Fax: (713) 745-1163
The influence of high body mass index on the prognosis of patients with esophageal cancer after surgery as primary therapy
Article first published online: 8 NOV 2010
Copyright © 2010 American Cancer Society
Volume 116, Issue 24, pages 5619–5627, 15 December 2010
How to Cite
Hayashi, Y., Correa, A. M., Hofstetter, W. L., Vaporciyan, A. A., Rice, D. C., Walsh, G. L., Mehran, R. J., Lee, J. H., Bhutani, M. S., Dekovich, A., Swisher, S. G. and Ajani, J. A. (2010), The influence of high body mass index on the prognosis of patients with esophageal cancer after surgery as primary therapy. Cancer, 116: 5619–5627. doi: 10.1002/cncr.25745
- Issue published online: 3 DEC 2010
- Article first published online: 8 NOV 2010
- Manuscript Accepted: 18 AUG 2010
- Manuscript Revised: 10 AUG 2010
- Manuscript Received: 3 MAY 2010
- body mass index;
- prognostic factors;
- esophageal cancer;
High body mass index (BMI), a prevalent condition in the United States, is associated with esophageal adenocarcinoma (EAC). Its influence on a patient's outcome remains unclear. In the current study, the authors examined the impact of BMI on survival and complications in patients with esophageal cancer (EC) who underwent surgery as their primary therapy.
The authors retrospectively reviewed 301 consecutive EC patients who underwent surgery but received no adjunctive therapy. Patients were segregated into 2 subgroups based on their baseline BMI: normal/low (<25 kg/m2) and high (≥25 kg/m2).
Seventy-six (25%) patients had a BMI <25 kg/m2 and 225 (75%) patients had a BMI ≥25 kg/m2. In the high BMI group, there were more men (P < .001), fewer upper ECs (P = .021), a lower baseline clinical stage (P = .006), and frequent EAC (P < .001). Postoperative morbidity was similar in both groups, with the exception of gastrointestinal complications (P = .016). The 5-year overall survival (OS) rates were 44% in the normal/low BMI group and 60% in the high BMI group (P = .017). The 5-year disease-free survival (DFS) rates were 41% in the normal/low BMI group and 60% in the high BMI group (P = .005). On multivariable analysis, age, weight loss, peripheral vascular disease (PVD), and both clinical and pathological stage of disease were found to be independent prognosticators for OS. Older age (hazard ratio [HR], 1.029; 95% confidence interval [95% CI], 1.009-1.049 [P = .004]), weight loss (HR, 1.525; 95% CI, 1.034-2.248 [P = .033]), and PVD (HR, 2.325; 95% CI, 1.039-5.204 [P = .040]) were found to be associated with poor OS.
High BMI is common in EC patients and the better OS/DFS noted in patients with a high BMI is because of the diagnosis of a low baseline clinical stage. Confirmation of these findings is warranted. Cancer 2010. © 2010 American Cancer Society.
High body mass index (BMI) is a serious health problem in the United States. According to the National Health and Nutrition Examination Survey, approximately 65% of adults were overweight (defined as a BMI of 25 to 30 kg/m2) and 30% were obese (defined as a BMI of ≥30 kg/m2).1 These numbers are increasing gradually. High BMI is associated with an increased risk of malignancies that are contributing to 14% to 20% of cancer-related deaths.2 The influence of high BMI on outcome in patients with cancer is an unsolved question. Recent studies demonstrated no significant correlation between BMI and survival rate in patients with ovarian, colon, and prostate cancer.3-6
In 2009, approximately 16,000 new cases and 14,000 deaths from esophageal cancer (EC) were estimated to have occurred in the United States.7 In the West, the incidence of esophageal adenocarcinoma (EAC) has been rising dramatically and that of squamous cell carcinoma (SCC) has been subsiding, with approximately 80% of new cases being diagnosed as EAC.8 High BMI is a risk factor for gastroesophageal reflux disease (GERD) and, consequently, for EAC.9-11 Several studies have reported that high BMI was not associated with improved or reduced overall survival (OS) after esophagectomy,12-14 although increased postoperative morbidity, such as respiratory complications, anastomotic leaks, and length of hospital stay, have been associated with high BMI in patients with gastroesophageal cancer who are undergoing surgery.14, 15 Many of these studies included small number of patients, in whom adjunctive therapy (preoperative chemotherapy and/or radiotherapy) was used.
