46.3)	102 (46.8)
RFA
Yes	8 (29.5)	53 (24.3)	0.51
No	33 (80.5)	165 (75.7)

SD, standard deviation; CRLM, colorectal liver metastasis; RFA, radiofrequency ablation.

Post-operative complications: influence of sarcopenia, obesity and sarcopenic obesity

Of the 259 patients who underwent surgery for CRLM, 60 patients developed a complication with a peri-operative morbidity rate of 23%. Morbidity after surgery was mostly associated with liver-related complications such as post-operative bleeding (n = 6), liver insufficiency (n = 4), liver abscess (n = 2), biloma (n = 2), sepsis (n = 1) and portal vein thrombosis (n = 1). Other complications included symptomatic pleural effusion requiring drainage (n = 6), pneumonia (n = 2), and hematological (n = 3), cardiovascular (n = 1), pulmonary (n = 2), gastrointestinal (n = 2), or other/non-specified (n = 6). Of the 60 complications, 26 (43%) were major in nature (Clavien grade ≥ 3). Sarcopenia was associated with overall morbidity risk [odds ratio (OR) 2.22; P = 0.02], as well as the risk of experiencing a major complication in the peri-operative period. Specifically, mean TPA/m² was significantly lower among patients who post-operatively developed Clavien grade ≥ 3 complications compared with those patients who did not (618 ± 212 mm²/m² vs. 712 ± 197 mm²/m², respectively; P = 0.02). Patients with sarcopenia had over a three-fold increased risk of developing major Clavien grade ≥ 3 complications after liver resection (Clavien grade ≥ 3 complications: non-sarcopenia 8% vs. sarcopenia, 22%, OR 3.44; P = 0.008). On univariate analysis, other factors, including gender, age, BMI and extent of liver resection were not associated with a risk of increased morbidity (all P > 0.05). In a multivariate logistic regression model that accounted for these other variables, sarcopenia remained independently associated with a risk of major Clavien grade ≥ 3 complications (OR 3.12, 95% CI 1.14–8.49; P = 0.02) (Table 3). Patients with sarcopenic obesity had a more pronounced risk of Clavien grade ≥ 3 complications compared with patients who did not have sarcopenia (sarcopenic obesity, 40% vs. non-sarcopenia 8%, P = 0.02). Only two patients died after surgery (peri-operative mortality rate of 0.8%) and one of these had sarcopenia.

Table 3.

Results of logistic regression analysis on the correlation between post-operative complications and sarcopenia (TPA mm²/m² ≤ 500)

	Univariate			Multivariate
	OR	95% CI	P-value	OR	95%CI	P-value
Age (per 1 year increase)	1.00	0.97–1.04	0.85	1.00	0.97–1.04	0.79
Gender (male)	0.53	0.24–1.22	0.14	0.78	0.31–1.97	0.60
BMI (≥30)	1.04	0.42–2.59	0.94	1.34	0.52–3.46	0.78
Resection (major)	1.37	0.61–3.10	0.44	1.40	0.60–3.23	0.43
Sarcopenia (present)	3.33	1.36–8.09	0.008	3.12	1.14–8.49	0.02

TAP, total psoas area.

Median ICU and overall length of hospital stay was 1 (range: 0–28 days) and 5 days (range: 2–41 days), respectively. Patients who had sarcopenia were more likely to have a prolonged ICU stay (i.e. > 2 days) (15%) vs. patients who did not have sarcopenia (4%) (P = 0.004). Similarly, patients who had sarcopenia were more likely to have a longer overall hospital stay (mean 6.6 ± 6.1 days) vs. patients who did not have sarcopenia (mean 5.4 ± 3.2 days) (P = 0.03). Patients with sarcopenic obesity were at an even higher risk of a prolonged ICU (20%) and overall hospital stay (12.6 ± 16.1 days) (P < 0.001).

Long-term outcome: does sarcopenia have an impact?

