Obesity is not associated with adverse outcome following surgical resection of oesophageal adenocarcinoma

Abstract

Objective: To study the impact of obesity on postoperative morbidity and outcome following surgical resection of primary oesophageal adenocarcinoma (EADC). Methods: From a prospective database, we compared clinicopathological findings (age, gender, surgical approach, tumour differentiation and stage), postoperative mortality, morbidity, length of hospitalisation, disease-free survival (DFS) and overall survival (OS) between 56 obese (body mass index (BMI) ≥ 30 mg kg⁻²) and 86 non-obese (BMI ≪ 30 mg kg⁻²) patients with EADC. Results: In this consecutive series, there were 118 male and 24 female patients with a median age of 63 years (range, 36–85 years). For all patients, the 5-year OS was 26.9%, with a median survival of 20 months. No significant differences (P > 0.05) were found between obese and non-obese patients, with respect to age, gender, surgical approach (transthoracic vs transhiatal), pT-stage, duration of hospital stay, postoperative mortality or morbidity. However, compared with non-obese patients, obese patients had a higher frequency of postoperative respiratory complications (odds ratio (OR), 3.05; 95% confidence intervals (CIs), 1.29–7.17). DFS and OS at 5 years were increased for patients who were obese at the time of oesophageal resection (P = 0.008). Conclusions: Obesity is not associated with increased postoperative complication rates or adverse outcome following oesophageal resection, and should therefore not be considered a relative contraindication to the surgical management of EADC. The improved survival of obese patients who underwent oesophageal resection for EADC suggests that further investigation of the association between obesity and oesophageal malignancy is now warranted.

Oesophagus, Adenocarcinoma, Obesity, Outcome

1 Introduction

In parallel with epidemiologic trends for cancer throughout Europe and North America, prevalence rates for obesity have also increased steadily over the past several decades [1,2]. Obesity is now recognised to be a risk factor for Barrett oesophagus [3] and oesophageal adenocarcinoma (EADC), an observation supported by a number of population-based case-control and cohort studies [4,5], and two recent meta-analyses [6,7]. As obesity is a strong predictor of mortality [8], a number of recent studies have evaluated outcomes for obese patients who underwent various high-risk procedures including cardiac surgery [9,10], pancreatic resection [11,12], and oesophageal resection for cancer [13–15]. It was reported that morbidly obese patients did not appear to experience substantially increased postoperative complication or mortality rates, observations consistent with a recent meta-analysis of patients requiring intensive care [16]. Furthermore, reduced mortality has been reported for obese patients with heart failure, giving rise to the concept of an ‘obesity survival paradox’ [17].

Recognising the importance of obesity in the pathogenesis of EADC, the clinical challenges associated with management of obese patients, and anticipating increasing numbers of obese individuals requiring surgery for EADC, the objective of this short report was to determine the impact of obesity on postoperative morbidity and outcome following surgical resection of primary EADC.

2 Materials and methods

2.1 Patients

A detailed description of the patient population has recently been reported [18]. Briefly, between 1991 and 2006, 142 consecutive patients (118 males and 24 females, with a mean age of 62 years) underwent oesophageal resection for EADC. Strict clinicopathological criteria were used to define adenocarcinomas of primary oesophageal origin (Siewert type I), thereby excluding cardia (type II) and subcardia gastric cancers (type III). No patient received induction chemotherapy or radiotherapy. Routine preoperative staging comprised oesophagogastroscopy and biopsy, radiologic contrast studies (barium swallow) and computed tomography (CT) of the chest and upper abdomen. Endoscopic ultrasound (EUS) and positron emission tomography (PET) were not available for staging. All patients were considered to have locally resectable tumours with no evidence of distant metastases. The operative approach [18,19] comprised a subtotal oesophagectomy performed using a right thoracotomy (77 patients) or transhiatally (65 patients), achieving a complete excision of the primary tumour en bloc with the thoracic and abdominal oesophagus, and the lesser curvature of the stomach to achieve a minimum 5 cm distal resection margin. A narrow gastric tube based on the right gastroepiploic artery was transposed through the posterior mediastinum, where an anastomosis to the cervical oesophagus was performed in the left neck [19]. A pyloromyotomy and a feeding jejunostomy (J-tube) were performed routinely. The surgical approach was not modified by individual patient body mass index (BMI). Postoperative follow-up comprised 3-monthly office visits for the first 3 years, 6-monthly monitoring (years 4 and 5), then yearly. All reasonable attempts were made to confirm tumour recurrence and/or metastasis cytologically or histologically, using radiologic-guided fine needle aspiration or endoscopic biopsy techniques.

