X-linked Tumor Suppressor Escape and Sex Dimorphism in Cancer: Implications for Risk Stratification and Prevention

Abstract

Epidemiological data demonstrate a marked male bias in incidence and mortality across most non-reproductive cancers. Beyond behavioral and hormonal factors, recent genomic insights highlight the critical role of X-chromosome genetics and epigenetics. Specifically, genes that escape X-chromosome inactivation (XCI)—termed Escape from X-Inactivation Tumor Suppressors (EXITS)—provide a female-specific "dosage protection" mechanism against oncogenesis. This review analyzes seven X-linked tumor suppressor genes associated with sex-biased cancer phenotypes (KDM6A, KDM5C, DDX3X, ATRX, PHF6, MAGEC3, and CNKSR2) and their roles in cancer etiology and tumor suppression. Under the two-hit hypothesis, female cells—with two X chromosomes—can offset a loss-of-function mutation on the active X chromosome through normal allele expression from the inactive X chromosome (Xi), lowering susceptibility. Conversely, males with only one X chromosome face a "single-hit vulnerability," in which a single loss-of-function mutation disrupts the tumor-suppressor network. We further note persistent knowledge gaps regarding the remaining X-linked tumor suppressors annotated in the TSGene 2.0 database, reflecting the facultative nature of XCI escape, limited statistical power in existing cohorts, and the underuse of single-cell multi-omic approaches. Elucidating XCI escape and X-linked gene dosage imbalances offers a framework for sex-specific precision oncology. For high-risk male populations, identifying EXITS gene mutations warrants more intensive, shorter-interval screening. In female patients, epigenetic modulators that induce X-reactivation of Xi may offer unique therapeutic benefits. Incorporating X-linked gene dosage parameters into clinical trials and drug development will be essential to overcome sex disparities in cancer prevention and therapy.

INTRODUCTION

Males generally exhibit a higher incidence and lower survival rates across the majority of non-reproductive cancers compared to females [1-3]. However, this trend is reversed in young adults aged 20 to 39, where endocrine cancers—accounting for 33% of tumors in females of this age group—serve as the primary driver elevating the female cancer incidence above that of males [3]. This observation supports the hypothesis that sex hormones, particularly estrogen, play a crucial role in determining sex dimorphism in carcinogenesis by regulating immune responses and cellular proliferation [4].

Nevertheless, the hormone hypothesis alone is insufficient to fully account for the entire spectrum of sex disparities in cancer. If estrogens were to directly suppress anti-tumor immune responses, biologically female individuals should be inherently more vulnerable to cancer than males; however, males exhibit a significantly higher risk in most cancer types [1]. Furthermore, the persistent disparity in cancer risk between sexes even after menopause, when sex hormone levels decline rapidly, strongly suggests that fundamental genetic and epigenetic factors beyond hormones actively contribute to sex differences in oncogenesis.

From this perspective, it is imperative to focus on sex chromosomes, which represent the most fundamental biological divergence between males and females. In particular, the genomic variation wherein females possess two X chromosomes while males possess only one creates a decisive difference in the expression patterns of X-linked tumor suppressor genes. The human X chromosome harbors approximately 1,000 genes, many of which are intimately involved in core biological processes, including immune regulation, DNA damage repair, cell cycle control, and chromatin remodeling [5,6].

Because not all genes on the X chromosome have functional counterparts on the Y chromosome [7], a mechanism is required to prevent expression dosage imbalance between sexes. To compensate for this gene dosage disparity, X-chromosome inactivation (XCI) occurs in females. XCI is an epigenetic mechanism that silences one of the two X chromosomes in female cells to achieve gene expression parity with males [8]. However, the phenomenon of "XCI escape," wherein a subset of X-linked genes evades silencing, provides females with an additional dosage of tumor suppressor genes. Conversely, this leaves males susceptible to genetic vulnerability, where a single mutational hit can lead to a complete loss of tumor suppressor function [9,10].

