Is BMI sufficient for predicting risk of postoperative complications in patients undergoing laparoscopic-assisted colectomy?

Introduction

Colon and rectal cancers collectively account for approximately 1.24 million new cancer cases per year and colorectal cancer (CRC) is the third most common cause of cancer-related death[1]. There is evidence to suggest colorectal cancer incidence will continue to rise rapidly in developed and developing countries as a result of the so- called ‘western lifestyle’ which is characterized by sedentarity together with diets high in fat and alcohol[2]. While obesity is an established risk factor for colorectal cancer, there is debate as to whether obesity portends a poorer outcome following laparoscopic-assisted colonic resection. A search of the literature is complicated by the varying criteria used for defining obesity. Some studies use BMI cutoffs of ≥25, 27 and 30kg/m² while others suggest measures of visceral adiposity are more reliable for predicting post-operative (PO) outcomes. This article critically examines studies using these diverse criteria in an attempt to identify the best approach to reduce risk of post-operative complications(POC) for CRC patients receiving laparoscopic-assisted colectomy (LAC).

Body Mass Index (BMI)

BMI is a standard measure of obesity used to identify risk of cardiovascular disease and diabetes. Most studies in this area use BMI as it is collected in most, if not all, cases of hospitalization. In a prospective study of 425 patients presenting for LAC, Poulsen and Ovesen found patients with BMIs ≥30 were not at significantly elevated risk of complications in a 30 day follow-up compared to a non-obese cohort (BMI<30), despite having more preexisting co-morbidities[3]. These data are broadly similar to those by Leroy et al who found no increased risk of post-operative complications (POCs) in obese (BMI ≥ 30) colorectal carcinoma patients undergoing left colon laparoscopic resection[4]. These results supporting the safety of LAC for obese patients have been replicated in Asian and South American countries, suggesting socioeconomic and cultural differences have little impact on the success of this procedure although these studies universally agree that operating time on obese patients is significantly elevated[5]. While a number of studies using BMI only record complications in the 30 days following surgery, other data suggest that laparoscopy on patients with BMI≥ 30 is safe in the long-term[6].

The majority of studies addressing the safety of laparoscopy in CRC patients indicate no significant difference in outcomes between obese and non-obese cohorts. However, of the complications most frequently reported in obese patients, higher risks of conversions to an open procedure and wound complications appear to be most common. For example recently published findings by Singh et al. suggest a 27% higher risk of conversion in patients with BMIs over 30, while Mustain and colleagues identified higher incidences of wound complications in obese and morbidly obese (BMI ≥40) patients compared to normal weight controls. It is not easy to identify reasons for these discrepancies, which may be related to the experience of the surgeon with LAC and with operating on obese patients[7, 8].

The problems of BMI in regards to defining obesity are well documented. For example, it does not take into account abdominal adiposity, high levels of which have been reported to pose a technical challenge in visualizing the target area with laparoscopy [9]. Also, it may be inappropriate to apply the same BMI cutoffs to different ethnic groups. Importantly, BMI was initially designed to measure health risks rather than body weight. Indeed, World Health Organisation (WHO) data indicates the current BMI cutoffs for Asian populations should be modified owing to inherently high incidences of cardiovascular disease and diabetes, together with differences in body fat distribution[10]. Thus, more sensitive measures of obesity may more accurately quantify PO morbidity for LAC in CRC patients.

Visceral Adiposity

The limitations of BMI have led to a search for new obesity indices to identify at-risk patients undergoing LAC. One of the most extensively used parameters is for visceral obesity is waist-to-hip ratio (WHR), cutoffs for which are commonly defined as >0.85 for men and >0.9 for women [11]. In a prospective international multicentre study involving 1349 patients WHR was associated with abdominal wall complications and risk of anastamotic leak. In contrast, high BMI (≥30) was highly sensitive for predicting abdominal wall complications, but not for anastamotic leak [12]. The value of WHR is further confirmed by Jung et al. who found that 76% of patients with WHRs of ≥0.9 went on to develop complications, especially wound infections and ileus[13]. Although WHR is demonstrated to be a better predictor of metabolic and cardiovascular events than BMI, limitations to WHR exist. In their study Chan et al. suggest that WHR and BMI cannot accurately measure posterior regional fat mass while recent feeding may temporarily increase abdominal size[14].

