The Relationships between Body Composition and the Systemic Inflammatory Response in Patients with Primary Operable Colorectal Cancer

1Approximately 1 in 3 people in the United Kingdom will develop cancer during their lifetime. Of these, almost half will experience a progressive involuntary weight loss with their disease, termed cancer cachexia. The degree of weight loss varies by tumour type but gastrointestinal tumours have a particularly high prevalence. Indeed, it is estimated that up to half of patients with colorectal cancer have experienced weight loss by the time of presentation.

2Cachexia has long been recognised as a marker of poor prognosis in cancer patients; associated with an increased risk of surgical complications, resistance to chemotherapy,, reduced quality of life and decreased survival –. The clear link between weight loss, reduced performance status, impaired response to treatment and poor prognosis in such patients may be due to the preferential loss of skeletal muscle. It has been suggested that, although the loss of adipose tissue accounts for the majority of the weight loss, it is the loss of muscle which impacts upon morbidity and mortality –. This has led some to describe the phenomenon of cancer-related weight loss as ‘sarcopenia’; a term originally employed to describe the gradual loss of skeletal muscle seen with ageing. The aetiological factors responsible for these changes in body composition are unclear but previous observations indicate there may be an association with inflammation. Indeed, there is now evidence that the systemic inflammatory response, already recognized as a marker of poor prognosis in patients with gastrointestinal cancer, is associated with the cardinal features of cachexia,. Previous work has demonstrated an association between systemic inflammation and a loss of lean tissue as measured using a total body potassium scanner although such equipment is not routinely available, is unlikely to be useful in clinical practice and has been superseded by the advent of cross-sectional imaging.

3The aim of the present study, therefore, was to examine the relationships between CT measured parameters of body composition and the systemic inflammatory response in patients with primary operable colorectal cancer.

4Patients with colorectal cancer who, on the basis of laparotomy findings and preoperative staging CT scan, were considered to have undergone potentially curative resection for colorectal cancer (Stage I – III) between January 1st 2003 and December 31st 2010 at Glasgow Royal Infirmary were identified from a prospectively maintained database. Of these, only patients with recorded height data and CT images taken preoperatively for diagnostic or staging purposes and stored in an electronic format suitable for image analysis were included in the study.

5Patient height and weight was recorded from preoperative assessment health records and included only if documented within 30 days of CT scan. Patients were classified by body mass index (BMI) as underweight (BMI<18.5), normal weight (BMI 18.5–24.9), overweight (BMI 25.0–29.9) or obese (BMI>30) according to World Health Organisation (WHO) criteria. The tumours were staged according to the 5th edition of the Tumour, Node and Metastases (TNM) classification. Additional pathological data were taken from reports issued at the time of resection.

6The systemic inflammatory response, as defined by a number of inflammation-based prognostic scores, has proven an important indicator of outcome in cancer patients. In particular, the modified Glasgow Prognostic Score (mGPS) has been shown to reflect clinically relevant alterations in acute phase protein production (CRP and albumin) and is an established prognostic marker in colorectal cancer. Preoperative systemic inflammatory response in the present study was thus assessed using three different measures (Table 1); (1) serum white cell count (WCC), (2) neutrophil to lymphocyte ratio (NLR) and (3) the modified Glasgow Prognostic Score (mGPS).

7The image analysis of CT scans was undertaken using medical imaging software. To test the reliability of different software packages, one commercially available program (Slice-O-Matic, version 4.3, Tomovision) and one governmental free-ware program (NIH ImageJ, version 1.44, http://rsbweb.nih.gov.ij/), were compared. Two trained investigators (CSDR and MTM) analysed a random sample of 50 cases using each of the software packages with the following results. (1) CSDR versus MTM using Slice-O-Matic software, mean difference of 4.51 cm2, limits of agreement −1.67 cm2 to 10.69 cm2, interclass correlation coefficient (ICC) = 0.977, (2) CSDR versus MTM using ImageJ software, mean difference of 1.52 cm2, limits of agreement −8.81 cm2 to 11.85 cm2, ICC = 0.987, (3) Slice-O-Matic versus ImageJ software, mean difference of 7.50 cm2, limits of agreement −13.63 cm2 to 28.64 cm2 ICC = 0.953. After establishing that both software packages provided reliable measurements, ImageJ was used for the entire cohort. Figure 1 provides an example of CT image analysis using NIH ImageJ software (Figure 1).

