Correlation between BMI (calculated as weight in kilograms divided by height in meters squared (A) and preoperative albumin level (to convert to grams per liter, multiply by 10) (B).
Overall survival (A) and disease-specific survival (B) curves for patients with and without sarcopenia.
eTable. Analysis of Patients With Sarcopenia Stratified by Obesity
Customize your JAMA Network experience by selecting one or more topics from the list below.
Stone L, Olson B, Mowery A, et al. Association Between Sarcopenia and Mortality in Patients Undergoing Surgical Excision of Head and Neck Cancer. JAMA Otolaryngol Head Neck Surg. 2019;145(7):647–654. doi:10.1001/jamaoto.2019.1185
Is sarcopenia associated with long-term survival after head and neck surgery?
In this cohort study with imaging analysis, 260 patients with head and neck cancer who had preoperative sarcopenia experienced decreased 2-year overall survival (71.9% vs 88.5%) compared with patients without sarcopenia. Decreased 5-year overall survival (36.5% vs 60.5%) was also noted.
Sarcopenia may be associated with decreases in long-term survival in patients with head and neck cancer who are undergoing surgery.
Sarcopenia, or the loss of muscle mass, is associated with poor treatment outcomes in a variety of surgical fields. However, the association between sarcopenia and long-term survival in a broad cohort of patients with head and neck cancer (HNC) is unknown.
To determine whether sarcopenia is associated with long-term survival in patients undergoing major head and neck surgery for HNC.
Design, Setting, and Participants
A retrospective medical records review was conducted at a tertiary care academic hospital. Two hundred sixty patients undergoing major head and neck ablative procedures with cross-sectional abdominal imaging performed within 45 days prior to surgery were included in the analysis. The study was conducted from January 1, 2005, to December 31, 2016. Data analysis was performed from June 1, 2018, to February 28, 2019.
Measurement of cross-sectional muscle area at the L3 vertebra level.
Main Outcomes and Measures
Two- and 5-year overall survival were the primary outcomes.
Of the 260 patients included in the study, 193 were men (74.2%); mean (SD) age was 61.1 (11) years. Sarcopenia was present in 144 patients (55.4%). Two-year overall survival was 71.9% of the patients (n = 82) in the sarcopenia group compared with 88.5% of the patients (n = 85) in the nonsarcopenia group (odds ratio [OR], 0.33; 95% CI, 0.16-0.70). At 5 years, overall survival was 36.5% in patients (n = 23) with sarcopenia and 60.5% in patients (n = 26) without sarcopenia (OR, 0.38; 95% CI, 0.17-0.84). On multivariate analysis, sarcopenia was a significant negative predictor of both 2-year (OR, 0.33; 95% CI, 0.14-0.77) and 5-year (OR, 0.38; 95% CI, 0.17-0.84) overall survival.
Conclusions and Relevance
Sarcopenia appears to be a significant negative predictor of long-term overall survival in patients with HNC undergoing major head and neck surgery. Sarcopenia may be accurately assessed on cross-sectional imaging and may be useful clinically as a prognostic variable and as an area for intervention to improve treatment outcomes.
Weight loss and nutritional deficiencies are common in patients with head and neck cancer (HNC). These patients experience dysphagia due to tumor location and effects, in addition to the consequences of treatment with surgery or radiation. Also, many patients with HNC are in poor health and may have chronic malnutrition. Although weight loss is common, the association between weight loss and survival outcomes is not clear.1-3 In the past decade, sarcopenia, or the loss of skeletal muscle mass, was identified as a negative prognostic or predictive factor in a variety of clinical settings, including cancer treatment. Originally coined as a term to describe functional impairment and physical disability relating to the frailty of an aging individual, sarcopenia was given an image-based diagnostic definition in 2008 and is an area of heavy research.4,5 Sarcopenia appears to be a more accurate marker for malnutrition than more conventional assessments, such as body mass index (BMI), albumin level, or prealbumin level.6 The utility of sarcopenia as a strong negative prognostic factor for wound complications or for mortality is established in a wide range of cancers including HNC, cancers of the gastroesophageal junction, cancers throughout the gastrointestinal and respiratory tracts, and pancreatic cancer.7-9
In the field of HNC, sarcopenia is a strong predictor of wound complications in total laryngectomies and esophagectomies.7,10 There were mixed results for sarcopenia being a predictor of short-term and long-term mortality in patients undergoing neoadjuvant chemoradiotherapy followed by surgical resection of esophageal cancer.10-12 However, sarcopenia is correlated with reduced survival in patients with HNC receiving radiotherapy alone.13 To our knowledge, the prognostic value of sarcopenia for mortality in a broad cohort of patients with HNC whose primary treatment was surgery has not been studied. Thus, the objective of this study was to evaluate the prognostic significance of preoperative sarcopenia on long-term mortality in patients undergoing surgical resection of HNC.