2.2 Data collection and statistical analysis

Routine clinicopathological and outcome data were recorded in a prospective research database, which included measured height and weight from which BMI was calculated [4]. Descriptive statistics were used to summarise data. We compared clinicopathological findings (age, gender, surgical approach, tumour differentiation and stage), postoperative mortality, morbidity, length of hospitalisation, disease-free survival (DFS) and overall survival (OS) between obese (BMI ≥ 30 mg kg⁻²) and non-obese (BMI ≪ 30 mg kg⁻²) patients using statistical tests as appropriate, and as reported in detail previously [18]. Statistical significance was set a priori at P ≪ 0.05. All analyses were performed using SPSS for Windows 15.0 (SPSS Inc., Chicago, IL, USA). Approval for data analysis was granted by the University of Saskatchewan College of Medicine Research Ethics Committee (M07-22).

3 Results

3.1 Overall mortality, morbidity, clinicopathological findings, staging and outcome

Postoperative mortality and morbidity, clinicopathological findings, staging and survival of all patients included in this series have recently been reported [18]. In summary, 30-day and in-hospital mortality was 2.8% (4/142), and 46.4% of surviving patients (64/138) experienced postoperative morbidity. As many as 53.6% of patients (76/138) had an uncomplicated postoperative course, with a median length of stay of 10 days (range, 7–27 days), compared with a median of 17 days (range, 7–63 days) if complicated (P ≪ 0.0001). A total of 41 EADCs were well differentiated (G1), 37 moderately differentiated (G2) and 64 poorly differentiated (G3), and were staged as I (n = 12), IIA (n = 41), IIB (n = 8), III (n = 76) and IVa (n = 5). A complete (R0) resection was confirmed histologically in 83.8% (119/142) of patients. Microscopic residual tumour (R1) was documented in 12.0% of patients (17/142) at the lateral tumour resection margin. Macroscopic residual tumour (R2) was documented in 4.2% of patients (6/142), resulting from extracapsular nodal disease (predominantly coeliac nodes). The median number of lymph nodes identified pathologically was 13 (range, 5–29). At last follow-up, 38 patients were alive and disease-free, with a median follow-up of 62 months (range, 15–137 months). Recurrent or metastatic disease was documented in 63.4% of patients (90/142), and was predominantly systemic (84.4%; 76/90). In general, 5-year survival was 26.9%, with a median survival of 20 months. As summarised in Table 1 , significant prognostic factors included complete resection, tumour grade, pT-stage, pN-stage, the number of positive lymph nodes resected, the ratio of positive to resected lymph nodes and BMI.

Table 1

Cox univariate regression analysis of selected prognostic factors [18], including BMI, for overall survival following surgical resection of primary oesophageal adenocarcinoma.

