Mechanism of Sex Differences in Bladder Cancer: Evident and Elusive Sex-biasing Factors

Androgens and Androgen Receptors (AR)

Sex steroid hormones and their respective receptors in BC have been studied for nearly half a century. In 1972, androgens were first identified as a BC risk factor when castrated male mice were better protected against bladder carcinogenesis induced by N-butyl-N-(4-hydroxybutyl)nitrosamine (BBN) than uncastrated males, and female mice treated with testosterone had increased BC risk compared to control female mice [15]. Preclinical studies further suggested that androgens induce while estrogens suppress carcinogen-induced BC development in rats [16]; clinical observation of men having decreased BC mortality after treatment with 5α-reductase inhibitors or dihydrotestosterone (DHT) blockers further solidified the relevance of androgens [17]. Ultimately, these studies were hypotheses-generating, as it remained unclear how exactly androgens initiated or potentiated BC. The next round of breakthrough studies came from molecular characterization of the androgen receptor (AR).

Using patient samples, Laor et al., observed greater levels of AR in tumor tissue compared to normal tissue, and females had relatively lower levels than males [18]. High grade tumors— although intuitively expected to exhibit increased AR— paradoxically had less AR than low grade tumors. It should be noted that other studies showed conflicting results, and this could be due to a lack of standardized staining and scoring [19, 20]. Usage of genetically engineered models (GEMs) subsequently led to a series of insightful studies. Mice with AR gene knockout (ARKO), either germline [21] or urothelium-specific knockouts [22], were less susceptible, while mice expressing extra AR (hAR^GoF) [23] were more susceptible to BC induced by a bladder-specific carcinogen. The genetic evidence convincingly demonstrated the essential role(s) of AR in promoting bladder tumorigenesis. Findings from Miyamoto et al., further implicated how AR and androgens did not completely rely on one another to mediate their cancer driving effects [21]. For example, DHT treated and untreated male ARKO mice developed bladder carcinoma at rates of 25% and 0%, respectively, suggesting the function of androgens via a non-AR pathway. Consistent with these implications, Chang and colleagues recently reported a membrane androgen receptor (mAR-SLC39A9) that functions through a noncanonical AR pathway Gαi/MAPK/MM9 to promote bladder cancer [24].

Mechanistic studies on AR have led to the identification of downstream target genes and molecular pathways. Initial characterizations of AR knockout mice suggested that tumor promoting activity of AR depends on the p53-PCNA DNA damage repair pathway [22]. Indeed, somatic loss-of-function of p53 tumor suppressor gene is frequently observed in BC [25, 26]. To validate whether AR promoting bladder tumorigenesis required the p53 pathway, Hsu et al used simian vacuolating virus 40 T (SV40T) to inactivate tumor suppressors p53 and Rb in mouse lines with or without urothelial AR; results showing no difference in bladder tumor initiation between the two groups supported the notion that AR bladder tumor initiating activity requires modulation of p53. The activity of AR in inducing BC cell proliferation in vitro [21, 27] also appears to be dependent on CD24, a glycosyl phosphatidylinositol-linked sialoglycoprotein— a direct downstream target of AR [27]. Interestingly, inactivation of CD24 is protective against BBN-induced BC in male but not in female mice, suggesting that the sex-biasing activity of AR is in part mediated by CD24 in vivo [27]. Moreover, Theodorescu and colleagues demonstrated that AR directly represses gene transcription of CD44 [28], the receptor for hyaluronic acid— a biomarker and potent driver of progressive disease in multiple tumor types, suggesting that AR may function in a stage dependent manner in promoting bladder tumor initiation but suppressing tumor progression.

Ectopic activation of the Wnt/ β-Catenin signaling pathway in the bladder urothelium induces luminal bladder tumor in mice [29]. Like the BBN-induced bladder carcinogenesis model, this GEM model of BC also displays a striking sex difference: the incidence is 45% in males vs. 3% in females when the Wnt/ β-Catenin pathway is constitutively activated in the bladder urothelium. Mechanistically, constitutive β-Catenin activation induces nuclear translocation of AR while conversely, activation of the AR pathway potentiates the Wnt/ β-Catenin signaling in bladder urothelial cells. Similar genetic interactions between androgen and Wnt/ β-Catenin signaling pathways are also found during the masculinization of external genitalia [30]. Together, these observations suggest that synergism between the AR and Wnt/ β-Catenin pathways may enhance sex differences in development and diseases.