The goal of the current analysis was to learn about the impact of high or low BMI on long-term outcome as well as surgical complications in 301 patients with EC who underwent surgery as their primary therapy at The University of Texas M. D. Anderson Cancer Center (MDACC) but did not receive any adjunctive therapy.
MATERIALS AND METHODS
We searched the EC database in the Department of Thoracic and Cardiovascular Surgery at MDACC to find 301 consecutive patients who were fully staged and had undergone surgery as their primary therapy and received no adjunctive therapy. Patients who did not undergo surgery at MDACC were not included in this database.
Patients were included if they had histologic documentation of EAC or SCC histology and computed tomography scans of the chest and abdomen. All eligible patients were aged ≥18 years. Staging was determined based on the American Joint Committee on Cancer TNM staging system.16 No other selection criteria were implemented. Stored information regarding alcohol consumption, smoking (ever smoker), surgical complications, and other parameters such as length of hospital stay and mortality were captured and analyzed. The Institutional Review Board of MDACC approved this analysis.
Types of esophagectomy included Ivor-Lewis, transhiatal, total esophagectomy, and minimally invasive esophagectomy.
Follow-Up and Survival
Patients were followed periodically for at least 5 years or until death. Additional follow-up data were obtained from review of the MDACC tumor registry and the hospital records or Social Security database. Follow-up time was calculated from the date of surgery to the event or date of the last contact.
The data in the EC database were collected in a prospective manner with a standardized approved protocol. All statistical analyses were performed using the SPSS Statistical Software Package (version SPSS 15.0; SPSS Inc, Chicago, Ill). The differences between groups were tested for significance using the Student t test for continuous variables and the Fisher exact test or chi-square test for categorical variables. Univariable Cox regression analyses were performed using death and disease recurrence or death as the outcomes with a significance level of P ≤ .05. Covariates that were significant at P ≤ .25 were included in the multivariable Cox regression analyses. Backward stepwise Wald elimination at P = .10 was used to find a final model. Death and disease recurrence or death functions were calculated according to the Kaplan-Meier method and differences were assessed using the log-rank test.
A total of 301 consecutive EC patients who underwent surgery as their primary therapy (and received no adjunctive therapy) were analyzed. Among these, 266 (88%) patients had EAC and 35 (12%) patients had SCC. Patient characteristics are summarized in Table 1. Men predominated the group as expected (P ≤ .001) and adenocarcinoma histology was common (P ≤ .001).
|BMI <25 kg/m2 N=76||BMI ≥25 kg/m2 N=225||P|
|≤60||24 (31.6%)||95 (42.2%)|
|61-70||25 (32.9%)||77 (34.2%)|
|>70||27 (35.5%)||53 (23.6%)|
|Weight loss in last 3 mo||34 (44.7%)||50 (22.2%)||<.001|
|Alcohol intake ≥4 oz/d||24 (31.6%)||62 (27.6%)||.502|
|Smoker||46 (60.5%)||104 (46.2%)||.031|
|ASA score 3-4||57 (75.0%)||171 (76.0%)||.860|
|Past medical history|
|Congestive heart failure||1 (1.3%)||2 (0.9%)||.584|
|COPD||6 (7.9%)||22 (9.8%)||.625|
|Coronary artery disease||10 (13.2%)||45 (20.0%)||.182|
|PVD||3 (3.9%)||6 (2.7%)||.406|
|Diabetes||5 (6.6%)||35 (15.6%)||.046|
|0||4 (5.3%)||24 (10.7%)|
|I||23 (30.3%)||104 (46.2%)|
|II||40 (52.6%)||79 (35.1%)|
|III||7 (9.2%)||18 (8.0%)|
|0||2 (2.6%)||23 (10.2%)|
|I||25 (32.9%)||110 (48.9%)|
|II||27 (35.5%)||46 (20.4%)|
|III||18 (23.7%)||36 (16.0%)|
|IV||4 (5.3%)||8 (3.6%)|
|Cervical/upper/middle||16 (21.1%)||24 (10.7%)|
