Functional genomics based on germline genome-wide association studies of endocrine therapy for breast cancer

Abstract

Breast cancer is the most common invasive cancer in women worldwide. Functional follow-up of breast cancer genome-wide association studies has led to the discovery of genes that regulate endocrine therapy response in a SNP- and drug-dependent manner. Here, we will present four examples in which functional genomic studies from breast cancer clinical trials led to novel pharmacogenomic insights and molecular mechanisms of selective estrogen receptor modulators and aromatase inhibitors. The approach utilized for studying genetic variability described in this review offers substantial potential for meaningful discoveries that move the field toward precision medicine for patients.

Breast cancer is the most common cancer and the leading cause of cancer death in women worldwide [1]. The majority (about 70%) of breast cancers are driven by the estrogen receptor (ER). Translocation of ER to the nucleus and subsequent transcription of ER-regulated genes requires the binding of estrogens, such as estradiol, to the ER ligand-binding pocket. Two major classes of drugs target the ER pathway in breast cancer. There are selective ER modulators (SERMs; tamoxifen and raloxifene) that competitively bind to ER ligand-binding pocket and the aromatase inhibitors (AIs; anastrozole, exemestane and letrozole) that block the enzyme responsible for estradiol and estrone production. While ER-targeting therapies have made substantial impacts on the survival of breast cancer patients, there remains a need to identify genetic biomarkers to identify which patients will not respond to therapy or will have adverse reactions.

The National Surgical Adjuvant Breast and Bowel Project P-1 and P-2 trials for SERMs prevention [2] and MA.27 trial for AI treatment [3] were two major breast cancer clinical trials examining drugs that target ER. A genome-wide association study (GWAS) were performed using these clinical trials [2,3], which identified SNPs that were associated with drug response including both efficacy and drug-induced side effects. To explore the function of these SNPs, a preclinical in vitro model system combined with various functional studies was applied in all the studies described subsequently.

Cell line model systems, such as the NCI-60 project [4–7] and HapMap lymphoblastoid cell line (LCL) panel (reviewed in [8]), have been widely used for pharmacogenomics (PGx) studies. Over 10 years ago, a similar LCL model system was developed in our lab for studying PGx expression-related hypothesis. This model system, known as the Human Variation Panel, consists of nearly 300 EBV-transformed human LCLs from healthy, unrelated individuals across three ethnicities (Caucasian, Han Chinese and African–American; reviewed in [8]). Utilizing this LCL panel has helped advance current understanding of mechanisms of chemotherapy resistance [8–24].

Furthermore, the LCL system was supplemented by additional technology that allows us to study SNP function in specific cell types. Genetic technology, such as CRISPR-Cas9 technology helps us to specifically determine individual SNP phenotypes via engineering of isogenic cells lines that differ from the parental cell line by a single nucleotide [25,26]. While the Human Variation Panel covers almost all of the common SNPs in humans, to study biology in more relevant tissue or cell types, such as cancer cells, previous functional studies were also performed in cell lines that were more appropriate for studying specific drug response phenotype (e.g., MCF7 and T47D and hormonal treatment) using CRISPR-Cas9 to engineer a given SNP genotype sequence. This was then followed by comparison of various cellular and drug response phenotypes between two cell lines with one nucleotide difference, while otherwise, the same genetic background.

The aforementioned Human Variation Panel model system combined with molecular biology techniques and CRISPR-Cas9 technology have been used to functionally explore signals from clinical GWAS. As a result, our laboratory has discovered a novel mechanism by which SNPs regulate therapeutic response to endocrine therapy which is both SNP and drug dependent. Specifically, expression of genes critical for endocrine therapy response is altered only under certain genotypes and in the presence of drug; a phenomenon our laboratory has coined as a PGx-expression quantitative trait loci (PGx-eQTL). PGx-eQTL expands the definition of eQTL to include the effect of a particular drug response. The subsequent paragraphs will describe the various examples of a clinical GWAS leading to the discovery of PGx-eQTLs as well as the mechanisms involved.