For the entire cohort, disease-free survival at 1-, 3- and 5-years was 65%, 28% and 26%, respectively. The median disease-free recurrence was 18 months. The presence of sarcopenia was not associated with the risk of recurrence. Specifically, the 5-year recurrence-free survival was 23% vs. 27% for sarcopenic and non-sarcopenic patients, respectively (P = 0.78) (Fig. 3a). To further explore any potential impact of TPA on recurrence, disease-free survival was analysed by stratifying TPA into quartiles; TPA, however, remained not predictive of disease-free survival (Fig. 3b). Furthermore, no difference in recurrence-free survival was noted among patients with sarcopenic obesity when compared with the rest of the cohort (P = 0.26). The median overall survival for the entire cohort was 46 months with a corresponding 5-year survival rate of 40%. Similar to disease-free survival, sarcopenia was not associated with long-term overall survival (Fig. 4). While patients with sarcopenic obesity tended to have a shorter median survival (30 months) compared with the rest of the cohort (46 months), this did not reach statistical significance (P = 0.05).

Figure 3.

Recurrence-free survival stratified by (a) sarcopenia [total psoas area (TPA)/m²≤ 500 vs. TPA/m² > 500] and (b) TPA/m² divided into quartiles

Figure 4.

Overall survival stratified by (a) sarcopenia [total psoas area (TPA)/m²≤ 500 vs. TPA/m² > 500] and (b) TPA/m² divided into quartiles

Discussion

In spite of the expansion in the utilisation of liver surgery, the mortality associated with hepatic resection has dramatically decreased to less than 5% over the past few decades. Peri-operative mortality is less than 5% even after extended hepatic resection,⁸ two-stage hepatectomy²⁷ and resection combined with radiofrequency ablation.^28,29 The morbidity associated with liver surgery, however, remains problematic with a peri-operative complication rate of up to 50% in some series.^4,7,10–13 While perhaps non-lethal, serious complications occur in a subset of patients, which can adversely impact patient quality of life, as well as increase hospital utilisation and costs. There is a need therefore to identify patients at highest risk for peri-operative complications after liver surgery for CRLM. The present study examines a large cohort of patients who underwent hepatic resection for CRLM to identify a possible predictor of peri-operative morbidity.

Many patients with cancer may experience involuntary weight loss.^30,31 Weight loss may be attributable to reduced nutritional intake (e.g. anorexia), which can be exacerbated by the side effects of anti-cancer therapies.³² In addition, a ‘cachexia’ syndrome can be associated with some advanced malignancies that is characterised by the release of endogenous transmitters, changes in inflammatory markers and other mediators that cause negative nitrogen balance, fatigue and weight loss.³² Interestingly, some data have demonstrated that body fat is lost more rapidly than lean tissue in progressive cancer cachexia, which may be related to alterations in circulating hormones and food intake.³³ As such, some investigators have proposed using sarcopenia as a predictor of outcome. Indeed, we found that the presence of sarcopenia was strongly associated with an increased risk of major post-operative complications. Moreover, patients with sarcopenia had both a higher chance of an extended ICU stay and a longer overall hospital stay. These data have important implications as they strongly suggest that measurements of sarcopenia, which can be easily obtained from pre-operative cross-sectional imaging, may help identify those patients at highest risk of peri-operative complications after resection of CRLM.

The ability to stratify patients at risk for peri-operative complications after CRLM has several benefits. The risk prediction of peri-operative morbidity can help provide prognostic information to the patient, guide treatment strategies and therefore allow more accurate comparisons of outcome among various studies/hospitals.³⁴ Most previous reports, however, have focused exclusively on post-operative mortality, with few studies examining factors that might be predictive of morbidity.^16,34 Breitenstein et al. proposed a pre-operative score to predict morbidity after liver resection that included ASA grade, alanine aminotransferase (AST), extent of liver resection and extrahepatic procedure. In the previous study, the authors ascribed a point system to these factors and noted that patients who had an increased score were at a higher risk of Clavien grade ≥ 3 complications.³⁴ This scoring system can be criticised, however, for relying too heavily on factors that may not be applicable to many patients undergoing liver resection for CRLM. For example, many patients undergoing liver resection for CRLM have a normal AST and do not undergo a simultaneous extrahepatic procedure. As such, a scale that utilises these two factors may not be applicable to many patients undergoing CRLM surgery. In addition, the score proposed by Breitenstein and colleagues relies heavily on procedure-related factors (e.g. extent of resection and extrahepatic procedure) and predominately incorporates patient-level factors only through the use of a ASA grade. The ASA score has, however, been criticised as a subjective estimate of organ disease and likelihood of survival.^16,35 As such, some investigators have suggested more global assessments of physiological reserve such as frailty.^16,36,37 In fact, Makary et al. reported that frailty independently predicted post-operative complications among a cohort of patients who underwent a wide variety of elective operations at a university hospital.¹⁶ While the use of frailty allows for a more global assessment of the patient, it similarly can be criticised as being potentially subjective and susceptible to recall bias. Specifically, frailty scores depend on a measurement of factors such as ‘exhaustion’ and ‘physical activity’, which are assessed by questioning the patient.¹⁶ One well-described component of frailty that can be measured more objectively, however, is muscle loss/sarcopenia.^38–40 While sarcopenia can be associated with the ageing process, it has also been shown to be accelerated with malnutrition and in chronic medical illness.^40–42 As such, sarcopenia may be an attractive objective measurement of frailty and overall physiological reserve that might help predict the risk of peri-operative morbidity.

