The Evidence Surrounding Non-Alcoholic Fatty Liver Disease in Individuals with Cancer: A Systematic Literature Review

Emerging evidence indicates an association between non-alcoholic fatty liver disease (NAFLD), cancer development and mortality. Cancer treatment-induced metabolic and hepatic dysfunction may be associated with increased rates of NAFLD. The review aims to investigate current evidence surrounding NAFLD in adults (≥18 years) with cancer including prevalence, effect of cancer treatments, metabolic co-morbidities, and mortality. Embase, Scopus, PubMed, and CINAHL were searched from inception to December 2021 including randomized controlled trials and observational studies. Twenty-three articles were included, comprising 142,218 participants. The overall risk of bias for observational studies was determined as low for 10 studies and neutral for 12 studies, and the RCT was determined as some concerns. The prevalence of NAFLD, based on imaging or histology, in adults with cancer ranged from 0.5 to 81.3%, with higher prevalence in breast, colorectal and gynecological cancers. Higher rates of NAFLD were also seen in patients who (i) underwent treatments—including chemotherapy and hormone therapy and/or who (ii) had higher BMI or other metabolic co-morbidities. NAFLD was associated with an increase in all-cause and cancer-related mortality. Based on review results, it is recommended that further assessment is carried out to determine whether liver screening in high-risk patients is cost effective and if interventions can be implemented to improve hepatic and health outcomes in adults with cancer.


Introduction
Non-alcoholic fatty liver disease (NAFLD) is the most common cause of chronic liver disease worldwide, affecting 5.5 million Australians, and 1 billion people globally [1,2]. The disease exists on a spectrum ranging from simple steatosis, non-alcoholic steatohepatitis (NASH), and with increased inflammation and fibrosis can lead to cirrhosis and elevated risk of liver cancer and cardiovascular diseases [3]. However, rates may be underestimated as the prevalence and incidence of recorded NAFLD diagnosis in healthcare records are lacking [4]. This may be as a result of underreported new and existing cases, differing definitions and diagnosis methods, and limited studies undertaken to elucidate rates [2]. NAFLD occurrence is reported to be highest in populations with metabolic syndrome (MetS) and existing chronic diseases affecting up to 80% of people with type 2 diabetes (T2DM) and up to 95% of obese people [2,5]. NAFLD co-exists with components of the MetS, such that it is often referred to as the hepatic manifestation of the metabolic syndrome [2]. Ninety percent of patients with NAFLD have more than one feature of metabolic syndrome, reflecting the high prevalence [6,7].

Parameter Criteria
Outcomes NAFLD and/or NASH measured by ultrasonography, abdominal computed tomography, liver magnetic resonance imaging, hepatic steatosis index, ICD-9 codes or histology Study design Randomized and non-randomized clinical trials, cohort, cross-sectional or case-control study design.

Eligibility Criteria
The inclusion criteria for this systematic review followed the PICOS framework [17]. Studies were considered eligible for inclusion if (a) the article was published in English; (b) participants were men or women aged 18 years and older; (c) participants had a diagnosis of cancer; and (d) the study reported liver outcomes for NAFLD and/or NASH (based on ultrasonography, Magnetic resonance imaging (MRI), abdominal Computed tomography scan (CT), ICD-9 codes or histology). We included the following study designs: randomized and non-randomized clinical trials, cohort, cross-sectional or case-control study design. Case studies were excluded.

Data Extraction and Quality Assessment and Risk of Bias
Data extraction was completed by one author (SS) and verified by two independent authors (ESG and BJB). Information extracted from each study included study design and duration of the study, participant characteristics (number of participants (n), age, BMI, sex), cancer type and treatment, NAFLD and/or NASH diagnosis and classification of liver disease, number of NAFLD participants (n, %), primary study outcomes and main findings. The quality assessment of included studies was conducted in duplicate by three independent authors (ESG, SS and BJB). Data were extracted from referenced protocols to inform the quality of the study when completing the risk of bias assessment. The Academy of Nutrition and Dietetics Evidence Analysis Library Quality Criteria Checklist was utilized for observational studies [20]. This checklist consists of an evaluation of studies' relevance (four questions) and validity (ten questions). Based on these criteria, the researchers assigned each article a quality rating of positive, neutral or negative (+, Ø, −) depending on the rating (Yes, No, Unclear, N/A) given to each individual question. Conflicts on ratings were resolved through consensus between at least two authors. Studies scoring a rating of positive were considered to be high-quality studies. The quality of randomized controlled trials was assessed using the Revised Cochrane Risk of Bias tool (RoB 2.0) [21]. This tool consists of five domains assessing the risk of bias arising from the randomization process, risk of bias due to deviations from the intended protocol, missing outcome data, measurement of the outcome, and selective reporting [21]. The researchers answered signaling questions to assign a domain-level judgement about the risk of bias (low risk of bias, some concerns, and high risk of bias).