Studies have also suggested the roles of androgens in dampening the host immunity against cancer, resulting in increased male cancer incidence [31] and mortality [32]. For example, androgens have been shown to be immunosuppressive by inhibiting the function of macrophages, NK cells and T cells [33]. When exposed to persistent antigen stimulation and hostile tumor microenvironment (TME) for tumor-infiltrating T cells, tumor-specific T cells are known to undergo programmatic dysfunction and exhaustion with diminished ability to mount effector function against cancer [34]. The mechanism of molecular programming of T cell exhaustion is not entirely clear, but the process involves key transcriptional and epigenetic changes centered around important transcription factors including Tcf7 (encoding Tcf1) and TOX [35, 36]. Using single cell RNA sequencing coupled by transcription studies, we recently discovered that androgens can promote CD8⁺ T cell exhaustion in the TME by directly transactivating Tcf7 [37].

Overall, this field has taken significant strides to characterize androgens/AR pathway in male BC incidence and progression. However, key questions remain: Are androgens and AR solely responsible for sex differences in BC development? Will targeting these pathways address sex differences more than existing therapies? While clinical trials (e.g., NCT02605863, NCT01234519, NCT02788201, NCT02300610)— including those testing the efficacy of existing hormone therapy for prostate cancer on recurrent NMIBC or advanced BC— had been initiated to test preclinical findings, trials consisted of small cohorts and in some cases were suspended due to insufficient enrollment. Meaningful clinical application would necessitate clinical trials with a larger patient cohort that includes both males and females.

Estrogens, Estrogen Receptors (ERα and ERβ), and Progesterone

Menopause provides a natural but indirect way to observe the effects of estrogen depletion in women, given that effects are confounded by age. In a 2006 prospective study, post-menopausal women had increased BC risk and early age menopausal onset yielded even greater risk, suggesting the protective effects of estrogens [38]. These findings are preceded by Bertram and Craig’s initial observation of estrogen’s protective effects in female mice contrasted with testosterone’s effects to increase BC risk in male mice [15]. As straightforward as this seems, a closer look at the relationships between estrogens, estrogen receptors, and progesterone reveals a more complex story.

Estrogens can function through two canonical nuclear receptors— ERα and ERβ— each responsible for potentially distinct functions, especially depending on tissue type. In rat and mice BC, ERα was inconsistently detected [39, 40]. Thus, unlike breast cancer, ERα did not seem to have prognostic value. However, in 2014, Hsu et al., showed that ER α knockout mice developed tumors faster than females with ER α, suggesting ERα’s inhibitory role in tumorigenesis [41]. ERβ, on the other hand, is expressed in human tumor tissues and increases with higher stage and grade BC, suggesting its role in tumor progression and metastasis [42]. In female mice, depleting ERβ slowed down tumor growth [43]. Thus, ERα may be the predominant driver of tumor initiation and ERβ of tumor progression. It has also been shown that Tamoxifen— a selective ER modulator (SERM) that inhibits ER activity— administered before, during, or after BBN treatment conferred protection from urothelial carcinogenesis in female mice [44]. Thus, existing studies imply: (1) ERα and ERβ serve different functions during BC initiation and progression, (2) Tamoxifen or SERM could act as a chemoprotective agent against BC, and (3) future studies should specify tumor type and stage when studying the mechanisms of ERα, ERβ, and SERMs.

Progesterone may also play a role. Parous female mice— those who gave birth and thus experienced changes in progesterone— developed smaller bladder tumors than nulliparous mice [45]. In humans, multiparous females had decreased BC risk [46, 47]. Thus far, there have not been follow-up studies to investigate the underlying mechanisms behind these observations; it would be interesting to see if progesterone mechanistically acts independently or interactively with ER receptors to affect BC risk in females. Finally, it is important to note that although males have lower levels of estrogens and progesterone, the actions of predominantly female sex hormones should not be ignored in males and vice versa.

Ultimately, sex hormones can only be partially responsible for BC sex disparities. Given that the median age of BC diagnosis is 73— an age with decreased sex hormone levels— how can sex hormones be the sole responsible player? Other critical sex-biasing factors must also be at play.

