Approximately 6% to 33% of the US population is estimated to have nonalcoholic fatty liver disease (NAFLD), the hepatic manifestation of metabolic syndrome. A subset of these individuals have a progressive form of NAFLD called nonalcoholic steatohepatitis (NASH), which can lead to advanced fibrosis and cirrhosis in approximately 10% to 20%.
The long-term risk associated with NAFLD is variable. Most patients with non-NASH NAFLD have significant metabolic disease (diabetes, dyslipidemia, and cardiovascular disease) but are not likely to develop advanced fibrosis of the liver. Patients with NASH, however, may have complications of liver disease in addition to cardiovascular disease. Current therapies for NAFLD include weight loss and overall cardiovascular risk modification (including treatment of metabolic syndrome). Because of the weight loss and improved glycemic control achieved with bariatric operations in appropriate medically complicated obese individuals, the utility of malabsorptive and/or restrictive bariatric procedures for NASH and non-NASH NAFLD is of current interest. The techniques used now are based largely on the preferences of patients and surgeons; all can be performed laparoscopically. Adjustable gastric banding is one of the most common bariatric procedures: an inflatable silicone band surrounds the upper stomach in order to create a small (30-cm3) gastric pouch, and it is accessed via a subcutaneous port later to constrict the stoma (Fig. 1). Currently, the gold-standard technique is the Roux-en-Y gastric bypass, in which a proximal gastric pouch is separated from the remainder of the stomach; this creates a proximal bypass (gastrojejunostomy) when it is anastomosed to the alimentary limb (jejunum; Fig. 2). A third technique, sleeve gastrectomy, separates the left lateral portion of the antrum, body, and fundus of the stomach from the medial aspect through the removal of the larger portion, with a narrow stomach remaining in the abdomen (Fig. 3). Other techniques exist, but they are beyond the scope of this review. In general, patients with a body mass index ≥ 40 kg/m2 are eligible for bariatric surgery; those with any features of metabolic syndrome may be eligible with a body mass index ≥ 35 kg/m2.
Several studies have attempted to address the potential risks and benefits of bariatric procedures with respect to hepatic histology. The contributions of bariatric surgery to fibrosis regression and progression and lipid accumulation have been studied in both retrospective and prospective studies, but no randomized controlled trials comparing the histological outcomes of bariatric patients with and without NAFLD have been conducted. In most of these studies, the follow-up interval has been variable and dependent on other indications for biopsy (subsequent operative interventions). Furthermore, the histological outcomes have been correlated not with the hard outcome of mortality (all-cause, cardiovascular, or liver-related) but with an accepted method of following disease improvement, including a validated scoring system (the NAFLD activity score). The question of foregut bariatric surgery as a primary treatment for NASH is far more controversial.
An exception to the relative paucity of large prospective studies is the well-designed study by Mathurin et al., who obtained clinical and metabolic data (including liver histology data) from 381 obese subjects before and 1 and 5 years after bariatric surgery. The types of operations were restrictive-malabsorptive (combination) and restrictive alone (gastric banding, bilio-intestinal bypass, and Roux-en-Y gastric bypass in 56%, 23%, and 21%, respectively). A significant improvement was seen in steatosis and ballooning during follow-up at 1 and 5 years. The baseline histology of 99 subjects revealed probable or definite NASH; the NAFLD activity score improved in these patients along with steatosis, ballooning, and the resolution of (probable or definite) NASH at both 1 and 5 years, with the majority of the improvement occurring within 1 year. This study also revealed a statistically significant increase in the mean fibrosis score at 5 years, but only 0.5% of the patients had advanced fibrosis (F3), and 96% had scores ≤ F1. No patients enrolled in this study had advanced fibrosis (F3 or F4) at the baseline. NASH patients demonstrated no worsening of fibrosis scores.
Data from meta-analyses and systematic reviews support regression in steatosis and inflammation, but they also raise questions about the risk of worsening fibrosis. In 2008, Mummadi et al. published a meta-analysis based on observational studies and found improvements in steatosis, steatohepatitis, and fibrosis, but the analysis was limited by study heterogeneity. Combining the most robust data for clinical recommendations, Chavez-Tapia et al. undertook a Cochrane systematic review in 2010. Twenty-one prospective or retrospective cohort studies were included (with various bariatric procedures): improvements in steatosis were seen in 18 studies, and 11 studies revealed evidence of improvements in inflammation, but 4 studies described some deterioration in the degree of fibrosis. No unbiased conclusions regarding the benefits and harm of bariatric surgery in individuals with NAFLD could be made because of the lack of randomized or quasi-randomized data.
Additionally, data are limited for outcomes of bariatric procedures in patients with cirrhosis. Recently, Mosko and Nguyen published perioperative mortality data for patients without cirrhosis and patients with compensated or decompensated liver disease undergoing bariatric surgery. Patients with decompensated disease, as expected, had a much higher perioperative mortality rate (16.3%) than patients with compensated cirrhosis (0.9%) and patients without cirrhosis (0.3%). The diagnosis of cirrhosis was established by diagnosis codes and not by histology. Similarly, evidence for portal hypertension was established by diagnosis codes but not by portal pressure measurements or a specific evaluation for the presence of portal hypertension. In a subset of 30 patients with cirrhosis (with the diagnosis made by an intraoperative inspection of the liver) among 2119 patients undergoing bariatric surgery, Dallal et al. showed no differences in perioperative mortality but found a higher incidence of transient renal dysfunction, longer operative times, and an increased potential for blood loss. In addition to those intraoperative complications, postoperative complications, including subacute liver failure, have been described. According to available data, hepatic decompensation is associated with malabsorptive procedures (Roux-en-Y gastric bypass and biliopancreatic diversion), which may be mechanistically related to bacterial overgrowth or protein malnutrition (and in proportion to the length of the roux limb). Restrictive operations (gastric banding and sleeve gastrectomy) may be preferable because of their nonmalabsorptive approach, but their safety in patients with cirrhosis or portal hypertension is not well studied.
On the basis of the available data regarding the safety of patients with NAFLD who have indications other than liver disease for weight-loss surgery, bariatric surgery is not contraindicated. However, the risks associated with bariatric surgery are not insignificant: death, reintervention, venous thromboembolism, and the failure to be discharged from the hospital. There are currently insufficient data regarding the safety and efficacy of bariatric procedures to recommend weight-loss surgery for patients with established cirrhosis who are otherwise eligible; however, the presence of NAFLD/NASH is not a contraindication in this population. The neurohormonal effects of bariatric surgery on intrahepatic lipid may prove beneficial in patients with NASH who undergo bariatric surgery; at this time, however, without randomized trials, it is premature to recommend foregut bariatric procedures for treating NASH alone.