Results
The literature search process is shown in Figure 1. The search identified a total of 10,891 articles from the databases. Duplicates (n = 2175) were removed, and the remaining 8716 records were identified for title and abstract screening. Among these, 8607 articles were deemed ineligible as a result of title or abstract screening. One hundred and nine studies from the search were eligible for full-text screening, and 86 were excluded for the following reasons: NAFLD or NASH not specified as an outcome (n = 73), abstract only/supplementary articles (n = 11) and wrong study design (n = 2). Therefore, 23 studies were included in this systematic review. following reasons: NAFLD or NASH not specified as an outcome (n = 73), abstract only/supplementary articles (n = 11) and wrong study design (n = 2). Therefore, 23 studies were included in this systematic review.
The frequency of fatty liver after chemotherapy was significantly higher in females than in males (52.4% and 34.7%, respectively, p = 0.04). No significant relationship between chemotherapy-induced fatty liver and age (p = 0.9), and the presence of metabolic syndrome (p = 0.4). The results indicate that chemotherapy was associated with a significantly increased risk of fatty liver, which was more in women than in men.
The highest frequency of fatty liver was observed in patients treated with paclitaxel, FOLFOX, and ECF with 53.5%, 42.9%, and 29.2%, respectively (p = 0.09).       The prevalence of NAFLD in all adults with cancer ranged from 0.5% (n = 259 out of 51,821) [28] in individuals in prostate cancer, to 81.3% (n = 52 out of 64) [44] in endometrial cancer. Breast, colorectal and hepatocellular cancer were the most frequently investigated cancer types reporting NAFLD prevalence in nine, four and three studies, respectively.
Chemotherapy and NAFLD Four studies assessed chemotherapy and NAFLD prevalence, including a total of 12,225 participants (range: n = 152 [35] to 11,187 [36]) in breast, gastric and hepatocellular types [35,36,38,40]. The studies included in this review reported on adjuvant chemotherapy with S-1 (80 mg/m 2 /day), systemic chemotherapy, trans arterial chemoembolization, and FOLFOX chemotherapy. Liver disease has been previously reported following treatment with chemotherapy such as methotrexate, and 5-FU [46,47]. One of these studies was conducted in breast cancer patients, reporting that patients treated with chemotherapy did not have a direct correlation with hepatic steatosis (r = 0.14, p = 0.17) [38]. Two studies in patients with gastric cancer receiving chemotherapy showed it was associated with a significantly increased risk of NAFLD as indicated by the frequency of fatty liver increasing from 2% to 46.7% in all patients after chemotherapy treatment (p = 0.0001) [35], and with adjuvant chemotherapy the risk of NAFLD and decompensated cirrhosis was increased (OR 28.26, 95% CI: 8.55-93.37; p < 0.001) [40]. Overall, three out of four studies concluded that chemotherapy treatment increases risk of NAFLD development. ii Hormone therapy and NAFLD Six studies evaluated hormone therapy and NAFLD including a total of 83,983 participants [26,28,31,32,43,44]. Studies conducted in breast and prostate cancer undergoing hormone therapy reported an increased risk of NAFLD, including aromatase inhibitors (HR: 15.92; 95% CI, 6.56-38.63; p = 0.0001) [32] and ADT (HR: 1.54, 95% CI, 1.40-1.68, p < 0.001) [28], respectively. Additionally, patients who underwent treatment with ADT had a higher prevalence of liver cirrhosis (HR 1.35, 95% CI 1.12-1.60, p = 0.015) and liver disease after treatment (HR: 1.84, 95% CI 1.73-1.96; p < 0.0001) [28]. Four out of six studies reported on patients undergoing cancer treatment with tamoxifen [26,31,43,44]. Two studies reported tamoxifen-related NAFLD in patients with a BMI of greater than or equal to 22 kg/m 2 (HR, 1.58; 95% CI: 1.00-2.48; p < 0.05) [26] and significantly higher risk of newly developed fatty liver (HR: 3.69, 95% CI 1.67-8.13, p < 0.001) [31]. Another study in overweight and obese patients with endometrial cancer receiving tamoxifen [44], indicated NAFLD occurrence was higher in the treatment group compared to placebo group (log rank p = 0.017). Conversely, one study with patients undergoing tamoxifen treatment as adjuvant endocrine therapy reported no significant results in patients with NASH [43].

