Aromatase Inhibitors and Newly Developed Nonalcoholic Fatty Liver Disease in Postmenopausal Patients with Early Breast Cancer: A Propensity Score‐Matched Cohort Study

Based on a retrospective analysis, this article examines the relationship between aromatase inhibitor use in postmenopausal breast cancer patients and nonalcoholic fatty liver disease.

Abstract

Background.

Unlike tamoxifen, the relationship between aromatase inhibitor use in postmenopausal patients with breast cancer and nonalcoholic fatty liver disease (NAFLD) has not been delineated.

Materials and Methods.

A retrospective analysis of 253 patients with early breast cancer without baseline NAFLD and treated with nonsteroidal aromatase inhibitors was performed. Among them, 220 patients were matched for sex, age, and menstruation status with healthy patients, and the prevalence of NAFLD was compared. NAFLD was determined by hepatic steatosis index in the absence of other known liver diseases. The presence of significant liver fibrosis in patients with NAFLD was determined noninvasively by AST‐platelet ratio index, FIB‐4 score, and NAFLD fibrosis score (NFS).

Results.

Postmenopausal patients with breast cancer undergoing treatment with aromatase inhibitors had higher prevalence of NAFLD independent of body mass index (BMI) and underlying diabetes mellitus (DM). Although the aromatase inhibitor group showed higher fibrotic burden by NFS, independent of BMI and DM, the proportion of advanced fibrosis did not show statistically significant differences between AI‐treated patients and the healthy patients. Those with abnormal baseline fasting glucose levels are suggested to have increased risk of NAFLD development after aromatase inhibitor treatment. In addition, patients with NAFLD developed after aromatase inhibitor use had significantly lower disease‐free survival than those without NAFLD, although there was no significant difference in overall survival.

Conclusion.

Results of this study suggest that inhibition of estrogen synthesis in postmenopausal women undergoing treatment with aromatase inhibitors could increase the risk of NAFLD, which might have some influence on the prognosis of patients with breast cancer.

Implications for Practice.

Unlike tamoxifen, the role of aromatase inhibitor treatment use in postmenopausal patients with breast cancer in development of fatty liver is not well known. In this propensity‐matched cohort study, postmenopausal patients with breast cancer treated with aromatase inhibitors had increased risk of nonalcoholic fatty liver disease compared with healthy women after menopause, independent of obesity and diabetes mellitus. The results show possible adverse influence of the newly developed fatty liver on breast cancer disease‐free survival and suggest a necessity for further validation. Fatty liver may need to be considered as an adverse event for aromatase inhibitor treatment.

Introduction

Postmenopausal women diagnosed with hormone receptor‐positive breast cancer are known to benefit from endocrine therapy with aromatase inhibitors [1], [2]. In the postmenopausal state, although ovarian function and pituitary control of estrogen production ceases, estrogen synthesis occurs in the body compartments, such as in the liver, muscle, and connective tissue, by converting circulating androgens into estrogens through aromatization [3]. Aromatase inhibitors are reported to suppress total estrogen synthesis by more than 90% [3], [4] and thus emerged as an alternative to tamoxifen, a selective estrogen receptor (ER) modulator, for postmenopausal patients with breast cancer [5], [6].

Nonalcoholic fatty liver disease (NAFLD) is becoming the most common cause of chronic liver disease around the world [7]. Although the exact proportion is not fully delineated, the subset of patients with NAFLD develop a progressive form of the disease, nonalcoholic steatohepatitis, that may result in advanced liver fibrosis and cirrhosis [8]. Although women at their reproductive age are reported to have lower risk of NAFLD compared with men at similar ages, women after menopause have increased prevalence of NAFLD [9], [10]. Postmenopausal women undergo metabolic changes, including visceral fat accumulation, dyslipidemia, and glucose intolerance, that are associated with insulin resistance and development of NAFLD [11], [12]. These changes seem to be due to the diminished protective effect of estrogen, and hormone replacement therapy alleviates NAFLD in postmenopausal women [13]. Furthermore, postmenopausal women with NAFLD are reported to be at higher risk of liver fibrosis, which is associated with the duration of estrogen deficiency [14], [15]. However, another study demonstrated that premenopausal women had an increased risk of hepatic inflammation when NAFLD developed, and this result seems to contradict the protective effect of estrogen against NAFLD progression [16].