Recent advances in large-scale multi-omics databases—such as The Cancer Genome Atlas Program (TCGA), The International Cancer Genome Consortium (ICGC), and The Genotype-Tissue Expression (GTEx)—alongside single-cell RNA sequencing (scRNA-seq) technologies have provided a robust foundation for exploring cell type- and cancer-specific XCI escape patterns, as well as the accumulation of somatic mutations in Escape from X-Inactivation Tumor Suppressors (EXITS). In particular, the use of the Tumor Suppressor Gene Database (TSGene), compiled from literature and genomic data, enables precise tracking of the functional networks of X-linked tumor suppressors [11]. Based on tumor suppressor gene information curated in TSGene, this review systematically searches academic literature databases, including PubMed, to synthesize the molecular mechanisms of X-linked tumor suppressor genes and comprehensively evaluate the associations between their XCI escape patterns and sex dimorphism in carcinogenesis and immune evasion.

MOLECULAR MECHANISMS OF XCI AND ESCAPE

Initiation and epigenetic regulatory network of XCI

XCI is initiated at the X-inactivation center (Xic), located in the Xq13.1–Xq13.2 region of the X chromosome [12]. In female eutherian mammals, the master regulator of XCI is Xist (X-inactive specific transcript), a long non-coding RNA (lncRNA) transcribed from the Xic [12,13]. Xist coats the X chromosome designated for silencing in cis, thereby triggering repressive chromatin remodeling [14].

A major functional antagonist in transcriptional opposition to Xist expression is its antisense lncRNA, Tsix, which finely regulates XCI by directly repressing Xist transcription [15]. Within this precise regulatory network, RNA exonuclease 1 (REX1) and Ring finger protein 12 (RNF12) exert opposing functions. REX1 prevents X-chromosome inactivation by activating Tsix and directly repressing Xist. Conversely, RNF12, an E3 ubiquitin ligase, promotes Xist expression and mediates XCI initiation by targeting REX1 for degradation via the ubiquitination and proteasomal degradation pathway [16,17].

Once fully inactivated, the inactive X chromosome (Xi) becomes blanketed with dense repressive epigenetic marks—including Polycomb Repressive Complex 2 (PRC2)-mediated accumulation of H3K27me3, H3K9me3 enrichment, promoter hypermethylation, and incorporation of the histone variant macroH2A—condensing into a transcriptionally silent Barr body structure [18].

Genomic architecture of XCI escape: pseudoautosomal regions (PAR) and non-PAR regions

Despite the inactivation of one of the two X chromosomes in female cells, not all X-linked genes are completely silenced. Approximately 15% of human X-linked genes consistently escape XCI and remain expressed, while an additional 10% to 32% exhibit heterogeneous or tissue-specific escape patterns depending on the individual, tissue, or cell type [19,20].

Unlike the prevailing repressive environment on the Xi, promoter regions of genes escaping XCI maintain a hypomethylated DNA state and feature an enrichment of active histone marks, such as H3K4me3 and H3K9ac. Furthermore, CCCTC-binding factor (CTCF) boundary proteins and variable chromatin looping protect their active state, preserving an open chromatin conformation [21].

A prominent class of XCI-escaping genes resides within the PARs, located at the terminal ends of the X and Y chromosomes, where meiotic recombination occurs [22]. Because males possess a single X chromosome, PAR genes must evade XCI to maintain gene dosage parity between sexes. PARs are divided into PAR1 (Xp22.33) and PAR2 (Xq28) based on their chromosomal position [22].

PAR1 harbors approximately 25 genes, nearly all of which show complete XCI escape across most cell types [20,23,24]. Genes positioned more centrally within PAR1 exhibit higher escape frequencies, whereas those near the PAR1 boundary, such as PLCXD1 and CD99, display lower escape frequencies and are silenced at higher proportions [24]. PAR2 contains approximately 10 genes, among which only a subset—including IL9R, WASH6P, and VAMP7—has been confirmed to escape XCI. Telomeric-adjacent genes like IL9R and WASH6P evade XCI, whereas the more internally located SPRY3 exhibits a unique pattern of silencing on both the X and Y chromosomes due to position effects [20,25].