Survival and Mortality
Six out of six studies (from overall 23 studies) including 52,073 participants, investigating mortality (range: 60 [23] to 39,840 [27]) determined that NAFLD increases the risk of all-cause mortality and cancer-related mortality in numerous cancers including HCC, breast cancer, endometrial, colorectal cancer, and a study including multiple types [22,23,27,29,32,36]. Two studies investigated patients with HCC indicating that the incidence of NAFLD significantly reduced survival time (9.43 months vs. 38.47 months) and increased mortality within two years (p < 0.001) [36]. Similarly, Prieto et al. reported that the survival time between those with and without liver cirrhosis was significantly different (9.4 months vs. 38.5 months, p ≤ 0.001) and survival differences were seen in treatment modalities (surgery, 13.17 months vs. transarterial chemoebolization and/or radiofrequency, 30.3 months vs. systemic chemotherapy, 26.7 months p ≤ 0.001) [29]. In two studies in patients with colorectal cancer, the first with 60 people reported that hepatocyte ballooning was linked with decreased hepatic disease-free survival (RR = 3.31, p = 0.003) [23] and the second study with 39,840 patients, with and without liver cirrhosis, showed that those with cirrhosis had a higher 30-day mortality at 24.1% in comparison to patients who were not cirrhotic (13.3%) and without liver disease (8.7%) (RR 2.59, 95% CI: 1.86-3.61) [27]. The remaining two studies assessed patients with breast cancer. Lee et al., including 253 patients undergoing aromatase inhibitor therapy which led to the development of NAFLD, reported a lower disease-free survival than those without NAFLD (HR, 2.8 95% CI: 1.26-6.23, p = 0.012) [32] and Brown et al. including 387 people with multiple cancer types reported an association with increased risk of all-cause (HR: 2.52, 95% CI: 1.47-4.34; p = 0.001) and cancer-specific mortality (and HR: 3.21, 95% CI: 1.46-7.07; p = 0.004) with NAFLD [22].

Metabolic Co-Morbidities
Five studies reported on BMI with a total of 2106 (range: 19 [34] to 875 [42]) participants investigating childhood-onset craniopharyngioma, gastric cancer, endometrial cancer, breast cancer and/or multiple cancer study (breast, gastrointestinal, genitourinary, gynaecological, lung and haematological) [22,33,34,40,42]. In four out of the five studies that reported on BMI as a risk factor for NAFLD in individuals with cancer, there was a positive association confirming the known association between increased BMI and NAFLD in the context of adults with cancer [22,33,34,40,42]. There were seven studies that assessed a range of metabolic co-morbidities in addition to NAFLD, including one or more of the following: type 2 diabetes mellitus, insulin resistance, higher fasting insulin levels, obesity, hypertension, hypercholesterolaemia, hypertriglyceridemia [22,24,27,33,34,38,42]. These included a total of 44,652 (range: 19 [34] to 39,840) participants [27] in studies including a combination of multiple cancer types, breast cancer, colorectal cancer, adults with a history of childhood onset craniopharyngioma and endometrial cancer. T2DM was a comorbidity reported in five studies in patients who developed NAFLD who had breast, colorectal and/or endometrial cancers [24,27,33,38,42]. Two out of five studies reported that T2DM was significantly associated with an increased risk of NAFLD in patients with breast cancer (OR = 11.87, 95% CI: 1.06-132.37; p = 0.004) [33] and endometrial cancer (HR, 1.41; 95% CI, 1.06-1.88) [42]. The remaining three studies suggested that patients with T2DM were more likely to have an increased risk of NAFLD, but no significance was reported [24,27,38]. Similarly, dyslipidaemia was a commonly reported factor in five studies and was associated with NAFLD in colorectal, endometrial, and breast cancer, and studies in multiple cancer types [22,27,33,38,42]. Two out of five studies confirmed that elevated triglycerides were significantly associated with NAFLD in multiple cancer types (2.6 vs. 1.6 mmol/L; p = 0.007) [22] and breast cancer (OR = 50.27; 95% CI: 4.41-573.03; p = 0.002) [33]. In addition, hypercholesterolaemia in endometrial cancer patients was significantly associated with NAFLD (HR: 1.90; 95% CI:1.26-2.87; p = 0.004) [42]. Two studies reported that no association was found between hypercholesterolaemia and steatosis [38,42]. One study investigated the link between blood pressure and NAFLD in participants across multi-ple cancer types, and this showed significantly higher systolic (130.1 vs. 121.8 mm Hg; p = 0.004) and diastolic (76.8 vs. 73.0 mm Hg; p = 0.029) blood pressure in patients with NAFLD [22]. Two studies demonstrated consistent evidence that obese individuals with breast [38] and colorectal cancer [24] were more likely to have an increased risk of NAFLD; however, no significance was reported.