A retrospective medical records review was performed for patients undergoing surgical resection for HNC at Oregon Health and Science University in Portland, Oregon, between January 1, 2005, and December 31, 2016. Patients were included if they underwent surgical excision as the primary treatment modality and had cross-sectional abdominal imaging completed within 45 days before surgery. Medical records were reviewed for data abstraction; data obtained included patient demographics, comorbidities, tumor staging and subsite, surgical details, adjuvant therapy details, treatment complications, and date of death or last follow-up. Data analysis was performed from June 1, 2018, to February 28, 2019. The study was approved by the institutional review board at Oregon Health and Science University. Owing to the retrospective nature of the study, requirements for informed consent were waived. After data abstraction and correlation with imaging, data were deidentified for analysis.
Sarcopenia was measured using previously described methods.14 Briefly, the cross-sectional area of skeletal muscle at the midpoint of the L3 vertebra was determined on abdominal computed tomographic scans, typically obtained as part of positron emission tomographic-computed tomographic imaging. Skeletal muscle was identified using the Hounsfield unit ranges of −29 to +150 (Slice-o-Matic software, version 5.0; Tomovision). Manual corrections were made in areas where fibrous tissue was incorrectly labeled as skeletal muscle. The muscles identified at the L3 level include transversus abdominus, internal oblique, external oblique, rectus abdominus, erector spinae muscles, psoas, and quadratus lumborum. The cross-sectional area was then normalized to patient height, and sarcopenia was defined using previously determined thresholds of less than 52.4 cm2/m2 for men and less than 38.5 cm2/m2 for women.5
Descriptive statistics were performed on all variables, and χ2 tests were used when appropriate. Univariate and multivariate analyses were conducted to determine differences between the sarcopenia and nonsarcopenia groups for 2- and 5-year disease-specific mortality and overall mortality. All variables with a P < .20 on univariate analysis were included in multivariate modeling using a binary logistic model with forward stepwise selection. Data are reported as mean (SD) or as a proportion. Effect size and 95% CIs around the effect size estimates are reported. For interpretation of Cohen d findings, we used a d value less than 0.2 as indicating a small effect, 0.5 as a medium effect, and greater than or equal to 0.8 as a large effect.15 All testing was performed as unpaired, 2-tailed analyses, and findings were considered significant at P < .05. Statistical analysis was performed using SPSS statistical software, version 25 (SPSS Inc).
In total, 260 patients met the inclusion criteria and were included in this study (Table 1). Of the 260 patients, 193 were men (74.2%) and 67 were women (25.8%), with a mean (SD) age of 61.1 (11) years. Using the Charlson Comorbidity Index (CCI), patients were stratified into low-risk (CCI <5, 219 patients [84.2%]) and high-risk (CCI ≥5, 41 patients [15.8%]) groups.16,17 Sarcopenia was identified on preoperative imaging in 144 patients (55.4%). Most of the patients were overweight; 61.2% of patients had a BMI greater than or equal to 25 (calculated as weight in kilograms divided by height in meters squared). The mean (SD) BMI of patients without sarcopenia was 30.0 (6.0), and the mean BMI of patients with sarcopenia was 24.5 (4.0). A scatterplot is shown in Figure 1 with an R2 value of 0.297. Sarcopenia was not well correlated with serum albumin level. Patients with sarcopenia tended to be older (≥70 years: 41 [28.5%] vs 9 [7.8%]), male (115 [79.9%] vs 78 [67.2%]), and smokers (31 [21.5%] vs 14 [12.1%]), with higher CCI scores (≥5: 29 [20.1%] vs 12 [10.3%]) than patients without sarcopenia. There were significant differences between the sarcopenic group and the nonsarcopenic group in age (odds ratio [OR], 4.73; 95% CI, 2.19-10.23), sex (OR, 1.93; 95% CI, 1.10-3.39), BMI, smoking status (OR, 2.77; 95% CI, 1.30-5.89), CCI score (OR, 2.19; 95% CI, 1.06-4.50), and preoperative albumin level (OR, 3.31; 95% CI, 1.15-9.54).
Kaplan-Meier analysis of overall survival and disease-specific survival was performed for patients with and without sarcopenia (Figure 2). Both overall survival and disease-specific survival were significantly better in patients without sarcopenia. There were 210 patients with 2-year follow-up data. Among these, overall survival was 71.9% in the sarcopenia group (n = 82) compared with 88.5% in the nonsarcopenia group (n = 85) (Table 2). Similarly, disease-specific survival at 2 years was 78.1% in patients with sarcopenia compared with 91.4% in patients without sarcopenia. There were 106 patients with 5-year follow-up data. Five-year overall survival rates showed trends similar to the 2-year survival rates. At 5 years, overall survival was 36.5% in patients with sarcopenia (n = 23) and 60.5% in patients without sarcopenia (n = 26). A subset analysis of patients with sarcopenia stratified by obesity (BMI>30) was performed (eTable in the Supplement). The findings demonstrated significantly decreased survival in patients with sarcopenia who were obese vs nonobese (5-year overall survival, 0% vs 40%; OR, 0.60; 95% CI, 0.48-0.74); a similar association of obesity with survival was not seen in patients without sarcopenia (5-year overall survival, 64% vs 59%; OR, 1.27; 95% CI, 0.34-4.45).