3.2 Comparison of variables between obese patients (BMI ≥ 30 mg kg⁻²) and non-obese patients (BMI ≪ 30 mg kg⁻²)

Fifty-six patients (39.4%) in this series had a measured preoperative BMI of ≥30 mg kg⁻² and were considered obese with a median BMI of 34.9 mg kg⁻² (range, 30.1–45.1 mg kg⁻²). Three patients (2.1%) had a BMI > 40 mg kg⁻². The median BMI of non-obese patients was 26.6 mg kg⁻² (range, 17.1–29.9 mg kg⁻²) with only one patient classified as underweight (BMI ≪ 19 mg kg⁻²). As summarised in Table 2 , no significant differences (P > 0.05) were found between obese and non-obese patients, with respect to age, gender, surgical approach (transthoracic vs transhiatal), pT-stage, duration of hospital stay, postoperative mortality or morbidity overall. However, compared with non-obese patients, obese patients had a higher frequency of postoperative respiratory complications (odds ratio, OR, 3.05; 95% confidence intervals, CIs, 1.29–7.17), and to have a higher frequency of well-differentiated tumours (OR 2.28; 95% CI 1.09–4.78; P = 0.03). Obese patients were also found to have fewer pN1-staged tumours (OR, 0.46; 95% CI 0.23–0.92; P = 0.03) than non-obese patients, although the number or lymph nodes resected did not differ substantially between non-obese (median 13, range 5–29) and obese patients (median 12, range 6–23). No differences in outcome were found for patients with pN0 tumours when stratified by BMI. However, when patients with pN1 tumours were considered, no differences in DFS (OR 0.68; 95% CI 0.42–1.09; P = 0.12) and only marginally improved OS (OR 0.62; 95% CI 0.38–0.99; P = 0.05) were found for obese, compared with non-obese, patients.

Table 2

Comparison of demographic variables between obese patients (BMI ≥ 30 mg kg⁻²) and non-obese patients (BMI ≪ 30 mg kg⁻²) with oesophageal adenocarcinoma.

In general, DFS and OS at 5 years were increased for patients who were obese at the time of oesophageal resection (P = 0.01, DFS; P = 0.0008, OS; Fig. 1 ) compared with non-obese patients. These findings were confirmed in multivariate Cox models which showed that obesity was associated with improved survival (OR 0.64; 95% CI 0.41–0.98; P = 0.03), after controlling for tumour grade, extent of resection and pT-stage.

Fig. 1

Survival curves for patients after surgical resection of primary oesophageal adenocarcinoma categorised by body mass index. Obese patients (with a body mass index ≥30 mg kg⁻²) had significantly improved 5-year overall survival compared with non-obese patients.

4 Discussion

In keeping with two recent surgical series [13,15], we have also confirmed that obesity is not associated with increased postoperative mortality or morbidity following oesophageal resection. Strengths of this study included our use of strict clinicopathological criteria to define primary EADC, and that a consistent operative approach was used to treat a consecutive series of 142 patients. The detailed study by Healy et al. [13] included only 64 type I EADCs (the remaining patients had type II and III adenocarcinomas), and tumour histology was not stated in the study by Kilic et al., which focussed on obesity-related outcomes associated with minimally invasive oesophagectomy [15]. This current series is also unique in that no patient received induction therapy, now standard practice in many centres, including the two earlier studies where 36%[15] to 67% [13] of obese patients received preoperative chemo- and/or radiation therapy. Interestingly, the use of neo-adjuvant therapies was recently reported to be a significant predictor for postoperative complications in overweight/obese patients with adenocarcinomas of the oesophagogastric junction [20].

Neither published study reported any significant difference in mortality between non-obese and obese patients who underwent oesophageal resection [13,15]. Mortality rates in our series of 2.3% (non-obese) and 3.6% (obese) are comparable with previously published rates of 2.5% for non-obese patients and 0% for obese patients [15], and 6% for both non-obese and obese patients [13]. Although we found no difference in overall postoperative morbidity between non-obese and obese patients, a higher frequency of respiratory complications was seen in obese patients (31.5%, 17/54) than in non-obese patients (13.1%, 11/84), similar to findings by Healy et al. [13], who reported respiratory complications in 58% (21/36) of obese patients and 38% (43/114) of non-obese patients in their series. It is possible that the lower respiratory complication rates reported by Kilic et al. [15], which did not differ between non-obese and obese patients, may well reflect the minimally invasive approach used exclusively at that centre. Cervical anastomotic leak rates were not statistically different in our series (10.5%, 9/86 for non-obese patients vs 16.1%, 9/56 for obese patients) and that reported by Kilic et al. [15] (12.6%, 25/198 for non-obese patients vs 13.1%, 11/84 for obese patients). However, Healy et al. [13] reported a higher percentage of anastomotic leaks in obese patients (14%, 5/36 vs 2%, 2/114, if non-obese), which included both thoracic and cervical anastomoses. Although multivariate analysis identified obesity as the only factor associated with anastomotic dehiscence, the use of water-soluble contrast alone to study the anastomosis may well account for the relatively low percentage of anastomotic leaks identified in non-obese patients [13]. We have previously recommended the use of dilute barium to avoid potential false-negative results, which may arise from the use of water-soluble contrast alone [19]. With respect to postoperative length of stay, whereas Healy et al. [13] reported a significantly longer hospital stay for obese patients (median, 23 days) compared with non-obese patients (median, 18 days), no significant differences were found by Kilic et al. [15] (median 7 days for obese, 8 days for non-obese patients) or in our series (median 10 days for obese, 11 days for non-obese patients).

From a well-characterised study population of 177 patients with EADC, who underwent surgical resection, Sundelof et al. reported borderline improved 5-year overall survival for obese patients (hazard ratio (HR), 0.6; 95% CI 0.3–1.0) compared with patients with normal BMI [14]. However, in this population-based retrospective analysis, few details regarding tumour stage and surgical approach were available, and BMI was calculated from estimated weight 20 years before diagnosis. In the study by Healy et al. [13], 5-year overall survival rates were reported for 46% of obese patients and 22% of non-obese patients, comparable with our findings of improved 5-year survival for obese patients (34.0% vs 23.0% for non-obese patients; P = 0.008; Fig. 1). The precise biologic mechanisms underlying these observations remain unknown, as do the mechanisms of obesity-related carcinogenesis. While there is an intuitive relationship between obesity, hiatus hernia and gastro-oesophageal reflux disease (GERD), also a significant risk factor for EADC [21,22], there has been increasing speculation that obesity may promote tumour development through various systemic inflammatory, metabolic and endocrine pathways involving active circulating biomarkers such as C-reactive protein, insulin-like growth factor, leptin and various adipokines derived from adipose tissues (reviewed in McTiernan) [23]. We recently reported that a common genetic polymorphism of the insulin-like growth factor I receptor gene (IGF-IR) modulates the risk of obesity for EADC [24], an effect mediated by altered IGF-IR gene expression [25], thereby representing a plausible molecular mechanistic link underlying the association between obesity and malignancy. The role of adipocyte-derived systemic inflammatory peptides and cytokines in modulating postoperative survival would appear to be a promising area of future research.

In conclusion, obesity is not associated with increased postoperative complication rates or adverse outcome following oesophageal resection, and should therefore not be considered a relative contraindication to the surgical management of EADC. The improved survival of obese patients who underwent oesophageal resection for EADC suggests that further investigation of these observations is now warranted for patients with oesophageal malignancy.

☆

Funding for these studies was obtained in part from the National Cancer Institute of Canada with funds from the Canadian Cancer Society and the Canadian Institutes for Health Research.

Acknowledgements

The following consultant pathologists were invaluable for expert independent clinical histopathologic review at each respective centre: Drs M. Troster (London, ON); Dr M. Redston (Toronto, ON); Dr J. Newman and Professor J. Crocker (deceased) (Birmingham, UK); and Drs D. Malatjalian and L. Geldenhuys (Halifax, NS). For follow-up, we wish to thank the following clinical trials nurses and administrative assistants: Bev Neskas, Sue Troyan and Natalie Zankowicz (London, ON); Susan Rosgen and Frances Hui (Toronto, ON); S. Jane Darnton and the Cancer Intelligence Unit (West Midlands, Birmingham, UK); Susan Winch and Dianne Russell (Halifax, NS) in addition to participating patients, their families and personal physicians. We thank Sarika Mann (Saskatoon, SK) for data management.