Risk of Bias
The risk of bias assessment for all included studies is shown in Tables 4 and 5. For the observational studies, 10 of the 23 articles received a positive quality rating, indicating a low risk of bias, and 12 articles received a neutral quality rating (Table 4). All observational studies were considered relevant and indicated applicability to practice. Validity in all 22 studies was also determined to be of high quality based on clear research questions, subject/patient selections, clearly defined research outcomes, use of valid and reliable measurements, and the reporting of limitations. For the single RCT study, we also evaluated the protocol referenced within the methods [48,49]. The risk of bias assessment was determined as some concerns. This was mainly due to missing information in one or more domains.
Were outcomes or condition or status of interest clearly defined and the measurements validand reliable?
Was the statistical analysis appropriate for the study design and type of outcome indicators?
Are conclusions supported by results with biases and limitations taken into consideration?

Discussion
This is the first systematic literature review, to our knowledge, to assess the evidence surrounding the prevalence of NAFLD in adults with cancer, the effect of therapy, and its impact on mortality. The main findings from this review were as follows: (1) the prevalence of NAFLD in adults with cancer varied widely but appeared highest in breast, gynaecologic and colorectal cancer; (2) a number of treatments were associated with an increased risk of NAFLD, including chemotherapy, tamoxifen for breast cancer, and hormone therapy in prostate cancer, and (3) NAFLD seems to poorly impact prognosis and increase mortality in people with cancer; finally, (4) individuals with higher BMI and other metabolic risk factors (irrespective of cancer diagnosis) appear to be at increased risk of NAFLD. Our review showed considerable heterogeneity in the methods of NAFLD diagnosis, sample size, study design, cancers and treatments and precludes definitive prevalence of NAFLD and associated risk factors in adults with cancer. Given the results from this review, targeted screening and/or assessment of NAFLD in those at higher risk appears to be warranted in individuals with cancer and should be assessed for cost effectiveness. It is recommended that future research prioritize supportive care survivorship interventions in adults with cancer.
Varying cancer treatments may be associated with the development of MetS phenotype and linked to pathophysiological factors that underpin the MetS. These dysfunctions include (i) insulin resistance and an increase in insulin-like growth factor 1, (ii) elevated adipokines secreted from visceral adipocytes, and (iii) free fatty acids and aromatase activity; these all collectively attribute to MetS and furthermore NAFLD [50]. These mechanisms as well as angiogenesis, glucose utilization, and oxidative stress with DNA damage, are thought to work together to increase the risk of NAFLD in adults with cancer [51]. Given the multiple metabolic alterations associated with NAFLD, high-risk patients that are diagnosed and treated for cancer and importantly present with multiple comorbidities may require liver screening, diagnosis and are likely to benefit from interventions for NAFLD.
MetS and obesity are associated with an increased risk of common cancers, albeit the risk seems to vary between populations and with the definition used for MetS [52]. Increased rates of MetS and CVD among patients diagnosed with cancer have been extensively reported in the literature [53]. Despite the large body of evidence assessing MetS in people with cancer [54], NAFLD and hepatic outcomes are not routinely assessed specifically. In this review, we have identified that there appears to be an increased risk of NAFLD in adults with cancer and this was amplified in those with a higher BMI. The use of chemotherapy in women with breast cancer and hormonal therapy for men with prostate cancer have been shown to contribute to an increase in body weight and fat mass, which are an established risk factor for NAFLD [55,56]. Multiple forms of chemotherapy have acute hepatocellular effects on liver dysfunction or toxicity, and previous studies have indicated that specific chemotherapy agents (i.e., 5-FU, platinum derivatives and taxanes) can lead to hepatic steatosis. Whilst steatosis composes NAFLD, the association of isolated chemotherapy agents or regiments (combination of agents) and risk of NAFLD is yet to be elucidated [57]. Pre-existing liver damage (from alcohol intake and/or poor lifestyle choices) prior to chemotherapy may have negative consequences in treatment tolerance [58]. Conversely, a reduction in lean (muscle) mass is often seen on presentation and over the duration of treatment in other cancer diagnoses including lung, upper and lower gastrointestinal cancers, prompting the National Cancer Institute to define Common Terminology Criteria for Adverse Events due to reduction in lean muscle mass (which mainly include altered liver enzymes) in all adults with cancer treated with chemotherapy [59]. Whilst adults with a higher BMI have an increased risk of hepatic steatosis, as reinforced by the studies in this review, we hypothesize that there may be a similar magnitude of liver damage in adults that experience reduced lean (muscle) mass, and are at risk of malnourishment as a consequence of chemotherapy. However, these cancer types have not been well investigated in the context of NAFLD and thus are not captured in this review. However, future studies may consider such assessments to determine whether the risk of NAFLD is also increased in these cancer types.
NAFLD has been described as a mediator for the obesity-cancer association [13], with the progression and onset of NAFLD underpinned primarily by insulin resistance. Subsequently, cancer treatment such as chemotherapy and hormone therapies are metabolized by the liver and therefore pose a risk of liver damage [59,60]. Therefore, whilst treatments are essential and efficacious, they appear to also elicit hepatotoxicity and the risk can be increased with pre-existing liver disease and extended treatment durations. Endocrine therapies in breast cancer and androgen depravation therapy in prostate cancer can last for many years (either continuously or intermittent). Even though these treatments are known to have a negative effect on the liver, this review has demonstrated that there are limited studies focused on NAFLD, and the cancer types investigated to date were heterogenous. Furthermore, the body of work assessing the impact of cancer treatment on metabolic risk factors and MetS is substantial [61,62], although these too overlook the hepatic implications and NAFLD. In addition to the direct adverse effects these therapies have on the liver, the treatments likely lead to an enhanced risk of NAFLD through an increase in body weight and/or reduction in lean mass, with lipid and glucose-insulin alterations [63]. Therefore, people with cancer and those with additional risk factors such as high BMI or reduced lean mass, who are at increased risk for NAFLD, should be considered for targeted screening and hepatic monitoring.
Damage to the liver increases the risk of liver-related and all-cause mortality [64]. Individuals with a history of cancer are a high-risk group for metabolic dysfunction and CVD, and the results from this review indicate an increased risk for NAFLD and its associated complications in cancer survivors. Therefore, it is important to further investigate this relationship and whether screening practices should be routinely carried out to monitor the liver in these individuals. Furthermore, rehabilitation involving lifestyle intervention, aimed at improving metabolic outcomes, should also consider hepatic benefits. This may prevent harmful side effects and damage to the liver in these individuals.
Future studies in individuals with cancer should consider assessment of liver outcomes and indeed long-term effects given the higher rates of MetS and increased mortality in these participants. Hormone therapy seen in breast and prostate cancer and some chemotherapy agents appear to be associated with increased liver steatosis and potentially NAFLD; however, future studies are warranted to evaluate the dose of treatment, pre-existing conditions, and liver outcomes. Furthermore, whether malnutrition, low lean muscle mass, and sarcopenia co-exist with NAFLD in individuals who have/had a cancer diagnosis requires additional investigation. Targeting high-risk adults with cancer and providing diet and exercise interventions which are simple, cost-effective ways proven to reduce metabolic health outcomes and mortality in other patient groups including those with heart disease, may be one approach [65]. Moreover, improving lifestyle is currently the only proven and safe way to improve hepatic outcomes [66]. Screening for liver disease and in particular NAFLD in high-risk cancer patients needs to be evaluated to determine the cost effectiveness (based on prevalence). However, to date, liver outcomes, the cost effectiveness of screening and the prevention and management of NAFLD in cancer patients have been largely overlooked.
The strengths of this systematic review are in its robust systematic methodology. Limitations include that the study designs were heterogenous, as were the cancer types captured. As a result, there was no scope for meta-analysis. Furthermore, NAFLD was assessed using a range of assessment and imaging techniques and no studies used the gold standard liver histology. However, such an invasive assessment for individuals already undergoing cancer treatment may not be appropriate. In addition, for those undergoing treatment, in many cases, the results could not be extrapolated due to the short follow-up time or lack of reporting. These are important outcomes to consider for future studies as the long-term effects of treatments in individuals with cancer are important to determine the effects of long-term liver and cardiovascular health.

Conclusions
People with cancer may have a higher risk of NAFLD and certain cancer treatments including chemotherapy, tamoxifen, and hormone therapies, seem to further exacerbate liver damage. High BMI and some metabolic risk factors appear to further increase the risk of NAFLD and those with NAFLD appear to have an increased risk of all-cause and cancer-related mortality. Further studies are needed to confirm if targeted hepatic screening in high-risk groups is cost effective and warranted to improve hepatic and health outcomes.

Conflicts of Interest:
The authors declare no conflict of interest.