Gastrointestinal Imaging
Original Research
Progressive Sarcopenia in Patients With Colorectal Cancer Predicts Survival
Audio Available | 
Share
OBJECTIVE. The purpose of this study was to evaluate the relationship between sarcopenia and overall and progression-free survival in patients with colorectal cancer.
MATERIALS AND METHODS. This study was retrospective and complied with HIPAA. Patients with colorectal cancer who underwent CT at the time of and 6–18 months after diagnosis were included. Patients were followed for at least 5 years after diagnosis. Skeletal muscle index (SMI) and mean muscle attenuation of the psoas and paraspinal muscles at the L4 level determined the degree of sarcopenia. Composite measurements combining psoas and paraspinal muscles (total muscle) were also obtained. Univariate and multivariate Cox proportional hazard analysis was performed to evaluate the association between survival and changes in SMI and changes in attenuation. Kaplan-Meier analysis was also performed.
RESULTS. A total of 101 patients were included (mean age ± SD, 63.7 ± 13.7 years; 68 men, 33 women). The hazard ratios for overall survival were 2.27, 1.68, and 1.54 for changes in SMI of the psoas muscle, paraspinal muscle, and total muscle (all p < 0.05). The hazard ratios for overall survival were 1.14, 1.18, and 1.24 for changes in attenuation of the psoas muscle, paraspinal muscle, and total muscle, respectively (all p < 0.05). The hazard ratios for progression-free survival were 1.33, 1.41, and 1.23 for changes in SMI of the psoas muscle, paraspinal muscle, and total muscle (not statistically significant). The hazard ratios for progression-free survival were 1.10, 1.21, and 1.23 for changes in attenuation of the psoas muscle, paraspinal muscle, and total muscle, respectively (p < 0.05). Kaplan-Meier analysis showed significant differences in overall and progression-free survival based on sex-specific quartiles of muscle quantity and quality.
CONCLUSION. Progressive sarcopenia after diagnosis of colorectal cancer has a significant negative prognostic association with overall and progression-free survival.
Keywords: colorectal cancer, sarcopenia, survival
Sarcopenia is defined as loss of muscle mass and has become increasingly recognized as a risk factor for adverse outcomes in benign and malignant conditions [1]. For several cancers, including gastric, esophageal, pancreatic, lung, bladder, and breast cancer, low muscle mass is associated with poorer short- and long-term clinical outcomes [2–9].
With particular regard to patients with colorectal cancer, diminution of muscle mass and muscle density bears a negative association with clinical outcome in terms of overall survival [10–13]. For example, in patients with colorectal liver metastases who undergo resection, sarcopenia is associated with worse overall and recurrence-free survival [14]. Poor muscle bulk and low muscle density is also associated with increased morbidity in patients with colorectal tumors. Sarcopenia is associated with an increase in postoperative complications such as infection and delayed recovery in patients undergoing colectomy for colon cancer, which may in part be related to increased inflammatory response [15–17]. Sarcopenia is also associated with toxicity and neuropathy in response to chemotherapy in patients with colon cancer [18–21].
The majority of the existing literature on sarcopenia and its relationship to clinical outcome considers muscle mass and quality at a single time point, typically a baseline measurement. Given the relationship to morbidity and mortality, this measurement may prove a useful biomarker, and changes in markers of muscle mass and quality over time may provide additional useful prognostic information. However, relatively little has been done to study these temporal changes. Recent work has shown changes in the prevalence of sarcopenia in patients undergoing neoadjuvant chemotherapy for esophageal cancer but did not find a relationship between changes in body composition and overall survival in a group of 35 patients [22].
The purpose of this study was to evaluate the relationship of progressive sarcopenia (as determined by changes in muscle area and attenuation on CT) to overall and progression-free survival in patients with colorectal cancer.
| Materials and Methods |   |
|---|
This HIPAA-compliant study was performed with approval of the institutional review board of Chang Gung Memorial Hospital. Requirement for informed consent was waived.
Patients with colorectal cancer who underwent CT at the time of diagnosis and between 6 and 18 months afterward were included. Patients with CT follow-up examinations performed outside of the 6- to 18-month window or for whom artifact precluded accurate CT measurement were excluded. All patients were followed clinically and with restaging CT examinations for at least 5 years after diagnosis to determine overall and progression-free survival. Survival data were retrospectively reviewed from the medical records.
CT images were acquired with a multidetector helical CT scanner (Fig. 1). All unenhanced CT scans of the lower thorax to the pelvis using 5-mm slice thickness were available. Unenhanced CT examinations obtained as part of clinical staging at time of diagnosis and 1-year follow-up were analyzed. For purposes of this analysis, the restaging CT closest to 1 year after diagnosis was used. Three consecutive slices at the L4 vertebral body level were included, and measurements were averaged over these three images. Segmentation of the psoas major and paraspinal muscles was performed with a combination of manual selection and attenuation segmentation, with a range of −30 HU to 150 HU [12, 14], using Matlab (version 8.3.0, Mathworks). First, attenuation segmentation was applied to the cross-sectional CT image at the L4 vertebral body level using the threshold of −30 HU to 150 HU; the pixel values within this range were identified in every slice. Then, the ROI for the bilateral psoas and paraspinal muscles was manually drawn for each individual slice. Regions with fat-containing pixels ranging from −190 HU to −30 HU were excluded from the ROI. For each slice, the cross-sectional area of the psoas and paraspinal muscles was measured in addition to the combined area (designated as “total muscle”) of the psoas and paraspinal muscles. For body composition analysis, normalizing measures of muscle area to patient stature is typically performed. Skeletal muscle index (SMI) was calculated for each area by dividing the cross-sectional area by the square of the patient's height in meters[23]. The mean attenuation was measured in each muscle compartment in addition to a mean attenuation of the psoas and paraspinal muscles combined. All measurements were taken by a 3rd-year radiology resident under the supervision of a musculoskeletal radiologist with 6 years' experience.
 View larger version (145K) | Fig. 1A —52-year-old man with stage IV colon cancer (T4N2M1). A, CT images obtained at (A) and 1 year after (B) diagnosis of colorectal cancer at L4 vertebral level. Using combination of manual selection and attenuation segmentation, with range of −30 HU to 150 HU, psoas muscles (purple) and paraspinal muscles (blue) are segmented for analysis of area and density. Although visually, we cannot be certain that there is loss of muscle volume, after segmented analysis, loss of measured value is seen from baseline to 1-year follow-up for skeletal muscle index in psoas muscle (from 5.02 to 4.59), paraspinal muscle (from 12.10 to 11.60) and attenuation in psoas muscle (from 38.63 HU to 26.03 HU) and paraspinal muscle (from 32.14 to 28.12). This patient died within 5-year follow-up period. |
 View larger version (144K) | Fig. 1B —52-year-old man with stage IV colon cancer (T4N2M1). B, CT images obtained at (A) and 1 year after (B) diagnosis of colorectal cancer at L4 vertebral level. Using combination of manual selection and attenuation segmentation, with range of −30 HU to 150 HU, psoas muscles (purple) and paraspinal muscles (blue) are segmented for analysis of area and density. Although visually, we cannot be certain that there is loss of muscle volume, after segmented analysis, loss of measured value is seen from baseline to 1-year follow-up for skeletal muscle index in psoas muscle (from 5.02 to 4.59), paraspinal muscle (from 12.10 to 11.60) and attenuation in psoas muscle (from 38.63 HU to 26.03 HU) and paraspinal muscle (from 32.14 to 28.12). This patient died within 5-year follow-up period. |
Patient demographics were recorded. Patients were followed clinically and by CT for a minimum of 5 years to determine overall survival and progression-free survival. Patient height and body mass index (BMI; calculated as the weight in kilograms divided by the height in meters squared) were recorded at baseline (preoperatively), and the tumor stage at presentation was recorded. The Glasgow Prognostic Score (GPS), a composite measure of host inflammatory response based on elevation in C-reactive protein and serum albumin that is a prognostic indicator for patients with colorectal carcinoma, was measured [24]. Carcinoembryonic antigen (CEA) was measured in all cases in addition to neutrophil-to-lymphocyte ratio (NLR), a further prognostic indicator [25, 26].
The change in SMI between baseline and follow-up CT was calculated by subtracting the follow-up SMI from the baseline SMI. This value was calculated for the psoas muscle, paraspinal muscle, and total muscle measurements. Each of these measurements was then correlated with outcome using univariate Cox proportional hazard analysis with overall survival as the dependent variable. Correlation with progression-free survival was also ascertained with univariate Cox proportional hazard analysis.
In a similar way, the change in muscle attenuation was calculated by subtracting the follow-up attenuation measurement from the baseline attenuation measurement. This value was calculated for psoas muscle, paraspinal muscle, and total muscle. Univariate correlation of these factors with respect to overall and progression-free survival was assessed using univariate Cox proportional hazard analysis.
Multivariate Cox proportional hazard analysis of overall and progression-free survival was performed with change in SMI, age, sex, tumor stage at diagnosis, BMI, CEA, GPS, and NLR as independent variables and was assessed with separate analysis of psoas, paraspinal, and total muscle. Similarly, multivariate Cox proportional hazard analysis of overall and progression-free survival was performed with change in attenuation, age, sex, tumor stage at diagnosis, BMI, CEA, GPS, and NLR as independent variables, with the analysis repeated for psoas, paraspinal, and total muscle.
Kaplan-Meier survival curves were generated for overall and progression-free survival, with change in total muscle SMI as the independent variable. Because there is no universally accepted definition of sarcopenia, change in SMI was divided into sex-specific quartiles. The significance of differences in survival was tested using a log-rank test for equality of survivor functions. We then repeated these procedures with attenuation change divided into quartiles as the independent variable.
To evaluate intraobserver variability, intra-class correlation was calculated using a two-way mixed-effect model. All analysis was performed using Stata software (version 13.1, Statacorp).
| Results | |
|---|
An initial group of 150 patients were identified, 47 of whom were excluded because their follow-up CT was performed outside of a 6- to 18-month interval, and two were excluded because of CT image degradation from artifacts related to spinal hardware. Thus, 101 patients were included in the study population (mean age ± SD, 63.7 ± 13.7 years; 68 men, 33 women).
Table 1 summarizes the results of univariate Cox proportional hazard analysis of overall survival. The hazard ratios were 2.27, 1.68, and 1.54 for change in SMI and 1.14, 1.18, and 1.24 for change in attenuation of the psoas muscle, paraspinal muscle, and total muscle, respectively (all p < 0.05).
Table 2 outlines the results of multivariate analysis of overall survival. Controlling for age, sex, tumor stage at diagnosis, BMI, CEA, GPS, and NLR, changes in SMI of the psoas, paraspinal, and total muscle were independent predictors of overall survival, with hazard ratios of 3.28, 1.95, and 1.74, respectively (p < 0.05). However, changes in attenuation were not statistically significant independent predictors of overall survival. In each regression model, tumor stage was associated with an independent increase in hazard ratio for death. NLR was also a significant independent predictor of overall survival in models that included decrease in SMI.
Figures 2 and 3 illustrate the effect on overall survival of changes in SMI and changes in attenuation, respectively, of total muscle in sex-specific quartiles. For both categories, the effect of different quartiles was significant when analyzed with a log-rank test for equality of survivor functions (p = 0.001 for changes in SMI and p = 0.0096 for changes in attenuation).
 View larger version (30K) | Fig. 2 —Kaplan-Meier survival curves for overall survival depending on sexspecific quartile of skeletal muscle index decrease between baseline and 1 year. |
 View larger version (30K) | Fig. 3 —Kaplan-Meier survival curves for overall survival depending on sexspecific quartile of muscle attenuation decrease between baseline and 1 year. |
Table 3 summarizes the results of univariate Cox proportional hazard analysis of progression-free survival. For changes in SMI, the hazard ratios for progression-free survival were 1.33, 1.41, and 1.23 of the psoas muscle, paraspinal muscle, and total muscle, respectively, but the results were not statistically significant. With respect to changes in attenuation, however, the hazard ratios for progression-free survival were statistically significant (1.10, 1.21, and 1.23 for psoas muscle, paraspinal muscle, and total muscle, respectively; p < 0.05).
Table 4 outlines the results of multivariate analysis of progression-free survival. Controlling for age, sex, tumor stage at diagnosis, BMI, CEA, GPS, and NLR, changes in SMI of the psoas and total muscle were independent predictors of progression-free survival, with hazard ratios of 1.74 and 1.33, respectively (p < 0.05), but change in SMI of the paraspinal muscle was not. Changes in attenuation in the paraspinal muscle and total muscle were independent predictors of overall survival with hazard ratios of 1.24 and 1.18, respectively (p < 0.05), but change in attenuation of the psoas muscle was not. In each regression model, tumor stage was associated with an independent increase in hazard ratio for death. NLR was also a significant independent predictor of overall survival in models that included decrease in SMI.
Figure 4 illustrates the effect of change in SMI of the total muscle on progression-free survival in sex-specific quartiles. The effect was significant when analyzed with a log-rank test for equality of survivor functions (p = 0.0005). Figure 5 illustrates the same analysis with respect to change in attenuation; the effect was also statistically significant (p = 0.0022).
 View larger version (31K) | Fig. 4 —Kaplan-Meier survival curves for progression-free survival depending on sex-specific quartile of skeletal muscle index change between baseline and 1 year. |
 View larger version (31K) | Fig. 5 —Kaplan-Meier survival curves for progression-free survival depending on sex-specific quartile of muscle attenuation change between baseline and 1 year. |
Regarding intraobserver variability, the intraclass correlation coefficients for a twoway mixed-effect model were 0.997 (95% CI, 0.995–0.998) for SMI of psoas muscle, 0.998 (95% CI, 0.996–0.999) for mean attenuation of psoas muscle, 0.997 (95% CI, 0.994–0.998) for SMI of paraspinal muscle, and 0.999 (95% CI, 0.998–0.999) for mean attenuation of paraspinal muscle.
| Discussion | |
|---|
The hypothesis of this study was that changes in CT measures of sarcopenia within the 6- to 18-month interval after diagnosis of cancer would have an association with overall and progression-free survival. Overall, a significant relationship between the change in muscle quantity (measured as SMI) and overall survival was established. For example, a hazard ratio of 2.3 of death within 5 years was found for each unit change in SMI of the psoas muscle between baseline and follow-up CT. Changes in muscle attenuation also had a significant association with worse prognosis, but with a lower magnitude than SMI, with, for example a hazard ratio of death by 5 years of 1.14 per unit change in attenuation of the psoas muscle. Interestingly, we found a strong relationship between change in SMI or attenuation and patient survival when the data were controlled for multiple clinical and laboratory findings and even for tumor stage. Furthermore, a remarkable difference in survival was shown using Kaplan-Meier analysis between patients in the lowest sex-specific quartile and those in the remaining quartiles (Fig. 2). One surprising result of our study was the lack of significant association of baseline SMI and attenuation with patient prognosis, which is at odds with prior reports [21]. We observed a trend toward worse overall and progression-free survival with lower baseline and follow-up SMI that did not reach statistical significance, but this may simply have been a result of the size of our patient population.
The study was limited by its retrospective design. A further limitation is the difficulty in determining whether patients have sarcopenia because there is no standard definition of the condition. The method we employed was designed to explore the relationship of CT measures of sarcopenia to prognosis and was supplemented with sex-specific quartile Kaplan-Meier analysis. We found a significant, adverse prognostic association for change in muscle SMI between baseline and follow-up CT for patients in the lowest sex-specific quartile. This approach mirrors that of Miyamoto et al. [13], who found a negative relationship between sarcopenia and outcome when sarcopenia was defined on the basis of the lowest sex-specific quartile SMI on baseline CT. Future research may establish generalizable threshold levels of skeletal muscle area and density or determine if measurements within specific populations are of greater prognostic importance.
Preventing cancer-related sarcopenia is challenging, and our results underpin the potential importance of future research in this area. Patients with cancer lose lean body mass loss for multiple reasons, which are already the subject of much investigation [27]. One component is the loss of appetite that may result from altered serotonin regulation, as well as altered hypothalamic gene expression of orexigenic neuropeptides [28, 29]. Furthermore, stimulation of catabolic mechanisms leads to decreased protein synthesis and increased muscle breakdown, mediated by a variety of cytokines including tumor necrosis factor–α, interleukin-1, and interleukin-6 [30]. As knowledge of these mechanisms expands, targets for therapy may emerge as potential avenues of treatment to decrease cachexia and improve patient outcomes. Research is also ongoing on the effect of nutritional intervention and lifestyle modification on sarcopenia and on overall outcome in patients with metastatic colorectal cancer [31, 32]. With potential treatment targets such as those described in this article, CT measures of sarcopenia may provide important bio-markers to monitor the effect of treatment interventions. Furthermore, additional investigation is warranted of the relationship of changes over time of CT-measured muscle quantity and quality to prognosis for patients with other cancers. As the importance of these measures in informing the clinical status of patients becomes more recognized, practical implementation of time-efficient means of deriving these data will be crucial to mainstream clinical utilization.
In this study, we included inflammatory and serologic tumor biomarkers in the analysis model. The elevation of systemic inflammatory response is associated with a poor outcome, independent of tumor stage [33, 34]. Markers of inflammation, such as GPS and NLR, are increasingly applied in clinical practice for predicting prognosis of colorectal cancer. Moreover, the CEA serologic tumor marker is produced by the tumor itself via abnormal oncogene expression, so it has also been used to monitor and predict the prognosis of colorectal cancer [35]. Of these markers, our study found only preoperative NLR to be an independent predictor of colorectal cancer survival. CEA level may in fact be more related to tumor size, thus offering limited prognostic predictive value [35].
In conclusion, the interval loss in muscle quality and quantity as measured on CT within a 1-year period after the diagnosis of colorectal cancer has a significant adverse prognostic implication in terms of both overall and progression-free survival.
Supported by grant CMRPG2E0103 from the Chang Gung Memorial Hospital of Keelung.
| References | |
|---|
The comprehensive book based on the ARRS 2017 Annual Meeting categorical course on A Practical and Current Approach for Managing Incidental Findings is now available! For more information or to purchase a copy, see www.arrs.org.