PGx of SERMs in breast cancer prevention

In the late 1990s, a large double-blind, placebo-controlled clinical trial (n = 13,388) was conducted to determine whether tamoxifen can be used to prevent breast cancer [27]. Several years later, another large double-blind trial (n = 19,747) was conducted to compare raloxifene and tamoxifen in the prevention of breast cancer [28,29]. These studies are referred to as the National Surgical Adjuvant Breast and Bowel Project P-1 and P-2 trials, respectively. While raloxifene and tamoxifen have been US FDA-approved for the prevention of breast cancer in high-risk patients after these trials, it is not widely used in this setting because of a large number needed to treat and serious adverse reactions. Therefore, a GWAS was performed to determine if there are potential biomarkers for breast cancer occurrence while on tamoxifen or raloxifene [2]. There were two signals from this breast cancer chemoprevention GWAS that were pursued in the laboratory: SNPs in ZNF423 (chromosome 16) and SNPs in CTSO (chromosome 4) [2]. The discussion given below will highlight the GWAS associations, the utility of LCLs in functional validation and the underlying molecular mechanisms that were discovered subsequently. ZNF423 will be discussed first followed by CTSO.

The variant alleles for two intronic ZNF423 SNPs (rs8060157 and rs11076499) were associated with decreased risk of breast cancer occurrence while on SERM therapy (odds ratio: 0.7) [2]. This led to the finding of 16 additional intronic SNPs (through imputation and subsequent genotyping) that were also associated with decreased risk. These SNPs are in tight linkage disequilibrium (r² > 0.9) and have minor allele frequencies of approximately 0.4 in the P-1 and P-2 cohorts. The SNP rs9940645 is one of these imputed SNPs and three estrogen response elements (EREs) are nearby.

Because SERMs such as raloxifene and tamoxifen, interact with the ER and because BRCA1 is a known breast cancer risk gene [30], the ZNF423 SNPs were investigated in the context of induction by estrogen and BRCA1 regulation. The Human Variation Panel of LCLs was used to select cell lines that were homozygous wildtype (WT), homozygous variant or heterozygous for the chromosome 16 ZNF423 SNPs. LCLs were treated with increasing doses of estrogen (E2) followed by tamoxifen and expression of ZNF423 and BRCA1 was assessed by quantitative reverse transcription PCR. In turn, it was discovered that E2 can significantly induce the expression of ZNF423 and BRCA1, but only in cells harboring the WT ZNF423 SNP. This SNP- and drug-dependent phenotype was reversed in the presence of tamoxifen treatment; ZNF423 and BRCA1 expression were induced under tamoxifen treatment in cells carrying the variant ZNF423 SNP (Error! Reference source not found.A, 1B) [2]. Chromatin immunoprecipitation (ChIP) assays revealed that the binding of ER to the EREs in the ZNF423 promoter in the presence of E2 and tamoxifen is affected by ZNF423 rs9940645 genotype (Error! Reference source not found.C) [2]. The variant genotype had more ER binding in the presence of tamoxifen (Error! Reference source not found.D) [2].

How the SNP, which is approximately 200 bp distant from the EREs, might regulate the gene expression in an E2-dependent fashion (Figure 1C) was further investigated. The underlying mechanism for this SNP- and drug-dependent phenotype is regulated by calmodulin-like protein 3, a sensor protein that binds directly to the rs9940645 SNP in the ZNF423 promoter and regulates ERα in ERE binding as a ER coregulator [31]. Additionally, all of these ZNF423 in vitro data are congruent with the clinical GWAS finding that the variant SNP was associated with a lower risk of recurrence.

Figure 1. . Tamoxifen induces ZNF423 and BRCA1 expression in lymphoblastoid cell lines with ZNF423 variant genotype.

