http://orcid.org/0000-0002-6383-4163
Figures
Abstract
Introduction
Aldo-keto reductase 1C3 (AKR1C3) is known for multiple functions including its catalytic activity towards producing extra-testicular androgen. The present study is towards understanding interaction between biological, lifestyle and genetic impacts of AKR1C3 and their influence on clinical factors in a prostate cancer (PC) cohort from New Zealand (NZ).
Method
Characteristics of 516 PC patients were collected from the Auckland Regional Urology Facility, NZ. These men were genotyped for the AKR1C3 rs12529 single nucleotide polymorphism (SNP). The leukocyte AKR1C3 activity was measured in a sub-cohort. Variability of leukocyte AKR1C3 activity between biological, lifestyle and clinical features as well as correlation between biological and clinical features were assessed with and without genetic stratification.
Results
The leukocyte AKR1C3 activity was associated with age at diagnosis (0.51 vs 0.34 μM coumberol units for >69y vs ≤69y, P = 0.03); and with anatomic stage/prognostic grouping among the AKR1C3 rs12529 CC genotype carriers (0.50 vs 28 μM coumberol units among low- and high-risk groups respectively, P = 0.02). Significant correlation between leukocyte AKR1C3 activity and age at PC diagnosis was also observed (correlation coefficient 0.20 and P = 0.02). Ever- smoking impacted both age and PSA at PC diagnosis among AKR1C3 rs12529 GG and CG genotype carriers respectively. Age at diagnosis significantly correlated with PSA at diagnosis in the main (correlation coefficient 0.29, and P<0.001) and sub-cohorts (correlation coefficient 0.24, and P = 0.01); and those carrying the AKR1C3 rs12529 CG and GG genotypes in both the main (correlation coefficient 0.30, and P<0.001 and correlation coefficient 0.35, and P<0.001 respectively) and sub-cohorts (correlation coefficient 0.43, and P<0.001 and correlation coefficient 0.39, and P = 0.06 respectively); but not with those carrying the CC genotype.
Citation: Karunasinghe N, Symes E, Gamage A, Wang A, Murray P, Zhu S, et al. (2019) Interaction between leukocyte aldo-keto reductase 1C3 activity, genotypes, biological, lifestyle and clinical features in a prostate cancer cohort from New Zealand. PLoS ONE 14(5): e0217373. https://doi.org/10.1371/journal.pone.0217373
Editor: Chih-Pin Chuu, National Health Research Institutes, TAIWAN
Received: March 27, 2019; Accepted: May 2, 2019; Published: May 24, 2019
Copyright: © 2019 Karunasinghe et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: Authors wish to acknowledge the following grantees from New Zealand for their support with the study since 2006 - the A+ Trust, Auckland District Health Board, Auckland, NZ (Grant No. 5461-PG-1204-007-JM), the GoodFellow Trust, Urology Department, Auckland Hospital, Auckland, NZ (Grant No. +3192-JM), the Cancer Society NZ (https://cancernz.org.nz/) (Grant No. 10/08-LRF); and the Auckland Northland Cancer Society, NZ (https://auckland-northland.cancernz.org.nz/) (NK). Funders played no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.
Competing interests: The authors declare that they have no competing interests.
Abbreviations: AJCC, American Joint Committee on Cancer; AKR1C3, Aldo-keto reductase 1C3; EDTA, Ethylenediaminetetraacetic acid; NZ, New Zealand; PBS, Phosphate buffered saline; PC, Prostate cancer; PCR, Polymerase chain reaction; PSA, Prostate-specific antigen; SNP, Single nucleotide polymorphism; TNM, Tumor–Node–Metastasis
Introduction
Prostate Cancer (PC) is the most common non-skin cancer among men in developed countries [1, 2]. In New Zealand (NZ) there were 3199 PC registrations and 607 PC-related deaths in 2012[3]. The status of PC risk varies between individuals in terms of the patient’s lifestyle and biological characteristics [4–6]. For PC management purposes, it is important to differentiate between men carrying indolent cancers from those with high-risk cancers. Routine assessments for PC include the digital rectal examination and the serum prostate-specific antigen (PSA) blood test, and a subsequent biopsy to confirm diagnoses. However, PC screening with PSA is currently debated due to its low specificity [7]. According to Merriel et al 2018 evaluations on performance of the PSA-based screening for diagnosis of both asymptomatic and symptomatic PC stands equivocal [8]. There are no specific serum PSA levels that are defined as normal/abnormal for men in any racial or ethnic group [2].
The serine protein PSA is produced subsequent to the androgen receptor (AR)-regulated expression of the PSA (kallkirein-related peptidase 3 (KLK3)) gene, while action of androgens are mediated by the AR [9]. Meanwhile, AKR1C3 is among the genes that show increased expression in advancing PC tissue [10]. The AKR1C3 enzyme is involved in reducing many aldehyde and ketone groups to their respective alcohols and is also known to catalyze the production of prostaglandins and extra-testicular androgens, including testosterone and dihydrotestosterone (DHT) [11]. AKR1C3 inhibitors have been shown to reduce PC cell growth in both in vivo and in vitro models, and making castrate resistant PC cell lines more sensitive to the AR antagonist enzalutamide treatment [12]. Recent research has shown a negative association with serum PSA levels in men carrying the G allele of the AKR1C3 rs12529 C>G polymorphism in exon 1 of the AKR1C3 gene which leads to a histidine to glutamine change in the AKR1C3 protein [4]. Meanwhile, allele C of this gene has been associated with increased PC-specific mortality among patients treated with androgen deprivation therapy (ADT) [13], while the same allele has been associated with increased cancer-specific symptoms if not treated with ADT [14].This same AKR1C3 rs12529 C allele has been associated with an increased risk of high-risk PC [5]. Using AKR1C3 promoter constructs of the HepG2 cells, it has been shown that the promoter activity of this gene is associated with promoter the SNP rs3763676,with allele A leading to a 2.2 fold increased activity when treated with DHT in comparison to allele G [15]. The rs11551177 SNP A>G in exon 2 of the AKR1C3 gene, leads to a glutamic acid to glycine change, and the G allele is associated with lower serum testosterone levels [16]. Increased testosterone and DHT dependant transactivation of AR has been recorded in in vitro cell cultures transfected with the AKR1C3 gene [17]. AKR1C3 is also found to be supporting the stability of factors that regulate AR activity [18]. It is reported that the TMPRSS2-ERG fusion protein, which is well known for its expression in 70% of PCs [19], drives AKR1C3 expression by binding to the AKR1C3 gene promoter region in PC cell cultures [10]. Our recent comparison of high-risk PC diagnosis pattern between US and NZ cohorts has shown a disparity between the two centres, especially among men with an ever-tobacco smoking lifestyle and carrying one or two G alleles on the AKR1C3 rs12529 polymorphism [20]. We are of the view that delayed diagnosis of high-risk PC among NZ ever-smokers carrying the AKR1C3 rs12529 G allele is due to its association with lower PSA levels compounded by lower PSA screening in NZ. Therefore, the PSA screening debate [7] could at least partially be due to variation of AKR1C3 activity among individuals, leading to difference in levels of extra-testicular androgen production.
AKR1C3 protein expression has been recorded in many human tissues including the leukocytes [21, 22]. The current study therefore aims to understand the interaction between AKR1C3 activity, AKR1C3 rs12529 SNP genotypes, biological and lifestyle features, and clinical factors in a PC patient cohort from NZ.
Material and methods
Patient recruitment and data collection
The patient cohort considered here was from the ‘Genomic studies on Prostate Cancer’ study (ethics reference NTY/05/06/037), carried out at the University of Auckland in collaboration with the Urology Department, Auckland City hospital. The recruitment process involved inviting men of any ethnicity with positive biopsies for PC from the Auckland Regional Urology Registry (Auckland, Middlemore, and North Shore hospitals). Recruitment was carried out at the Green Lane Outpatient’s Clinic, in Green Lane, Auckland, the Manukau Super Clinic in Manurewa, and the North Shore hospital in Takapuna. Recruitment was restricted to men between 45–90 years of age attending the clinics for follow up before or after the surgery, hormonal or radiation therapy, chemotherapy, or those on active surveillance or watchful waiting. Patient recruitment took place between October 2006 and December 2013. Initially patients were recruited within one year of diagnosis, if they had not undergone any treatment for PC other than radical prostatectomy (RP). In 2008, the criterion was relaxed to include all patients with malignancies but within one year of diagnosis. In September 2010, the timeframe for recruitment was altogether removed. A total of 517 men were recruited from NZ to the study from which a sub-cohort of 155 recruited between September 2010 and December 2013 was considered for leukocyte AKR1C3 activity measurement. Patient recruitment from NZ was carried out with informed written consent under the Northern B (former Northern Y) ethics approval NTY/05/06/037. Clinical and pathology records of patients were evaluated at the hospital databases to collect age and PSA level at diagnosis, Gleason grade and disease stage [tumor-node-metastasis (TNM)].
Subsequently, patients were further stratified based on the disease stage/prognostic grouping followed the criteria defined by the 7th edition of the American Joint Committee on Cancer (AJCC) abbreviated as I, IIA,IIB,III and IV as mentioned previously [20]. D’Amico et al retrospectively monitored a PC patient cohort that had undergone RP, and radiation implant with or without neoadjuvant ADT towards an outcome measure of PSA failure [23]. Based on these outcome measures, these authors were the first to stratify a high-risk PC patient category as those having a clinical tumour stage ≥T2C, or a PSA level of >20ng/ml, or a Gleason grade of ≥8 (equivalent to ≥ Stage IIB).
SNP genotyping
At recruitment, patients provided a blood sample in an Ethylenediaminetetraacetic acid (EDTA) Vacutainer tube from Beckton Dickinson. DNA was extracted from 300μl of EDTA blood using a QIAamp genomic DNA kit (Qiagen, Hilden, Germany) following the manufacturers’ protocol with the aid of a fully automated QIAcube (Qiagen, Hilden, Germany). SNP genotyping was carried out using either the Sequenom MassArray system (Sequenom, San Diego, CA, USA) as described in Ferguson et al. 2010 [24] and subsequently the TaqMan SNP Genotyping Assay from Applied Biosystem (AB) using AB 7900 Real-Time PCR system [14, 24].
Leukocyte AKR1C3 activity measurement
AKR1C3 activity measurement was based on a fluorometric coumberol assay adapted from Jamieson et al 2014 [21]. Coumberol (SN32425) used for the standard measures and the coumberone substrate (SN32424) and the AKR1C3 inhibitor (SN34037) were kindly provided by the in-house team at the Auckland Cancer Society Research Centre (ACSRC). All stock solutions for the coumberone substrate, the AKR1C3 inhibitor and coumberol were dissolved in DMSO, while coumberol standards were further diluted in PBS. Aliquots of an in-house produced AKR1C3 plasmid inserted LNCaP PC cell line (LNCaP+ cell line) was used as a positive AKR1C3 control at 1 million cells per ml concentration. This LNCaP+ cell line has been produced following protocols from Guise et al 2010 [25], to insert a plasmid encoding sequence-confirmed open reading frame for AKR1C3, which has been kindly provided by the AKR1C3 team. As the LNCaP prostate cancer cell line is known to have undetectable levels of AKR1C3 activity [26], 1 million cells per ml concentration of the LNCaP cell line was used as a negative control in each assay plate. These cells grown to confluence were stored in freeze media containing 75% phenol red free Minimum Essential Medium α from Gibco (Cat#41061–029), 15% sterile filtered fetal calf serum (from Moregate Australia & NZ) and 10% dimethyl sulfoxide (ECP Analytical reagent) in -80°C freezer until used.
A total of 155 blood samples collected with heparin anti-coagulant at recruitment were assayed for AKR1C3 activity. These samples were stored frozen at -80°C for up to 8 years. Blood samples were thawed at room temperature and immediately transferred onto an ice bath. These samples were mixed both by pipetting and vortexing with 4ml of cold phosphate buffered saline (PBS) pH 7.4, containing 2mM EDTA and 5% fetal calf serum to avoid clumping of the final cell pellet. This mixture was centrifuged at 1500 x g for 5min at 4°C. The supernatant was removed and the resultant cell pellet was washed twice more with PBS pH 7.4, containing 2mM EDTA and 5% fetal calf serum. Each cell pellet was mixed with 750μl of cold phenol red free Minimum Essential Medium α from Gibco (Cat#41061–029) and 180μl of sample was loaded in to four wells each on a black solid bottom Nunc F96 MicroWell Plate (Cat.# 137101). Positive and negative controls were also loaded to four wells each. To two of the wells with sample or positive/negative controls, the AKR1C3 inhibitor was added to provide a final concentration of 10μM in 200μl final volume. The plates were placed in a plate shaker (Eppendorf Thermomixer C) for 5sec at 300 rpm and were incubated at 37°C for 1 hr in the dark before the coumberone substrate was added to give a final concentration of 10μM in final 200μl volume in each well. The plates were placed in a plate shaker for 5sec at 300 rpm and incubated at 37°C overnight in the dark. 200μl aliquots of coumberol standards containing 0.5μM, 0.25μM, 0.125μM, 0.0625μM, 0.0312μM, 0.0156μM and blank PBS were made fresh prior to fluorescence measurements and were loaded in the standards wells in duplicate.
The fluorescence was measured at 390nm excitation and 510nm emission using the PelkinElmer Enspire Multimode Reader (PerkinElmer, Inc. USA) at room temperature, and the AKR1C3 activity was estimated as the amount of coumberol produced in wells with and without AKR1C3 inhibition. The difference in coumberol produced in wells with and without AKR1C3 inhibitor was considered as due to the total AKR1C3 activity produced by the leukocytes. Using a cell counting protocol, trypan blue stained cell suspension (10μl trypan blue: 10μl cell suspension) was assessed using Ventriplast 10 chamber counting slides, Cat.# 211710 to estimate the number of live cells per 1μl of cell suspension. Samples with less than 10,000 cell per 1ml were not considered in the final activity assessment. Therefore 14% of the initial samples assessed for AKR1C3 activity measurements were removed from analysis. Estimated coumberol produced per million leukocytes were considered equivalent to the AKR1C3 activity in tested samples. LNCaP cell line produced no detectable AKR1C3 activity in this assay. The median AKR1C3 activity in the LNCaP+ cell line was 0.13μM coumberol per million cells with 25th and 75th percentiles at 0.10 and 0.15. The coefficient of variation for inter-plate assay was 23.3% while that for the intra-plate assay was 13.8%
Statistical analysis
In this analysis, those with a current or past tobacco smoking lifestyle were considered as ever-smokers, while the others were considered as never-smokers. Patients receiving luteinizing hormone agonists or anti-androgens were classified under ADT regardless of whether the treatment was short- (< six months) or long-term (≥six months). Brachytherapy on its own or with other radiation treatment were considered under radiation therapy (RT). Any type of prostatectomy (radical or robot assisted) was considered under RP. Active surveillance and watchful waiting categories were categorised per standard nomenclature.
Continuous variables were compared using the Kruskal-Wallis One Way Analysis of Variance on Ranks test as most data types were not normally distributed. Measurements for non-normally distributed data were provided as medians and 25% and 75% points. The Spearman Rank Order Correlation was used to analyse the non-linear correlation between continuous data sets. Categorical variables were tested with the Chi Square test. Trend lines for age and PSA at diagnosis was derived from the Polynomial, Cubic function f = y0+a*x+b*x^2+c*x^3. All statistical analysis were performed using SigmaPlot version 14.0 (Systat Software Inc.).
Results
Patient characteristics
General characteristics and clinical details of the main PC cohort and the cohort selected for the AKR1C3 activity assessment are given in Table 1.The main and sub-cohorts consisted of >96% Caucasians. Ethnicity data, BMI (at study entry), tobacco smoking lifestyle and alcohol consumption frequencies between the main and sub-cohorts are comparable. However, the sub-cohort was significantly older than the main cohort (median age 69y vs 66.6y, P<0.001). The clinical characteristics showed that the sub-cohort had a higher median PSA at prostate cancer diagnosis as compared to the main cohort (10.6ng/ml vs 8.6ng/ml, P = 0.02); and with a higher frequency of those with Gleason sum ≥8 compared to the main cohort (31.6% vs 20.9%, P = 0.02). TNM staging data was not available for 35.8% and 22.6% of the main and sub-cohorts respectively. According to the available data, the sub-cohort had a lower frequency of patients with ≥T2C (30.8% compared to the main cohort with 37.8%, P<0.001). However, the anatomic stage/prognostic grouping was comparable between the sub- and main cohorts with high-risk ≥IIB category being 45.9% and 49.1% of the respective cohorts (P = 0.48). Due to missing data with a TNM classification, further analyses with data stratified based on TNM was not tested. PC management options were also significantly different between the main and sub-cohorts (P<0.001). The sub-cohort recorded ADT as the main therapy (54.9%), while the main cohort had RP as the most common management option (54.6%). The time lag between recruitment and diagnosis was similar between the sub- and the main cohort (median 1y and 25th and 75th percentile being at 0 and 2y respectively for both groups, P = 0.73).
Genetic data
Genotype data recorded for the AKR1C3 rs12529 SNP genotype (Table 2) for 380 men in the main cohort showed that genotype and allele frequency data were within the Hardy-Weinberg equilibrium.
AKR1C3 activity variation in the sub-cohort
The median leukocyte AKR1C3 activity level was 0.41 (25th and 75th percentiles at 0.21 and 0.73 respectively). When data were stratified between the AKR1C3 rs12529 SNP genotypes, no statistically significant difference was seen between genotypes, except for the heterozygous genotype recording 24% higher median activity compared to the CC genotype and 19.8% higher median activity compared to the GG genotype (Table 3). The median PSA level at prostate cancer diagnosis in this sub-cohort also showed no statistically significant difference between genotypes (Table 3).
AKR1C3 activity data were further stratified between median age at diagnosis ≤69 and >69y, BMI>25 and ≤25, never- and ever-smokers, alcohol consumers and non-alcohol consumers (Table 4). Those with age at diagnosis at ≥69y showed a significantly higher leukocyte AKR1C3 activity compared to those at <69y (0.51 [25th and 75th percentiles at 0.25 and 0.78] vs 0.34 [25th and 75th percentiles at 0.17 and 0.65], P = 0.03), however, this variation was not significant after stratifying between the AKR1C3 rs12529 genotypes. However, none of the other features (BMI, alcohol consumption, and tobacco smoking behavior) showed any significant variation in AKRIC3 activity both with and without the AKR1C3 rs12529 genotype stratification (Table 4).
This sub-cohort was further stratified between PSA level at diagnosis ≤20 or >20, Gleason sum <8 and ≥8, PC anatomic stage/prognostic group <IIB and ≥IIB, those receiving no ADT vs those receiving ADT, and those receiving RP or no RP, both with and without further stratification, based on the AKR1C3 rs12529 genotype (Table 5). AKR1C3 activity was marginally higher among those carrying the AKR1C3 rs12520 GG genotype and having PSA at diagnosis >20ng/ml compared to those with ≤20ng/ml group (0.55 [25th and 75th percentiles at 0.28 and 0.75] vs 0.25 [25th and 75th percentiles at 0.16 and 0.51], P = 0.06). Those with the AKR1C3 rs12529 CC genotype recorded a lower median AKR1C3 activity among those with high-risk PC (≥Stage IIB) group compared to those with low-risk (stage <IIB) group (0.28 [25th and 75th percentiles at 0.14 and 0.49] vs 0.50 [25th and 75th percentiles at 0.27 and 1.03], P = 0.02). None of the other features showed any significant variation in AKRIC3 activity.
Clinical characteristics of the main cohort
Median age at diagnosis was marginally higher among ever-smokers compared to never-smokers (68y [25th and 75th percentiles at 62 and 72y] vs 65y [25th and 75th percentiles at 59 and 71y, P = 0.05)] (Table 6). When stratified by genotypes, median age at PC diagnosis were 68.0y [25th and 75th percentiles at 61, 73], 67.5y [25th and 75th percentiles at 62, 72] and 66.0y [25th and 75th percentiles at 62.5, 71] respectively among men with the AKR1C3 rs12529 CC, CG and GG genotypes (P = 0.76) (Table 6). Age at diagnosis was not different between those with different smoking status among the AKR1C3 rs12529 CC and CG genotypes either. However, for those with the GG genotype, a significantly higher age at diagnosis was recorded in ever- smokers compared to never-smokers (68 [25th and 75th percentiles at 64, 72] vs 64.5 [25th and 75th percentiles at 57.8, 68] y, P = 0.01). PSA levels at PC diagnosis recorded for the main cohort showed no significant difference between the AKR1C3 rs12529 genotypes (Table 6) (CC = 9.6 [25th and 75th percentiles at 6.0, 15.6], CG = 8.6 [25th and 75th percentiles at 5.9, 14.6], GG = 9.3 [25th and 75th percentiles at 6.0, 15.9], P = 0.22). There was no significant difference in median PSA level between never- (7.5 [25th and 75th percentiles at 5.5, 13.6]) and ever- smokers (9.1 [25th and 75th percentiles at 6.3, 17.0], P = 0.22) in the main cohort. When data were stratified by both smoking status and genotype, those with the CG genotype recorded a significantly higher median PSA level at diagnosis among ever-smokers (8.9 [25th and 75th percentiles at 6.2, 16.7] compared to never-smokers (7.2 [25th and 75th percentiles at 5.8, 12.2]), P = 0.03). Although the AKR1C3 rs12529 GG genotype carriers recorded a higher PSA level at diagnosis among ever-smokers (11.4 [25th and 75th percentiles at 7.5, 20.0]) compared to never-smokers 6.4 [25th and 75th percentiles at 5.2, 11.6], this was statistically non-significant (P = 0.50).
Correlation between clinical and non-clinical factors with and without genetic stratification of the sub- and main cohorts
Increasing age was negatively correlated with BMI in the sub-cohort (correlation coefficient -0.30 and P<0.01) as well as when stratified by the AKR1C3 rs12529 CC (correlation coefficient -0.32, P = 0.04) and CG (correlation coefficient -0.33 and P = 0.01) genotypes (Table 7). In the main cohort, a significant correlation between age at diagnosis and BMI was not seen (Table 8). Age at diagnosis of the sub-cohort showed a correlation with AKR1C3 activity (correlation coefficient 0.20 and P = 0.02), PSA level at PC diagnosis (correlation coefficient 0.24 and P = 0.01) and Gleason sum (correlation coefficient 0.22 and P = 0.01) (Table 7). No correlation between the AKR1C3 activity and age at PC diagnosis was observed among stratified genotypes except for the AKR1C3 rs12529 GG genotype showing a non-significant marginal correlation (correlation coefficient = 0.35 and P = 0.09). The AKR1C3 rs12529 CG genotype group of the sub-cohort showed a significant correlation between age and PSA at diagnosis (correlation coefficient 0.43 and P<0.01) (Table 7). The AKR1C3 rs12529 GG genotype group of the sub-cohort showed a correlation trend between age and PSA at diagnosis (correlation coefficient 0.39 and P = 0.06). A significant correlation between the age and PSA at PC diagnosis was reproduced in the main cohort (correlation coefficient = 0.29, P<0.01) and the CG (correlation coefficient = 0.30, P<0.01) and GG genotypes (correlation coefficient = 0.35, P<0.01) (Table 8). As in the sub-cohort age and PSA at diagnosis did not correlate in the AKR1C3 rs12529 CC genotype group. Trend lines for the variation between age and PSA at diagnosis stratified by the AKR1C3 rs12529 genotype groups in the main cohort are shown in Fig 1A–1C. Those with the AKR1C3 rs12529 CC genotype, a concentration of correlation points towards lower PSA levels around ages 65-75y was noted with the trend line. For the AKR1C3 rs12529 CG and GG genotypes, a trend of increase is seen from 60y up to around 80y. Age at diagnosis also showed a correlation with the Gleason sum in the AKR1C3 rs12529 CG genotype (correlation coefficient = 0.46 and P<0.01) in the sub-cohort. This was not reproduced in the main cohort, although in the main cohort, age at diagnosis and Gleason sum showed a marginal correlation within the GG genotype group (correlation coefficient 0.21 and P = 0.05) (Table 8). BMI and AKR1C3 activity showed no correlation with either PSA at diagnosis or the Gleason sum in the sub-cohort (Table 7). PSA at diagnosis showed a correlation with the Gleason sum in the sub-cohort (correlation coefficient 0.46 and P<0.01) as well as in the main cohort (correlation coefficient 0.27 and P<0.01) (Tables 7 and 8). Such significant correlations between PSA at diagnosis and Gleason sum were seen among those with the AKR1C3 rs12529 CC and CG genotypes (correlation coefficient 0.48 and P<0.01 and correlation coefficient 0.45 and P<0.01 respectively) in the sub- cohort (Table 7). The correlation between PSA at diagnosis and Gleason sum was reproduced in the main cohort (correlation coefficient = 0.27 and P<0.01) and also remained when stratified by CC and CG genotypes (correlation coefficient 0.23 and P = 0.01 and correlation coefficient 0.29 and P<0.01 respectively) in the main cohort (Table 8). In addition in the main cohort the AKR1C3 rs12529 GG genotype group also showed a correlation between PSA at diagnosis and Gleason sum (correlation coefficient 0.25 and P = 0.02).
(a—CC genotype, b—CG genotype and c—GG genotype).
Statistics are given in the order of correlation coefficient, P value and the number assessed under each comparison.
Statistics are given in the order of correlation coefficient, P value and the number assessed under each comparison.
Discussion
This study presents an analysis of interaction between AKR1C3 activity, AKR1C3 rs12529 genotypes, biological and clinical features in a PC cohort from Auckland, NZ. The leukocyte AKR1C3 activity was measured only in a sub-cohort. Due to the sub-cohort showing significantly higher age and PSA at PC diagnosis as well as a higher PC severity as shown by Gleason sum data, the former is not entirely representing the main cohort. The sub-cohort and those with the AKR1C3 rs12529 CC and CG genotypes within the sub-cohort showed a negative correlation of BMI with age at PC diagnosis, although this was not seen in the main cohort except for a marginal negative trend shown among those carrying the AKR1C3 rs12529 CC genotype in the main cohort. As age dependent BMI change is not directly relevant to the theme of this manuscript, it is considered towards the end of the discussion.
Genotype frequency
The frequencies between the AKR1C3 rs12529 genotypes are similar to that of the European Americans and African Americans recorded before [20, 28]. The frequency of the AKR1C3 rs12529 G allele in our cohort was 0.463 which is marginally lower than the frequencies recorded in the Genome Aggregation Database (0.499); but similar to the Trans-Omics for Precision Medicine Database (0.476); and higher than the TWINSUK Database (0.385) and Avon Longitudinal Study of parents and Children Database from the University of Bristol (0.382) (https://www.ncbi.nlm.nih.gov/snp/rs12529). According to a study from Taiwan, the AKR1C3 rs12529 G was the major allele (G = 0.988) in an Asian cohort [13].
AKR1C3 activity
The median AKR1C3 activity measured in leukocytes was 0.41μM coumberol per million cells (25th and 75th percentiles at 0.21 and 0.73 respectively) and was approximately three times the level produced by the positive control LNCaP+ cell line (0.13μM coumberol per million cells, and 25th and 75th percentiles at 0.10 and 0.15 respectively).To our knowledge, this is the first recording of the AKR1C3 activity measurements made in leukocytes extracted from -80°C stored blood samples as well as in leukocytes from PC patients. When stratified by median age at PC diagnosis, men in ≥69y age group showed higher AKR1C3 activity compared to those at <69y. This is the first ever age based variation in leukocyte AKR1C3 activity recorded in men. However, when data were further stratified for the AKR1C3 rs12529 genotype, none of the genotypes showed significant variation of AKR1C3 activity based on this age stratification. In this sub-cohort, AKR1C3 activity in leukocytes showed no variability with lifestyle factors (tobacco smoking, alcohol consumption and BMI). The AKR1C3 promoter constructs of the HepG2 cells have previously shown that the promoter activity of this gene is associated with promoter SNP rs3763676 [15]. However, such information is not yet available for the AKR1C3 rs12529 SNP.
Those with anatomic stage/prognostic group <IIB showed a higher AKR1C3 activity compared to those with ≥IIB group, when data were stratified by the AKR1C3 rs12529 CC genotype. However, previous studies on AKR1C3 expression in PC tissue or PC cell lines have shown higher levels with PC progression and severity [29]. It is a possibility that the increase in AKR1C3 levels with disease severity are either confined to PC tissue/cell lines or such hierarchical increases are delegated to PC tissue at the expense of AKR1C3 activity elsewhere such as is seen in the leukocytes. If the latter is the case, the reduced levels in AKR1C3 activity in leukocytes among men with the AKR1C3 rs12529 CC genotype may be a reflection of increased levels of AKR1C3 in PC tissue. However, this has to be verified in future studies. Meanwhile, hierarchy in AKR1C3 production between various tissue types has been recorded before [30]. It has been shown that castration resistance is associated with increased expression of genes including that of AKR1C3 [17]. In the current analysis, AKR1C3 activity in leukocytes of men stratified between those who have received ADT and those managed without ADT showed no variation. It could be either due to AKR1C3 over-expression with castration by ADTs being limited to PC cells or due to such increases being limited to castration resistant PC patients. However, as these men were recruited from Urology clinics, there was only 1.8% and 0.8% of patients from the main and sub-cohorts recording castrate -resistant PC. Due to the small sample size used for the leukocyte AKR1C3 activity measurement, the ADT group included both short- and long-term treatment groups and thereby, diluting the possible long-term ADT impacts on AKR1C3 activity. AKR1C3 is also known to be produced in subcutaneous fat deposits, especially in obese women and those with polycystic ovarian syndrome, and considered to be a factor towards intra-adipose testosterone and DHT [31]. However, as men in this sub-cohort showed a decrease in BMI with increasing age, which could be associated with a decrease in fat deposits, AKR1C3 produced in subcutaneous fat compartments would have diminished with age. It is interesting to know in the future whether this decline in subcutaneous fat-based AKR1C3 production is compensated by the production in leukocyte based AKR1C3, with increasing age. An interesting observation made in this analysis is that the leukocyte AKR1C3 activity was significantly associated with age in the sub-cohort. Genetic stratification shows that this trend in age at diagnosis and leukocyte AKR1C3 activity correlation is seen only in men with the AKR1C3 rs12529 GG genotype (correlation coefficient 0.35, P = 0.09). It is a possibility that with increasing age, men with the AKR1C3 rs12529 GG genotype have a potential to produce higher AKR1C3 activity compared to those with other genotypes, and subsequently support a higher proportion of extra-testicular androgen production that increases the androgen pool.
Clinical parameters and AKR1C3 rs12529 genotypes
In the main cohort ever-smokers showed a marginally significant increase in age at diagnosis compared to never-smokers. When this feature was stratified by genotypes, it is the AKR1C3 rs12529 GG genotype which shows a prominent increase in age at diagnosis among ever-smokers compared to never- smokers. In the main cohort, PSA at diagnosis was not significantly different between ever- and never-smokers. However, when stratified by genotypes it was only the AKR1C3 rs12529 CG genotype that showed a significantly higher PSA at diagnosis among ever-smokers compared to never-smokers. Involvement of AKR1C3 in metabolizing polycyclic aromatic hydrocarbons (PAH) leading to the formation of pro-reactive oxygen species such as catechols and quinones have been documented [32–34]. Meanwhile, Lan et al 2004 have recorded that the AKR1C3 rs12529 GG genotype was more susceptible to lung cancer risk in those exposed to smoke derived from coal burning [35].Tobacco smoke is known for its PAH content and the tobacco smoke exposure-related PAH kinetics have been studied in humans previously [36]. Impacts of a 3 mg/kg dose of tobacco smoke constituent bezo-a pyrene given on five occasions within a 26 day period to male Tilapia fish is reported by Colli-Dula et al 2018 [37]. The authors report a gene ontology analysis of PAH effects and report changes including that of AR to PSA signalling pathway that was shown to get up-regulated in the liver samples (1.35 fold) and decrease in testis samples (1.05 fold). They also show a 1.8 fold increase in glutathione peroxidase transcripts in liver tissue by BaP treatment without any changes in the testes. Rybicki et al 2008 [38] have evaluated the proportion of PAH-DNA adduct levels in both tumour and non-tumour cells from surgical prostate tissue. These authors record that after one year follow up from surgery, there is a transient association in the number of men with biochemical recurrence (BCR) with that of PAH-DNA adduct levels in both tumour and non-tumour tissue. These authors also report that a higher level of adducts in non-tumour tissue compared to tumour tissue led to stronger association with BCR, reflecting an innate ability of the non-tumour tissue to activate carcinogens. It is possible that PC patients carrying the AKR1C3 rs12529 G alleles have a higher impact of tobacco smoking affecting their PSA based PC diagnosis. We have previously recorded delayed diagnosis of high-risk PC in ever-smokers from NZ carrying the AKR1C3 rs12529 G allele when compared to similar cohorts from the US [20]. Elimination of those with the AKR1C3 rs12529 G allele for screen detected PC due to their lower PSA levels is a possibility with a subsequent delayed diagnosis with high-risk PC. Those carrying the AKR1C3 rs12529 CG genotype getting diagnosed at a higher PSA level especially among ever-smokers as reported here could also be due to these men not getting captured earlier with low risk PC at a lower PSA level. For those carrying the AKR1C3 rs12529 G allele and are ever-smokers, their AKR1C3 activity may not be able to catalyse the production of extra-testicular androgens at the same rate as their never-smoker counterparts and those with the CC genotype and are ever-smokers.
In both the sub- and main cohort, the age at diagnosis significantly correlated with the PSA at diagnosis. However, it was shown to be relevant only to those with the AKR1C3 rs12529 CG and GG genotypes and not the CC genotype. It is a possibility that for at least the AKR1C3 rs12529 GG genotype carriers, this may be related to their potential to produce higher levels of AKR1C3 activity with age. This can be also interpreted as those with the AKR1C3 rs12529 G alleles could have significant PCs that go undetected at lower ages due to lower PSA levels. It has been shown that physiological levels of DHT (10nm) treatment in LNCaP PC cell line, causes a 70% reduction in AKR1C3 activity [39] that can be interpreted as lower levels promoting AKR1C3 activity. Therefore, a hypothesis that can be derived from these results is that for those with the AKR1C3 rs12529 CG and GG genotypes, general reduction in androgen around 65-75y [40, 41] promote AKR1C3 activity based extra-testicular androgen production while for the CC genotype carriers, this cannot take place in this age range. If this correlation between age and PSA at diagnosis can be proven in larger cohorts elsewhere, this may prove to be among the solutions to improve age based PSA thresholds for PC screening. If this fact cannot be proven in larger cohorts elsewhere, it could be a NZ specific factor/s. We have previously recorded that our NZ PC cohort were diagnosed at higher age and PSA levels compared to that of African and Caucasian PC cohorts from US [20]. NZ has very specific conditions that might especially impact ever-smokers, including the relatively lower levels of available dietary selenium (Se) when compared to most other regions in the world. Serum Se levels are relatively lower in both PC patients and healthy men in NZ compared to levels in certain other parts of the world [42, 43]. Additionally, both low serum Se and tobacco smoking have come up as risk factors for PC incidence in our studies with NZ men [43] while the latter came up as a risk factor for its high-risk PC forms in our NZ cohort [5]. A systematic review and meta-analysis show that tobacco smoking was associated with PC incidence in the era prior to PSA based screening for PC but not since then [44], meaning that NZ may carry a unique PC risk status with regards to the environmental factors such as tobacco smoking. Our studies with NZ cohorts have also shown that current smokers among men with no diagnosis of PC have lower levels of serum Se compared to never- smokers [45] and that the seleno-enzyme glutathione peroxidase level increases almost two fold when 200μg Se is supplemented for six months particularly in ever-smokers compared to never-smokers in NZ [46].
In both the sub- and main cohorts, PSA at diagnosis and the Gleason sum correlated significantly including when stratified by the AKR1C3 rs12529 CC and CG genotypes. This was also the same with the GG genotype group in the main cohort. A similar PSA at diagnosis and Gleason sum correlation has been reported by Yarney et al 2013 for an African cohort with a mean age at presentation at 65.4y [47]. Thompson et al 2006 have reported PSA level as a predictive factor for high-grade disease (Gleason score ≥ 7) [48]. A study with a cohort of African men with a mean age at diagnosis >70y, Gleason sum has been assessed for linear correlation with PSA at diagnosis [49]. Although these authors record no linear correlation, they have not assessed the non-linear correlation between the Gleason sum and PSA at PC presentation in this group. Correlation coefficients between PSA at diagnosis and Gleason sum were stronger in the current sub-cohort, indicating that this correlation increases with increasing severity of the disease.
BMI and age
BMI increases from young age to middle age due to increasing fat mass and subsequent decrease from middle to old age due to decrease in lean mass are well known [50–52]. Therefore, the decrease in BMI with increasing age in the older sub-cohort is as expected. Additionally, the discrepancy of age based BMI decline between the sub- and the main cohorts can be attributed to variation in recruitment criteria between initial and late stages of the study. The time lag between diagnosis to recruitment being similar in the sub- and main cohorts, we cannot assume that the age at diagnosis dependent BMI decline in the sub-cohort to be associated with this factor. Instead, the age at diagnosis dependent BMI decline could also be associated with the clinical features and PC management options between the sub- and main cohorts. van Londen et al 2008 [53] have monitored body fat and lean mass in a group of PC patients either receiving ADT or not and a group of healthy men, over a period of 24 months. They have shown that % body fat from total body mass increases by approximately 2% after acute ADT (ADT initiation <six months prior to enrolment) and approximately 1% in the chronic ADT (treated with ADT for ≥six months at enrolment) after 24 months. They also report a reduction in the proportion of lean body mass to total body mass by 2% in the acute ADT group and 1% in the chronic ADT group after 24 months of enrolment. In their study the impact on PC patients not receiving ADT were less than 0.5% gain in body fat from total body mass as well and less than 0.5% loss in lean body mass proportion from total body mass, 24 months from enrolment. However, there was no deficit between weight gain by fat mass and decline in lean mass due to ADT. However, BMI of these different groups at enrolment in van Londen et al [53] study has remained comparable. A systematic review also shows an increase in percentage body fat and decrease in percentage lean mass in PC patients treated with ADT [54]. Age-related bone loss is reported in PC patients not receiving ADT [55]. Bone mineral density decline in PC patients receiving ADT is also reported [56, 57]. However, the majority of studies indicate that BMD is negatively associated with BMI, while some showing a ‘U’ shaped relationship between BMI and bone fracture risk [58, 59]. Therefore, it is not possible to relate BMI decline with increasing age in the current sub-cohort as due to ADT.
Conclusions
Although PSA was considered as the gold standard for screening for PC, it has reached a controversial status since 2008 [60, 61]. This is due to both over-diagnosis and over-treatment of men as well as under-diagnosis of others when screen diagnosed with PC. Our assessment shows that age based PSA increase in men carrying PC is limited to those with the AKR1C3 rs12529 CG and GG genotypes. This means for those men carrying the AKR1C3 rs12529 CC genotype (which is 32% of our cohort), an age based PSA increase is not a valid concept. Age dependant PSA correlation was stronger in the sub-cohort that recorded more severe disease. When this sub-cohort was stratified, it was only the AKR1C3 rs12529 CG (correlation coefficient = 0.46, P<0.01) and the GG (0.39, P = 0.06) genotypes that showed correlation between age and PSA at diagnosis. In this sub-cohort and the AKR1C3 rs12529 CG genotype of this sub-cohort, as well as the GG genotype of the main cohort, age dependant correlation of Gleason sum was also significant. It is a possibility that with increasing age, men carrying the AKR1C3 rs12529 G allele has an increasing potential to produce more AKR1C3 activity, thereby adding a higher proportion of adrenal derived extra-testicular androgen to the androgen pool. Such increases could be the reason behind age dependant PSA increases seen in the AKR1C3 rs12529 G allele carriers. The AKR1C3 rs12529 G allele is the major allele in Asian, Maori and Pacific men in our cohort [62]. Therefore, it will be of major importance to come up with a new set of age-based PSA cut-off thresholds for PC screening especially for these men as well as all NZ men carrying the AKR1C3 rs12529 G allele. Among the draw backs in our study are the small sample size used for the leukocyte AKR1C3 measurement; pooling all types of ADT based management options under one category in our assessment and non-availability of TNM staging for 36% and 23% respectively for the main and sub-cohorts. It is possible with a significantly larger sample size, with better patient stratification, a better assessment between genotype based leukocyte AKR1C3 activities can be confirmed and compared against biological, lifestyle and clinical features and pave way for a more stringent genetically stratified PSA-based PC screening.
Supporting information
S1 Table. Summary of details related to batch genotyping for the AKR1C3 rs12529 SNP.
(Details provided as a requirement of the Standard Strengthening the Reporting of Genetic Association Studies (STREGA)–An Extension of the STROBE Statement.).
https://doi.org/10.1371/journal.pone.0217373.s001
(DOCX)
S2 Table. Supplementary microsoft office excel data sheet.
https://doi.org/10.1371/journal.pone.0217373.s002
(XLSX)
Acknowledgments
We wish to thank Associate Prof. Jeff Smaill from the ACSRC for providing the coumberol (SN32425) and the coumberone substrate (SN32424) and Dr. Adrian Blaser for providing the AKR1C3 inhibitor (SN34037) for the AKR1C3 analysis. We also wish to thank Associate Prof. Adam Patterson from the ACSRC for providing the AKR1C3 plasmid and Dr.Chris Guise from the ACSRC for overlooking the AKR1C3 plasmid insertion into the LNCaP prostate cancer cell line. Contributions made for the initial genotyping made by Dr. Katja Lange is also acknowledged.
References
- 1. Ferlay J, Shin HR, Bray F, Forman D, Mathers C, Parkin DM. Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008. Int J Cancer, 2010. 127(12): p. 2893–917. pmid:21351269
- 2. Attard G, Parker C, Eeles RA, Schroder F, Tomlins SA, Tannock I, et al. Prostate cancer. Lancet, 2016. 387(10013): p. 70–82. pmid:26074382
- 3.
Ministry-of-Health, Cancer: New registrations and deaths 2012. 2015: Ministry of Health. Wellington, New Zealand.
- 4. Karunasinghe N, Lange K, Han DY, Goudie M, Zhu S, Wang AH, et al. Androgen pathway related gene variants and prostate cancer association in Auckland men Curr Pharmacogenomics Person Med, 2013. 11 (1): p. 22–30.
- 5. Vaidyanathan V, Naidu V, Kao CH-J, Karunasinghe N, Bishop KS, Wang A, et al. Environmental factors and risk of aggressive prostate cancer among a population of New Zealand men—a genotypic approach. Molecular BioSystems 2017, 2017.
- 6. Vaidyanathan V, Naidu V, Karunasinghe N, Kao CH, Pallati R, Jabed A, et al. Effect of ageing and single nucleotide polymorphisms associated with the risk of aggressive prostate cancer in a New Zealand population. Mol Biosyst, 2017.
- 7. Fleshner K, Carlsson SV, and Roobol MJ. The effect of the USPSTF PSA screening recommendation on prostate cancer incidence patterns in the USA. Nat Rev Urol, 2017. 14(1): p. 26–37. pmid:27995937
- 8. Merriel SWD, Funston G, and Hamilton W. Prostate Cancer in Primary Care. Adv Ther, 2018. 35(9): p. 1285–1294. pmid:30097885
- 9. Kim J. and Coetzee GA. Prostate specific antigen gene regulation by androgen receptor. J Cell Biochem, 2004. 93(2): p. 233–41. pmid:15368351
- 10. Powell K, Semaan L, Conley-LaComb MK, Asangani I, Wu YM, Ginsburg KB, et al. ERG/AKR1C3/AR Constitutes a Feed-Forward Loop for AR Signaling in Prostate Cancer Cells. Clin Cancer Res, 2015. 21(11): p. 2569–79. pmid:25754347
- 11. Ishizaki F, Nishiyama T, Kawasaki T, Miyashiro Y, Hara N,Takizawa I, et al. Androgen deprivation promotes intratumoral synthesis of dihydrotestosterone from androgen metabolites in prostate cancer. Sci Rep, 2013. 3: p. 1528. pmid:23524847
- 12. Verma K, Gupta N, Zang T, Wangtrakluldee P, Srivastava SK, Penning TM et al. AKR1C3 Inhibitor KV-37 Exhibits Antineoplastic Effects and Potentiates Enzalutamide in Combination Therapy in Prostate Adenocarcinoma Cells. Mol Cancer Ther, 2018. 17(9): p. 1833–1845. pmid:29891491
- 13. Yu CC, Huang SP, Lee YC, Huang CY, Liu CC, Hour TC, et al. Molecular markers in sex hormone pathway genes associated with the efficacy of androgen-deprivation therapy for prostate cancer. PLoS One, 2013. 8(1): p. e54627. pmid:23359804
- 14. Karunasinghe N, Zhu Y, Han DY, Lange K, Zhu S, Wang A, et al. Quality of life effects of androgen deprivation therapy in a prostate cancer cohort in New Zealand: can we minimize effects using a stratification based on the aldo-keto reductase family 1, member C3 rs12529 gene polymorphism?. BMC Urology, 2016. 16(1): p. 48. pmid:27485119
- 15. Schulze JJ, Karypidis H, and Ekstrom L. Basal and Regulatory Promoter Studies of the AKR1C3 Gene in Relation to Prostate Cancer. Front Pharmacol, 2012. 3: p. 151. pmid:22888320
- 16. Jakobsson J, Palonek E, Lorentzon M, Ohlsson C, Rane A, Ekstrom L et al. A novel polymorphism in the 17beta-hydroxysteroid dehydrogenase type 5 (aldo-keto reductase 1C3) gene is associated with lower serum testosterone levels in caucasian men. Pharmacogenomics J, 2007. 7(4): p. 282–89. pmid:16983398
- 17. Yepuru M, Wu Z, Kulkarni A, Yin F, Barrett CM, Kim J, et al. Steroidogenic enzyme AKR1C3 is a novel androgen receptor-selective coactivator that promotes prostate cancer growth. Clin Cancer Res, 2013. 19(20): p. 5613–25. pmid:23995860
- 18. Fan L, Peng G, Hussain A, Fazli L, Guns E, Gleave M, et al. The Steroidogenic Enzyme AKR1C3 Regulates Stability of the Ubiquitin Ligase Siah2 in Prostate Cancer Cells. J Biol Chem, 2015. 290(34): p. 20865–79. pmid:26160177
- 19. Mehra R, Tomlins SA, Shen R, Nadeem O, Wang L, Wei JT, et al. Comprehensive assessment of TMPRSS2 and ETS family gene aberrations in clinically localized prostate cancer. Mod Pathol, 2007. 20(5): p. 538–44. pmid:17334343
- 20. Karunasinghe N, Ambs S, Wang A, Tang W, Zhu S, Dorsey TH, et al. Influence of lifestyle and genetic variants in the aldo-keto reductase 1C3 rs12529 polymorphism in high-risk prostate cancer detection variability assessed between US and New Zealand cohorts. PLoS One, 2018. 13(6): p. e0199122. pmid:29920533
- 21. Jamieson SM, Gu Y, Manesh DM, El-Hoss J, Jing D, Mackenzie KL, et al. A novel fluorometric assay for aldo-keto reductase 1C3 predicts metabolic activation of the nitrogen mustard prodrug PR-104A in human leukaemia cells. Biochem Pharmacol, 2014. 88(1): p. 36–45. pmid:24434189
- 22. Karunasinghe N, Masters J, Flanagan JU and Ferguson LR. Influence of Aldo-keto reductase 1C3 in prostate cancer -a mini review. Curr Cancer Drug Targets 10.2174/1568009617666170330115722, 2017.
- 23. D'Amico AV, Whittington R, Malkowicz SB, Schultz D, Blank K, Broderick GA, et al. Biochemical outcome after radical prostatectomy, external beam radiation therapy, or interstitial radiation therapy for clinically localized prostate cancer. JAMA, 1998. 280(11): p. 969–74. pmid:9749478
- 24. Ferguson LR, Han DY, Fraser AG, Huebner C, Lam WJ, Morgan AR, et al. Genetic factors in chronic inflammation: single nucleotide polymorphisms in the STAT-JAK pathway, susceptibility to DNA damage and Crohn's disease in a New Zealand population. Mutat Res, 2010. 690(1–2): p. 108–15. pmid:20109474
- 25. Guise CP, Abbattista MR, Singleton RS, Holford SD, Connolly J, Dachs GU et al. The bioreductive prodrug PR-104A is activated under aerobic conditions by human aldo-keto reductase 1C3. Cancer Res, 2010. 70(4): p. 1573–84. pmid:20145130
- 26. Byrns MC, Mindnich R, Duan L and Penning TM. Overexpression of aldo-keto reductase 1C3 (AKR1C3) in LNCaP cells diverts androgen metabolism towards testosterone resulting in resistance to the 5alpha-reductase inhibitor finasteride. J Steroid Biochem Mol Biol, 2012. 130(1–2): p. 7–15. pmid:22265960
- 27. Bastian PJ, Boorjian SA, Bossi A, Briganti A, Heidenreich A, Freedland SJ, et al. High-risk prostate cancer: from definition to contemporary management. Eur Urol, 2012. 61(6): p. 1096–106. pmid:22386839
- 28. Sausville LN, Jones CC, Aldrich MC, Blot WJ, Pozzi A, Williams SM. Genetic variation in the eicosanoid pathway is associated with non-small-cell lung cancer (NSCLC) survival. PLoS One, 2017. 12(7): p. e0180471. pmid:28704416
- 29. Tian Y, Zhao L, Zhang H, Liu X, Zhao L, Zhao X, et al. AKR1C3 overexpression may serve as a promising biomarker for prostate cancer progression. Diagn Pathol, 2014. 9: p. 42. pmid:24571686
- 30. Nakamura Y, Hornsby PJ, Casson P, Morimoto R, Satoh F, Xing Y, et al. Type 5 17beta-hydroxysteroid dehydrogenase (AKR1C3) contributes to testosterone production in the adrenal reticularis. J Clin Endocrinol Metab, 2009. 94(6): p. 2192–8. pmid:19336506
- 31. O'Reilly MW, Kempegowda P, Walsh M, Taylor AE, Manolopoulos KN, Allwood JW, et al. AKR1C3-Mediated Adipose Androgen Generation Drives Lipotoxicity in Women With Polycystic Ovary Syndrome. J Clin Endocrinol Metab, 2017. 102(9): p. 3327–3339. pmid:28645211
- 32. Flowers L, Bleczinski WF, Burczynski ME, Harvey RG, Penning TM. Disposition and biological activity of benzo[a]pyrene-7,8-dione. A genotoxic metabolite generated by dihydrodiol dehydrogenase. Biochemistry, 1996. 35(42): p. 13664–72. pmid:8885846
- 33. Flowers L, Ohnishi ST and Penning TM. DNA strand scission by polycyclic aromatic hydrocarbon o-quinones: role of reactive oxygen species, Cu(II)/Cu(I) redox cycling, and o-semiquinone anion radicals. Biochemistry, 1997. 36(28): p. 8640–8. pmid:9214311
- 34. Palackal NT, Lee SH, Harvey RG, Blair IA and Penning TM. Activation of polycyclic aromatic hydrocarbon trans-dihydrodiol proximate carcinogens by human aldo-keto reductase (AKR1C) enzymes and their functional overexpression in human lung carcinoma (A549) cells. J Biol Chem, 2002. 277(27): p. 24799–808. pmid:11978787
- 35. Lan Q, Mumford JL, Shen M, Demarini DM, Bonner MR, He X, et al. Oxidative damage-related genes AKR1C3 and OGG1 modulate risks for lung cancer due to exposure to PAH-rich coal combustion emissions. Carcinogenesis, 2004. 25(11): p. 2177–81. pmid:15284179
- 36. St Helen G, Goniewicz ML, Dempsey D, Wilson M, Jacob P 3rd and Benowitz NL. Exposure and kinetics of polycyclic aromatic hydrocarbons (PAHs) in cigarette smokers. Chem Res Toxicol, 2012. 25(4): p. 952–64. pmid:22428611
- 37. Colli-Dula RC, Fang X, Moraga-Amador D, Albornoz-Abud N, Zamora-Bustillos R, Conesa A, et al. Transcriptome analysis reveals novel insights into the response of low-dose benzo(a)pyrene exposure in male tilapia. Aquat Toxicol, 2018. 201: p. 162–173. pmid:29913432
- 38. Rybicki BA, Neslund-Dudas C, Bock CH, Rundle A, Savera AT, Yang JJ, et al. Polycyclic aromatic hydrocarbon—DNA adducts in prostate and biochemical recurrence after prostatectomy. Clin Cancer Res, 2008. 14(3): p. 750–7. pmid:18245535
- 39. Wang JH and Tuohimaa P. Regulation of 17beta-hydroxysteroid dehydrogenase type 2, type 4 and type 5 by calcitriol, LXR agonist and 5alpha-dihydrotestosterone in human prostate cancer cells. J Steroid Biochem Mol Biol, 2007. 107(1–2): p. 100–5. pmid:17627817
- 40. Rubens R, Dhont M and Vermeulen A. Further studies on Leydig cell function in old age. J Clin Endocrinol Metab, 1974. 39(1): p. 40–5. pmid:4600506
- 41. Ricciuti A, Travison TG, Di Dalmazi G, Talor MV, DeVincentiis L, Manley RW et al. A Subset of Men With Age-Related Decline in Testosterone Have Gonadotroph Autoantibodies. J Clin Endocrinol Metab, 2016. 101(4): p. 1535–41. pmid:26963952
- 42. Vogt TM, Ziegler RG, Graubard BI, Swanson CA, Greenberg RS, Schoenberg JB et al. Serum selenium and risk of prostate cancer in U.S. blacks and whites. Int J Cancer, 2003. 103(5): p. 664–70. pmid:12494476
- 43. Karunasinghe N, Han DY, Goudie M, Zhu S, Bishop K, Wang A, et al. Prostate Disease Risk Factors among a New Zealand Cohort. J Nutrigenet Nutrigenomics, 2013. 5(6): p. 339–51.
- 44. Islami F, Moreira DM, Boffetta P and Freedland SJ. A systematic review and meta-analysis of tobacco use and prostate cancer mortality and incidence in prospective cohort studies. Eur Urol, 2014. 66(6): p. 1054–64. pmid:25242554
- 45. Karunasinghe N, Han DY, Zhu S, Yu J, Lange K, Duan H, et al. Serum selenium and single-nucleotide polymorphisms in genes for selenoproteins: relationship to markers of oxidative stress in men from Auckland, New Zealand. Genes Nutr, 2012. 7(2): p. 179–90. pmid:22139612
- 46. Ferguson LR, Karunasinghe N, Zhu S, Han DY, Triggs CM, Wang AH, et al. Understanding Heterogeneity in Supplementation Effects of Selenium in Men: A Study of Stratification Variables and Human Genetics in a Prospective Sample from New Zealand. Current Pharmacogenomics and Personalized Medicine, 2012. 10: p. 204–216.
- 47. Yarney J, Vanderpuye V and Mensah J, Clinicopathologic features and determinants of Gleason score of prostate cancer in Ghanaian men. Urol Oncol, 2013. 31(3): p. 325–30. pmid:21441045
- 48. Thompson IM, Ankerst DP, Chi C, Goodman PJ, Tangen CM, Lucia MS, et al. Assessing prostate cancer risk: results from the Prostate Cancer Prevention Trial. J Natl Cancer Inst, 2006. 98(8): p. 529–34. pmid:16622122
- 49. Nnabugwu II, Udeh EI, Ugwumba FO and Ozoemena FO. Predicting Gleason score using the initial serum total prostate-specific antigen in Black men with symptomatic prostate adenocarcinoma in Nigeria. Clin Interv Aging, 2016. 11: p. 961–6. pmid:27486316
- 50. Hughes VA, Roubenoff R, Wood M, Frontera WR, Evans WJ, Fiatarone Singh MA. Anthropometric assessment of 10-y changes in body composition in the elderly. Am J Clin Nutr, 2004. 80(2): p. 475–82. pmid:15277173
- 51. Malhotra R, Ostbye T, Riley CM and Finkelstein EA. Young adult weight trajectories through midlife by body mass category. Obesity (Silver Spring), 2013. 21(9): p. 1923–34.
- 52. Silventoinen K, Jelenkovic A, Sund R,Yokoyama Y, Hur YM, Cozen W, et al. Differences in genetic and environmental variation in adult BMI by sex, age, time period, and region: an individual-based pooled analysis of 40 twin cohorts. Am J Clin Nutr, 2017. 106(2): p. 457–466. pmid:28679550
- 53. van Londen GJ, Levy ME, Perera S, Nelson JB and Greenspan SL. Body composition changes during androgen deprivation therapy for prostate cancer: a 2-year prospective study. Crit Rev Oncol Hematol, 2008. 68(2): p. 172–7. pmid:18706829
- 54. Haseen F, Murray LJ, Cardwell CR, O'Sullivan JM and Cantwell MM. The effect of androgen deprivation therapy on body composition in men with prostate cancer: systematic review and meta-analysis. J Cancer Surviv, 2010. 4(2): p. 128–39. pmid:20091248
- 55. Conde FA, Sarna L, Oka RK, Vredevoe DL, Rettig MB and Aronson WJ. Age, body mass index, and serum prostate-specific antigen correlate with bone loss in men with prostate cancer not receiving androgen deprivation therapy. Urology, 2004. 64(2): p. 335–40. pmid:15302490
- 56. Wang A, Obertova Z, Brown C, Karunasinghe N, Bishop K, Ferguson L, et al. Risk of fracture in men with prostate cancer on androgen deprivation therapy: a population-based cohort study in New Zealand. BMC Cancer, 2015. 15: p. 837. pmid:26525985
- 57. Kim DK, Lee JY, Kim KJ, Hong N, Kim JW and Hah YS, et al. Effect of Androgen-Deprivation Therapy on Bone Mineral Density in Patients with Prostate Cancer: A Systematic Review and Meta-Analysis. J Clin Med, 2019. 8(1).
- 58. Palermo A, Tuccinardi D, Defeudis G, Watanabe M, D'Onofrio L, Lauria Pantano A, et al. BMI and BMD: The Potential Interplay between Obesity and Bone Fragility. Int J Environ Res Public Health, 2016. 13(6).
- 59. De Laet C, Kanis JA, Oden A, Johanson H, Johnell O, Delmas P, et al. Body mass index as a predictor of fracture risk: a meta-analysis. Osteoporos Int, 2005. 16(11): p. 1330–8. pmid:15928804
- 60. Andriole GL, Crawford ED, Grubb RL 3rd, Buys SS, Chia D, Church TR, et al. Prostate cancer screening in the randomized Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial: mortality results after 13 years of follow-up. J Natl Cancer Inst, 2012. 104(2): p. 125–32. pmid:22228146
- 61. Schroder FH, Hugosson J, Roobol MJ, Tammela TL, Ciatto S, Nelen V, et al. Prostate-cancer mortality at 11 years of follow-up. N Engl J Med, 2012. 366(11): p. 981–90. pmid:22417251
- 62.
Karunasinghe N, Zhu Y, Han DY, Lange K, Wang A, Zhu S, et al. Can we minimize androgen deprivation therapy-related quality of life effects in Māori & Pacific prostate cancer survivors using a genetic stratification?. in The Science of Cancer Health Disparities in Racial/ethnic Minorities and the Medically Underserved. 2015. Atlanta, Georgia GA: American Association of Cancer Research.