Outside the PARs, genes with functional Y-linked gametologs (NLGN4X, TXLNG, EIF1AX, ZFX, USP9X, DDX3X, KDM6A, KDM5C, RPS4X, etc.) predominantly evade XCI [22]. However, numerous non-PAR genes lack Y-homologs yet still escape XCI (ARSD, STS, RBBP7, etc.) [26]. Non-PAR escapees are primarily enriched on the short arm of the X chromosome (Xp); protein-coding escapees on the long arm (Xq) are relatively scarce, limited to genes such as RPS4X and HTR2C. Among these, HTR2C represents a classic Xq escapee lacking a Y counterpart, and non-coding lncRNA genes like JPX are also known to escape XCI on Xq. Notably, non-coding RNA genes exhibit a higher overall proportion of XCI escape compared to protein-coding genes [20].

Dosage effects and disease susceptibility driven by XCI escape

Sex-specific gene dosage disparities arising from complete and variable XCI escape represent a key epigenetic determinant of physiological immune function, central nervous system development, and sex-biased disease susceptibility [27].

1). Immune function & autoimmunity

Immune-related genes that escape XCI are expressed at higher levels in females, markedly enhancing antigen recognition and immune defense capabilities against pathogen infections compared to males [12]. However, this elevated immune dosage can disrupt immune tolerance, imposing a trade-off that substantially increases the risk of autoimmune diseases in females [28]. Furthermore, aberrations in the XCI process resulting in the preferential expression of a mutant allele (skewed XCI) can accelerate disease pathogenesis [29].

2). Neurodevelopmental disorders & mosaic buffering

The X chromosome is highly enriched with genes critical for cognitive function and brain development, and their XCI escape or variable expression drives significant sex disparities in the incidence and phenotypic presentation of neurodevelopmental disorders (NDDs) [12].

In hemizygous males (XY), a single X-linked loss-of-function mutation leads to a complete absence of normal gene function, resulting in higher incidence rates and severe phenotypic manifestations. In contrast, female (XX) heterozygotes display somatic mosaicism due to random XCI, where cells expressing the wild-type allele coexist alongside those expressing the mutant allele. This cellular mosaicism, combined with dosage compensation via XCI escape of the wild-type allele, exerts a protective "buffering" effect that substantially mitigates disease severity in female patients [12].

Cancer relevance: EXITS genes and the two-hit hypothesis

Analogous to its roles in the nervous and immune systems, XCI escape serves as a pivotal genetic driver of sex disparities in cancer incidence and prognosis [1,10]. In particular, genes escaping XCI that exert tumor-suppressive functions are designated as EXITS [10].

EXITS genes are expressed biallelically in female cells, maintaining two functional alleles [10,12]. From the perspective of Knudson's "Two-Hit" hypothesis, this grants females a robust "dosage protection" mechanism [10,30]. Specifically, even if a somatic mutation impairs one allele in a female cell, tumor suppressor activity is preserved because the corresponding allele on the Xi escapes XCI and remains functionally expressed [10,12,22] (Fig. 1A).

Figure 1. Dosage-protection model conferred by XCI escape at EXITS tumor-suppressor loci.

(A) In female (46,XX) cells, XCI escape at the inactive X (Xi) preserves tumor-suppressor function despite a somatic loss-of-function mutation on the active X (Xa), so a “Two-Hit” is required for complete loss of function. (B) In male (46,XY) cells, no compensating allele is available: the Y-linked locus either has no homolog or, when present, has insufficient activity (e.g., KDM6A–KDM6C), so a single mutation (“Single-Hit”) is sufficient to abolish tumor-suppressor function. The lower panels indicate the corresponding precision-oncology implication for each sex. ✕, somatic loss-of-function mutation; ★, XCI escape (expression maintained). Chromosome and locus positions are schematic and do not represent actual genomic coordinates. XCI, X-chromosome inactivation; EXITS, escape from X-inactivation tumor suppressors; Xa, active X chromosome; Xi, inactive X chromosome.