(A & B) mRNA expression for ZNF423 (A) and BRCA1 (B) for LCLs with WT/WT (8 cell lines), WT/V (7 cell lines), and V/V (8 cell lines) genotypes for the chromosome 16 ZNF423 SNPs after exposure to E2 alone or E2 with increasing concentrations of 4-OH-TAM. (C) schematic depiction of the ZNF423 intron 2 area near the rs9940645 SNP. The locations of EREs are shown as boxes, and arrows show the locations of primers used to conduct the ChIP amplifications. (D) ChIP assay for the area of ZNF423 that contains the rs9940645 SNP. V, variant. (E) dual luciferase reporter-gene assays showing the effect of 4-OH-TAM (left) and raloxifene (right). A 500 bp DNA sequence that included either a WT or variant SNP sequence for rs9940645 and all of the EREs shown in Figure 4 was cloned into the pGL3 basic reporter plasmid. A 1500 bp region of the ZNF423 promoter was then cloned downstream of the 500 bp segment, and these constructs were used to transfect IGROV-1 cells that were treated with 0.01 nmol/l E2, with or without 10−7 mol/l 4-OH-TAM (left) or 10−7 mol/l raloxifene (right). Error bars represent SEM.

ChIP: Chromatin immunoprecipitation; ERE: Estrogen response element; LCL: Lymphoblastoid cell line; OH-TAM: Hydroxytamoxifen; SEM: Standard error of the mean; WT: Wild type.

Reproduced with permission from [2], © Cancer Discovery (2013).

The clinical implication of this finding is that patients at high risk for breast cancer with the variant ZNF423 SNP would likely benefit from tamoxifen to prevent the occurrence of breast cancer due to increased BRCA1 activity and dsDNA breaks repair. Clinicians and patients that are weighing the benefit of using tamoxifen to prevent breast cancer versus the risk of serious adverse side effects may be more inclined to the former given a ZNF423 variant genotype. Moreover, patients with this variant ZNF423 SNP might benefit from the combination treatment of tamoxifen and docetaxel based on the mitosis-related genes VRK1 and PBK which are identified as downstream genes of ZNF423 other than BRCA1 [32].

The other signal from the chemoprevention GWAS is CTSO. Two genotyped SNPs in CTSO (rs10030044 and rs4256192), as well as imputed SNPs, were found to be associated with an increased risk of breast cancer occurrence while on SERM therapy (odds ratio 1.42 and 1.44) [2]. Using the Human Variation Panel of LCLs, CTSO and BRCA1 expression were shown to be estrogen inducible in cells with the WT, but not the variant, CTSO SNPs.

The mechanism of this CTSO SNP-dependent phenotype, however, is not appreciated after adding a pharmacologic agent as in the ZNF423 example. Rather, the variant rs6810983 CTSO SNP disrupts an ERE. Additional experiments revealed that CTSO regulates BRCA1 transcription factors, including ZNF423, and protein degradation [33]. This explains the previous observation that CTSO knockdown resulted in increased double-strand breaks that could be rescued with BRCA1 overexpression [2]. In summary, this example has shown how we were able to fine-tune the molecular understanding of both SNP- and drug-dependent transcriptional regulation as well as SNP-disrupting ER-binding transcriptional regulation in the context of tamoxifen response.

PGx of adjuvant AIs in early breast cancer

A second example of PGx regulation came from MA.27. MA.27 was a large clinical trial (n = 4658) that compared anastrozole and exemestane as an adjuvant therapy for early-stage breast cancer. While this trial did not find a significant difference between the two AIs [34], a large meta-analysis found that nearly one-fifth of patients on an AI have a breast cancer recurrence within 10 years [35]. Therefore, a GWAS was performed using MA.27 patient data to identify potential biomarkers of AI response using breast cancer-free interval as an end point [36].