Several studies reported that both pre‐ and postmenopausal patients with breast cancer who received endocrine therapy with tamoxifen may have increased risk of developing NAFLD, which would negatively affect the survival outcome [17]. Our study aimed to evaluate the role of aromatase inhibitors on the development of NAFLD and liver fibrosis in postmenopausal patients with early breast cancer by comparing them with a propensity score‐matched healthy cohort. We also sought to evaluate whether development of NAFLD after aromatase inhibitor treatment would affect the recurrence of breast cancer and survival.

Materials and Methods

Patient Cohort

In this retrospective cohort study, the study population was obtained from a cohort of 1,480 consecutive adult women with breast cancer who initiated adjuvant endocrine therapy with nonsteroidal aromatase inhibitors, anastrozole (1 mg per day) or letrozole (2.5 mg per day), at Gangnam Severance Hospital, Yonsei University College of Medicine, an academic tertiary referral hospital in Seoul, Korea, between July 1, 2005, and June 30, 2011. Patients with early breast cancer who underwent curative resection were selected. Patients were excluded if they met any of the following criteria: metastatic breast cancer, experience of tamoxifen treatment, positive serologic markers for hepatitis B virus or hepatitis C virus, alcohol consumption >140 g per week, insufficient baseline information, presence of fatty liver disease defined by abdominal ultrasound, or by hepatic steatosis index (HIS) >36, calculated from the baseline data (supplemental online Fig. 1). Insufficient information included not having exact body mass index (BMI) parameters, lipid profiles, and other biochemical parameters either at baseline or during the follow‐up period so as not to be able to assess NAFLD or liver fibrosis status.

The control group was obtained from the cohort of 138,171 adult patients who underwent a health checkup at Gangnam Severance Hospital between July 1, 2006, and June 30, 2011. The patients with history of any malignant diseases were excluded. This study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Institutional Review Board of Gangnam Severance Hospital, Yonsei University College of Medicine, Seoul, Korea (3‐2016‐0319).

Measurement of Clinical Parameters and Biochemical Analysis

The index date for the study entry was defined as the first prescription date of adjuvant endocrine therapies. The baseline data of the study patients within 3 months before the index date were collected from electronic medical records. The follow‐up data were collected from the results obtained within 3 months of the last visit. Patients were considered to be obese when BMI was ≥25 kg/m², based on the criteria for the Asian‐Pacific region [18]. Diabetes mellitus (DM) was diagnosed with use of insulin or oral hypoglycemic agents or with fasting plasma glucose ≥126 mg/dL. Patients were considered to be hypertensive if blood pressure was ≥140/90 mmHg or if antihypertensive medication was currently prescribed.

Definition of Hepatic Steatosis and Significant Liver Fibrosis

In clinical practice, NAFLD is diagnosed based on biochemical tests and increased echogenicity on abdominal imaging such as ultrasonography [19]. In addition, liver biopsy is still the gold standard in determining progression of NAFLD and degree of liver fibrosis, although it cannot be casually performed because of possible complications [20]. Instead, noninvasive tests using serum biomarkers are popularly used after validations [21].

In this study, development of NAFLD during follow‐up was determined by the previously validated fatty liver predicting model, HIS [22].

$$\begin{array}{c}\mathrm{HIS}=8\times (\text{aspartate aminotransferase}\left[\mathrm{AST}\right]/\\ \text{alanine aminotransferase}\left[\mathrm{ALT}\right])\\ +\mathrm{BMI}\left(+2\text{if female},+2\text{if diabetes mellitus}\right)\end{array}$$

The cutoff value of HIS >36 was used to detect NAFLD with specificity of 92.4% [22].

Among the individuals with NAFLD, defined by HIS, serum markers of fibrosis were used to assess advanced liver fibrosis. These included AST‐platelet ration index (APRI) [23], Fibrosis‐4 (FIB‐4) [24], and NAFLD fibrosis score (NFS) [25] and were calculated according to the following published formulas. These formulas include serum platelet count as one of the parameters. The inverse correlation of platelet count and liver fibrosis stage has been recognized, which can be explained by portal hypertension leading to pooling of platelets in an enlarged spleen [26].

$$\text{APRI}=\left(\left[\mathrm{AST}/\text{upper limit of normal}\right]/\text{platelet count}\left[{10}^9/\mathrm{L}\right]\right)\times 100$$

The cutoff value of APRI ≥1.5 was applied to detect high probability of advanced fibrosis, as previously published [23].

$$\begin{array}{c}\mathrm{FIB}-4=\left(\mathrm{age}\left[\text{years}\right]\times \mathrm{AST}\left[\mathrm{U}/\mathrm{L}\right]\right)/\\ \left(\text{platelet count}\left[{10}^9/\mathrm{L}\right]{\left(\mathrm{ALT}\left[\mathrm{U}/\mathrm{L}\right]\right)}^{1/2}\right)\end{array}$$

The cutoff value of FIB‐4 ≥2.67 was used to detect intermediate and high probability of advanced fibrosis [24].

$$\begin{array}{c}\mathrm{NFS}=-1.675+0.037\times \mathrm{age}\left(\text{years}\right)+0.094\times \mathrm{BMI}\left(\mathrm{kg}/{\mathrm{m}}^2\right)\\ +1.13\times \text{diabetes}(\mathrm{yes}=1,\mathrm{no}=0)+0.99\times \mathrm{AST}/\mathrm{ALT}\text{ratio}\\ -0.013\times \text{platelet count}\left(\times {10}^9/L\right)–0.66\times \text{albumin}\left(\mathrm{g}/\mathrm{dL}\right)\end{array}$$

The cutoff value of NFS >0.676 was applied to detect high probability of advanced fibrosis [25].

Statistical Analysis

Statistical analyses were performed using IBM SPSS version 23 (IBM; Armonk, NY) and R package version 3.2.2.

Propensity analysis was carried out using logistic regression in order to create a propensity score for female patients who received aromatase inhibitors and healthy female patients who underwent a health checkup. The variables entered into the propensity model were sex, age (at last follow‐up), and menstruation status.

Continuous variables are expressed as means and SD or medians and ranges. Categorical variables are expressed as the number of cases and proportions. For the propensity score‐matched cohorts, differences in quantitative variables were compared with the Student's t test or nonparametric Mann‐Whitney U test. Multivariable logistic regression analysis was used to determine independent association between administration of aromatase inhibitors and occurrence of NAFLD as well as significant liver fibrosis. The variables that were used in calculation of HSI or fibrosis models were not tested in multivariable logistic regression analysis because they could not be considered as independent variables. To evaluate factors predicting occurrence of NAFLD after application of aromatase inhibitors in postmenopausal women, overall treatment duration was adjusted using multivariable Cox regression models with factors tested on Kaplan‐Meier estimates. The variables that were used in calculation of HSI or fibrosis models were excluded in analysis.

All reported p values are two‐sided, and p values less than .05 were considered significant. However, variables with p values less than .1 in univariate analysis were included in the multivariate models. Hazard ratios (HRs) were presented with 95% confidence interval (CI).

Results

Characteristics of the Study Population After Propensity Score Match Analysis

Postmenopausal women with early breast cancer who had their estrogen levels further suppressed by aromatase inhibitors were compared with age‐matched postmenopausal female patients without endocrine therapy. Incidence of NAFLD and liver fibrosis were compared. From 343 patients with early breast cancer undergoing adjuvant endocrine therapy with aromatase inhibitors, 253 patients without evidence of fatty liver disease by ultrasound and HIS were selected for propensity matching (supplemental online Fig. 1). The patients were followed up for 8.4 years (median, 0.8–11.5) and given aromatase inhibitors for 4.4 years (median, 0.3–7.8; supplemental online Table 1). Finally, 220 patients were matched for sex, age, and menstruation status. The demographic and clinical characteristics of the two groups are described in Table 1. Postmenopausal woman who underwent aromatase inhibitor treatment had lower BMI than the postmenopausal women who did not receive endocrine therapy (p < .001). There were no significant differences in the incidence of diabetes and hypertension between the two groups (p = .190 and p = .629, respectively). HIS was significantly higher in the aromatase inhibitor‐treated group (33.15 ± 4.35 vs. 38.08 ± 8.03; p = .001), and the proportion of patients with HIS >36 who were considered to have high probability of NAFLD was significantly larger in the aromatase inhibitor‐treated patients (25.9% vs. 53.6%; p = .001).

Table 1. Characteristics of the propensity score‐matched cohorts.

Calculation of liver fibrosis score, APRI, FIB‐4, and NFS were performed only in subjects with NAFLD assessed by HSI model.

The results are expressed as mean ± SD, except described otherwise.

Abbreviations: AI, aromatase inhibitor; ALT, alanine aminotransferase; APRI, AST‐platelet ratio index; AST, aspartate aminotransferase; BMI, body mass index; FIB‐4, Fibrosis‐4; HDL, high density lipoprotein; HSI, hepatic steatosis index; NAFLD, nonalcoholic fatty liver disease; NFS, NAFLD fibrosis score.

Relative Risk of NAFLD in Postmenopausal Patients with Breast Cancer

As described earlier, HIS is a NAFLD model that includes AST, ALT, BMI, and existence of diabetes. Significant NAFLD is defined as HIS >36. To evaluate the relative risk of NAFLD as predicted by giving aromatase inhibitors without confounding influence of obesity, we stratified the study patients using the cutoff value for BMI of 25 kg/m² (n = 308, 70% with BMI <25 kg/m²; n = 132, 30% with BMI ≥25 mg/m²). We found that postmenopausal patients with breast cancer who received aromatase inhibitors (n = 220) had higher prevalence of NAFLD than those who did not undergo endocrine therapy (n = 220), regardless of BMI (46.3% vs. 4.5% in patients with BMI <25 kg/m²; HR, 15.922; 95% CI, 6.562–38.633; p = .000; 87.7% vs. 58.8% in patients with BMI ≥25 kg/m²; HR, 4.483; 95% CI, 1.576–12.751, p = .002; Fig. 1A). We also evaluated relative risk of NAFLD as predicted by giving aromatase inhibitors without confounding influence of DM. We found that postmenopausal patients with breast cancer who underwent aromatase inhibitor treatment (n = 220) had higher prevalence of NAFLD than the matched control (n = 220), regardless of underlying DM (78.8% vs. 44.8% in patients with DM; HR, 5.222, 95% CI, 1.504–18.137, p = .009; 49.2% vs. 23.2% in patients without DM; HR, 2.898, 95% CI, 1.832–4.585, p = .000; Fig. 1B). To evaluate whether the use of aromatase inhibitors in postmenopausal women would affect NAFLD after adjusting confounding covariates other than BMI, diabetes, and AST/ALT, multivariable logistic regression analysis was performed. Variables that composed HIS, namely BMI, DM, AST/ALT, were not included in the multivariable logistic regression model. Receiving aromatase inhibitors was significantly associated with having NAFLD in postmenopausal patients with breast cancer (HR, 2.977; 95% CI, 1.956–4.531; p = .000), along with having hypertension (HR, 1.935; 95% CI, 1.167–3.209; p = .011; Table 2).

Figure 1.

Prevalence of NAFLD using HSI according to AI use, stratified by BMI and diabetes mellitus (DM). Prevalence of NAFLD was significantly higher in postmenopausal women with inhibited estrogen synthesis by aromatase inhibitor than in those with circulating estrogen independent of BMI (p < .001) (A) and underlying DM (p = .002) (B).

Abbreviations: AI, aromatase inhibitor; BMI, body mass index; CI, confidence interval; HR, hazard ratio; HIS, hepatic steatosis index; NAFLD, nonalcoholic fatty liver disease

Table 2. Factors associated with nonalcoholic fatty liver disease in the propensity score‐matched cohorts^a .

^aNonalcoholic fatty liver disease (NAFLD) was diagnosed using hepatic steatosis index (HIS), and HIS >36 was diagnosed as having NAFLD. Variables included in calculation of HIS were excluded from the analysis.

Abbreviations: AI, aromatase inhibitor, CI, confidence interval; HIS, hepatic steatosis index = 8 × (aspartate aminotransferase/alanine aminotransferase ratio) + body mass index (+2, if female; +2, if diabetes mellitus); HR, hazard ratio.

Degree of Liver Fibrosis and Aromatase Inhibitor Use in Patients with Menopause with Early Breast Cancer

Among the patients whose HIS was >36 (total, n = 175; aromatase inhibitor group, n = 118; control, n = 57), serum markers of fibrosis, APRI, FIB‐4, and NFS were used to assess severity of fibrosis. The aromatase inhibitor group showed higher scores on APRI (0.16 ± 0.02 vs. 0.59 ± 1.53, p = .008), FIB‐4 (1.45 ± 0.60 vs. 2.75 ± 2.94, p = .000) and NFS (−1.55 ± 1.17 vs. −0.39 ± 1.83, p = .000; Table 1). The proportion of patients with significant fibrosis by FIB‐4 and NFS was larger in the aromatase inhibitor group (FIB‐4, 5.3% vs. 21.2%, p = .000; NFS, 3.5% vs. 19.5%, p = .001; Fig. 2). No statistically significant difference was shown by APRI in the proportion of patients with fibrosis (0.0% vs. 5.9 %, p = .080). To assess the association between degree of liver fibrosis and aromatase inhibitor use without the confounding influence of obesity, the study patients with NAFLD were stratified using the cutoff value for BMI of 25 kg/m² (n = 75, 42.9% with BMI <25 kg/m²; n = 100, 57.1% with BMI ≥25 mg/²). When we evaluated the degree of liver fibrosis using NFS, postmenopausal patients with breast cancer who received aromatase inhibitors (n = 118) demonstrated higher fibrotic burden than those without the endocrine therapy (n = 57; aromatase inhibitor group vs. control, −0.24 ± 1.87 vs. −2.84 ± 1.63 in patients with BMI <25 kg/m², p = .001; −0.73 ± 1.71 vs. −1.37 ± 0.99 in patients with BMI ≥25 kg/m², p = .044; Fig. 3A). After stratifying the study patients by existence of DM (n = 130, 74.3% without DM; n = 45, 25.7% with DM), the aromatase groups still demonstrated higher fibrotic burden regardless of underlying DM (aromatase inhibitor group vs. control, −0.68 ± 1.84 vs. −1.86 ± 1.08 in patients without DM, p = .000; 0.39 ± 1.56 vs. −0.50 ± 0.84 in patients with DM, p = .023; Fig. 3B).

Figure 2.

Association of aromatase inhibitor use and fibrotic burden in postmenopausal females. Fibrotic burden is assessed using APRI, FIB‐4, and nonalcoholic fatty liver disease fibrosis score (NFS). Receiving AI and having estrogen synthesis near completely inhibited has significant positive association with fibrosis, evaluated by FIB‐4 (p = .000) and NFS (p = .001).

Abbreviations: AI, aromatase inhibitor; APRI, AST‐platelet ratio; FIB‐4; NFS, nonalcoholic fatty liver disease fibrosis score.

Figure 3.

Differences in fibrotic burden assessed by NFS according to AI use stratified by BMI and diabetes mellitus (DM). Fibrotic burden assess by NFS was significantly higher in postmenopausal females with inhibited estrogen synthesis by aromatase inhibitor than in those with circulating estrogen independent of BMI (A) and underlying DM (B) (all p < .05).

Abbreviations: AI, aromatase inhibitor; BMI, body mass index; NFS, nonalcoholic fatty liver disease fibrosis score.

However, when the proportion of significant liver fibrosis using the NFS model was compared after stratifying the patients by BMI, the differences failed to show statistical significance (aromatase inhibitor group vs. control, 23.5% vs. 14.3% in patients with BMI <25 kg/m²; HR, 1.807; 95% CI, 0.183–17.814; p = .612; 10.8% vs. 4.0% in patients with BMI ≥25 kg/m²; HR, 2.119; 95% CI, 0.243–18.509; p = .497). Stratifying patients by existence DM did not provide statistically significant differences in the proportion of significant fibrosis either (aromatase inhibitor group vs. control, 15.1% vs. 0% in patients without DM: HR, 152,522,841.6; 95% CI, 0; p = .997; 27.1% vs. 22.8% in patients with DM: HR, 2.074; 95% CI, 0.371–11.596; p = .406).

The Effect of NAFLD Development on Survival in Postmenopausal Patients with Early Breast Cancer

Of the 253 patients without evidence of fatty liver disease at baseline and who underwent aromatase inhibitor (AI) treatment, 11.5% (33/253) died. Of those 33 patients, 3 patients died from a cause other than breast cancer‐related events (1 from acute myocardial infarction, 1 from pancreas cancer, and 1 from pneumonia). Of the patients with NAFLD developed after aromatase inhibitor treatment, 25 patients (25/139, 18.0%) died, whereas 8 patients (8/114, 7.1%) died in the group without NAFLD. The causes of death in the NAFLD and non‐NAFLD group after AI treatment are described in supplemental online Table 3. However, Kaplan‐Meier analysis demonstrated that there was no significant differences between the two groups on overall survival (p = .059; Fig. 4A). In contrast, from the patients with NAFLD developed after aromatase inhibitor treatment, 25 patients (25/139, 18.0%) had breast cancer recurrences, whereas 8 patients (8/114, 7.0%) without NAFLD had recurring breast cancer. Kaplan‐Meier analysis demonstrated that patients with NAFLD developed after aromatase inhibitor use had significantly lower disease‐free survival (DFS) than those without NAFLD (p = .010; Fig. 4B). In multivariate analysis, positive ER and progesterone receptor (PR) showed independent factors predicting longer DFS, whereas having developed NAFLD after aromatase inhibitor treatment suggested shorter DFS in postmenopausal patients with early breast cancer who underwent aromatase inhibitor therapy (Table 3).

Figure 4.

Kaplan‐Meier survival curves in early breast cancer with aromatase inhibitor treatment after curative resection. (A) Overall survival and (B) disease‐free survival are compared between the patients without the NAFLD after aromatase inhibitor treatment (in bold line) and those with newly developed NAFLD after the treatment (in dashed line).

Abbreviation: NAFLD, nonalcoholic fatty liver disease.

Table 3. Factors associated with disease‐free survival in patients with breast cancer who underwent aromatase inhibitor treatment.

^a p value <.1 is considered to be significant.

^b p value <.05 is considered to be significant.

Abbreviations: AI, aromatase inhibitor; CI, confidence interval; ER, estrogen receptor; HIS, hepatic steatosis index = 8 × (aspartate aminotransferase/alanine aminotransferase) + body mass index (+2, if female; +2, if diabetes mellitus); NAFLD, nonalcoholic fatty liver disease; PR, progesterone receptor.

Factors Predicting NAFLD in Postmenopausal Patients with Breast Cancer Treated with Aromatase Inhibitors

To evaluate factors predicting significant NAFLD in postmenopausal women treated with aromatase inhibitors (n = 253), we applied Cox proportional hazard model to adjust variable treatment duration and follow‐up period. Postmenopausal patients with early breast cancer whose baseline fasting glucose was greater than 110 mg/dL had significant risk for developing NAFLD after aromatase inhibitor treatment (HR, 1.921; 95% CI, 1.229–3.003, p = .004; supplemental online Table 2). In addition, NAFLD occurrence was not associated with duration of AI treatment, suggesting that NAFLD development after AI use might not be dose dependent.

Discussion

This study demonstrated that endocrine therapy using nonsteroidal aromatase inhibitors in postmenopausal patients with breast cancer increased the risk of NAFLD development independent of obesity and DM. In addition, those with newly developed NAFLD after aromatase inhibitor treatment may have decreased DFS, although this result needs to be validated by larger cohort study. Those with abnormal fasting glucose level at the initiation of the endocrine therapy might have increased risk of newly developed NAFLD after aromatase inhibitor treatment.

NAFLD encompasses a spectrum of liver disease from simple steatosis to advanced fibrosis and may progress to liver cirrhosis or hepatocellular carcinoma [27], [28]. However, with all these possible liver‐associated morbidities and mortalities, pathogenesis of NAFLD and its progression is yet to be delineated. Studies support that estrogen might be protective against development of NAFLD [29], [30], [31], and treating patients with breast cancer with an ER modulator, tamoxifen, is suggested to increase the risk of NAFLD. However, studies investigating the role of estrogen on NAFLD are often complicated by age factors and accompanying metabolic traits that might facilitate NAFLD [32], [33]. Our study demonstrates that when estrogen that is the product of aromatization of circulating androgens is nearly inhibited in postmenopausal women, occurrence of NAFLD increases independently of BMI and underlying DM. Our matched cohort had no significant difference in the proportion of patients with diabetes compared with the aromatase‐treated group, and BMIs were even lower in the aromatase inhibitor group. Nevertheless, NAFLD was more prevalent in the aromatase‐treated group. The results further support the beneficial role of estrogen against NAFLD development.

