NAFLD is prevalent in adolescent females with PCOS. A retrospective chart review by Barfield et al[28] found a 15.4% prevalence rate of elevated aminotransferases within an adolescent PCOS population. Michaliszyn et al[29] demonstrated that in obese adolescent girls with PCOS, liver fat is associated with increasing age, even in the narrow adolescent age range, increasing abdominal adiposity, worsening insulin sensitivity, and dyslipoproteinemia; age and total testosterone made the greatest contribution to liver fat in this study.

The data are less clear on whether or not NAFLD is also more prevalent in lean PCOS patients. Importantly, PCOS has been noted to affect 28% of obese women, but only 5% of lean women[20,30]. Of the previous studies discussed thus far, Gambarin-Gelwan et al[24] was the only one to include both lean and obese patients. They found a prevalence of NAFLD in 39% of lean PCOS women to suggest that PCOS may confer a risk for NAFLD that is independent of obesity. However, these findings have not been confirmed in other studies including a small study of 17 young women with PCOS as well as an additional study that included a subgroup of 34 lean PCOS women compared to lean controls[31,32].

Insulin resistance is detected in up to 80% of cases of NAFLD and there is a near universal association between NAFLD and IR irrespective of obesity[33,38]. While the exact pathogenesis of NAFLD is still unclear, evidence supports IR as a primary factor in its development[39]. The “two-hit theory” that was initially suggested about 15 years ago suggested that IR is the first “hit” to cause hepatic steatosis, when insulin resistant visceral adipocytes release free fatty acids that flow to the liver and accumulate; cytokine stresses and apoptosis are the second “hit”, which mediate the progression to NASH[7,40,41].

More recently, two additional concepts, “multiple-hit” and “distinct hit” theories, are being considered to explain the pathogenesis of NAFLD/NASH, and both still include IR as an integral component[41,42]. The “multiple-hit” theory conceptualizes NAFLD/NASH pathogenesis in the same way as the “two-hit” theory, where NASH is generally a condition preceded histologically by simple steatosis and pathophysiologically by IR and its associated metabolic disturbances[41]. However, fatty infiltration of the liver then leads to a series of parallel multiple hits, such as cytokines, adipokines, and oxidative stresses, which mediate the progression to NASH and fibrosis. The “distinct hit” hypothesis was generated by evidence that NASH and pure fatty liver could arise as two independent conditions, since inflammation occasionally precedes steatosis and patients with NASH may present with very little steatosis[41,42]. Therefore, according to this model, distinct pathways are activated (potentially by IR) which lead to either simple steatosis or NASH, rather than an accumulation of multiple parallel hepatotoxic injuries[41].

Similarly, IR is thought to play a central pathophysiological role in PCOS that is not explained alone by its frequent association with obesity[43]. The landmark 1989 study by Dunaif et al[44] first showed that IR is present in both obese and non-obese women with PCOS and this was confirmed in later studies[45,46]. There are several proposed reasons why PCOS women may have a decreased sensitivity to insulin, which include both insulin receptor defects and insulin signaling defects at the level of glucose transport into skeletal muscle[47].

A systematic review determined that IR is present in 50%-80% of women with both PCOS and NAFLD[47] and multiple studies have shown that PCOS women with hepatic steatosis have elevated levels of IR compared to PCOS women without steatosis[24,25,48-50]. Those with steatosis showed higher BMI, waist-hip ratio, fasting insulin, and homeostasis model assessment-estimated insulin resistance (HOMA-IR) scores than PCOS women without steatosis[25]. Further, Targher et al[51] found that insulin sensitivity by euglycemic hyperinsulinemic clamp measures markedly decreased in PCOS women with abnormal ALT levels (defined as ALT over 19 U/L) whereas it was similar between PCOS women with normal ALT levels and matched healthy controls.