Conversely, males (XY), possessing only a single X chromosome, face distinct genetic vulnerability: a single loss-of-function hit in an EXITS gene completely abolishes its tumor suppressor activity (Fig. 1B) [9,10].

This sex-differential vulnerability carries direct clinical implications: female carriers of EXITS-gene alterations may be candidates for therapies that pharmacologically reactivate the compensatory allele on the Xi, whereas male carriers may benefit from shortened-interval, high-risk cancer screening protocols tailored to their single-hit susceptibility (Fig. 1).

Furthermore, recent cancer genomics studies demonstrate that "XCI erosion"—a phenomenon where repressive epigenetic marks on the Xi are lost within the tumor microenvironment, leading to incomplete silencing [31]—alongside tumor-type-specific fluctuations in XCI escape, collectively contributes to malignant progression, gene expression imbalance, and immune evasion in female cancers [31,32].

Beyond its somatic consequences within the tumor microenvironment, this male–female asymmetry in tumor-suppressor dosage protection also mirrors a broader pattern of sex-differential disease susceptibility, in which autoimmune diseases skew toward females and most cancers skew toward males. One evolutionary explanation, the pregnancy compensation hypothesis, attributes this trade-off to selection for a female immune system tolerant of antigenically foreign fetal tissue during pregnancy, with heightened autoimmune susceptibility and enhanced tumor immunosurveillance emerging as correlated consequences under modern conditions [33]. Extending this logic to EXITS loci, the persistence of XCI escape at tumor-suppressor genes may likewise reflect sex-specific selective pressures rather than a purely incidental epigenetic outcome, offering a plausible evolutionary basis for the sex-biased cancer risk discussed throughout this review [9].

X-LINKED TUMOR SUPPRESSOR GENES AND SEX DIMORPHISM IN CANCER

Tumor suppressor genes (TSGs) broadly govern essential molecular mechanisms required to maintain normal cellular homeostasis, including cell cycle regulation, DNA damage repair, RNA metabolism, chromatin remodeling, signal transduction pathway control, and apoptosis induction. Consequently, the inactivation or loss-of-function (LoF) mutations of these genes serve as critical drivers of tumorigenesis and malignant progression [34,35]. Regarding the chromosomal distribution of TSGs across the human genome, 1,180 are located on autosomes, 38 on the X chromosome, and 1 on the Y chromosome [11].

X-linked TSGs are likewise subject to regulation by epigenetic XCI and XCI escape [10,12]. For instance, genes such as KDM6A and DDX3X evade XCI to balance expression dosage with their Y-linked gametologs or maintain a doubled gene dosage through biallelic expression in female cells [10]. However, heterogeneous or random patterns of XCI escape and the accumulation of somatic mutations induce disparities in the expression levels and functional activities of TSG-encoded proteins between sexes, ultimately leading to distinct sex dimorphism in cancer initiation and progression [1,10].

According to TSGene 2.0, a database dedicated to TSGs, X-linked TSGs in the human genome comprise a total of 38 genes, including 23 protein-coding genes and 15 non-coding genes [11]. Nevertheless, among the 23 protein-coding genes, only three (DDX3X, KDM6A, and PHF6) are annotated within TSGene 2.0 itself as having documented sex-specific characteristics [10,36,37]. For the remaining 20 genes not flagged as such in the database, although their basal protein functions and general involvement in cancer have been reported, systematic research on sex-specific mutation frequencies, expression patterns, and prognostic disparities remains an unaddressed gap for most of these genes; four exceptions, identified through an independent PubMed literature search rather than through TSGene 2.0's own sex-specific annotations, are introduced below.

