Introduction

Obesity is one of the most important public health problems worldwide and is associated with a wide variety of cancers [1, 2]. It is closely correlated with colorectal cancer (CRC) recurrence and increased mortality [3,4,5]. In particular, because surgery for mid and low rectal cancer (MLRC) is performed within the narrow pelvic cavity and requires precise total mesorectal excision (TME), it is thought to be even more affected by visceral obesity. Although body mass index (BMI) is the most widely used index to assess obesity, it does not always accurately reflect the degree of obesity within the abdominal cavity [6,7,8]. Computed tomography (CT) and magnetic resonance imaging were recently acknowledged as the gold standards for measuring intra-abdominal fat volume [7, 9, 10], such as visceral fat area (VFA), subcutaneous fat area (SFA), total fat area (TFA) and VFA/SFA ratio (V/S ratio). We think that it is very important to examine whether visceral obesity is associated with the oncological outcomes of MLRC and identify an obesity index that accurately predicts such outcomes.

The objective of this study was to examine whether visceral obesity measured by CT can predict the oncological outcomes of patients with MLRC. We also aimed to identify the obesity index that most accurately reflects clinical outcomes.

Method

Patients and clinical data collection

We retrospectively analysed patients diagnosed with MLRC between September 2004 and December 2010. MLRC was defined as a tumour within 10 cm from the anal verge. Only those patients who underwent elective curative resection after a preoperative abdominopelvic CT were included. A total of 158 patients underwent surgery for MLRC. Thirty-three patients were excluded from the study. A total of 125 patients were enrolled in this study; none of them received neoadjuvant chemoradiation therapy. Patients with stage II or III disease received 5-fluorouracil-based adjuvant chemoradiation therapy after surgery.

All of the surgeries were performed by the same experienced colorectal surgeon and surgical team. The patients were followed up every 3 months for the first 3 years, every 6 months for the following 2 years and once per year thereafter. The median follow-up time was 60.3 months (range, 38.2–122.6 months).

BMI measurement

We retrieved the patients’ heights and weights from the anaesthesia data recorded on the day of the operation. BMI was calculated by dividing weight in kilogrammes by the squared height in meters (kg/m2).

Volumetric measurements of abdominal adipose tissue

We measured VFA and TFA on the CT scans, which were obtained with the patient in a supine position using one of two scanners (Somatom Sensation 16, Siemens Medical Solutions, Forchheim, Germany; Brilliance 64, Philips Healthcare, Cleveland, OH, USA). The images for VFA measurement were obtained at the umbilical level on a single 5-mm-thick slice. The acquired images were analysed using commercially available software (Rapidia version 2.8; INFINITT, Seoul, Korea). The attenuation level of the software was set between 190 and 30 Hounsfield units, the specific CT range for adipose tissue [11]. We defined VFA as the intra-abdominal adipose tissue area within the parietal peritoneum, excluding the paraspinal muscle, intervertebral bodies and intramuscular fat. SFA was defined as the adipose tissue external to the peritoneum and back muscle. A region of interest drawn along the external margin of the dermis was used to calculate TFA. The SFA was obtained by subtracting the VFA from the TFA. All measurements were made by one experienced radiologist to minimise inter-examiner variation and reported in squared centimetres.

Statistical analysis

The statistical analyses were conducted using SPSS version 21.0 (SPSS Inc., Chicago, IL, USA). Student’s t test and two-tailed Chi-squared tests were used for univariate analysis. Obesity indices were evaluated as continuous measures and are expressed as mean ± standard deviation (SD). Pearson’s correlation coefficients (ρ) were used to assess the correlation between obesity indices. On multivariate analysis, Cox’s proportional hazard model was used to assess the predictive factors for recurrence. The Kaplan-Meier analysis and log-rank tests were used to compare differences in disease-free survival (DFS) and overall survival (OS). P values < 0.05 were considered statistically significant.

Results

Demographics and clinical features

A total of 125 patients (mean age, 63 ± 11 years) were included. There were more male (83; 66.4%) than female (42; 33.6%) patients. The numbers of patients with stage I, II and III cancers were 32 (25.6%), 27 (21.6%) and 66 (52.8%), respectively.

Obesity indices

The mean BMI, TFA, VFA, SFA and V/S ratio were 23.4 kg/m2, 260.6 cm2, 118.4 cm2, 142.2 cm2 and 0.89, respectively. Female patients had greater TFA and SFA values than male patients. Male patients had a higher V/S ratio than female patients (1.00 ± 0.37 versus 0.69 ± 0.27; P < 0.001). BMI was correlated with TFA (ρ = 0.708, P < 0.001), VFA (ρ = 0.663, P < 0.001) and SFA (ρ = 0.582, P < 0.001) but not with V/S ratio (ρ = 0.105, P = 0.24).

