Several strategies directed at the different steps of pathogenesis can be used for the treatment of NAFLD. Key drugs used are insulin sensitizers, hypolipidemics, antioxidants and cytoprotectants. Weight loss is the only measure that acts at the source of the pathogenesis and can offer complete cure versus drugs. This can be achieved by diet, exercise and bariatric surgery. NAFLD per se is not an indication for bariatric surgery. Bariatric surgery is the best alternative option for weight reduction if diet/exercise fails [20].

Numerous studies have reported significant histological improvement after roux-en Y gastric bypass (RYGB) [21–33]. Few studies have reported improvement after vertical banded gastroplasty (VBG) and adjustable gastric banding (AGB) [34–40]. Keshshian et al. and Kral et al. reported improvement after duodenal switch (DS) and bilio-pancreatic diversion (BPD) respectively [41, 42]. The authors have reported improvement after sleeve gastrectomy (SG) [12]. Though majority of studies report a consistent beneficial effect on liver histology, some studies have reported a few cases of worsening or new onset steatosis, inflammation or fibrosis. Silverman et al., Mattar et al., Mottin et al., Csendes et al. and Furuya et al. reported a few cases of worsening or new fibrosis after RYGB [21, 23, 25, 28, 29]. Luyckx et al. found significant regression of hepatic steatosis but an increase in the incidence of hepatocellular inflammation after gastroplasty [38]. Stratopoulos et al. found significant improvement in steatosis and steatohepatitis with an overall decrease in fibrosis but 11.7 % had increased fibrosis after gastroplasty [34]. AGB was one of three interventions used by Mathurin et al., the other two being RYGB and biliointestinal bypass [36]. Percentage of steatosis fell but inflammation remained unchanged and a significant increase in fibrosis was seen in 20 % of patients. Kral et al. after biliopancreatic diversion showed decreased steatosis but postoperative increase in fibrosis in 40 % [42]. Some showed decreasing mild inflammation, whereas some developed new onset mild inflammation. Keshishian et al. after duodenal switch (DS) operation showed improved steatosis and inflammation but liver function tests/inflammation had slightly worsened by the 6-month period, but then normalized by 12 months [41]. Fibrosis was not examined in this study. Most data on effects of bariatric surgery on NAFLD is from western literature. Two histological studies from Asia exist. In 2008 Huang et al. in 21 RYGB patients and in 2015 Praveenraj et al. in 20 SG and 10 RYGB patients demonstrated dramatic histological improvement in steatosis, steatohepatitis and fibrosis.

Histologic studies comparing the effects of the various bariatric interventions are lacking in literature. Caiazzo et al. recently reported superior NAFLD improvement after RYGB versus AGB. This favorable liver outcome occurred earlier and was more profound after RYGB which was suggested to be related to better weight loss outcomes after RYGB. Other potential influencing factors suggested were greater improvement in insulin sensitivity and other hormonal changes as AGB is an exclusively restrictive operation [43]. The authors reported better NAFLD improvement after sleeve gastrectomy versus a RYGB though comparison did not reveal statistically significant difference [11]. Further studies are needed to confirm superiority of the various bariatric procedures. While most studies demonstrated improvement by repeating liver biopsy at 1–2 years after surgery we demonstrated improvement as early as 6 months after surgery [11].

The mechanism of how bariatric surgery plays a role as potential treatment of NAFLD is complex and not fully understood. Bariatric surgery is associated with sustained and significant weight loss. It leads to improvement of insulin sensitivity. It leads to improvement/resolution of hyperglycemia, dyslipidemia and hypertension. It reduces obesity-related low-grade chronic inflammation resulting from excess production tumor necrosis factor-α, interleukin-1, interleukin-8, interleukin-18, monocyte chemoattractant protein-1, C-reactive protein etc. Last, it acts by producing complex hormonal changes in ghrelin, glucagon-likepeptide-1, peptide YY, oxyntomodulin, adiponectin, etc. [44].

Though a majority of studies report a consistently beneficial effect on liver histologic examination, the exact mechanism of worsening reported in some series is unclear. This could probably be related to the rapid weight loss resulting in increased free fatty acid levels derived from extensive fat mobilization causing liver injury. Exposure to toxins from bacterial overgrowth from intestinal diversion, nutritional deficiencies, and protein malnutrition from malabsorption probably could act as a “second hit”. Another possibility that has been suggested is a ‘misdiagnosis’ of worsening in many reported cases that may occur as steatosis is a prominent feature in specimens obtained before weight loss and careful assessment and reporting of necro-inflammatory and portal tract changes may often not be done by the pathologist if the latter is minimal [35]. A dramatic reduction in steatosis post-surgery may enhance or unmask the view of these inflammatory cells which would then be reported [35].

There is exists few series in literature that suggests that bariatric surgery can be performed safely in selected patients with Childs A cirrhosis. They can be safely subjected to a sleeve gastrectomy or a RYGB. These patients may have a greater incidence of transient renal dysfunction (acute tubular necrosis) and an increased potential for blood loss. Patients achieved excellent weight loss and improvement in obesity-related co-morbidities [45–47]. Several issues however need to be taken into consideration in day to day practice. RYGB is the most widely accepted bariatric procedure with good effect on metabolic syndrome and minimal degree of malabsorption. Although we have ourselves reported that it can be safely done even in cirrhotics, one of the main concerns of its use in patients with cirrhosis is that the bypassed stomach will be inaccessible should variceal bleeding develop [12]. Biliopancreatic diversion/duodenal switch (BPD/DS) is a less commonly used bariatric procedure which has a high risk of complications by induced malabsorption and there are a few reports about hepatic dysfunction. Restrictive operations – LAGB and SG are quick procedures and less invasive than a RYGB or BPD/DS which makes these more practical options.

There may be two scenarios in which such a choice needs to be made –unexpected cirrhosis at bariatric surgery and patients with known cirrhosis and medically complicated obesity. The intraoperative decision making for unexpected cirrhosis (found in 1–2 % cases) may involve changing the planned procedure to a different one e.g. a RYGB or BPD/DS to a SG if cirrhosis was suspected preoperatively and this was discussed with the patient. If extent of cirrhosis or portal hypertension is not known or if the patient had not consented to have an alternative procedure, it would be right to perform only a liver biopsy and defer the operation until full workup.

In referred patients with known cirrhosis and medically complicated obesity the cause of cirrhosis should be elucidated (i.e. NASH, hepatitis, alpha 1-antitrypsin deficiency). Childs A cirrhosis with normal synthetic function can be safely subjected to a sleeve gastrectomy or a RYGB. The recommendation would be preferably a restrictive procedure, such as SG. However if liver disease is secondary to steatohepatitis, and there is no portal hypertension, a RYGB should be considered because of the added benefits to the metabolic syndrome. Patients who have decompensated liver function or severe portal HT should be managed in partnership with the liver transplant service. In severe portal hypertension. with an otherwise preserved liver function, some consideration can be given to placement of transjugular intrahepatic posrtosystemic shunts (TIPSS) to decrease the portal pressure and then proceed to a safer, technically possible surgical procedure possibly as a bridge to liver transplantation. Most surgeons would not consider a RYGB in these circumstances because of the possibility of variceal bleeding.