It has been reported that one of the most important prognostic factors for NAFLD is associated advanced liver fibrosis [34]. In this study, liver fibrosis scores that are calculated using the NFS model showed statistically significant differences between postmenopausal patients with breast cancer who received aromatase inhibitor and the age‐matched control patients independently of obesity and DM. However, when the proportions of significant fibrosis were compared, they failed to demonstrate statistical significances. Significant liver fibrosis was assessed in those with HIS >36, and the number of the study patients were inevitably decreased to n = 57 in the control group and n = 118 in the aromatase inhibitor‐prescribed patients with breast cancer. The absolute numbers for the proportion of significant fibrosis were greater in those who underwent aromatase inhibitors treatment even when the influences of obesity and DM were controlled. Because it cannot be ruled out that the failure in showing the statistical significances might be due to the inadequate sample size, the role of aromatase inhibitors in progression of significant liver fibrosis should be further evaluated in another study with larger cohort.

In this study, postmenopausal patients with breast cancer, in whom NAFLD developed after aromatase inhibitor treatment, showed decreased DFS, and NAFLD seemed to be one of the significant factors affecting DFS, along with PR and ER positivity. A recent large cohort study demonstrated association of NAFLD and increased breast cancer incidence rate, suggesting that NAFLD might be a negative prognostic factor for breast cancer [35]. In addition, a retrospective cohort study showed that NAFLD development after tamoxifen treatment had a negative effect on survival outcome of patients with breast cancer [17]. Although this study comprised relatively small number of patients, with breast cancer recurring in 7.5% (19/253) of patients, and the results, especially, regarding the survival should be validated in another larger cohort, it can be suggested that development of NAFLD after the endocrine therapy should not be neglected.

This study is limited by several factors. First, NAFLD was determined using a comprehensive NAFLD score instead of ultrasonography, which is shown to have an acceptable diagnostic accuracy [36]. Although the patients who underwent treatment with aromatase inhibitors underwent abdominal ultrasound as a part of the breast cancer staging workup before the treatment, and those with underlying NAFLD either by ultrasound or HIS were excluded from the study, only 29.5% (65/220) of the patients had follow‐up ultrasound. Positive predictive value, compared with ultrasound‐defined moderate‐severe fatty liver, was 96.6% (28/29) in these patients. In contrast, all the patients in the control underwent ultrasound as a part of health checkup, and positive predictive value in this group was 91.1% (52/57). Second, this is a single‐center, retrospective study, and sample size is relatively small. The study results, especially regarding the effect of estrogen inhibition on significant fibrosis and the survival outcome, should be re‐evaluated by another study with larger sample size. Third, plasma estrogen levels have not been evaluated in patients undergoing treatment with aromatase inhibitors, and differences might exist accordingly. However, many studies indicate that aromatase inhibitors may block total body estrogen synthesis by >98%, although several studies report that estrogen levels may be sustained at 20%–40% of pretreatment levels [37]. Fourth, duration of AI treatment was evaluated by medical records without interviewing the patients. Therefore, compliance to AI medication could not be assessed in this study.

Conclusion

Even with these limitations, this study suggests that near total inhibition of estrogen synthesis in postmenopausal women would increase the risk of NAFLD development compared with those with sustained estrogen synthesis even if the level of circulating hormone might be remarkably lower than that at the reproductive ages. Impact of NAFLD development after aromatase inhibitor treatment on liver fibrosis and survival outcome in patients with breast cancer should further be investigated.

See http://www.TheOncologist.com for supplemental material available online.

Author Contributions

Concept/design: Jung Il Lee, Kwan Sik Lee

Provision of study material or patients: Sung Gwe Anh, Joon Jeong

Collection and/or assembly of data: Jung‐Hwan Yu, Hyun Woong Lee

Data analysis and interpretation: Jung Il Lee, Jung‐Hwan Yu

Manuscript writing: Jung Il Lee

Final approval of manuscript: Jung Il Lee, Jung‐Hwan Yu, Sung Gwe Ahn, Hyun Woong Lee, Joon Jeong, Kwan Sik Lee

Disclosures

The authors indicated no financial relationships.