|Lower/GEJ/gastric||60 (78.9%)||201 (89.3%)|
|Adenocarcinoma||55 (72.4%)||211 (93.8%)|
|Squamous cell carcinoma||21 (27.6%)||14 (6.2%)|
|Well||3 (4.1%)||17 (8.5%)|
|Moderately||41 (56.2%)||96 (48.0%)|
|Poorly||28 (38.4%)||85 (42.5%)|
|Undifferentiated||1 (1.4%)||2 (1.0%)|
Patient Characteristics by BMI
Patients were divided into 2 groups according the traditional BMI cutpoints (<25 kg/m2 or ≥25 kg/m2). There were 76 patients (25.2%) with a pretreatment BMI <25 kg/m2 and 225 patients (74.8%) with a pretreatment BMI ≥25 kg/m2. As shown in Table 1, the 2 groups were well matched with regard to age, Zubrod performance status, past medical history except for sex, diabetes, and histological type. However, in the high BMI group, male sex, a primary tumor in the lower esophagus (with or without gastroesophageal junction [GEJ] involvement), EAC, and lower baseline clinical stage were represented at a statistically significantly higher rate than in patients in the low BMI group (P ≤ .001, P = .021, P ≤ .001, and P = .006, respectively). Patients who lost weight during the preceding 3 months from the date of surgery or who smoked tobacco were more likely to be in the normal/low BMI group compared with the high BMI group (P ≤ .001 and P = .031, respectively).
Ivor-Lewis esophagectomy was performed in 132 patients, transhiatal esophagectomy was performed in 125 patients, total esophagectomy was performed in 24 patients, and minimally invasive esophagectomy was performed in 17 patients. Outcomes of surgery are shown in Table 2. The 30-day surgical mortality rate was 2.6% (2 patients) in the normal/low BMI group and 1.8% (4 patients) in the high BMI group (P = .473). Although gastrointestinal complications (diarrhea, dumping syndrome, gastric outlet/conduit obstruction, ileus, and small bowel obstruction) were more frequent in the high BMI group, there was no significant difference noted with regard to other postoperative complications, such as pulmonary (P = .796), cardiovascular (P = .271), and wound-related/infectious complications (P = .509) and anastomotic leak (P = .786). The median postoperative hospital stay was 11 days in the normal/low BMI group and 12 days in the high BMI group.
|BMI <25 kg/m2 n = 76||BMI ≥25 kg/m2 n = 225||P|
|Pulmonary||31 (40.8%)||88 (39.1%)||.796|
|Cardiovascular||13 (17.1%)||52 (23.1%)||.271|
|GI events||10 (13.2%)||60 (26.7%)||.016|
|Anastomotic leak||10 (13.2%)||32 (14.4%)||.786|
|Wound/infection||18 (23.7%)||62 (27.6%)||.509|
|Blood transfusion||11 (15.4%)||31 (13.8%)||.880|
|30-d mortality||2 (2.6%)||4 (1.8%)||.473|
|Recurrence||25 (32.9%)||53 (23.6%)||.108|
|Local/regional recurrence||15 (19.7%)||29 (12.9%)||.144|
|Distant recurrence||21 (26.9%)||45 (15.3%)||.164|
Disease recurrence after surgery developed in 25 patients (32.9%) in the normal/low BMI group and 53 patients (23.6%) in the high BMI group. The type of recurrence in the 2 groups was similar (P = .144 and P = .164, respectively).
OS and Disease-Free Survival
The 5-year OS rates were 43.9% for patients with normal/low BMI and 60.2% for patients with high BMI (P = .017) (Fig. 1). The 5-year disease-free survival (DFS) rates were 41.0% for patients with normal/low BMI and 59.9% for patients with high BMI (P = .005) (Fig. 2). In the univariable analysis, the factors that were found to be significantly associated with OS are shown in Table 3. High BMI and a lower esophageal/gastric/GEJ tumor location were associated with longer OS. Patients with weight loss before surgery, a history of alcohol and/smoking, peripheral vascular disease (PVD), a tumor diameter >3 cm, SCC histology, poor histologic differentiation, a larger number of cancer-positive lymph nodes, and stage II to IV disease were found to have a significantly shorter OS.