Therefore, to thoroughly investigate the genetic factors underlying sex dimorphism in cancer, this chapter focuses on analyzing seven key X-linked tumor suppressor genes: three genes registered in the TSGene 2.0 database (DDX3X, KDM6A, and PHF6) together with four additional tumor suppressors identified through an independent PubMed literature search — genes whose sex disparities and XCI escape characteristics have been clearly established in the literature and clinical studies despite their absence from TSGene 2.0's own sex-specific annotations (KDM5C, alpha-thalassemia mental retardation X-linked [ATRX], MAGEC3, and CNKSR2) [10,11]. The molecular functions and XCI escape characteristics of these seven genes are summarized in Table 1, and their chromosomal locations and Y-linked homologs are mapped in Figure 2.

Table 1. Molecular characteristics of the seven X-linked tumor suppressor genes reviewed in this article.

Gene	Molecular function	Y-Homolog	XCI escape status	Major cancer types	Key sex-biased finding [Reference]
DDX3X	DEAD-box RNA helicase; mRNA processing, translation initiation, RNA transport, telomere maintenance	DDX3Y	Escapes XCI	Breast, colorectal, HNSC, KIRC, LUAD, LUSC, GBM, LGG	LoF mutations occur significantly more often in males; because DDX3Y lacks tumor-suppressive activity equivalent to DDX3X, mutational loss more readily abolishes DDX3X function in males [10,39,40,43].
KDM6A	Lysine-specific demethylase 6A; removes repressive H3K27me3 mark to promote differentiation- and tumor-suppression-related transcription	KDM6C (low enzymatic activity)	Escapes XCI (protein dosage still sex-biased)	Bladder (muscle-invasive), melanoma	~26% of muscle-invasive bladder cancers harbor KDM6A mutations; decreased expression negatively affects survival in female but not male patients [47,50].
KDM5C	Lysine-specific demethylase 5C; removes methyl groups from histone H3 lysine 4 (H3K4)	KDM5D	Escapes XCI	ccRCC, GBM, thyroid	Mutated in ~20% of ccRCC patients with a strong male bias; KDM5C mutation counteracts the tumor-suppressive effect of KDM5D mutation, reflecting functional disparity between the two gametologs [54].
ATRX	ATP-dependent chromatin remodeler; H3.3 deposition at telomeres and PRC2 recruitment to XIST-targeted chromatin	—	Escapes XCI (inferred from biallelic expression in female cells)	IDH-mutant astrocytoma, neuroblastoma, gastric	Mutations are frequent in IDH-mutant astrocytoma, a subtype with higher male incidence, driving ALT-pathway activation; biallelic ATRX expression in female cells is thought to confer enhanced protection [60,61].
PHF6	PHD-like zinc finger protein; transcriptional regulation via chromatin binding	—	Rarely escapes XCI	T-ALL, AML	Mutations occur exclusively in male T-ALL cells; biallelic expression is absent in female T-ALL cells, implying rare XCI escape [35,67].
MAGEC3	Cancer-testis antigen (MAGEC family); linked to DNA-repair- and cell-cycle-related chromatin accessibility	—	Escapes XCI	Colorectal, esophageal, ovarian	Across 21 cancer types, mutations and loss-of-function variants occur more frequently in male patients [10].
CNKSR2	Multidomain Ras-signaling scaffold protein; inhibits apoptosis in specific cancer cells	—	Undetermined (evidence insufficient and tissue-dependent; possible escape in LUAD)	Cervical, papillary thyroid, LUAD	LoF mutations occur more frequently in male cancer patients [10,72].

XCI, X-chromosome inactivation; LoF, loss-of-function; ccRCC, clear cell renal cell carcinoma; GBM, glioblastoma multiforme; HNSC, head and neck squamous cell carcinoma; KIRC, kidney renal clear cell carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; LGG, low-grade glioma; T-ALL, T-cell acute lymphoblastic leukemia; AML, acute myeloid leukemia; ALT, alternative lengthening of telomeres. Reference numbers correspond to the same in-text citation numbers already used at each fact’s location in the manuscript body.