Insulin resistance in the ovaries may be central to the pathogenesis of PCOS[52]. The ovaries are abundant with insulin receptors and the dysregulation of insulin signaling may augment the production of androgens in theca cells, the primary source of excessive androgen biosynthesis in women with PCOS[47]. IR leads to compensatory hyperinsulinemia, which stimulates theca cells in LH-sensitized ovaries to secrete testosterone and androstenedione[53]. Interestingly, if a patient with PCOS and hyperandrogenism has chemical or surgical ablation of the ovaries, the hyperandrogenism is reversed but elevated insulin levels persist[22,54-56].

This combination of hyperandrogenism and IR may contribute to the pathogenesis of NAFLD in PCOS, but there is still speculation as to their independent effects and relationship to each other. A case report of a woman with hyperthecosis and biopsy-proven NAFLD documented the reversal of abnormal liver tests and hirsutism after bilateral oophorectomy[57]. Because an oophorectomy should not improve IR, this report suggests an independent contribution of hyperandrogenism to fatty liver in these women[22,57].

One theory for how hyperandrogenism puts PCOS women at risk for NAFLD regards its effects on the low-density lipoprotein receptor (LDLR), which is down-regulated 0.51 fold in PCOS women[33]. Androgens may suppress LDLR gene transcription to prolong the half-life of very low-density lipoprotein (VLDL) and LDL and thus induce lipid accumulation in the liver. The LDLR gene is a subject of estrogen control that is maintained through the estrogen-responsive region adjacent to the sterol response element within the LDLR promoter; androgen receptor agonists have been shown to attenuate the estrogen-induced up-regulation of LDLR in hepatocytes[33,58]. It is possible that hyperandrogenemia may have a direct suppressive influence on LDLR levels in both adipocytes and the liver, thus making PCOS women more prone to liver disease[33].

Observational findings from multiple studies suggest that to better understand and control NAFLD in women with PCOS, not only metabolic variables, but also hyperandrogenism should be taken into consideration. For instance, hirsutism may be a risk factor for elevated ALT[22]. In addition, hyperandrogenemia has been associated with elevation of ALT levels in PCOS women independent of obesity[49,58]. Further, Jones et al[45] showed that after statistical adjustment for BMI, HOMA-IR, and visceral adipose tissue, 19 hyperandrogenic PCOS women had a significantly higher quantity of liver fat, as measured by proton magnetic resonance spectroscopy (H-MRS) compared with 10 non-hyperandrogenic PCOS or 22 controls (3.7% and 2.1% respectively; P < 0.05).

Further evidence that androgens may exert harmful effects to the liver was noted by Chen et al[59]: (1) men have been shown to have higher serum levels of ALT and a higher risk of NAFLD than women, after adjusting for obesity[60-64]; (2) long-term administration of androgens for female-to-male transsexuals has been noted to be associated with borderline elevated ALT levels[65]; (3) androgens and androgen receptors are known to be involved in the carcinogenesis of hepatocellular carcinoma (HCC) both in men and women[66,67]; and (4) androgen ablation therapy is one of the potential therapeutic treatment options for HCC[68,69].

Obstructive sleep apnea (OSA), while common in women with PCOS, may contribute to NAFLD development and accelerate its progression to NASH[70-74]. A recent meta-analysis found that NAFLD is 2.6 times more frequent in OSA patients and these patients have a 2.6-fold higher risk of progressing to liver fibrosis[74]. The presence of OSA has been shown to increase the chance of NAFLD by 7.6 times in women with PCOS[70]. The mechanism through which this occurs is likely related to IR. Chronic intermittent hypoxia increases production of inflammatory cytokines, increases oxidative stress, and decreases adiponectin, all of which contribute to IR and to fat accumulation in the liver. This process may be androgen-driven, since generally men are at a higher risk for OSA[75]. Therefore, the high free testosterone levels in women with PCOS may be one of the predisposing factors leading to OSA in this group[70].

There is evidence for a pro-apoptotic environment contributing to the progression from NAFLD to NASH[7]. Hepatic apoptosis triggers regenerative mechanisms to replace dead hepatocytes; however, aberrant responses may occur in some individuals, resulting in the activation of hepatic stellate cells to myofibroblasts and the hepatic recruitment of pro-inflammatory, pro-fibrogenic immune cells. Induction of apoptosis occurs through the activation of effector caspases that cleave a host of intracellular substrates including cytokeratin 18 (CK18), a member of the intermediate filament family of cytoskeletal proteins[76,77]. It has been shown that the plasma-borne caspase-generated CK18 fragment levels correlate with the magnitude of hepatocyte apoptosis and independently predict NASH in multivariate analysis[78].

Two studies examined CK18 fragment levels (measured as “M30”) as both a marker of apoptotic cell death and a qualifier for NAFLD/NASH in women with PCOS[33,76]. M30 levels were significantly elevated in PCOS patients compared to non-PCOS age-matched controls after correction for BMI[76]. Tan et al[76] found that 27.6% of PCOS patients compared to only 1.4% of the non-PCOS controls showed M30 levels considered high enough to qualify for NASH though no confirmatory biopsies were done. A separate study of biopsy-proven NAFLD patients undergoing bariatric surgery found higher M30 levels in the PCOS-NAFLD group compared to the non-PCOS-NAFLD group[33]. These data by Baranova et al[33] are interesting, because patients with histologically confirmed NASH were excluded, thus indicating that an increase in apoptosis is an early feature of NAFLD observed in PCOS subjects.

Some have hypothesized that the pro-apoptotic environment is exacerbated by androgens[33]. Androgens are known as pro-apoptotic agents and have been shown to act upon many types of peripheral cells, including hepatocytes[33,79]. The hyperandrogenism of PCOS may therefore contribute to an increased risk of NAFLD progression through its effects on apoptosis.

Visceral fat accumulation and adipose tissue dysfunction have been implicated in the pathogenesis of NAFLD in PCOS women. Central obesity has been shown to be associated with both hepatic steatosis and hepatic IR in women with PCOS[80,81]. Women with PCOS have higher global adiposity and increased amounts of visceral adipose tissue, especially in the intraperitoneal and mesenteric depots[80,82]. Furthermore, on multivariate logistic regression, fasting insulin and mesenteric fat thickness were identified as independent predictors of fatty liver among subjects with PCOS[80].

Villa et al[83] suggests that the distribution of fat to abdominal/visceral sites is not the only explanation for the metabolic abnormalities in women with PCOS. These women also exhibit clear abnormalities in their adipose tissue, compared to similarly obese women without PCOS. PCOS women have an adipocyte diameter that is generally increased by 25% relative to that in comparably obese control women who do not have PCOS[83,84]. Their obesity is therefore characterized by an increase in fat cell size, termed “hypertrophic obesity”, rather than an increase in number, or “hyperplastic obesity.” Hypertrophic obesity is associated with IR, because the hypertrophic adipocyte cells themselves are insulin resistant[83,85]. Androgens again may play an extrinsic role, since they have been shown to induce IR in subcutaneous adipose cells in vitro[83,86].

A number of adipokines have been implicated in the pathogenesis of PCOS: leptin, adiponectin, vaspin, visfatin, chimerin, and acute-phase amyloid A[87]. Adiponectin may have anti-inflammatory and insulin-sensitizing effects. Secreted by adipocytes, it is the only adipokine downregulated by obesity[47]. PCOS women have been shown to have lower levels of adiponectin when compared to weight-matched controls and low levels of adiponectin were correlated with IR[88]. In cases where lean women with PCOS maintain normal sensitivity to insulin, adiponectin biosynthesis deficiency or chronic low level inflammation is often present[47]. However, Baranova et al[47] found no difference in levels of adiponectin between PCOS and non-PCOS obese women undergoing bariatric surgery with biopsy-proven NAFLD.

A number of inflammatory and macrophage-derived factors have also been studied in the pathogenesis of PCOS: resistin, tumor necrosis factor-alpha (TNFα), interleukins, and C-reactive protein (CRP)[87]. TNFα impairs insulin action and individuals with IR show higher serum levels of TNFα[7]. Its expression is up regulated in both PCOS and NASH, independent of obesity[87]. The effects of TNFα may actually lie in part with its biological relationship to adiponectin. Higher TNFα may activate downstream kinases that induce cytokine production, while attenuating the expression and activity of adiponectin[7].

One study looked at the role of estrogens in NAFLD and PCOS[89]. Gutierrez-Grobe et al[89] found that women without NAFLD had significantly higher levels of serum estradiol (100 +/- 95.4) compared to women with NAFLD (55.5 +/- 66.6). The estradiol level specifically in PCOS women with NAFLD was 64.9 compared to 101.36 in PCOS women without NAFLD. The prevalence of NAFLD in premenopausal, postmenopausal, and PCOS women was 32.2%, 57.9%, and 62% respectively. These results suggest that NAFLD is more prevalent in postmenopausal and PCOS women than premenopausal women and estrogens may have a protective effect. Furthermore, it has been shown that NAFLD severity worsens after menopause and this is believed to be due to the loss of estrogens[90,91].

One study suggested a potential role for bisphenol A (BPA)[92]. BPA at very low doses alters several metabolic functions, including oxidative stress, a key component of inflammatory reactions. Studies have implicated BPA in the global rise of obesity, type 2 diabetes, cardiovascular disease, and most recently, PCOS[93-95]. Tarantino et al[92] showed that in premenopausal women with PCOS, there was an association of serum BPA levels with hepatic steatosis and markers of low-grade inflammation, particularly spleen size. Increased spleen volume has been proposed as an index of chronic inflammation associated with fatty infiltration of the liver[92,96]. It is possible that BPA acts as a pro-inflammatory primer, via macrophage activation and pro-inflammatory cytokine hypersecretion; within this context, the spleen enlargement could represent a marker of this inflammatory process, and the inflammation may contribute to NASH[92]. Further studies are needed to investigate this hypothesis.

A number of genes have been implicated in both PCOS and NAFLD and some of them are linked to the pathogenetic mechanisms already discussed above. Some examples of these genes are the following: LDLR, genes involved in steroidogenesis, 5-alpha reductase, sex hormone binding globulin, fat mass gene, inflammatory mediators (such as hsCRP, which is increased by interleukin 6 produced by macrophages and adipocytes), and adiponectin[47].

Early recognition of NAFLD in PCOS patients is warranted, but as discussed above, routine screening is not currently recommended and our screening tests are limited. Therefore, screening in this population of women should potentially target those more likely to develop NASH, such as those with metabolic syndrome[5,87]. The presence of metabolic syndrome is a strong predictor for the presence of steatohepatitis in patients with NAFLD, so guidelines by AASLD recommend using the presence of metabolic syndrome to target patients for liver biopsy[5].

Gangale et al[97] studied a group of 140 hyperinsulinemic overweight women with PCOS and showed that all cases of the metabolic syndrome in their cohort were detected in women with NAFLD. Furthermore, NAFLD affected up to nearly half of their overweight hyperinsulinemic patients, so they suggested NAFLD screening in all PCOS women who are overweight and hyperinsulinemic. On the other hand, Ciotta et al[102] found a NAFLD incidence rate of 40% in lean PCOS women and therefore felt that it would be prudent to screen all PCOS women, regardless of BMI. Some suggest screening PCOS women with both liver biochemistries and ultrasound, while others feel that an ultrasound is warranted regardless of the serological evaluation of the liver[32,97,102,103]. Although the Endocrine Society clinical practice guidelines[104] for PCOS recommend against routine screening for NAFLD, they suggest obtaining serum markers of liver dysfunction in women with metabolic risk factors and/or IR. If serum markers are elevated, ultrasound and liver biopsy can then be considered.

The fatty liver index (FLI) might serve as a simple and accurate tool to help physicians select subjects for liver ultrasonography and intensified lifestyle counseling, because it has been shown to correlate well with hepatic steatosis on ultrasound[105,106]. The FLI is an algorithm based on BMI, waist circumference, triglycerides, and GGT. PCOS women had an odds ratio of 2.52 for elevated FLI compared to non-PCOS BMI matched controls. The FLI may therefore identify PCOS patients at high risk for hepatic disturbances, though future confirmatory studies are needed.

Finally, a hepatology referral should be considered if there is concern that the patient has steatohepatitis, her ALT is double the upper limit of the reference range, or there is evidence of cirrhosis[37]. On the contrary, an endocrinology referral should be considered and the existence of PCOS should be investigated in a young woman with NAFLD[26].

Weight loss is a key treatment for both PCOS and NAFLD. A systematic review of 23 studies supports the link between weight reduction and/or increased physical activity and reduced liver fat and improved glucose control/insulin sensitivity in patients with NAFLD[109]. We reported a young woman with PCOS and NASH who showed histopathologic improvement on liver biopsy after lifestyle modification[21]. However, no large studies to date have examined the effects of lifestyle interventions on NAFLD specifically in women with PCOS. A Cochrane review found that lifestyle therapy gave modest benefits in women with PCOS compared with minimal treatment for a variety of other surrogate outcomes, including lower total testosterone, improved hirsutism, weight, lower waist circumference, and fasting serum insulin; however, markers of liver disease were not reviewed[110]. No specific diet has been shown to improve PCOS and generally the recommendation from experts has been for a hypo-caloric diet with a 500 kcal/d deficit and a composition likely to increase adherence[52,111]. Similarly, no specific diet has been proven superior for treatment of NAFLD; in fact, studies published to date do not allow a clear differentiation of the effects of diet composition or of diet relative to physical activity[109].

A systematic review and meta-analysis[112], which examined 78 randomized trials for the treatment of NAFLD found that although weight loss of greater than or equal to 7% improved histological disease activity, this was only achieved in less than 50% of patients. Hurdles to implementation of lifestyle changes specifically in the PCOS population may include the following: (1) high dropout rates; (2) difficulty in morbidly obese women to adapt to significant lifestyle changes; and (3) the overall lack of clinical infrastructure to implement lifestyle therapy in many settings throughout the United States and the world[52,113,114].

Metformin is a widely utilized diabetes drug that works by suppressing hepatic gluconeogenesis and improving insulin-sensitizing ability[52]. Metformin is generally not recommended as a specific treatment for liver disease in adults with NASH, because studies have shown mixed results[5]. In women with PCOS, metformin has been shown to improve markers related to liver disease and the metabolic syndrome[97]. In this study by Gangale et al[97], long-term treatment with metformin in 70 overweight women with PCOS and NAFLD demonstrated a significant reduction in AST levels and AST/ALT ratio at 6 mo despite no significant change in BMI; however, no change in hepatic steatosis was detected by ultrasound at 12 mo. Another study showed significantly reduced levels of ALT and GGT in 66 obese women with PCOS after 8 mo of treatment with metformin[115]. Data are limited on metformin treatment in lean PCOS patients with NAFLD. Gastrointestinal discomfort is a common side effect, but metformin otherwise has a low risk of severe adverse effects and has been studied fairly extensively in women of reproductive age[52,116].

Thiazolidinediones (TZDs) are known to enhance insulin sensitivity by acting on peroxisome proliferator-activated receptor gamma and increasing circulating adiponectin[117]. They also reduce fat accumulation in the liver through stimulation of free fatty acid storage in adipose tissue. TZDs have shown promising effects on treating NAFLD/NASH in the general population. A multicenter study of troglitazone for PCOS provided important proof of concept for the use of TZDs in the treatment of these women, due to the resultant improvements in ovulatory function, hirsutism, hyperandrogenism, and insulin resistance[52,117-119]. However, this class of medication has not been specifically studied with regards to its effects on NAFLD in PCOS.

There are a number of limitations in using TZDs in patients with NAFLD and PCOS, particularly the associated weight gain[52,107,108]. Patients are likely to relapse after the drug’s discontinuation, so it would have to be taken indefinitely[108]; however, the significant weight gain detracts from its long-term usefulness. Additionally, the safety concerns regarding this class of medication have restricted their availability or removed them from the market: toxicity from troglitazone, increased cardiovascular events with rosiglitazone, and bladder cancer with pioglitazone. The risk-benefit ratio is inadequate for women with PCOS when taking all these factors into account on top of concerns for teratogenicity[52].

Liraglutide is the first glucagon-like peptide-1 analogue shown to improve liver fibrosis markers in obese women with PCOS and NAFLD[120]. This medication may be an attractive option for obese women with PCOS, because it causes glucose dependent insulin secretion, promotes weight loss, and may subsequently improve IR. In a case control study of young obese PCOS women with age and weight matched controls, 6 mo of liraglutide treatment resulted in a mild reduction in weight and significant improvement of IR, inflammation, and liver fibrosis markers. However, liraglutide is not FDA approved for NAFLD treatment and further studies are needed to confirm these findings before use of this medication is recommended in this setting.

Antioxidant supplements may potentially protect cellular structures against damage from oxygen-free radicals and reactive products of lipid peroxidation[121]. Vitamin E is thought to be the “last antioxidant defense” in lipid metabolism, thus justifying its use in preventing the progression of NAFLD to NASH. A 2007 Cochrane review[121] found insufficient data to either support or refute the use of antioxidant supplements for treatment of NAFLD and data from subsequent clinical trials have shown mixed results[107,122]. However, no studies have examined the effects of Vitamin E on NAFLD in women with PCOS.

Omega-3 fatty acids may treat NAFLD through a number of mechanisms. In addition to their anti-inflammatory and antioxidant properties, they inhibit triglyceride synthesis, alter hepatic genes involved in lipid and glucose metabolism, decrease postprandial lipemia, and increase lipoprotein lipase activity[123]. It is therefore not surprising that omega-3 fatty acid supplementation has been shown to significantly reduce liver fat, triglycerides, fasting insulin, and HOMA-IR scores in PCOS women with hepatic steatosis. A recent systematic review and meta-analysis further supports the role of omea-3 fatty acids in reducing liver fat, but suggests that the optimal dose is not currently known[124]. Omega-3 fatty acids are an attractive option for women with PCOS and NAFLD, but they are not recommended as first line treatment for NAFLD unless the patient has concurrent hypertriglyceridemia[5].

Other pharmacotherapies include ursodeoxycholic acid, probiotics therapies[125], pentoxifylline[126], traditional chinese herbal medicine[127], and most recently the farnesoid X receptor ligand Obeticholic Acid[128]. None of these therapies have been studied specifically in women with PCOS.

Controversy surrounds the use of statins in both PCOS and NAFLD. Although some open-label trials have suggested improvement in liver enzymes with statins, there is no convincing histological data supporting their use for treating NAFLD[5]. The AASLD guidelines[5] state that there is a lack of evidence to show that patients with NAFLD and NASH are at an increased risk for serious drug-induced liver injury from statins and suggest that they can be used to treat dyslipidemia in patients with NAFLD and NASH. Statins may have intrinsic antioxidant properties that justify treatment for PCOS and many of these women have dyslipidemia that will benefit from treatment with statins. However, these benefits must be weighed with the possibility that statin therapy will worsen insulin sensitivity in this group of women who are at a higher risk of developing type 2 diabetes[129]. The teratogenicity of statins must also be kept in mind when treating women of reproductive age.

Finally, bariatric surgery may improve steatosis, steatohepatitis, and fibrosis; some cases of NAFLD/NASH may even resolve[108,130]. The AASLD guidelines suggest that it is premature to consider this surgery as an established option to specifically treat NASH, but it is important to note that foregut bariatric surgery is not contraindicated in otherwise eligible obese individuals with NAFLD or NASH (but without established cirrhosis)[5]. A study of 389 PCOS women who underwent Roux-en-Y gastric bypass surgery resulted in a significant decrease in liver aminotransferases for Hispanic women only[131]. It is possible that cardio-metabolic disease risk improvements vary by ethnicity. Thus, obesity may impact liver function changes more in Hispanic vs non-Hispanic women with PCOS.