|P||Hazard Ratio||95% CI|
|BMI ≥25 kg/m2||.018||0.640||0.442-0.928|
|Weight loss in last 3 mo||<.001||3.161||2.216-4.507|
|Alcohol intake ≥ 4 oz/d||.029||1.488||1.041-2.126|
|ASA score 3-4||.087||1.467||0.946-2.277|
|Past medical history|
|Congenital heart failure||.049||3.159||1.003-9.949|
|Type of surgery||.002|
|Minimally invasive esophagectomy||.077||0.168||0.023-1.213|
|Tumor diameter >3 cm||<.001||6.211||4.226-9.128|
|Tumor location (lower/GEJ/gastric)||.021||0.592||0.379-0.925|
|Histological grade: poorly||<.001||2.711||1.887-3.897|
|No. of positive lymph nodes||<.001||1.115||1.085-1.145|
BMI was eliminated (P = .40) after Step 1 of the 9-step multivariable analysis. Age, weight loss before surgery, PVD, and both baseline clinical and pathological stage of disease were recognized as independent prognostic factors on the multivariable Cox regression analysis (P = .004, P = .033, P = .040, P = .043, and P ≤ .001, respectively) (Table 4). Hazard ratios (HRs) for factors associated with poor OS were as follows: older age (HR, 1.029), weight loss before surgery (HR, 1.525), and PVD (HR, 2.325).
|P||Hazard Ratio||95% CI|
|Weight loss in last 3 mo||.033||1.525||1.034-2.248|
Patients in the high BMI group were more likely to have a clinical stage of disease that was 0 to I than those in the normal/low BMI group, in whom clinical stage was more likely to be II to IV (P ≤ .001). Therefore, we performed subgroup analysis in patients with EAC (because these patients were preponderant) who had a clinical T classification of ≥T2. In this subgroup of 115 patients (Table 5), there were 32 patients (27.8%) with normal/low BMI and 83 patients (72.2%) with high BMI. There was no difference in patient baseline characteristics noted between the 2 groups, including for the factors that were significant in the entire cohort. There were no differences noted with regard to the postoperative complications in the 2 groups (Table 6). The BMI of the patients in this subgroup, whether <25 kg/m2 or ≥25 kg/m2, did not appear to have any relation to OS on either the log-rank test or univariable Cox regression analysis (P = .882 for both) (Fig. 3) (Table 7). Pathological stage of disease and number of dissected and cancer-positive lymph nodes were recognized as independent prognostic factors on the multivariable Cox regression analysis (P ≤ .001, P = .003, and P = .029, respectively) (Table 8). This finding suggests no relation between BMI and the aggressiveness of EC, but BMI certainly appears to be associated with the diagnosis of early stage EC that leads to an improved OS in the group with high BMI, as discussed earlier.
|BMI <25 kg/m2 n=32||BMI ≥25 kg/m2 n=83||P|
|≤60||8 (25.0%)||34 (41.0%)|
|61-70||9 (28.1%)||30 (36.1%)|
|>70||15 (46.9%)||19 (22.9%)|
|Weight loss in last 3 mo||17 (53.1%)||34 (41.0%)||.239|
|Alcohol intake ≥4 oz/d||11 (34.4%)||27 (32.5%)||.850|
|Smoker||23 (71.9%)||53 (63.9%)||.416|
|ASA score 3-4||26 (81.3%)||67 (80.7%)||.949|
|Past medical history|
|Congestive heart failure||1 (3.1%)||2 (2.4%)||.628|
|COPD||3 (9.4%)||6 (7.2%)||.481|
|Coronary artery disease||5 (15.6%)||17 (20.5%)||.553|
|PVD||2 (6.3%)||5 (6.0%)||.630|
|Diabetes||5 (15.6%)||14 (16.9%)||.872|
|II||26 (81.3%)||67 (80.7%)|
|III||5 (15.6%)||16 (19.3%)|
|I||5 (15.6%)||16 (19.3%)|
|II||14 (43.8%)||28 (33.7%)|
|III||10 (31.3%)||32 (38.6%)|
|IV||3 (9.4%)||7 (8.4%)|
|Cervical/upper/middle||2 (6.3%)||7 (8.4%)|
|Lower/GEJ/gastric||30 (93.8%)||76 (91.6%)|
|Well||1 (3.1%)||4 (4.9%)|
|Moderately||17 (53.1%)||25 (30.5%)|
|Poorly||14 (43.8%)||51 (62.2%)|
|BMI <25 kg/m2 n=32||BMI ≥25 kg/m2 n=83||P|
|Pulmonary||15 (46.9%)||28 (33.7%)||.192|
|Cardiovascular||7 (21.9%)||15 (18.1%)||.642|
|GI events||1 (3.1%)||22 (26.5%)||.005|
|Anastomotic leak||3 (9.4%)||9 (10.8%)||.559|
|Wound/infection||6 (18.8%)||20 (24.1%)||.539|
|Blood transfusion||6 (18.8%)||12 (14.5%)||.570|
|30-d mortality||1 (3.1%)||2 (2.4%)||.628|
|Recurrence||13 (40.6%)||44 (53.0%)||.234|
|Local/regional recurrence||6 (18.8%)||25 (30.1%)||.218|
|Distant recurrence||12 (37.5%)||39 (47.0%)||.359|
|P||Hazard Ratio||95% CI|
|BMI ≥25 kg/m2||.882||0.965||0.606-1.538|
|Weight loss in last 3 mo||.001||2.109||1.364-3.260|
|Alcohol intake ≥4 oz/d||.337||0.803||0.513-1.257|
|ASA score 3-4||.960||1.014||0.588-1.747|
|Past medical history|
|Congenital heart failure||.434||1.585||0.500-5.026|
|Type of surgery||.508|
|Minimally invasive esophagectomy||.409||0.433||0.059-3.159|
|Tumor diameter >3 cm||<.001||3.663||2.151-6.237|
|Tumor location (lower/GEJ/gastric)||.116||0.555||0.267-1.155|
|Histological grade: poorly||.007||1.869||1.190-2.935|
|No. of dissected lymph nodes||.082||0.980||0.958-1.003|
|No. of positive lymph nodes||<.001||1.069||1.033-1.106|
|P||Hazard Ratio||95% CI|
|Tumor location (lower/GEJ/gastric)||.062||0.466||0.209-1.041|
|Histological grade: poorly||.094||1.525||0.931-2.499|
|No. of dissected lymph nodes||.003||0.955||0.927-0.984|
|No. of positive lymph nodes||.029||1.072||1.007-1.141|
The literature is not clear regarding the impact of BMI on the outcome of patients with EC. Intuitively, one would imagine that high BMI would be associated with a higher rate of postoperative complications, but there may not be substantive influence on OS or DFS. A few reports that specifically focused on BMI suggested that pretreatment BMI was not associated with differential outcome with regard to OS or DFS.12-14 However, the largest study to date is by Trivers et al17 and included 1142 EAC patients who were segregated in 3 groups: those with a BMI <25 kg/m2, those with a BMI of 25 to29.9 kg/m2, and those with a BMI ≥30 kg/m2. Patients with a BMI ≥30 kg/m2 did not appear to have a survival advantage or disadvantage but <18% of patients had a BMI ≥30 kg/m2. Patients with a BMI between 25 and 29.9 kg/m2 survived longer than those with a BMI <25 kg/m2. Precise stage of disease at diagnosis was not reported in this analysis and the authors, unfortunately, did not analyze the relation between “stage” at diagnosis and BMI. In a much smaller study, Di Fiore et al18 demonstrated that a BMI of >18 kg/m2was one of the independent prognostic factors in 105 patients treated with definitive chemoradiotherapy (P = .003). However, in this study, the cutpoint for BMI was nontraditional and the sample size was small.
In the current study, 74.8% of EC patients had high BMI and had better OS and DFS compared with those with normal/low BMI. However, high BMI, per se, was not found to be an independent prognosticator. Further analyses of our data revealed a finding that has not yet been reported in the literature, namely that high BMI was associated with early baseline stage EC. Thus, the better OS and DFS reported in the high BMI group is attributed to lower EC stage and high BMI does not appear to influence the aggressiveness of established EC. Why would high BMI be associated with a diagnosis of early stage EC? An answer to this intriguing finding remains unclear. We speculate that because patients with high BMI are diagnosed with GERD more often than those with low BMI,19-23 patients with high BMI are more likely to undergo endoscopic surveillance and this could result in the diagnosis of early EC. Lagergren et al11 reported an odds ratio of 7.7 for EAC among patients with GERD symptoms compared with individuals without GERD symptoms. The hypothesis that high BMI patients tend to have GERD, which is associated with EAC, is compatible with our results. In our cohort of patients with high BMI, EAC and lower esophageal localization were significantly more frequent than for patients with low BMI, who had a lower frequency of EAC and tended not have to primary tumors in the lower esophagus (P < .001 and P = .021, respectively).
We found that the patients with normal/low BMI were more likely to have an advanced stage (stage II-IV) of EC (P = .001). Greater than half of the patients with high BMI (56.9%) were diagnosed with clinical stage 0 or I disease, whereas 35.6% of patients with normal/low BMI had an early clinical stage of disease at diagnosis. One explanation for this trend could be that the patients with advanced disease have low appetite and/or significant dysphagia (and weight loss). In the current study cohort, 44.7% of patients with normal/low BMI had lost weight in the 3 months before surgery whereas only 22.2% of patients with high BMI lost weight (P < .001). Our other finding with regard to weight loss, which correlates with a higher tumor burden, as a poor prognosticator is consistent with previously published literature.
High BMI was considered to be a risk factor for surgical complications such as cardiac insufficiency, respiratory failure, anastomotic leak, and wound dehiscence. However, we were not able to confirm these findings and our findings are similar to more recent reports.24, 25 We did observe more gastrointestinal complications in patients with high BMI (P = .016) compared with those with low BMI. These findings need confirmation and the reason for this observation is not clear.
The current study included patients with both EAC and SCC but EAC was diagnosed in approximately 93.8% of patients with high BMI and 72.4% of patients with normal/low BMI (P < .001). Our study could be biased because patients with EAC have been reported to have a better prognosis than those with SCC.26, 27 Holscher et al26 and Meltzer27 have reported that SCC invades the submucosa and lymph nodes at an early stage and the 5-year OS rate of patients with stage I disease who undergo surgery was 82.5% in patients with EAC and 59.2% in patients with SCC. However, this finding is not generalizable and further studies are warranted.
There are weaknesses to the current study: 1) it is a retrospective review; 2) we included a relatively small number of patients; 3) we cannot explain why patients with high BMI had earlier stage EC at diagnosis; and 4) some of our findings are not consistent with the literature. The strengths of the current study are 1) a homogeneous study population (surgery only as therapy and no adjunctive therapy received); 2) detailed analysis is provided; 3) many findings are consistent with the literature; and 4) it has novel findings.
BMI was not an found to be an independent prognostic factor, but it was associated with a lower baseline stage of EAC. It appears that BMI itself has no influence on the clinical aggressiveness of EC and in the current analyses; it was found to have no adverse influence with regard to surgical complications as a whole. The incidence of GERD and EAC is high in patients with high BMI and our novel findings present an opportunity for detecting EAC at an early stage if patients with high BMI can be surveyed more aggressively.
In conclusion, high BMI is common in patients with EC and patients with high BMI are diagnosed with early stage EC and therefore tend to have better survival than those with low/normal BMI.
CONFLICT OF INTEREST DISCLOSURES
Supported in part by grants from The University of Texas M. D. Anderson Cancer Center; the Dallas, Park, Cantu, Smith, and Myers families; and the River Creek Foundation. Also supported by National Cancer Institute (NCI) Grants RO1CA138671, R21CA127672, and R21CA129906. Tissues from some patients presented in this article were used to generate preliminary data for the above NCI grants.
- 16GreeneFL, ed. AJCC Cancer Staging Atlas/American Joint Committee on Cancer. New York, NY: Springer; 2006.
- 18Baseline nutritional status is predictive of response to treatment and survival in patients treated by definitive chemoradiotherapy for a locally advanced esophageal cancer. Am J Gastroenterol. 2007; 102: 2557-2563., , , et al.Direct Link:
- 20The association between gastroesophageal reflux disease and obesity. Am J Gastroenterol. 2008; 103: 2111-2122., , ,Direct Link:
- 21Adenocarcinoma of the esophagus and Barrett's esophagus: a population-based study. Am J Gastroenterol. 1999; 94: 86-91., , , ,Direct Link: