Deletions of the Androgen-Metabolizing UGT2B Genes Have an Effect on Circulating Steroid Levels and Biochemical Recurrence after Radical Prostatectomy in Localized Prostate Cancer

Abstract

Context:

The prognostic relevance of inherited variations in hormone-related genes in the context of prostate cancer (PCa) progression has not been well studied. Of these, UDP-glucuronosyltransferase (UGT) gene products lead to inactivation of steroids.

Objective:

Our objective was to determine whether polymorphisms in five UGT genes, involved in steroid metabolism, are associated with the risk of biochemical recurrence after radical prostatectomy (RP) and to examine their relationship with hormonal exposure.

Design:

The study included 526 Caucasian and 320 Asian men who underwent RP for clinically localized PCa. The relationship between genotypes and biochemical recurrence were assessed with multivariate Cox proportional hazard models. Plasma steroids were measured using specific and sensitive mass spectrometry-based methods.

Results:

The presence of at least two deleted copies of UGT2B17 and UGT2B28 genes resulted in a hazard ratio of 2.26 (95% confidence interval = 1.41–3.61; P = 0.0007) for Caucasians and 2.16 (95% confidence interval = 1.24–3.73; P = 0.006) for Asians. A positive association was observed only between UGT2B17 deletion and the Gleason score in Asians, whereas no other interaction was shown with prostate-specific antigen, Gleason score, and TNM (tumor node metastasis) staging. Patients carrying UGT2B17 deletions and those with three deleted UGT2B copies had significantly lower androgen glucuronides, in support of an altered androgen metabolism.

Conclusion:

This study is the first to recognize the prognostic significance of common deletions in steroid inactivation pathways in localized PCa after RP. Alteration of circulating steroid levels associated with UGT2B gene deletions further support the notion that such inherited genomic deletions have the potential to modify hormonal exposure and risk of recurrence.

Prostate cancer (PCa) is the sixth most common cancer in the world and the second leading cause of cancer death in North American men (1). The specific mortality of PCa is directly related to both the stage at time of diagnosis and age at onset (2). Of the several known risk factors, the most significant are age, ethnicity, and genetic and dietary factors (3, 4). In the early stages of carcinogenesis, androgens are centrally implicated in the development and progression of this hormone-dependent cancer. However, an association between plasma androgen levels and PCa susceptibility has not been consistently observed across studies (5). Currently, PCa recurrence risk is assessed based on clinical stage, prostate-specific antigen (PSA) level, and Gleason scores. However, the heterogeneity observed in clinical behavior emphasizes the need to identify additional tools suitable for use in clinical practice, such as molecular biomarkers.

Over the last several decades, a number of studies have addressed the role of inherited variations in hormone-related genes and the associated risk of PCa. However, to our knowledge, few studies have looked at the prognostic value of hormone-related gene polymorphisms in PCa progression and mortality. Lindström and colleagues (6) studied the role of germline variations in androgen receptor (AR) gene and the biosynthetic genes CYP17 and SRD5A2, and suggested that polymorphisms in the AR gene affect hormonal treatment and ultimately PCa-related death. More recently, overexpression of the androgen/estrogen inactivating enzyme HSD17B4 was associated with a poor clinical outcome (7). Despite the fundamental role of androgens in PCa development, few data are available regarding the consequence of genetic variations in steroidogenic pathways with the risk of PCa progression after treatment, particularly for early-stage disease.

UDP-glucuronosyltransferases (UGT) are conjugating enzymes that, in concert with other steroidogenesis enzymes, contribute to the maintenance of intracellular levels of sex-steroid hormones in target cells (8). This enzymatic process generates glucuronide derivatives that are biologically inactive and subsequently excreted in bile or urine (9). UGT are expressed in almost all tissues, including hormone-targeted organs, such as prostate (9, 10). Five UGT isoforms are mainly responsible for the inactivation of androgenic and estrogenic sex steroids (Fig. 1). The UGT1A1 protein isoform is primarily involved in the glucuronidation of estradiol, whereas the UGT2B class of enzymes (namely UGT2B7, UGT2B15, UGT2B17, and UGT2B28) is mainly involved in androgen catabolism in hormone-dependent cells (9). Common polymorphisms in two of these genes, UGT2B15 and UGT2B17, have been associated with PCa risk in some, but not all, studies (11–16). Furthermore, deletions of the UGT2B17 and UGT2B28 genes have been reported, and over 50% of Caucasians and 75% of Asians have a reduced copy number in at least one of these genes (17–19). However, their impact has not yet been addressed in PCa recurrence.

Fig. 1

Reported specificity of UGT enzymes for sex steroids. †, The regioselectivity of UGT2B28 for the glucuronidation of 3α-diol has not been assessed (35). Unpublished data support reactivity of UGT2B28 against estrogens.

In this study, we initially assessed whether functional genetic variations or deletions of the major sex steroid-metabolizing genes UGT1A1, UGT2B7, UGT2B15, UGT2B17, and UGT2B28 are associated with susceptibility to biochemical PSA recurrence after prostatectomy. Because the majority of PCa cases are now diagnosed at a localized stage, two populations of clinically localized PCa patients comprising Caucasian and Asian men were studied; accounting for 846 men followed for biochemical recurrence (BCR). Besides, the rationale for selecting the UGT genes was based on the following concepts: 1) the early phase of prostate carcinogenesis is hormone dependent, and any genes modifying the level of active hormones might have an impact upon cancer recurrence and progression; 2) UGT genes are expressed in target cells where hormone action occurs; 3) polymorphisms in UGT genes are known genetic factors associated with PCa risk; and 4) UGT genes are hormone-regulated genes that are differentially expressed in some cancers. We also established the link between genotypes positively associated with BCR and circulating steroids in plasma available from the Caucasian population.

Findings reveal that UGT2B17 and UGT2B28 gene deletions are predictors of PSA recurrence in clinically localized PCa after surgical treatment. These deletions are linked to altered levels of circulating sex steroids, supporting the hypothesis that reduced androgen glucuronidation by UGT-mediated inactivation would lead to an increased PSA recurrence.

Materials and Methods

Studied cohorts

Clinical characteristics of 846 PCa patients from two independent cohorts are shown in Table 1. The first cohort was composed of 526 Caucasians recruited at the CHUQ-Hôtel-Dieu de Québec hospital (Québec, Canada) between 1999 and 2002, and the second was composed of 320 Asian men as described previously (20). All patients were diagnosed with localized PCa and were followed postoperatively with serial PSA measurements. Detailed clinical information was available from medical records. All participants provided a written consent before surgery for the analysis of their genome. The local research ethical committees approved the research protocol.

Table 1

Clinical and pathological characteristics of study populations

Characteristics	Caucasians (n = 526)	Asians (n = 320)
Age at diagnosis (yr)
Mean	63.3	65.7
sd	6.8	6.6
Range	43.5–80.7	44.0–79.0
Follow-up median (months)	88.8	30.8
BCR (%)	130 (24.7)	116 (36.3)
PSA at diagnosis (ng/ml)
≤10	362 (69)	132 (43)
>10–20	103 (20)	105 (34)
>20	56 (11)	72 (24)
Biopsy Gleason score (%)
≤6	318 (61)	122 (39)
7	140 (27)	149 (48)
≥8	67 (13)	41 (13)
T stage at diagnosis (%)
T1	259 (50)	209 (67)a
T2	247 (48)	95 (30)a
T3	14 (3)	8 (3)a
Nodal invasion (%)
N0	481 (92)	290 (93)
N+	44 (8)	21 (7)
Neoadjuvant hormonotherapy (%)
Yes	31 (6)	0
No	495 (94)	320
Adjuvant hormonotherapy (%)
Yes	30 (6)	0
No	496 (94)	320
Androgen-deprivation resistance (%)	12 (2)	NA

NA, Not available; N, node; T, tumor.

^a

Pathological staging. In Asians, numbers are for pathological staging T2, T3, and T4, respectively.

Definitions

The primary outcome variable was BCR. BCR was defined as the period of time elapsed between the surgical procedure and the date of PSA recurrence. In Asians, PSA recurrence was defined as two consecutive PSA measurements of more than 0.2 ng/ml at an interval of more than 3 months, and PSA levels of more than 0.2 ng/ml at the first follow-up was considered as the time of recurrence (20). In Caucasians, PSA recurrence was defined as two consecutive PSA values of at least 0.3 μg/liter, one PSA value of at least 0.3 μg/liter followed by androgen-deprivation therapy (ADT) or radiation therapy, and a single last-recorded PSA value of at least 0.3 μg/liter after prostatectomy.

Genotyping

Genomic DNA was prepared from peripheral blood mononuclear cells collected from patients on the morning of a preoperative ambulatory clinical visit. All samples were kept frozen at −80 C until the time of study. Genomic DNA was extracted using the QIAamp DNA Blood mini kit (QIAGEN Inc., Mississauga, Ontario, Canada) and stored at −80 C. For the Asian cohort, DNA was prepared as previously described (20).

Genotyping for UGT1A1 (rs34815109), UGT2B15 (rs1902023), and UGT2B7 (rs7439366) were performed by resequencing of PCR amplification products. Briefly, specific PCR amplification products were cleaned up from unused oligonucleotides and sequenced with the Big Dye v3 terminator cycle sequencing kit (Applied Biosystems, Foster City, CA). Sequencing reactions were then resolved onto an ABI 3730xl genetic analyzer (Applied Biosystems) in Caucasians. The genotyping for UGT2B17 and UGT2B28 deletions was performed as previously described (19). Negative controls were included in each run. Quality controls (random replicates of known genotypes) were successfully performed in 5% of the study cohorts.

Measurement of steroid levels

Five hundred plasma samples were available from the Caucasian cohort and collected on the morning of the surgical procedure. Measurement of steroid levels was carried out by validated gas chromatography-mass spectrometry or liquid chromatography-tandem mass spectrometry methods for unconjugated steroids or glucuronides, respectively (21). Internal standards (deuterated steroids) were added to all samples, and quality controls were included in each run. The lower limit of quantification for each metabolite is as follow: testosterone (0.025 ng/ml), dihydrotestosterone (DHT) (0.010 ng/ml), ADT (0.025 ng/ml), ADT-glucuronide (ADT-G) (1 ng/ml), and 3α-diol-3G and 3α-diol-17G (0.25 ng/ml). Coefficients of variation for intra- and interassays for these methods were 10.0% or below, and accuracies for testosterone, DHT, ADT, ADT-G, 3α-diol-3G, and 3α-diol-17G were 100.5, 103.5, 96.5, 99.5, 89.2, and 100.2%, respectively.

Statistical analysis

The null hypothesis of no deviation from the Hardy-Weinberg equilibrium was tested for each polymorphism using a χ² test. Each variant was treated as a categorical variable, with a common (reference) homozygote group, a heterozygote, and homozygote for the minor allele, or combined, when the minor allele frequency was below 0.10. For instance, the UGT2B28-null genotype was combined with heterozygous because of its low frequency (Caucasians, n = 1; Asians, n = 6). Initial PSA level, Gleason score at biopsy, and tumor clinical stages were treated as categorical variables (22).

Kaplan-Meier BCR analysis was performed to compare genotypes using the log-rank test. Polymorphisms positively associated with BCR in univariate analysis (P < 0.05) were then evaluated in a multivariate Cox regression analysis. Covariates of Table 1 were included in the models along with UGT2B17 and UGT2B28 genetic status. The proportional hazard assumption of Cox models was verified graphically by log-minus-log vs. log (time) plots for each variable. Clinical variables were compared across genotypes using the Pearson's χ² test. A P value <0.05 was considered statistically significant, and we considered the two-sided assumption for all tests. Statistical analyses were performed using SAS Statistical Software version 9.2 (SAS Institute, Cary, NC) and using PASW statistics version 17 (SPSS Inc., Chicago, IL). Analysis of covariance was used to compare means of log₁₀-transformed hormonal variables between two or more independent groups, adjusted for age, but only untransformed data are shown to facilitate the interpretation. For comparisons among three or more groups, any significant differences revealed by analysis of covariance were further investigated using post hoc pairwise comparisons.

Results

Overall, 130 patients (24.7%) experienced PSA failure in Caucasians and 116 (36.3%) in Asians. The preoperative PSA values and the pathological biopsy Gleason scores were both associated with BCR. In Caucasians, men with a PSA value over 20 were associated with BCR in a multivariate analysis [hazard ratio (HR) = 2.49; 95% confidence interval (CI) = 1.56–3.97; P = 1 × 10⁻⁴), a Gleason score of at least 8 is associated with a HR of 3.25 (95% CI = 2.05–5.15; P = 5 × 10⁻⁷), whereas clinical tumor stage of at least T2c was not significantly associated (95% CI = 0.68–2.54; P = 0.411), similar to previous reports (Supplemental Table 1, published on The Endocrine Society's Journals Online web site at http://jcem.endojournals.org) (23, 24). Deletion of either UGT2B17 or UGT2B28 gene is significantly associated with BCR (Table 2), whereas no significant association was observed for additional UGT variants tested in Caucasians (Supplemental Table 2). The distribution of genotype frequencies are shown in Supplemental Table 3 and are consistent with previous reports (13, 17, 19).

Table 2

Multivariate Cox proportional hazard models for BCR

Number of deleted gene copies	Overall n (%)	PSA failure n (%)	Multivariate
			HR	95% CI	P
Caucasians (n = 526)
UGT2B17
0	279 (53)	60 (22)	1.00
1	217 (41)	60 (28)	1.47	1.02–2.13	0.040
2	30 (6)	10 (33)	1.96	1.00–3.85	0.051
UGT2B28
0	390 (74)	84 (22)	1.00
1	136 (26)	46 (34)	1.49	1.03–2.14	0.035
UGT2B17 + UGT2B28
0	213 (41)	41 (19)	1.00
1	226 (43)	57 (25)	1.23	0.82–1.86	0.318
≥2	87 (17)	32 (37)	2.26	1.41–3.61	0.0007
≤1	439 (83)	98 (22)	1.00
≥2	87 (17)	32 (37)	2.02	1.34–3.03	0.0007
Asians (n = 320)
UGT2B17
0	51 (18)	17 (33)	1.00
1	53 (18)	16 (30)	0.67	0.31–1.45	0.309
2	185 (64)	65 (35)	1.48	0.83–2.65	0.187
≤1	104 (36)	33 (32)	1.00
2	185 (64)	65 (35)	1.79	1.11–2.90	0.018
UGT2B28
0	200 (71)	67 (34)	1.00
1	80 (29)	25 (31)	0.97	0.59–1.61	0.918
UGT2B17 + UGT2B28
≤1	82 (30)	24 (29)	1.00
≥2	194 (70)	66 (34)	2.16	1.24–3.73	0.006

Multivariate analyses indicated that individuals with one and two deleted alleles of UGT2B17 had an increased likelihood of PSA recurrence with HR values of 1.47 (95% CI = 1.02–2.13; P = 0.04) and 1.96 (95% CI = 1.00–3.85; P = 0.051), respectively, in support of a gene dosage effect (Table 2). For UGT2B28, one deletion was associated with HR values of 1.49 (95% CI = 1.03–2.14; P = 0.035). However, too few individuals with the null genotype for UGT2B28 were found to establish a potential gene-dosage effect for this gene. It should be noted that over 28% of PCa patients with at least one deletion of UGT2B17 were also carriers of one UGT2B28 deletion, whereas 51% of heterozygous cases for UGT2B28 were carriers of at least one deletion of UGT2B17. Thus, HR values likely represent the compound effect of simultaneous UGT2B17 and UGT2B28 genetic deletions.

Individuals of Asian ancestry were then studied because this population presents the highest reported frequencies of UGT2B17 deletion with a frequency of 73%, compared with 27% in Caucasians (Supplemental Table 3) (17). In Asians, 64% (n = 185) of PCa patients had the UGT2B17-null genotype. For UGT2B28, the minor allele frequency was 15%, similar to the frequency of 13% observed in Caucasians. Data analysis supports that deleted copies of UGT2B gene also lead to a higher risk of PSA recurrence after prostatectomy in Asian patients. Carriers of two deleted copies of the UGT2B17 gene had a HR of 1.79 (95% CI = 1.11–2.90; P = 0.018) whereas two or more deletions of UGT2B17 and UGT2B28 resulted in a more significant HR of 2.16 (95% CI = 1.24–3.73; P = 0.006) (Table 2). Figure 2 displays the Kaplan-Meier curves of BCR for no more than one or at least two UGT2B deleted genes. The UGT2B17 gene deletion was associated with a higher Gleason score in Asians, whereas no other significant interaction was noted between gene polymorphisms, PSA, Gleason score, and TNM (tumor node metastasis) staging for both Asians and Caucasians (Supplemental Table 4).

Fig. 2

Kaplan-Meier curves of BCR-free analysis for deletions in UGT2B genes. Log-rank (LR) P values are shown in each frame. BCR-free, Biochemical recurrence-free; RP, radical prostatectomy.

Sex steroid hormone levels and UGT2B genetic deletions in Caucasian PCa patients

We initially evaluated circulating levels of androgen and their glucuronide metabolites in carriers of the null genotype for the UGT2B17 gene and carriers of one UGT2B28 deleted copy and carriers of one UGT2B28 deleted copy (Table 3). PCa patients with the UGT2B17-null genotype had a significant decrease in levels of a main DHT metabolite 3α-diol-17G by 42% (P = 0.0005) and higher levels of androsterone (26% increase; P = 0.015). The same trends were observed for heterozygous UGT2B17 patients. Furthermore, patients with three deleted copies of the UGT2B17/2B28 genes, also presented significantly higher levels of androsterone (39% increase; P = 0.037), lower levels of 3α-diol-17G (40% decrease; P = 0.018) and a trend toward lower testosterone (23% decrease; P = 0.125) (Table 3). As for carriers of the UGT2B28 deletion, no significant alteration in the hormonal profile was observed. It was not possible to study patients with the null UGT2B28 genotype because only one patient was found. Patients receiving ADT were also excluded from these analyses.

Table 3

Circulating steroid levels in Caucasians stratified by UGT2B deletion genotypes

	Deletion genotype stratification
Number of deleted copies
UGT2B17	0	1	2	0	2
UGT2B28	0	0	0	1	1
Number of patients	194	138	16	61	12
Steroids (median ± sd)
Testosterone (pg/ml)	3675.00 ± 1484.51	3710.00 ± 1534.07	3640.00 ± 1478.40	4010.00 ± 1421.85	2830.00 ± 1239.19
DHT (pg/ml)	315.04 ± 149.47	313.29 ± 155.46	288.32 ± 168.23	335.60 ± 156.90	251.62 ± 120.72
Androsterone (pg/ml)	164.49 ± 103.12	188.24 ± 81.72a	207.39 ± 113.38b	173.30 ± 240.77	228.92 ± 112.20b
Androsterone-G (ng/ml)	30.20 ± 41.19	29.00 ± 19.68	32.15 ± 20.98	31.80 ± 20.20	27.80 ± 14.71a
3α-Diol-3G (ng/ml)	1.62 ± 1.60	1.42 ± 0.99b	1.43 ± 1.33	1.49 ± 1.15	1.74 ± 1.63
3α-Diol-17G (ng/ml)	3.61 ± 2.55	3.15 ± 1.79a	2.09 ± 1.63c	3.26 ± 1.89	2.15 ± 2.52b

It was not possible to study the null genotype for UGT2B28 given its very low frequency (n = 1). P values are vs. carriers of no deletion.

^a

P < 0.10.

^b

P < 0.05.

^c

P < 0.001.

Discussion

The purpose of this study was to assess the prognostic significance of common genetic variants in five major sex-steroid-metabolizing UGT gene isoforms in clinically localized PCa patients. To our knowledge, there is no previous study reporting this association. Studying genes involved at the end of the androgenic signal is particularly biologically relevant in the initial phase of disease development because PCa cells are hormone responsive at this stage, and hormonal manipulation is a cornerstone of disease management. We observed that both Caucasian and Asian cancer patients carrying common deletions of the UGT2B17 and UGT2B28 genes have an increased risk of BCR after surgical treatment and that these genotypes are associated with reduced circulating levels of inactive androgen metabolites.

Over the last several decades, several studies have addressed the association between inherited variations in hormone-related genes and the risk of developing PCa (25–27). However, few studies have looked at the prognostic value of hormone-related gene polymorphisms in PCa progression and mortality, and none with sex-steroid-inactivating UGT. Few predictive or prognostic markers have been identified in the steroidogenic pathway (6, 7, 28, 29), although several prognostic biomarkers have been identified in other cellular pathways, including cell cycle genes (30) and tumor suppressor genes (20).

Of all genetic processes involved in PCa progression, gene deletions are likely one of the most drastic molecular events. Recently, copy-number variations (CNV) in UGT2B17, one of the best characterized and most frequent CNV in humans (17, 19), was shown to be associated with osteoporosis in Asians (31), linking CNV with a complex metabolic disease. Deletion of UGT2B17 was shown to be associated with PCa risk in some studies but not in others (13–15). However, none of these studies looked simultaneously at other UGT2B family members. This is relevant because genes encoding UGT2B enzymes involved in steroid metabolism are all clustered on chromosome 4q13, which may contribute to the discrepancy between studies.

Because PSA is a well-established androgen-responsive gene, loss of function of a sex-steroid-metabolizing enzyme is likely to affect systemic and intracellular levels of active hormones that may lead to PSA secretion and disease recurrence. UGT2B genes are expressed in human prostate and encode enzymes that convert steroids into inactive glucuronide derivatives in prostate cells and several other tissues (10). Because the products of UGT genes function at the end of the steroid signaling pathways in target cells, it seems reasonable that alteration of these metabolic genes might significantly affect the inactivation of hormones in a wide range of tissues including in residual PCa cells after the surgical procedure. The level of active hormone is expected to be enhanced in UGT-deficient cells, potentially driving hormone-dependent cells into cell replication and leading to a higher risk of PCa recurrence in these individuals. UGT2B17 possesses glucuronidation activity toward androgens at the hydroxyl group at position 17 of 3α-diol (Fig. 1). Despite the fact that the number of patients with at least two deleted copies of UGT2B17 and UGT2B28 genes was limited, we showed that circulating levels of inactive androgens, namely the 3α-diol-17G, are altered in individuals carrying deletions. Steroid glucuronides such as 3α-diol-17G, are more accurate markers of hormonal actions than active hormones, because they reflect the tissue conversion of potent hormones (DHT and testosterone) to less-potent forms, which occurs in multiple tissues including the prostate. It is thus not unexpected to detect no significant change in testosterone or DHT levels in carriers of UGT2B deletions. It is similarly observed in healthy individuals carrying UGT2B17 genetic deletions (32–34) and lower intraprostatic androgen metabolite levels in PCa individuals with a deleted copy of the UGT2B17 gene (32).

At the time of biochemical relapse, the disease is particularly androgen dependent for growth and progression, and UGT2B haploinsufficiency, reflected by lower circulating glucuronide levels, is thought to promote the availability of intracellular androgens, required for AR activation. Further studies are needed to carefully evaluate the impact of specific UGT2B deletions on prostatic hormone levels. Nevertheless, in support of our observation, a previous study showed that repression of UGT2B17 expression by RNA interference in prostatic cancer LNCaP cells led to an increase in media and in intracellular concentrations of bioactive DHT, with a greater transcriptional activation of AR and subsequent enhanced PSA secretion (8).

Due to the mainly localized features of the tumors in this study, there were too few clinical events after PSA recurrence, and therefore, determinations regarding long-term outcome could not be assessed such as androgen-deprivation resistance or cancer-specific death. Additional studies are also required to assess their role in more advanced diseases and their relationship with androgen-deprivation resistance to better understand the mechanisms involved on PCa disease progression.

In conclusion, this study provides the first evidence of deletions in sex-steroid-metabolizing genes as potential biomarkers for PCa recurrence and supports the significance of sex steroid metabolism in the pathogenesis and recurrence of this prevalent cancer.

Abbreviations:

 
• ADT

  Androgen-deprivation therapy

 
• ADT-G

  ADT-glucuronide

 
• AR

  androgen receptor

 
• BCR

  biochemical recurrence

 
• CNV

  copy-number variations

 
• DHT

  dihydrotestosterone

 
• PCa

  prostate cancer

 
• PSA

  prostate-specific antigen

 
• UGT

  UDP-glucuronosyltransferase.

Acknowledgments

We thank Céline Veilleux and Hélène Hovington for their help with data collection, Mario Harvey for helpful discussion, and the Biostatistics Services of the CHUQ research center.

This work was supported by the Canadian Prostate Cancer Foundation (to E.L.)., the Canadian Institutes of Health Research (CIHR) (to C.G.), Canadian Urological Association-Canadian Uro-Oncology Group-Abbott Oncology research grant (to L.L.), and the Canada Research Chair Program (to C.G.). J.B., E.A.-W., and C.F. are recipients of a Frederick Banting and Charles Best Canada Graduate Scholarship award from CIHR. E.L. is recipient of a CIHR clinician-scientist salary award. C.G. holds the Canada Research Chair in Pharmacogenomics.

Disclosure Summary: G.N., J.B., E.A.-W., C.F., S.-P.H., B.-Y.B., P.D., and P.C. have nothing to declare. E.L., C.G., L.L., and Y.F. have been named inventors on a patent application owned by Laval University in work related to this study.

Author notes

^*

G.N., J.B., and E.A.-W. contributed equally.