In the present study, while peri-operative mortality was rare, morbidity occurred in up to one-quarter of patients undergoing surgery for CRLM. Many complications were live related including bleeding, liver insufficiency and biloma. A full one-third of complications were, however, non-liver related, including cardiac and pulmonary morbidities – among others. Even more noteworthy was our finding that of the 60 complications noted, nearly one-half (43%) were major in nature. While factors such as gender, age, BMI and extent of liver resection were not associated with morbidity, the presence of sarcopenia was strongly associated with the risk of serious complications after surgery. Patients who experienced a serious complication had a mean TPA mm²/m² about 100 mm²/m² lower than those patients who did not. This translated into over a three-fold (OR 3.44; P = 0.008) increased risk of developing a major Clavien grade ≥ 3 complication among patients with sarcopenia. As such, our data strongly suggest that the finding of muscle loss (sarcopenia) on cross-sectional imaging may be a valuable pre-operative characteristic to stratify a patient's risk of peri-operative morbidity.

The impact of sarcopenia on long-term outcome remains ill defined. Previous data have suggested that sarcopenia was associated with worse outcomes among patients being treated with chemotherapy for a range of malignancies.^17–21 In addition, Englesbe et al. reported that sarcopenia was predictive of long-term mortality among patients undergoing liver transplantation.²² Specifically, in the last study, the authors noted that when stratified into quartiles based on psoas area, 1-year survival ranged from 50% for the quartile with the smallest psoas area to 87% for the quartile with the largest. Survival at 3 years among these groups was 26% and 77%, respectively.²² In contrast, in the present study, we did not find an association between the presence of sarcopenia and long-term outcome. In fact, when we stratified psoas area into quartiles, there was no discernible difference in either disease-free or overall survival (Figs 3,4). These data emphasise that the impact of sarcopenia on long-term outcomes may be a function of the underlying patient populations being studied. Whereas the impact of sarcopenia may be more pronounced on a patient's ability to tolerate an acute stress such as surgery, the role that sarcopenia has to play in long-term outcome is probably more complicated and may involve a complex interplay between overall general health status and the natural history of the patient's specific disease process.

The present study had several limitations. The study was based on a retrospective analysis, and data on other frailty parameters such as grip strength, walking speed or levels of exhaustion were not captured.¹⁶ These would have allowed examination of the impact of sarcopenia relative to other parameters of frailty. In addition, despite being a major hepatobiliary centre, the number of patients included in the study was somewhat limited due in part to the incomplete availability of scans for re-review, which required exclusion of 130 patients who had imaging performed at outside institutions.

While mortality after hepatic resection of CRLM was rare, one-quarter of patients experienced a peri-operative complication after surgery and almost one-half of these were serious. Sarcopenia, a measurement of muscle mass that can easily be obtained from cross-sectional imaging, accurately predicted those patients at highest risk of serious complications. In addition, the presence of sarcopenia was associated with both a prolonged ICU and overall hospital stay. As such, assessment of sarcopenia may provide an easy pre-operative tool to help identify those patients at highest risk of peri-operative morbidity after liver surgery for CRLM.

Conflicts of interest

None declared.