Risk factors for recurrence

Of the 125 patients, 28 (22.4%) experienced recurrence. The recurrence group had a higher preoperative carcinoembryonic antigen (CEA) level (10.5 versus 6.8 ng/mL; P = 0.02) and more frequent lymphatic (P = 0.022) and perineural (P = 0.019) invasions. Higher T and N stage were associated with more frequent recurrence (P = 0.033 and P = 0.005, respectively), and stage III patients had the highest recurrence rate (45 of 97, 46.4% versus 21 of 28, 75.0%; P = 0.015). Only the V/S ratio differed significantly between the recurrence and no recurrence groups (0.86 ± 0.34 versus 1.02 ± 0.45; P = 0.046). There were no significant differences in gender, age, differentiation, vascular invasion, circumferential resection margin (CRM), TME, distal margin or time to chemotherapy between the two groups.

On a multivariate Cox regression analysis, only V/S ratio (hazard ratio 3.323, 95% confidence interval 1.22 to 9.09; P = 0.019) was found to be an independent risk factor associated with recurrence.

Survival analysis

We analysed the patients’ DFS and OS using the V/S ratio, an identified risk factor for recurrence. We defined V/S ratio = 1.0 as the cut-off value and analysed the survival rates of the obese and non-obese groups. The obese group had a shorter mean DFS (P = 0.02) and OS (P = 0.01) (Fig. 1).

Fig. 1
figure 1

Disease-free (a) and overall survival (b) in obese and non-obese patients when a visceral fat area/subcutaneous fat area ratio of 1.0 is defined as the cut-off value. a P = 0.021, b P = 0.011 (log-rank test)

Full size image

Discussion

We found two studies that investigated long-term outcomes of rectal cancer and visceral obesity. However, these studies either included patients with upper rectal cancer, which is less affected by visceral obesity than MLRC [12] or defined visceral obesity using an arbitrary V/S ratio of 0.4 as cut-off value [2]. There is no established set of volumetric parameters to define obesity. In our study, the mean V/S ratio was higher in the recurrence group (mean V/S ratio, 1.0), while the other obesity indices were not associated with recurrence. Moreover, multivariate analysis showed that only the V/S ratio was an independent risk factor for recurrence. To establish the criteria for visceral obesity, we performed additional analyses using the V/S ratio. We analysed the survival rates of the obese and non-obese groups using V/S ratio values of 0.7, 0.8, 0.9 and 1.0 as cut-off values. The threshold value that resulted in different DFS and OS outcomes between the two groups was 0.8 (P = 0.01 and P = 0.01, respectively). In addition, the DFS (P = 0.02 and P = 0.02, respectively) and OS (P = 0.01 and P = 0.01, respectively) of the obese group were significantly shorter than those of the non-obese group at cut-off values of 0.9 and 1.0. Considering that the mean V/S ratio of the entire subject pool was 0.89, it would be acceptable to identify a V/S ratio = 1.0 as the most appropriate cut-off value for dividing patients into obese and non-obese groups while best reflecting the intergroup differences in DFS and OS. These findings have several important clinical implications. First, when predicting the recurrence rate for patients with MLRC, it is more appropriate to consider the distribution of adipose tissue than to only consider general or visceral obesity. Second, this study provides objective evidence for defining visceral obesity in future studies by establishing the mean V/S ratio and threshold that results in the difference in survival rate for patients with MLRC.

It should be mentioned that our study has several limitations. First, the patients in this study were enrolled in a single university hospital in Korea. Therefore, the number of participants was small, and all were Korean. We think that a further study is needed to validate whether our results apply to large numbers of Western populations. Second, our study did not include patients who underwent preoperative chemoradiation therapy. Although neoadjuvant chemoradiation therapy has been established as a standard treatment in patients with advanced MLRC, this treatment modality was not applied routinely to all patients during the study period in our institution. We assume that these differences in treatment modality may have contributed to the higher recurrence rate. Further studies to clarify the relationship between the V/S ratio and recurrence in patients with MLRC treated with neoadjuvant chemoradiation therapy may be needed in the future.

An increased V/S ratio was significantly associated with a higher recurrence rate and shorter DFS and OS in patients with MLRC who underwent curative resection. Of a variety of obesity indices, only V/S ratio accurately predicted the recurrence and survival outcomes of patients with MLRC.