The top SNPs were located on chromosome 8 and subsequent imputation and deep sequencing around that signal was performed. Three of the top six SNPs were related to a long noncoding RNA called MIR2052HG. The top SNP, rs13260300, was 5′ of the lncRNA and the other two SNPs, rs4735715 and rs3802201, were located within the lncRNA, the latter of which was located 16 bp away from a putative ERE [36].

Using the Human Variation Panel, it was discovered that LCLs with the variant MIR2052HG SNP genotypes showed a dose-dependent increase in MIR2052HG expression and ER binding in the presence of androstenedione, which is catalyzed by aromatase to estrone (E1) (Figure 2A & C). MIR2052HG expression was increased in the MIR2052HG WT genotypes but dramatically decreased in the variant genotypes after exposure to addition of exemestane (Figure 2A) or anastrozole (Figure 2C). Furthermore, the expression pattern of ESR1 (the gene which encodes ERα) was similar to that of MIR2052HG (Figure 2B & D). Finally, knockdown of MIR2052HG decreased the mRNA and protein levels of ERα as well as decreased cell proliferation and colony formation. MIR2052HG altered ERα levels by regulating AKT/FOXO3-mediated ESR1 transcriptional activity and modulating proteasome-dependent degradation of ERα protein [36].

Figure 2. . The mRNA expression of MIR2052HG and ESR1 is SNP and androstenedione/exemestane dose-dependent.

(A) and (B) MIR2052HG mRNA expression in LCLs with WT and V genotypes for both rs4476990 and rs3802201 after exposure to androstenedione alone and with increasing concentrations of exemestane or anastrozole. (C) and (D). mRNA expression for ESR1 in LCLs with the same conditions as (A) and (B).

*p < 0.05; **p < 0.01.

LCL: Lymphoblastoid cell line; WT: Wild type.

Reproduced with permission from [36], © Cancer Research (2016).

Further mechanistic studies supported a model (Figure 3) in which the protective MIR2052HG variant SNP genotype regulates LMTK3 by enhancing the recruitment of early growth response protein 1 to the promoter region of LMTK3 [37]. To study how MIR2052HG and LMTK3 might regulate ERα protein degradation, ERα pSer167 levels, which is known to be involved in the ERα protein degradation [38] were examined after knockdown of MIR2052HG. As a result, knockdown of MIR2052HG increased the ERα pSer167 levels as well as WT ERα ubiquitination. Since the pSer167 of ERα is regulated by pp90 (RSK1) which is activated by MAPK, MEK/ERK/p90RSK1 activity (ERα pSer167, pMEK, pERK and pRSK1) was tested after knockdown of MIR2052HG. The results support that MIR2052HG regulates LMTK3 leading to regulation of ERα stability via the MEK/ERK/RSK1 axis. At the same time, we also found that LMTK3 regulation of PKC activity led to the ERα protein degradation and AKT-FOXO3 regulation on ESR1 transcription. Taken together, MIR2052HG regulates ERα mRNA and protein level through the axis of LMTK3/PKC/MEK/ERK/RSK1 (Figure 3) [37]. Similar to ZNF423, MIR2052HG regulates LMTK3 in a SNP- and AI-dependent fashion: compared with wild-type genotype, the variant SNP genotype increased the activity of LMTK3 in response to androstenedione, the substrate for aromatase [37]. Adding an AI can reverse this pattern.

Figure 3. . Hypothetical model illustrated how MIR2052HG might regulate LMTK3 transcription.

The red arrows indicate the transcription direction. The transcribed MIR2052HG interacts with EGR1 protein and brings EGR1 to the LMTK3 locus. Together with other transcription machinery, binding of EGR1 to the LMTK3 promoter initiates transcription. LMTK3 protein inhibits the PKC, therefore downstream MAPK and AKT/FOXO3 pathways, leading to regulation of ERα degradation and ESR1 transcription.

Reproduced with permission from [37], © Breast Cancer Research (2019).

PGx studies of AI-related musculoskeletal adverse events

Notably, the SNP- and drug-dependent regulation not only involved in the outcome of drug treatment, but also, the serious side effect of drug treatment, see TCL1A gene in Table 1 [3,39,40]. This is the third example of PGx regulation came from the MA.27 clinical trial noted above. In the MA.27 AI trial, about half of patients treated with anastrozole have joint-related complaints and the major reason for anastrozole discontinuation is musculoskeletal adverse events. To reveal the genetic variations in patients developing these adverse events, a nested, matched and case–control GWAS was performed. The top SNPs near the TCL1A gene on chromosome 14 were identified with promising associations (p < 1 × 10^-6) with the musculoskeletal adverse events. Among them, one imputed TCL1A SNP rs1184958 (p = 6.67 × 10^-7) which creates an ERE, resulted in significantly greater TCL1A expression in LCLs with estradiol exposure [3]. The expression of TCL1A in LCLs with the variant genotype dropped dramatically with the addition of ERα downregulator ICI 182780. However, in the LCLs with wild-type genotype, TCL1A expression increased dramatically (Figure 4) [3]. Interleukin receptors IL17RA, IL12RB2 and IL1R2 were then found to be the ‘downstream’ of TCL1A, presenting the SNP- and estrogen-dependent manners as TCL1A. Remarkably, the activity of NF-κB, which is involved in joint inflammation, showed a significant difference in the LCLs harboring the variant and the WT genotypes after adding ICI 182,780 [3].

Table 1. . Examples of lymphoblastoid cell line-led discoveries of SNP- and drug-dependent transcriptional regulation.

Clinical GWAS	Clinical phenotype	Top candidate gene from GWAS	LCL-led discovery	Ref.
NSABP P-1 and P-2 breast cancer prevention trials	Breast occurrence while on SERM therapy	ZNF423	ZNF423 SNPs regulate BRCA1 expression and DNA-damage response through a sensor protein called CALML3 in the presence of tamoxifen	[28,34]
MA.27 aromatase inhibitor trial	Breast cancer-free interval	MIR2052HG	MIR2052HG SNPs regulate ERα in the presence of an aromatase inhibitor through increased ESR1 transcription and decreased ERα degradation	[32,33]
Postmenopausal women on anastrozole	Plasma estradiol concentration	TSPYL	TSPYL5 SNPs regulate CYP19 expression under estrogen exposure TSPYL1 SNP regulate CYP17 and 3A4 which affects abiraterone concentration	[35] [36]
MA.27 aromatase inhibitor trial	Musculoskeletal adverse events	TCL1A	TCL1A SNPs regulate chemokine and cytokine expression through increased NF-κB signaling in the presence of endocrine therapy drugs	[37–40] [41]

GWAS: Genome-wide association study; LCL: Lymphoblastoid cell line; NSABP: National Surgical Adjuvant Breast and Bowel Project; SERM: Selective estrogen receptor modulator.

Figure 4. . SNP-related variation in T-cell leukemia 1A expression and NF-κB activity in lymphoblastoid cell lines.

SNP-related variation in TCL1A expression and NF-κB transcriptional activity in three LCLs with variant (V) genotypes and three LCLs with WT genotypes for the chromosome-14 SNPs after exposure to increasing concentrations of estradiol (E2), or E2 plus the ER antagonist ICI-182780. (A) Quantitative RT-PCR of TCL1A expression after exposure to increasing concentrations of E2 for 24 h, followed by 0.01 nM E2 plus increasing concentrations of ICI-182,780 for an additional 24 h. (B) The same six LCLs were transfected with an NF-κB reporter construct. After exposure to increasing concentrations of E2 for 24 h, followed by 0.01 nM E2 plus increasing concentrations of ICI-182,780 for an additional 24 h, luciferase activities were determined. All values represent the mean ± standard error of the mean of triplicate determinations.

ER: Estrogen receptor; LCL: Lymphoblastoid cell line; RT: Real time; TCL: T-cell leukemia; WT: Wild type.