On univariate analysis, sarcopenia was a negative predictor for 2-year overall survival (72% vs 88%; OR, 0.33; 95% CI, 0.16-0.70), 2-year disease-specific survival (78% vs 91%; OR, 0.34; 95% CI, 0.14-0.79), and 5-year overall survival (36% vs 60%; OR, 0.38; 95% CI, 0.17-0.84). Body mass index was a predictor for 2-year overall survival (38% for BMI <18.5, 76% for BMI of 18.5-24.9, and 85% for BMI ≥25.0; OR, 0.20 for BMI <18.5; 95% CI, 0.06-0.70) and 5-year overall survival (18% for BMI <18.5, 39% for BMI 18.5-24.9, 56% for BMI ≥25.0; OR, 0.35 for BMI <18.5; 95% CI, 0.07-1.86). Age, sex, CCI score, and adjuvant radiotherapy or chemoradiotherapy were not significant predictors of mortality on univariate analysis (Table 2).
Further multivariate analysis with a model including BMI, smoking status, human papillomavirus status, preoperative albumin level, and sarcopenia (Table 3) demonstrated that sarcopenia was the only predictor for 2-year disease-specific survival (OR, 0.32; 95% CI, 0.12-0.84), 2-year overall survival (OR, 0.33; 95% CI, 0.14-0.77), and 5-year overall survival (OR, 0.38; 95% CI, 0.17-0.84). A low BMI (<18.5) was not found to be a predictor of 2- or 5-year overall survival.
Although sarcopenia is relatively well established as a negative predictor of surgical and oncologic outcomes in a variety of surgical specialties, its role in HNC is not as recognized. Sarcopenia is a predictor of mortality in patients with HNC undergoing definitive radiotherapy.18 However, all studies examining patients with HNC who underwent surgery focused on total laryngectomy7,19,20 and have generally evaluated surgical complications and pharyngocutaneous fistula. To our knowledge, the importance of sarcopenia in outcomes of patients with HNC undergoing the breadth of head and neck surgical oncologic procedures has not been examined. To address this gap in the literature, we analyzed sarcopenia in 260 patients undergoing a variety of head and neck ablative surgical procedures. We found that sarcopenia appears to be a significant predictor of mortality at both 2 years and 5 years after treatment. Indeed, at 5 years after treatment, sarcopenia was the only significant predictor of overall survival on multivariate analysis. These results highlight the importance of preoperative sarcopenia and the potential clinical utility of sarcopenia measurement.
It is well established that patients with lower pretreatment weight often have worse outcomes than those who are normal weight or overweight.21 This difference could be due to a number of factors, including a more advanced stage of disease at initial presentation. However, weight loss alone is not necessarily indicative of outcome18,22; instead, it must be coupled with other measures, such as BMI, to give a more accurate prognosis. A number of studies demonstrated that sarcopenia is a more critical factor than BMI when determining prognosis of patients with solid-tumor HNC.23,24 Grossberg et al18 suggested that BMI and sarcopenia were colinear variables that could not be separated when determining prognosis. They further asserted that, because increased BMI accounted for increased skeletal muscle index, BMI was a superior prognostic indicator for HNC. Our results concurred with the assertion that, as BMI rises, skeletal muscle index rises as well, with an expected drop in instances of sarcopenia. However, in our data set, there was a significant percentage of overweight patients who exhibited sarcopenia, and these values were not colinear variables. Body mass index was not an independent negative prognostic indicator of survival at 2 or 5 years. Sarcopenia was the only variable that was predictive of both 2- and 5-year survival.
These results are consistent with those of other studies demonstrating that sarcopenia is a negative predictor of outcomes regardless of BMI5,25; in other disease states, sarcopenic obesity may be especially deleterious.26 Our data generally support this idea; although only about 5% of patients in our study demonstrated sarcopenia and obesity (BMI >30), these patients demonstrated significantly worse 5-year overall and disease-specific survival than all other patients with sarcopenia. Thus, preoperative sarcopenia assessment may be useful to identify this particularly high-risk subgroup that might not be identified by other clinical assessments.