Figure 2. Cytogenetic map of the seven X-linked tumor suppressor genes reviewed in this article on the X (A) and Y (B) chromosomes.

Red text indicates that a gene has a Y-linked homolog; black text indicates that it does not. Bold text indicates that the gene escapes XCI; regular weight indicates that it rarely or does not escape; † indicates that XCI-escape status is undetermined for CNKSR2: current evidence remains insufficient and may be tissue-dependent, with expression data suggesting possible escape only in lung adenocarcinoma. PAR1 (Xp22.33/Yp11.32) and PAR2 (Xq28/Yq12 termini) are pseudoautosomal regions that recombine between the X and Y chromosomes at each meiosis, so genes located within them retain two copies in both sexes regardless of XCI-escape status. XCI, X-chromosome inactivation; PAR1, pseudoautosomal region 1; PAR2, pseudoautosomal region 2.

DDX3X

DDX3X encodes a DEAD-box RNA helicase, and its expressed protein participates in critical RNA metabolic processes, including post-transcriptional mRNA processing, translation initiation, RNA transport, and telomere maintenance [38]. It escapes XCI and functions as a gametolog to DDX3Y on the Y chromosome. DDX3X is implicated in various malignancies, including breast, colorectal, lung, liver, prostate, and squamous cell carcinomas [39]. In cancer, DDX3X mutations exhibit distinct sex dimorphism, being discovered more frequently in males. Beyond oncology, sex disparities associated with DDX3X are also observed in IFN-I expression within immune cells and in intellectual disability phenotypes driven by DDX3X mutations [40,41].

In breast cancer, DDX3X is overexpressed and acts as an oncogene that promotes the proliferation of mammary epithelial cells. Hypoxia-inducible factor 1-alpha (HIF-1α) binds to the HIF-response element located in the DDX3X promoter, activating transcription and thereby upregulating DDX3X expression. Furthermore, DDX3X interacts directly with estrogen receptor alpha (ERα) independently of estrogen presence, enhancing ERα activity and target gene expression. This suggests that DDX3X can substitute for estrogen signaling to accelerate tumor progression even in estrogen-deprived environments [39,42].

In colorectal cancer (CRC), conflicting findings coexist regarding whether DDX3X functions as an oncogene or a tumor suppressor. DDX3X promotes the binding of the SP1 transcription factor to the KRAS promoter, inducing KRAS expression. In cancer cells harboring wild-type KRAS, elevated KRAS levels trigger reactive oxygen species (ROS) generation, which subsequently increases HIF-1α expression, thereby promoting tumor progression similarly to the mechanism observed in breast cancer [43]. Conversely, other studies demonstrate that CRC patients with lower DDX3X expression experience worse prognoses and higher frequencies of distant metastasis [44,45]. Loss of DDX3X activates the mitogen-activated protein kinase (MAPK) pathway, reducing E-cadherin expression and enhancing β-catenin signaling, which ultimately increases the proliferation and motility of CRC cells [45].

In male tumors, DDX3X is frequently downregulated even in the absence of somatic mutations, and LoF mutations are observed significantly more often in males than in females [10]. Higher DDX3X expression levels in females compared to males have been documented in head and neck squamous cell carcinoma (HNSC), kidney renal clear cell carcinoma (KIRC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), glioblastoma multiforme (GBM), and low-grade glioma (LGG) [10]. Because females express DDX3X from both X chromosomes due to XCI escape, whereas males express it from a single X chromosome, mutational events in males result in a complete loss of DDX3X expression. Moreover, because the Y-linked gametolog DDX3Y lacks equivalent tumor-suppressive activity to DDX3X, this expression disparity accounts for the sex differences in cancer incidence driven by DDX3X mutations [10].