Reproduced with permission from [3], © Breast Cancer Research (2012).

Furthermore, to investigate the transcriptional regulation of across the genome, RNA sequencing (RNA-seq) and TCL1A ChIP sequencing (ChIP-seq) was performed in LCLs with homozygous WT or variant TCL1A SNP genotypes treated with vehicle, E2 or E2 plus tamoxifen [39]. A total of 357 genes were significantly increased in LCLs with TCL1A variant SNP genotypes after E2 treatment. Moreover, these same 357 genes were significantly increased with the addition of tamoxifen. Gene ontology analysis showed that these genes were in pathways related to transcriptional regulation and T-cell activation. ChIP-seq revealed that TCL1A bounds mostly to the promoter region of genes that were involved in transcription regulation and immune-related pathways. Most importantly, NF-κB p65-binding motif was highly enriched in TCL1A-binding sites across the genome, suggesting that TCL1A and NF-κB form a transcription complex that might modulate downstream gene expression in a TCL1A SNP- and estrogen-dependent fashion.

PGx of baseline estrogen level in women with early stage-breast cancer treated with AI anastrozole

The fourth example of PGx SNP eQTL came from a study that examined the baseline estrogen levels in women with breast cancer [41,42]. The SNP- and drug-dependent regulation were also investigated in a GWAS for baseline plasma estradiol concentration in 772 postmenopausal women with breast cancer before anastrozole treatment [43]. The SNPs across the TSPYL5 gene were found to have genome-wide significance (<5 × 10^-8). Among them, the variant rs2583506 SNP created an ERE which can recruit ERα binding, resulted in an increased TSPYL5 expression after E2 exposure in LCLs with heterozygous rs2583506 SNP genotype, but not in those with WT rs2583506 SNP genotype. Further functional studies revealed the positive regulation of TSPYL5 on CYP19A1, which was also known as aromatase enzyme, critical for the synthesis of estradiol. These experimental results supported the finding that patients with homozygous variant TSPYL5 genotypes had higher estradiol concentrations than patients who were homozygous WT genotypes [43].

Interestingly, other TSPYL family members (TSPYL1, TSPYL2 and TSPYL4) were found to regulate the expression of multiple CYP family members including CYP3A4 and CYP17A1 [44,45]. Moreover, a nonsynonymous SNP of TSPYL1, rs3828743, altered TSPYL1’s suppression of CYP3A4 and CYP2C19 but not CYP17A1, mediating the intracellular level of CYP17A1 inhibitor, abiraterone, a drug that is commonly used for the treatment of castration-resistant prostate cancer and is catalyzed by CYP3A4 into inactive metabolites. This rs3828743 SNP was associated with poor abiraterone response and progression-free survival in patients with castration-resistant metastatic prostate cancer [45]. Additionally, this TSPYL1 SNP was shown to be associated with the metabolism of selective serotonin reuptake inhibitors in patients with major depressive disorder via SNP-dependent regulation of CYP2C19 expression [44]. In summary, a common theme has emerged with the above examples of SNPs in genes shown in Table 1 [2,36] – SNPs, can alter transcription regulation of downstream genes, however, these effects are only observed under certain pharmacologic or physiologic conditions, such as tamoxifen or estrogen exposure, respectively.

Clinical implications & future PGx directions

Notably, GWAS was still relatively new when the clinical trials related to AI-induced musculoskeletal adverse events were published in 2010 [46]. However, we were able to take advantage of the samples collected from these large clinical trials with rich clinical datasets to generate new PGx hypotheses, Here, we presented a series of studies in PGx of breast cancer with a focus on hormonal treatment that enable us to discover many PGx-eQTL SNPs that might not regulate gene expression at baseline, but can alter the gene expression and downstream effectors in the presence of different exposures, leading to a significant impact on therapeutic responses. Increasing evidence suggest that many of these GWAS signals of drug responses are within the regulatory regions of the genome. Additional tools, such as ATAC-seq, ChIP-seq and methyl-seq to understand transcription regulation as related to drug response should be applied in future studies. The current GTEx (https://www.gtexportal.org/home/) mainly provides eQTL relationship at baseline. SNP–gene expression relationship post exposure of xenobiotics should be considered to gain full picture of the contribution of genetic variation in drug response. Finally, studies to investigate PGx eQTL at genome-wide level are essential for us to gain new knowledge and to understand the impact of these SNPs on cellular function as well as on clinical treatment response. The knowledge would also help us to take advantage of individual genetic background to design better personalized therapy through altering critical gene expression. Therefore, identifying and understanding these PGx-eQTL SNP–gene relationships [47] have the potential to generate new hypothesis and have significant impact on future clinical practice.

Executive summary.

Background

• Hormonal therapy (selective estrogen receptor modulators [SERMs] and aromatase inhibitors [AIs]) is the standard care for patients with the most common breast cancer subtype, estrogen receptor-positive (ER+) disease. However, large variations of response to these drugs exist. Using genome-wide association study (GWAS) approach together with functional studies suggest that host genetic variation have been shown to contribute significantly to variation in response observed in clinic through a novel mechanism, pharmacogenomic (PGx)-expression quantitative trait loci (SNP- and drug-dependent gene transcription regulation).

PGx in SERM prevention

• GWAS study has identified SNPs in ZNF423 and CTSO associated with breast cancer occurrence in the setting of SERM prevention. Further studies indicated that these SNPs regulate ZNF423 and CTSO in a SNP- and tamoxifen-dependent fashion, leading to downstream differentially regulated genes and pathways that contribute to cell proliferation and response to SERMs.

PGx studies of AIs as adjuvant therapy for early-stage breast cancer

• Using samples from the largest MA.27 clinical trial that randomized to two AIs, anastrozole and exemestane has identified SNPs in a long noncoding RNA, MIR2052HG regulating ERα through regulation of LMTKs pathway. MIR2052HG regulates LMTK3 in a SNP- and AI-dependent fashion: compared with wild-type genotype, the variant SNP genotype increases the activity of LMTK3 in response to androstenedione, the substrate for aromatase. Adding an AI can reverses this pattern.

PGx studies of AI-related musculoskeletal adverse events

• This study focusing on the most common AI-induced side effect has identified SNPs in TCL1A that influence multiple downstream genes involved in chemokine and cytokine pathways, which result in symptoms like musculoskeletal adverse events. During the process, it was also discovered that TCL1A in lymphoblastoid cell lines (LCLs) with the variant genotype dropped dramatically with the addition of ERα downregulator, ICI, while in the LCLs with wild-type genotype, leading to similar changes of TCL1A downstream genes such as NF-κB activity.

PGx of baseline estrogen level in women with early-stage breast cancer treated with AI anastrozole

• GWAS of SNPs with baseline estrogen level in 772 postmenopausal women with breast cancer has identified SNPs across the TSPYL5 gene as the most significant signal. Further functional studies confirmed the TSPYL5 regulation of CYP19, the gene encoding aromatase and the target for AI. The variant rs2583506 TSPYL5 SNP creates an estrogen response element which can recruit ERα binding, resulted in an increased TSPYL5 expression after E2 exposure in LCLs with variant rs2583506 SNP genotype, but not in those with wild-type SNP genotype. This finding has been extended to other TSPYL family genes (TSPYL 1, 2 and 4) with regard to their regulation of additional CYPs, such as CYP3A4 and CYP17A1, both of which play important roles in abiraterone response in the treatment of castration-resistant prostate cancer.

Clinical implications & future PGx directions

• The concept of PGx-expression quantitative trait loci SNPs has significant impact not only on our basic understanding of how these SNP biomarkers might regulate gene expression and downstream signaling pathways involved in drug response, but also on our ability to use these information to better individualize the therapy.