8Total fat, subcutaneous fat, visceral fat and skeletal muscle cross-sectional areas (cm2) were measured at the level of L3 using standard Hounsfield unit ranges (adipose tissue: −190 to −30; skeletal muscle: −29 to +150). Each parameter was then normalized for patient stature, as is conventional for BMI, and designated as total fat index (cm2/m2), subcutaneous fat index (cm2/m2), visceral fat index (cm2/m2), skeletal muscle index (cm2/m2). To further test inter-observer agreement, each parameter was again measured independently by two investigators in a random sample of 50 cases (total fat index, ICC = 0.982; subcutaneous fat index, ICC = 0.992; visceral fat index, ICC = 0.955; skeletal muscle index, ICC = 0.987).

9The authors confirm that this study was approved by the West of Scotland Research Ethics Committee, Glasgow with written informed consent obtained from all participants.

10Body composition parameters are presented as mean values with standard deviation (SD) and are categorised into sex-specific tertiles (low/medium/high). Grouping of other variables was carried out using standard or previously published thresholds. Relationships between continuous and categorical variables were examined using X 2 linear-by-linear analysis, non-parametric tests and Pearson correlation coefficients (r) as appropriate. P values of less than 0.05 were considered statistically significant. Statistical analysis was performed using SPSS software (Version 18.0. SPSS Inc., Chicago, IL, USA).

11A total of 548 patients underwent potentially curative resection of colorectal cancer during the study period. Of these, 374 patients were excluded (314 patients did not have an electronic version of their CT scans available for image analysis and 60 patients did not have any height data recorded) and 174 patients were included. Figure 2 summarises the study selection process. Baseline clinico-pathological characteristics of the included cohort are shown in Table 2. Approximately one third of patients were 75 years or older with a similar number of males and females. The majority of patients had no evidence of a systemic inflammatory response prior to surgery. According to WHO BMI classification, 3% of patients were underweight, 36% normal weight, 33% overweight and 28% obese. The operations were carried out for colon cancer in 66% of cases and rectal cancer in 34%. Pathology reports classified 16% of the tumours as stage I, 44% as stage II and 40% as stage III (Table 2).

12The body composition parameters of the patients are shown in Table 3. There were no sex differences in BMI. Females had significantly more total fat (150.3 cm2/m2 versus 124.1 cm2/m2, p<0.001) and subcutaneous fat (104.4 cm2/m2 versus 73.7 cm2/m2, p<0.001) while males had significantly more skeletal muscle (46.2 cm2/m2 versus 36.9 cm2/m2, p<0.001). These differences justified the use sex-specific tertiles in the study i.e. data relating to body composition is thus corrected for sex (Table 3).

13The relationships between parameters of body composition and measures of the systemic inflammatory response in patients with primary operable cancer are shown in Table 4. There were no relationships between any parameter of body composition and serum WCC or NLR. However, there was a significant relationship between an elevated mGPS and a low skeletal muscle index (p = 0.001) (Table 4).

14To further examine this relationship, absolute values of C-reactive protein and albumin were correlated with each parameter of body composition. With regard to C-reactive protein, there were no relationships with total fat index, subcutaneous fat index or visceral fat index but there was a significant negative correlation with skeletal muscle index (r = −0.21, p = 0.005). With regard to albumin, there were no relationships with total fat index or subcutaneous fat index but there were significant positive correlations with visceral fat index (r = 0.18, p = 0.02) and skeletal muscle index (r = 0.31, p<0.001). Scatterplots demonstrating these correlations are shown in Figure 3.

15The relationships between skeletal muscle index and clinicopathological characteristics of the patients are shown in Table 5. There were significant associations between a low skeletal muscle index and increasing age (p<0.001) and presence of anaemia (p = 0.029). There were no associations between skeletal muscle index and any of the tumour-related variables (Table 5).

16The relationships between BMI classification and skeletal muscle index are illustrated in Figure 4. At least some patients from all the BMI categories fell within the lowest tertile of skeletal muscle index. In females, this meant a total of 24 patients (30%) with a normal, overweight or obese BMI were within the lowest tertile of skeletal muscle index. In males, 31 patients (33%) with a normal, overweight or obese BMI were within the lowest tertile of skeletal muscle index (Figure 4).

17The results of the present study demonstrate a strong association between low skeletal muscle mass and the presence of a systemic inflammatory response, as measured by mGPS, in patients with primary operable colorectal cancer. Furthermore, there were no direct relationships between skeletal muscle mass and any tumour-related variables, including tumour stage or nodal status. Taken together, these results would suggest that the loss of lean tissue in cancer cachexia may be driven by the host systemic inflammatory response.