A recent study by Watanabe et al. compared BMI with visceral fat area (measured by CT scan at the level of L4 verterbrae) as a means of examining the impact of obesity on PO morbidity in LAC for 338 patients[15]. Interestingly, 54% of patients with an obese VFA (≥100cm²) had a non-obese BMI (<25). Obese VFA patients in this study were at significantly increased risk of anastamotic leak and surgical site infections compared to those with non-obese VFA (<100cm²). Importantly, obese BMI was not associated with anastamotic leak. These results concur with those of Matsumoto et al. who also found a higher rate of anastamotic leak and wound complications in high VFA patients, albeit using a smaller sample (unpublished poster data). Another Japanese study found a striking 16% increased risk of wound complications in VFA-obese compared to VFA non-obese patients [16]. However, this study used a VFA cutoff for obesity of ≥130 cm², making comparison to the aforementioned studies difficult. Future work must use standard VFA cutoffs to facilitate national and international interpretation of findings.

In one of the few European studies to examine visceral fat parameters, Cecchini et al. found visceral/subcutaneous fat ratio (VAT/SAT) to be a sensitive indicator for PO LAC complication, supporting the importance of measuring visceral obesity[17]. In addition, a recent study found perirenal fat surface area (PFSA)(≥40cm²) area to be superior to both BMI and WHR for predicting postoperative complications, although a LAC approach was associated with a better outcome than laparatomy in patients with high PFSA[13]. A number of studies have shown PFSA can be measured effectively using ultrasound[18] which is estimated to be 3-10 times cheaper than CT and does not involve ionizing radiation, thereby reducing risk of secondary malignancies[19].

Hyperglycemia as an additional risk factor

In many of the studies cited above, a significant proportion of patients undergoing LAC had a number of existing co-morbidities which may be related to excess abdominal fat, with cardiovascular disease (CVD) and diabetes being most common. It is therefore important to assess the predictive significance of these factors in POCs. For example, in the study by Watanabe, 15% if patients with obese VFA had diabetes, compared to only 8% of those with non-obese VFA[15]. Similarly , In Jung’s study almost a quarter of patients who developed complications had diabetes compared to under 14% for those that did not develop complications[13].

An independent retrospective study of 200,000 patients by Anand et al surprisingly found that diabetics undergoing surgery for CRC displayed significantly reduced risk of infected wounds, and reduced risk of urinary, pulmonary and gastrointestinal complications. As this is the first published study to show this,it is unclear if diabetes in itself is a predictor of postoperative complication[20]. Other work has found diabetes in itself was not sufficient to predict complications in general surgery, but patients at risk of developing pneumonia, urinary tract and wound infections on average had a PO blood glucose level 10 mg/DL higher than patients without complications[21]. Indeed, recent data by Kiran reveal that perioperative hyperglycemia increases risk of postoperative wound infections, anastamotic leak and sepsis even for patients classed as non-diabetic[22]. In addition, hyperglycemia and obesity together are associated with increased detection of the cytokines IL-1β, IL-6 and TNF-α in serum[23-25]. Elevation of the latter two cytokines may be predictive for PO ileus[26]. It is likely that new evidence will emerge linking cytokine profiles to particular postoperative complications for LAC.

Conclusions

It is clear from literature that BMI is not sufficient to define the risk of postoperative morbidity in CRC patients undergoing LAC. Visceral adiposity measured by fat area or using PFSA as a surrogate marker seems to be a better predictive measure, and is likely to be more accurate in future due to advances in technology. However, more international studies with larger patient cohorts are needed to validate the predictive value of visceral adiposity. More work is also necessary to assess the value of blood glucose and cytokine signatures as independent biomarkers of surgical outcome. Perhaps combining these indicators of postoperative complication risk may help in identifying patients most at risk and take steps to minimize complications. Figure 1 below is a schema to stratify patients according the variables mentioned above. The risk scores can be used by surgical teams pre-operatively to decide whether patients with moderate, high or very high POC risk should be more closely monitored after surgery.

Figure 1: The postoperative complication risk score: BMI is assessed together with either WHR, VFA or PFSA. Next, serum levels of pro-inflammatory cytokines are assessed together with relative blood glucose levels (RBGL) and other co-morbidities. RBGL indicates mg/DL higher than population average. Scores for each parameter are in brackets. M= male F=female