Non-alcoholic fatty liver disease increases the risk of biochemical recurrence in high-grade metastatic prostate cancer patients


 Background

Non-alcoholic fatty liver disease (NAFLD) has been reported to be helpful to identify high-risk individuals of developing prostate cancer. In the retrospective cohort study, our aim is to investigate the relationship between NAFLD and biochemical recurrence in metastatic prostate cancer patients.
Methods

We retrospectively investigated 602 patients with metastatic prostate cancer receiving androgen deprivation therapy. Liver fat was estimated with liver-to-spleen ratio by CT scans. The relationship between NAFLD and biochemical recurrence was investigated with Cox models. The model for biochemical recurrence was adjusted for multiple variables.
Results

NAFLD was significantly associated with biochemical recurrence in patients with Gleason score ≥ 4 + 3 when adjusting for each of body mass index (hazards ratio [HR] = 1.38; 95% confidence interval [CI] = 1.08–1.77; p = 0.01), visceral adipose tissue (HR = 1.36; 95% CI = 1.07–1.74; p = 0.01), hypertension (HR = 1.41; 95% CI = 1.10–1.80; p = 0.01) and diabetes mellitus (HR = 1.42; 95% CI = 1.11–1.82; p = 0.01), using age and prostate-specific antigen level as potential confounder. The 2-year biochemical recurrence rate in Gleason score ≥ 4 + 3 patients with and without NAFLD was 84.0% (100/119) and 72.2% (130/180), respectively (p = 0.018). The median biochemical recurrence free survival of Gleason score ≥ 4 + 3 patients with and without NAFLD were 17 and 21 months, respectively (p = 0.005).
Conclusions

NAFLD is an independent risk factor for biochemical recurrence in patients with high-grade metastatic prostate cancer. If validated in prospective studies, future research should test whether treatment of NAFLD can lead to better prognosis.


Background
The incidence rates of prostate cancer (PCa) have increased in most countries with increasing ageing populations. Although the ve-year relative survival rates of local PCa are approximately 100%, metastatic PCa remains a fatal disease with poor prognosis [1]. Androgen deprivation therapy (ADT) is commonly applied in patients with metastatic PCa as primary treatment. Biochemical recurrence (BCR) constantly leads to treatment failure.
The previous studies on the relationship between obesity and PCa incidence have shown inconsistent results [2][3][4][5]. However, obesity has been veri ed to be associated with an increased risk for BCR and PCaspeci c mortality [6]. Non-alcoholic fatty liver disease (NAFLD), the leading cause of chronic liver disease worldwide, is closely associated with obesity. The association of PCa with NAFLD has been paid more attention. NAFLD has been reported to be helpful in identifying high-risk PCa in the elderly and in the absence of obesity or metabolic syndrome [7]. However, to the best of our knowledge, the question of whether NAFLD is related to the risk of BCR has been less well studied. Only a study by Choi et al investigated the association of NAFLD and BCR and reported that the former was negatively related to the risk of BCR after radical prostatectomy. However, the sample size was small, only 32 patients developed BCR [8]. In the retropective study, we investigated the association between NAFLD and the risk of BCR in metastatic PCa patients.

Patients and diagnosis
The 667 patients in three hospitals were diagnosed as metastatic PCa and received ADT between January 2010 and December 2015. We excluded 65 patients including (1) history of radical prostatectomy, orchiectomy, chemotherapy, radiation therapy (n = 21); (2) records with only two prostate-speci c antigen (PSA) values and more than one year period between the rst and second measures (n = 3); (3) increased PSA level after ADT initiation (n = 5); (4) Alcohol use being de ned as > 20 g/day (n = 16); (5) other hepatic steatosis causes (n = 15); and (6) missing patient data (n = 5). Consequently, we included 602 patients in the study. All patients were treated with complete androgen blockade (50 mg bicalutamide daily plus a luteinising hormone-releasing-hormone agonist), and no Chinese medicine was used. All cases were con rmed through transrectal ultrasound (TRUS)-guided core needle biopsy or transurethral resection of the prostate and histopathological examination.
Patient follow-up and data collection Subjects were followed up from the time of initial ADT until the date of BCR, loss to follow-up or study endpoint (31st December 2018). BCR was de ned as rst PSA increase that is ≥ 25% and ≥ 2 ng/ml above the nadir, which is con rmed by a second value obtained ≥ 3 week later [9]. BCR-free survival was de ned as the time between the date of initial ADT and the date of BCR or last follow-up. Data were collected from the medical records including age, Gleason score, metastasis position, initial PSA level, baseline testosterone level, weight, height, body mass index (BMI), subcutaneous adipose tissue (SAT), visceral adipose tissue (VAT), blood lipid parameters, liver-to-spleen (L/S) ratio, hypertension, fasting glucose and diabetes mellitus. The height and body weight of patients were measured, and BMI was calculated as the weight in kilograms divided by the height in square metre. Abdominal fat distribution was assessed through a non-contrast CT scan performed with GE (GE LightSpeed VCT 64, GE Healthcare, Waukesha, WI) multi-detector CT scanners. The VAT volume was de ned as the sum of fat voxels in the area enclosed by the peritoneal membrane centred at the L4 and L5 disks with 10 mm slice thickness.
VAT and SAT were assessed using commercially available software. NAFLD was de ned as the L/S ratio < 1.0 after exclusion of other hepatic steatosis causes. The L/S ratio was graded as mild (0.7-1.0), moderate (0.5-0.7, including 0.7) and severe (≤ 0.5) [10]. Liver and spleen attenuations were measured using unenhanced abdominal CT scans [11]. The intraclass correlation coe cient between readers in randomly selected samples of 150 participants was 0.98 for liver attenuation, indicating high reproducibility of liver attenuation for assessment of NAFLD in this study.

Statistical analysis
The χ 2 test was used to assess the relation between clinical variables and NAFLD for categorical variables. The continuous variables were analysed using t-test or Wilcoxon rank-sum test. The Kaplan-Meier method and log-rank test were used to determine the differences in BCR-free survival. Cox proportional hazards regression models were used to test the association between NAFLD and BCR. The model for BCR was adjusted for BMI, VAT, hypertension and diabetes mellitus, respectively, using age, PSA level and Gleason score as potential confounder. Statistical tests were two-sided, and p < 0.05 was considered indicative of a statistically signi cant difference. All statistical analyses were performed using SPSS, version 19.0 (SPSS, Chicago, IL, USA).
Results 602 patients were included in the study. Table 1 shows the baseline patient characteristics. Amongst the 602 patients, 549 (91.2%) developed BCR, and 188 (31.2%) were diagnosed with NAFLD. A total of 3 (1.60%) patients showed moderate NAFLD, and 185 (98.4%) showed a mild diagnosis. The median BCRfree survival was 21 (7-86) months. Approximately 12.8% of patients were obese (BMI > 30 kg/m 2 ), and 50.8% were overweight (25-29.9 kg/m 2 ). We analysed the relationship between the clinical variables and the presence of NAFLD. The patients with NAFLD presented higher Gleason score, weight, BMI, serum triglyceride level, SAT and VAT volume and lower serum HDL cholesterol level compared with those without NAFLD. Hypertension, diabetes mellitus, fasting glucose and hyperlipidaemia were signi cantly related to NAFLD (all p < 0.001) (Table S1).

Discussion
In this study, metastatic PCa patients were treated with ADT only. No obesity measurement, such as BMI and VAT, as well as hypertension and diabetes mellitus were found to be associated with BCR. However, we did nd that the presence of NAFLD was an independent risk factor for BCR in patients with highgrade PCa. Gleason score ≥ 4 + 3 was always regarded as high-grade prostate cancer, as previous reported [12].
Given the rapidly increasing incidence of NAFLD [13], it is important to identify and understand the association between NAFLD and PCa. To the best of our knowledge, only two studies have speci cally investigated the relationship between NAFLD and PCa. One study investigated 10,516,985 subjects, amongst which NAFLD based on fatty liver and hepatic steatosis indexes was identi ed in 19% and 25% of patients, respectively. The results indicated that NAFLD might help in identifying elderly men at high risk of developing PCa, including in the absence of metabolic syndrome or obesity [7]. Another study showed NAFLD as an independent negative predictive factor of BCR after radical prostatectomy. However, the sample size was small, only 32 patients developed BCR [8]. The current study has the advantage of a lager sample size of 602 metastatic PCa patients, amongst which 549 (91.2%) developed BCR, and 188 (31.2%) were diagnosed with NAFLD through multi-detector CT. In the univariate analysis, NAFLD and the high Gleason score were associated with BCR. By contrast, NAFLD showed no signi cant association with BCR in the multivariate analysis. The positive association in the univariate analysis may have been confounded by other predictors. Further analysis revealed that NAFLD was signi cantly associated with a high Gleason score. Thus, the positive association between NAFLD and BCR in the univariate analyses might have been confounded by Gleason score. NAFLD has been regarded as a common aspect of metabolic syndrome [14]. Several studies demonstrated that metabolic syndrome is associated with high-grade PCa [15,16], which may explain the positive relation between NAFLD and high-grade PCa in our study.
We further assessed the association of NAFLD with the risk of BCR in patients with Gleason score ≥ 4 + 3 and ≤ 3 + 4, respectively. A total of 299 patients manifested Gleason score ≥ 4 + 3 in the study, amongst which 119 were diagnosed with NAFLD, and 117 patients with NAFLD developed BCR. The high-grade PCa patients with NAFLD exhibited a signi cantly increased risk of BCR compared with those without NAFLD. The 2-year BCR rate in high-grade PCa patients with NAFLD signi cantly increased compared with that of patients without NAFLD. The median BCR-free survival of these patients with and without NAFLD was 17 and 21 months, respectively. However, no association between NAFLD and the risk of BCR was observed in patients with low-grade PCa (Gleason score ≤ 3 + 4). Thus, NAFLD might help in identifying men at high risk of developing castration-resistant prostate cancer (CRPC) among patients with high-grade cancer. Over the past few decades, metabolic syndrome (MetS) has been thought to in uence PCa etiology. Individual components of MetS, such as obesity, hypertension and diabetes mellitus have been reported to be associated with BCR after radical prostatectomy or radical radiotherapy [17][18][19][20]; however, our study failed to nd such an association between obesity markers, including BMI, VAT or hypertension or diabetes mellitus and BCR. The associations between MetS and PCa risk are different between ethnic groups. It has been reported that MetS was associated with PCa risk in African-American men, but not in white men [21]. The study from Guiming et al found the presence of MetS was associated with an increased risk of high-grade PCa in a Chinese population [22]. Conversely, Jeon et al found that the presence of MetS was correlated with decreased risk of high-grade PCa in a Korean population [23]. In addition, most patients in our study exhibited overweightness or normal weight but not obesity. These may help to explain the negative relations observed. Thus, NAFLD may be a positive predictive factor for BCR in patients with high-grade PCa, especially in the absence of obesity. The causal mechanisms underlying these nding require further study. If there is indeed a causal association between NAFLD and BCR, then future research should investigate whether treatment of NAFLD can lead to better PCa prognosis, and the association between the presence of NAFLD and PCa prognosis in other ethnic populations warrant further study.
In subjects with NAFLD, liver fat is closely correlated with some measures of insulin resistance such as fasting serum insulin and glucose level [24]. It is interesting to note that there are increasing data that suggest that insulin may play a signi cant role in PCa. A study by Gucalp et al demonstrated that periprostatic white adipose tissue in ammation, which is related to high insulin level, is associated with high-grade PCa [25]. Another study by Lehrer et al demonstrated that patients at the highest risk of PCa recurrence exhibited higher levels of serum insulin when comparing with those at medium and low risk [26]. These observations suggest that the higher levels of serum insulin required to counter insulin resistance in patients with NAFLD may promote the progression of CRPC. In the study, the 2-year BCR rate in Gleason score ≥ 4 + 3 patients with NAFLD signi cantly increased compared with that of patients without NAFLD, but 1-year BCR rate had no signi cant difference between the two groups. A study from Sebastiano et al found prostate cancer patients receiving ADT had an increased risk of metabolic syndrome, which may be driven by increased serum insulin levels [27]. Thus, we suggest ADT may aggravate insulin resistance in patients with NAFLD. After a period of ADT, the aggravated insulin resistance may promote the progression of CRPC. In addition, In the Chinese and Japan population, the incidence of NAFLD is 15% and 14%, respectively, while NAFLD occur in 17%-33% of Western population due to metabolic syndrome prevalence which increases to 57.5-74% in obese people [28]. Many studies have reported that variants in or near TM6SF2, PNPLA3, NCAN, GCKR, LYPLAL1 are associated with the development of NAFLD in multiple ethnic groups. The study from Wang identi ed the variants of TM6SF2 and PNPLA3 were the most signi cant risk alleles of NAFLD in Chinese population [29]. Thus, the relationship between TM6SF2 and PNPLA3 variants and the risk of BCR after ADT will be investigated in the future.
This study featured several limitations. First, we investigated a limited number of patients treated in three hospitals. The way in which the patients were enrolled may have introduced selection bias. Second, Obesity is much less common in China than in the west. The study only included 12.8% of the patients with obesity and most of them showed mild NAFLD, which may limit the generalizability of our ndings. Third, the results based on Chinese people, which cannot be generalized to other ethnic populations.
Fourth, this work is a retrospective observational study, so additional prospective studies are necessary to elucidate the potential associations between NAFLD and the prognosis of PCa.

Conclusions
NAFLD is an independent risk factor for BCR in patients with high-grade metastatic PCa. Future studies are needed to understand whether NAFLD is of greater clinical signi cance. Prospective studies with longer follow-up are also needed to validate the nding of the current study.

Consent for publication
Not applicable Availability of data and materials The datasets used during the study could be provided by the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.