18The negative impact of systemic inflammation on cancer outcome has been reported previously; associated with an increased risk of septic complications, functional decline and decreased survival. The present study confirms that, using a different methodological approach, systemic inflammation plays a role in the development of muscle wasting in patients with colorectal cancer. This is supported by experimental models whereby pro-inflammatory cytokines, including interleukin-1 (IL-1), IL-6 and tumour necrosis factor-ά (TNF), have been reported as mediators of both anorexia and skeletal muscle proteolysis, the key components of weight loss in patients with cancer. Furthermore, the present study points to such inflammatory mediators having an effect on the liver, key to the elaboration of the systemic inflammatory response. In addition to the hepatic production of acute phase proteins and their influence on skeletal muscle metabolism, there is also an increase in liver enzyme activity associated with an elevated mGPS,. Overall, these results highlight the potential importance of a liver-derived systemic inflammatory response in the progressive nutritional and functional decline of patients with colorectal cancer. It should be emphasised that these findings may also be applicable to benign disease. Indeed, similar observations regarding the depletion of skeletal muscle being associated with activation of the systemic inflammatory response have been made in non-cancer cohorts, including patients with renal failure and chronic obstructive airways disease,.

19Several previous studies investigating the clinical impact of cancer cachexia have focused specifically on the loss of lean tissue,. However, in cancer patients, muscle wasting can occur with or without the loss of adipose tissue while in non-cancer patients there is evidence that obesity and visceral adipose tissue in particular are associated with a low grade inflammatory state,. In order to examine these relationships in detail we included measures of both adipose tissue and skeletal muscle and can now report that a systemic inflammatory response in patients with colorectal cancer is associated with a reduction in skeletal muscle as opposed to an increase in visceral adiposity.

20It is clear from the present study that a simple measure of BMI is insufficient to detect the changes in body composition associated with malignant disease. This is particularly true in populations with an increasing prevalence of obesity; it is of interest that only 3% of patients in the present study were classified as underweight according to WHO classification. Even the application of a cutoff value of <20, as suggested by Fearon and co-workers as a more sensitive indicator of cachexia, increased this figure to only 5%. It is evident that traditional descriptors of body composition, such as BMI, do not have the capacity to adequately identify patients with reduced levels skeletal muscle. The present study, therefore, supports the use of cross-sectional imaging to assess the body composition of patients with malignant disease. By comparing two widely-available software packages, we have demonstrated that such analysis of CT scans is an objective and reproducible method of quantifying body composition.

21In the present study we chose to use sex-specific tertiles rather than specific cutoff values to define levels of adiposity and sarcopenia. The most common current definition of sarcopenia is an appendicular skeletal muscle index more than two SDs below that of healthy adults (5.45 kg/m2 for females and 7.26 kg/m2 for males). These values relate to dual-energy x-ray (DEXA) scanning and may not be readily applied to cross-sectional imaging. Prado and co-workers, using CT image analysis, defined a skeletal muscle index of 52.4 cm2/m2 in men and 38.5 cm2/m2 in women as associated with mortality. However, the population on which these cutoff values were developed was highly selective, consisting of 250 patients with an obese BMI (≥30) and a heterogeneous selection of respiratory tract and gastrointestinal cancers. Application of these cutoff values to the present cohort would have resulted in over 70% of patients being classified as ‘sarcopenic’; a figure which highlights the need for additional reference values for cross-sectional imaging modalities. Indeed, an international consensus group on the diagnostic criteria for cancer cachexia concluded that definitive cutoffs for the diagnosis of sarcopenia still need to be determined from large contemporary datasets.

22This study has a number of limitations. Height and weight data were primarily based on patient-reported values, although these have proven reliable in previous studies,. Electronic records of CT images were difficult to access prior to 2006 and only routinely available after 2008, meaning long term outcomes could not be assessed. In addition, although cancer-related weight loss is a continuous process, this study only assessed body composition at a single point in time. The changes in adipose tissue and skeletal muscle mass which occur over time and the relationships with cancer survival are of considerable interest and will be the subject of future work.

23The present study adds important objective evidence to what is often empirically accepted; that patients with cancer preferentially lose lean tissue during the cachectic process. In addition, these results highlight a direct relationship between low levels of skeletal muscle and the presence of a systemic inflammatory response in patients with primary operable colorectal cancer.