Estimates of the prevalence of sarcopenia in the HNC population vary significantly across studies, ranging from 16% to as high as 70%.7,11 In a previous study,7 the incidence may have been elevated owing to the high rate of recurrent and T3/T4 disease in the patient population; however, a prevalence of 70% is not significantly elevated from other reports. In the present study, 55.4% of patients met the criteria for sarcopenia, adding to the growing list of research that suggests the prevalence of sarcopenia could be over 50% in HNC,7,18,27 as it is in esophageal carcinoma.8,10,28-30 There remains some uncertainty regarding the optimal area for determination of cross-sectional muscle area. Although measurement at L3 is the most well-established procedure in the literature, abdominal imaging is not as readily available in the HNC population as neck imaging; thus, other studies have utilized the C3 level for measurement. Wendrich et al27 established cutoff values for sarcopenia at the C3 level; however, these were determined as a marker for the development of chemotherapy dose-limiting toxic effects, not as a marker of survival outcomes, as were the established cutoffs at L3.5 Thus, we used the L3 vertebra level in this study; future studies to determine the optimal sarcopenia cutoff values at the C3 veterbra level that stratify for survival could provide a significant benefit to the future study of sarcopenia in HNC patients.
In the overall survival analysis presented herein, sarcopenia was associated with a decrease of 13 percentage points at 2 years and 24 percentage points at 5 years. This finding is similar to the difference in survival seen in patients undergoing total laryngectomy20 and similar to the effect seen in patients undergoing definitive radiotherapy or chemoradiotherapy.18 This outcome is larger than that of many well-established treatments, such as the addition of chemotherapy to adjuvant radiotherapy in patients with high-risk features after surgical resection.31 Furthermore, although the association between sarcopenia and survival at 2 years is not as notable as that of cancer staging, sarcopenia remains the only predictive variable of overall survival at 5 years, surpassing even cancer staging information. It is possible that sarcopenia is a marker of other variables, such as maladaptive inflammatory responses (ie, cachexia), dysphagia, overall poor health, or other tumor/treatment effects that may lead to poor outcomes. However, these data suggest that nutritional interventions both before and after surgery in patients with sarcopenia would be beneficial. Future interventional studies along these lines have the potential to significantly improve outcomes for these patients with otherwise dismal long-term prospects.
This study has important limitations. As previously mentioned, abdominal computed tomographic imaging is not always a part of standard HNC evaluation, although it is often obtained as part of positron emission tomographic-computed tomographic imaging. However, many patients were excluded owing to a lack of imaging available for this analysis, which could introduce a selection bias. As with any retrospective study, missing data or lack of follow-up can also introduce bias into the results. In addition, as this study was conducted at a single tertiary care institution, these results may not be generalizable to all patients with HNC.
Despite the study’s limitations, these data may provide an important benchmark for the role of sarcopenia in patients with HNC. To our knowledge, this is the first study to suggest that sarcopenia is a significant negative prognostic factor in a broad population of patients with HNC undergoing ablative surgery. Sarcopenia was present in 55.4% of the patients with HNC and was a significant predictor of decreased survival at both 2 years and 5 years after treatment. These data suggest that sarcopenia measurement is a useful prognostic tool in patients with HNC and may serve as a marker to target patients for intensive nutritional support or other intervention to improve outcomes.
Accepted for Publication: April 11, 2019.
Corresponding Author: Daniel Clayburgh, MD, PhD, Department of Otolaryngology/Head and Neck Surgery, Oregon Health and Science University, 3181 SW Sam Jackson Park Rd, PV01, Portland, OR 97239 (firstname.lastname@example.org).
Published Online: June 6, 2019. doi:10.1001/jamaoto.2019.1185
Author Contributions: Mr Stone and Dr Clayburgh had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Stone, Krasnow, Jiang, Li, Wax, Andersen, Marks, Achim, Clayburgh.
Acquisition, analysis, or interpretation of data: Stone, Olson, Mowery, Krasnow, Jiang, Li, Schindler, Wax, Marks, Achim, Clayburgh.
Drafting of the manuscript: Stone, Jiang, Wax, Achim, Clayburgh.
Critical revision of the manuscript for important intellectual content: Stone, Olson, Mowery, Krasnow, Li, Schindler, Wax, Andersen, Marks, Achim, Clayburgh.
Statistical analysis: Stone, Mowery, Clayburgh.
Administrative, technical, or material support: Olson, Krasnow, Wax, Marks, Clayburgh.
Supervision: Li, Schindler, Wax, Andersen, Achim, Clayburgh.
Conflict of Interest Disclosures: Dr Marks received personal fees from Pfizer and grants from the National Institutes of Health during the conduct of the study. No other disclosures were reported.
Meeting Presentation: The study was presented at the American Head & Neck Society Annual Meeting held during the Combined Otolaryngology Spring Meetings; May 2, 2019; Austin, Texas.
Create a personal account or sign in to: