A Systematic Review of the Risk of Non-alcoholic Fatty Liver Disease in Women With Polycystic Ovary Syndrome

Polycystic ovary syndrome (PCOS) is a complex hormonal disorder associated with complications throughout various body organs. Previous studies have shown evidence of liver disease in some women with PCOS. In this study, we attempted to explore the risk of non-alcoholic fatty liver disease (NAFLD) in PCOS women and the specific factors involved in its development. We searched PubMed, PubMed Central, Medline, and ScienceDirect for articles related to the topic, screened those articles according to our inclusion/exclusion criteria, and conducted a thorough quality check using various quality appraisal tools to select articles relevant to our research. The process was conducted according to Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) Checklist 2020. We selected 11 high-quality observational studies for our review. Studies from various countries were included, and all studies demonstrated an increased prevalence of NAFLD in PCOS patients compared to healthy controls. Although insulin resistance, obesity, and increased androgens contribute to the increase in the risk of NAFLD in these patients, hyperandrogenism was the most influential risk factor in four of these studies. Two studies explored the degree of NAFLD in these patients using transient elastography (TE). They concluded that PCOS was significantly associated with hepatic steatosis (HS) rather than hepatic fibrosis in most patients. PCOS patients have an increased risk of developing NAFLD, particularly HS, and hyperandrogenism seems to be the main determinant. Therefore, effort should be put into screening and monitoring these patients to manage the disease. TE may be a useful method for monitoring the natural history of NAFLD in these patients, which requires further exploration.


Introduction And Background
Polycystic ovary syndrome (PCOS) is an endocrine disorder that affects one in 10 women of reproductive age [1]. In these women, the ovaries produce excessive androgens leading to hormonal imbalance. This disorder is characterized by hyperandrogenism, polycystic ovaries, and anovulation. While the exact cause of PCOS is not known, experts have linked it to genetics, hyperandrogenism, insulin resistance (IR), and low-grade inflammation [2]. Its diagnosis requires a complete medical history, a physical exam, pelvic exam, pelvic ultrasound, and blood tests [1]. The treatment varies greatly but may include metformin, clomiphene citrate, letrazole, aromatase inhibitors, laparoscopic ovarian drilling, and gonadotropins [3]. Studies on the efficacy of such treatments are ongoing as there is no known cure for the disease. PCOS has been linked to many complications, including infertility, metabolic syndrome (MetS), cardiovascular disease, obstetric cancers, and psychological disorders [3]. In addition, many studies have tried to show a link between PCOS and nonalcoholic fatty liver disease (NAFLD), undoubtedly because the risk factors of NAFLD are also co-morbidities found in PCOS [4].
NAFLD is the fat accumulation in the liver independent of alcohol consumption greater than 5% to 10% of the liver's total weight [5]. Fat accumulation results in an inflammatory response that damages the parenchyma and may progress to non-alcoholic steatohepatitis (NASH), cirrhosis, and eventually liver failure or cancer [6]. NAFLD is called a silent disease because it is largely asymptomatic. Diagnosis of the condition involves a medical history, physical exam, imaging studies like ultrasound and CT scan, and lastly, liver biopsy [7]. Treatment is yet to be established, but it can be managed by changing to a healthier lifestyle.
It is well-known that people with obesity, MetS, and type 2 diabetes mellitus often develop NAFLD [7]. Thus, the early studies that established an association between PCOS and NAFLD attributed the association to MetS, which can be present in patients with PCOS and those with NAFLD. Studies regarding the pathogenesis of NAFLD in PCOS patients are inconsistent and inconclusive. Thus, an updated systematic

Search Outcomes
We identified a total of 983 articles from the selected databases. Eight hundred seventy-two of these articles were from ScienceDirect, of which 500 were removed after the following filters were applied: review articles, research articles, medicine, and dentistry. This left us with a total of 483 articles. Thirty-one duplicates were then removed. The screening began by selecting articles with titles that were relevant to the question. This resulted in the removal of 365 articles. Next, 36 articles were removed due to a lack of access to abstracts or abstracts that were unrelated to the research question. Nineteen of the remaining 51 articles were removed because we could not access the full-text articles. Finally, 32 articles were screened via the chosen inclusion and exclusion criteria. Twenty-two articles underwent quality appraisal, leaving only 11 to be evaluated. The PRISMA flowchart in Figure 1 demonstrates the article selection process.

Results of Quality Appraisal
The remaining articles were then critically assessed using quality appraisal tools to develop a final list of high-quality, relevant articles to be reviewed. Most of the studies included in this review were observational studies for which the Joanna Briggs Institute (JBI) critical appraisal tool was used. Assessment of multiple systematic reviews (AMSTAR) checklist was used for a few systematic reviews and meta-analyses. This systematic review contains all articles that satisfied at least 70% of the criteria in each critical appraisal tool. Tables 1-4 show the results of the quality appraisal.     11. In case of meta-analysis, did the authors use appropriate methods for the statistical combination of results?

Case-control studies
No metaanalysis Yes 12. In case of meta-analysis, did the authors assess the potential effect of RoB in individual studies on the results of the meta-analysis or other evidence synthesis?
No metaanalysis Yes 13. Did the authors account for RoB in individual studies when interpreting or discussing the results? No Yes 14. Did the review authors provide a reasonable explanation for and discussion of any heterogeneity observed in the results of the review? Yes Yes 15. If they performed quantitative synthesis, did the review authors carry out a satisfactory investigation of publication bias (small study bias) and discuss its likely impact on the results of the review?
No metaanalysis Yes 16. Did the authors report any sources of conflict of interest, including any funding they received for conducting the review? Yes No  Additionally, fibrosis-4 (FIB-4), aspartate aminotransferase (AST)-to-Platelet Ratio Index (APRI), BMI-Age-Alanine aminotransferase (ALT)-Triglycerides (BAAT), and BMI AST/ALT Ratio Diabetes (BARD) were used for the measurement of fibrosis [14]. This study concluded that PCOS was more likely associated with HS than hepatic fibrosis. On the other hand, two studies attempted to use transient elastography (TE) to assess NAFLD. Nevertheless, all studies demonstrate a significant risk of NAFLD associated with PCOS. Table 5 shows the types of studies included in this review, and Table 6 summarizes the results of each of the studies in further detail.

Discussion
PCOS is not simply a disease of the ovaries as some women present without polycystic ovaries on ultrasound as per the name. It is thus considered a multi-organ disease affecting organs like the pancreas, adrenal glands, heart, and liver [29]. In this study, we aimed to understand its consequences in the liver as several studies have suggested an association of PCOS with NAFLD.

PCOS: Independent Risk Factor?
The first case of NAFLD in PCOS patients was discovered in 2005 by Brown et al. in a 24-year-old obese woman with PCOS who underwent investigations for elevated liver enzymes [30]. It was thus speculated that NAFLD might occur in PCOS women since NAFLD and PCOS have common risk factors such as MetS. The research was then directed at confirming the correlation between the two and investigating the primary risk factors involved. Although several studies have suggested that PCOS patients have a higher risk for developing NAFLD, the debate regarding whether PCOS is an independent risk factor for the outcome is ongoing. Vassilatou et al. conducted a study in 2010 amongst 57 Caucasian PCOS patients and 60 Caucasian controls in which they reported that the odds of discovering NAFLD in PCOS patients were 3.55 times higher than in the controls. They also noticed that all subjects with MetS, including the controls, had NAFLD [10]. Therefore, this suggests that MetS may be the main link between the two rather than the diagnosis of PCOS alone.
Other factors like age, obesity, IR, WC, and androgen level also contribute to the development of PCOS. Similarly, Macut et al. reported that HOMA-IR and LAP were the most significant factors involved in the association of NAFLD with PCOS [15]. Based on these two studies, it can be concluded that although the risk for NAFLD is increased in PCOS patients, the reason is not solely the diagnosis of PCOS. Although Vassilatou's study only included 117 subjects and was published more than a decade ago, Macut's study was more recent and included more participants. Nonetheless, their conclusions imply that PCOS is not an independent risk factor for NAFLD.
On the other hand, Asfari et al. conducted a very large cross-sectional study including 77 115 PCOS patients identified from the National Inpatient Sample database from 2002 to 2014. They found that women with PCOS had a four times higher risk of having NAFLD and a lower incidence of hypertension, dyslipidemia, and diabetes mellitus [4]. Since these three conditions are all features of MetS, this implies that MetS may not have such a significant role in the prevalence of NAFLD in PCOS patients as once thought. Additionally, Salva-Pastor et al. came to a similar conclusion while studying Mexican PCOS patients. The study results demonstrated that despite having a lower prevalence of NAFLD in non-obese PCOS patients, PCOS patients still have a higher risk for NAFLD than other women overall [6]. This indicates that while we cannot dismiss the role of obesity in PCOS for developing NAFLD, there are still factors specific to PCOS that lead to NAFLD. Furthermore, they implied that different phenotypes of PCOS have different degrees of risk involved. Interestingly, this study used TE to diagnose NAFLD, whereas most used ultrasonography. Nevertheless, both studies present a strong argument that PCOS is, in fact, independently associated with the prevalence of NAFLD.

Role of Hyperandrogenism in Pathophysiology of NAFLD
Although some studies in the past have argued that PCOS is an independent risk factor for NAFLD, further studies were required to determine which feature of PCOS is the most significant in the progression of NAFLD in these women. In this systematic review, we came across several studies that determined hyperandrogenism in PCOS patients may be the most influential factor in the development of NAFLD. For example, a study conducted in 2012 by Jones et al. presented evidence that hyperandrogenic PCOS women had more liver fat content than PCOS women with normal androgens or women who served as controls [16]. This observation was held even after adjusting for BMI, HOMA-IR, and adipose tissue volume in the internal and visceral organs. This suggests that hyperandrogenic PCOS women can develop NAFLD despite having normal weight and insulin sensitivity. However, this study only included 51 participants, whereas Kim et al. demonstrated the same observation in 2017 with 1167 participants. Hyperandrogenism, signified by elevated free testosterone and free androgen index (FAI), was credited as a significant individual risk factor for NAFLD in the 275 non-obese PCOS patients studied compared to 892 non-obese healthy controls [9]. This conclusion seems plausible as some studies have suggested that the level of androgens and their receptors play an important role in lipid metabolism in the liver [9]. The results of this study were significant because previous studies on the subject had only discussed the prevalence of NAFLD in obese PCOS patients, therefore connecting the association to elements of MetS like obesity, dyslipidemia, and IR. However, it is important to note that this study used ultrasonography to assess NAFLD instead of liver biopsy (the gold standard for diagnosis), although they argued that it was cost-effective, sufficiently sensitive, and specific for a study of this scale.
In 2018, Kumarendran et al. decided to further explore the idea of hyperandrogenism as the driver behind NAFLD in PCOS patients. They conducted a wide-scale retrospective cohort study involving 1,84,184 participants selected from a general practice electronic database in the United Kingdom called The Health Improvement Network (THIN). The study found that among the 63,000 PCOS patients, there was a two-fold greater risk of NAFLD development than in the controls [22]. Additionally, they argued that although the reason for NAFLD development in this cohort may be a complex interplay between obesity, IR, and hyperandrogenism, hyperandrogenism played a more significant role. They attempted to solidify this argument by measuring NAFLD risk in two independent cohorts that did not have PCOS. They discovered that women with serum testosterone levels greater than three nmol/L and sex hormone binding globulin (SHBG) less than 30 nmol/L had an increased rate of NAFLD occurrence [22].
Nonetheless, one major shortcoming in this study was minimal documentation of criteria used to diagnose PCOS and NAFLD. This is concerning because it could mean that the two independent cohorts which demonstrated an increased risk of NAFLD in hyperandrogenic women could be women who had PCOS but were inadequately diagnosed. Additionally, a meta-analysis conducted by Wu et al. in 2018 suggested that hyperandrogenism was a major risk factor irrespective of BMI and geographic differences in the selected study population [28]. Their reasoning behind this argument was that hyperandrogenism could lead to NAFLD directly by promoting visceral fat deposition or indirectly by promoting IR and an inflammatory state [28]. Despite the few studies included in the meta-analysis, the idea that hyperandrogenism is the most significant factor involved in the early development of NAFLD persists.

Extent of NAFLD
Most of the studies included in this systematic review measured NAFLD using ultrasonography despite acknowledging that it is inferior to liver biopsy. Although understandably, performing a liver biopsy on each participant in a study is unfeasible, perhaps another method exists which will not only diagnose but also stage and monitor the extent of liver disease. For example, Polyzos et al., 2014, attempted to use noninvasive indices to measure steatosis and fibrosis, representing different stages in liver disease. They discovered that the indices used to measure HS were significantly higher in PCOS patients, especially those with MetS, whereas fibrosis indices correlated poorly with PCOS [13]. Therefore, this suggests that PCOS may only lead to a mild, reversible form of liver disease rather than a much more severe form like cirrhosis. It may even suggest that these patients' natural history of liver failure is prolonged. However, the extent of liver disease in PCOS patients and the frequency and speed of progression to cirrhosis are areas of research that need further exploration.
Another method used to diagnose NAFLD in some studies is TE. This non-invasive imaging technique measures liver fibrosis by determining liver stiffness measurement (LSM) and controlled attenuation parameter (CAP). Chakraborty et al. in 2020, for instance, set out to determine TE's effectiveness in assessing NAFLD in a cohort of Indian women with PCOS. They discovered that both LSM and CAP were higher in PCOS women, but only CAP significantly correlated with the other measures of liver fat content like liver fat score (LFS) and HIS [14]. Because LSM values indicate liver fibrosis and CAP is a measure of HS, this reinforces the idea that PCOS may only lead to a mild NALFD.
Similarly, Taranto et al. also used TE and other non-invasive indices to diagnose and stage NAFLD in a subset of Brazilian women with PCOS. Like, Chakraborty et al., they found strong evidence for the correlation but were unable to find a strong correlation between PCOS and liver fibrosis [12]. Thus, it can be argued that TE may be a more favorable alternative for NAFLD measurement than ultrasonography; however, further study is still warranted.

Strengths and limitations
Strengths of this systematic review include the utilization of studies that contain subjects from different countries like Greece, Brazil, India, South Korea, Mexico, and the United Kingdom. This allows the comparison of results among different populations and enhances the significance of the results. Additionally, various qualitative studies were included with varying sample sizes to analyze the significance of the data. On the other hand, limitations include a lack of access to a few relevant full-text papers despite an attempt to contact authors. Additionally, the limited number of databases searched further decreases the number of valuable articles that could have contributed to enhancing the systematic review.

Conclusions
In conclusion, our systematic review aimed to determine whether PCOS increases the risk of developing NAFLD. The results of the reviewed studies have demonstrated an increased risk of NAFLD in PCOS patients of reproductive age in various countries. Despite the significance of risk factors like obesity and IR in developing NAFLD, hyperandrogenism seems to be the most influential factor in PCOS patients. Therefore, early recognition is warranted to control and potentially reverse liver disease as most cases are limited to HS. The younger age of the patients may explain this. However, the natural history of NAFLD in PCOS patients requires further study. Furthermore, TE may be used as an alternative for ultrasonography to diagnose better and monitor this population's natural history of NAFLD.
no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.