X Chromosome Inactivation (XCI) Escapees

One copy of the X chromosome in XX female cells is normally inactivated in a random manner through a process called X Chromosome Inactivation (XCI). This dosage compensation mechanism ensures that X-linked genes are expressed at comparable levels between XX and XY cells. However, XCI is an imperfect process— a fraction of genes escape XCI, leading to higher levels of XCI escapees expressed in XX females than in XY males. In humans, nearly 23% of X-linked genes escape XCI [57], which supports the hypothesis that XCI escapees can contribute to sex differences in cancer. Because females effectively have two functional copies of these escapees while males have only one copy, losing one functional copy in males is sufficient to cause a phenotypic change while females will still have one functional copy to compensate the loss. Based on this premise, Lane and colleagues analyze paired tumor–germline exome sequencing data from 4,126 patients across 21 tumor types from TCGA and Broad Institute data sets and identified that loss-of-function mutations of six X-chromosome genes (ATRX, CNKSR2, DDX3X, KDM5C, KDM6A, and MAGEC3) are more frequently found in males than females [5]. Of the 6 named genes, only KDM6A exhibited relevance in affecting sex differences in bladder cancer, which we will further discuss in the section titled Epigenetics: The Sex Epigeneome.

Sexually Dimorphic Genetic Architecture

It is worth noting that in addition to X and Y sex chromosomes, autosomes may also contribute to sex differences in cancers. By interrogating the ICGC/TCGA data, Li et al. showed that somatic mutation frequency of an autosomal gene CTNNB1 in liver hepatocellular cancer is significantly higher in males than in females, implying that there are sex differences in oncogenic mutational processes beyond the mutational differences of X-chromosome-encoded genes [7, 8]. Genetic variations can also occur in non-coding regions to influence gene regulation and expression. Collectively, these genetic variations, linked to observed quantitative differences in gene expression, can be referred to as an individual’s genetic architecture. Here, we will discuss how differences in genetic architecture contribute to sexual dimorphism.

Advancement of genomic techniques and curation of large genomic databases, such as the Genotype-Tissue Expression (GTEx) project [58], has made studying sex differences of genetic architecture— down to the level of a single nucleotide— feasible. Using GTEx data, Lopes-Ramos et al. interestingly observed more sex differences on the level of transcription regulation than gene expression in a tissue specific manner: 87% of genes were differentially targeted, meaning they experienced different regulatory pressures, but 70% of these differentially targeted genes, did not show differential expression [59]. This means that certain phenotypes may manifest in a latent manner as time, age and stress differentially impact male and female specific transcription factors. Meaning, at any given time point, the potential for differential expression may be higher than observed due to differences in regulation hidden beneath more uniform expression. Oliva et al. extended upon these finding by analyzing the cis sex-biased expression quantitative trait loci (eQTLs) and performed allelic expression analysis between males and females [60]. Their findings reiterated the concept that a seemingly similar phenotype between males and females is supported by a diverse set of sex-specific checks and balances. They identified sex-biased genes in pathways such as epigenetic methylation and xenobiotic metabolism including CYP450 genes.

One limitation of these recent studies using the GTEx dataset is the lower population data for bladder, endocervix, ectocervix, fallopian tube, renal medulla than the rest of the tissue types (n < 25 vs. n = 73–670), resulting in the exclusion of these five tissues from the analyses. In general, this avenue of research has not yet been extensively applied to study sex differences in BC. Previous efforts of genome-wide association studies (GWAS) have identified several loci that are tightly associated with BC risk [61–67]. We believe that analyzing gene regulation such as eQTLs could be insightful in explaining male and female differences in BC incidence and mortality by revealing the sex-specific variability in response to the environmental signals and pathological states.

Sex-specific selection constraints may be another potential mechanism contributing to a sex-specific genetic architecture. In two different studies, Gershoni et al., found that both men and women had reduced selection pressures on sex-specific genes compared to non sex-specific genes [68, 69]. Moreover, the magnitude of decreased selection was greater in men than women [69]. They postulate that the reduced negative selection in male-biased genes may contribute to a greater accumulation of deleterious mutations in men and in turn yield different patterns of disease incidence. Whether this mechanism is relevant in explaining BC sex differences is not yet known and should be further explored.