DDX3X also drives sex differences in IFN-I expression, which plays a central role in antiviral immunity. IFN-I enhances T cell-mediated immunity and suppresses metastatic signaling pathways in cancer cells, exhibiting potent anti-tumor effects such as apoptosis induction and angiogenesis inhibition [46]. Higher IFN-I expression is therefore associated with augmented anti-tumor efficacy and reduced oncogenic risk. In plasmacytoid dendritic cells (pDCs)—the primary source of IFN-I in humans—females exhibit higher DDX3X expression than males, with elevated DDX3X levels inducing a stronger IFN-α response in female pDCs. Conversely, inhibition of DDX3X results in lower IFN-I concentrations in males compared to females [40]. These findings suggest that sex differences in DDX3X-mediated IFN-I responses partially contribute to the higher incidence of cancer observed in males.

DDX3X is also implicated in X-linked intellectual disability. In males, LoF mutations are often lethal, whereas in females, they cause muscle spasticity and autism spectrum disorders. Meanwhile, heterozygous females carrying inherited hypomorphic variants remain unaffected, whereas males inheriting these variants present with non-syndromic intellectual disability [41].

KDM6A

KDM6A encodes lysine-specific demethylase 6A, a chromatin remodeler that erases the repressive histone mark (H3K27me3) to promote gene transcription [38]. Cooperating with the SWI/SNF complex, KDM6A activates the transcription of genes involved in cellular differentiation and tumor suppression [38]. KDM6A possesses a Y-linked gametolog, KDM6C, and is known to escape XCI. However, despite escaping XCI, sex differences in protein expression levels persist, largely because the lysine demethylase activity of the Y-linked KDM6C is inherently low [47,48]. Specifically, whereas females achieve full KDM6A protein production through the expression of both X alleles, males rely on a single KDM6A allele alongside the enzymatically less active KDM6C, leading to a functional disparity in protein dosage. The presence of KDM6A mutations in cancer cells leads to a loss of histone lysine demethylase activity, indicating that KDM6A mutations promote oncogenesis by disrupting its tumor-suppressive functions [49]. In a 2009 study analyzing female cancer cell lines, cases harboring a KDM6A mutation exhibited a significantly higher tendency for the remaining wild-type allele to be lost as well [49]. This demonstrates that complete inactivation of both alleles is required for tumorigenesis in females. Conversely, although males possess only a single copy of KDM6A, it has been suggested that the Y-linked gametolog KDM6C may functionally compensate for it; notably, male cancer cell lines with KDM6A mutations exhibited a significantly higher frequency of concurrent KDM6C loss [49].

Clinically, bladder cancer occurs more frequently in male patients than in female patients, and KDM6A mutations significantly influence bladder cancer development. Approximately 26% of patients with muscle-invasive bladder cancer harbor KDM6A mutations, and mutant tumor tissues show a reduced presence of tumor-infiltrating immune cells, thereby dampening immune responses [48]. KDM6A upregulates the expression of the p53 target gene cyclin-dependent kinase inhibitor 1A (CDKN1A) and p53 apoptosis effectors, suggesting that it protects females against bladder cancer through a KDM6A-dependent mechanism [48]. Furthermore, loss of KDM6A inhibits glycolytic flux and reduces intratumoral lactate accumulation, thereby attenuating histone lactylation (H3K9la and H3K18la) in regulatory T cells (Tregs). Consequently, KDM6A mutations induce the downregulation of key immunosuppressive genes such as FOXP3, TGFB, and PDCD1, and suppress PD-1 expression in Tregs [50]. While KDM6A mutations and decreased expression had no significant impact on the overall survival of male bladder cancer patients, they negatively affected survival outcomes in female patients [51].

In melanoma, KDM6A expression was similarly higher in female patients compared to male patients. KDM6A exhibits elevated expression in female endocrine and lymphoid tissues. Gene Set Enrichment Analysis (GSEA) revealed that the expression of genes associated with the IFN-γ pathway was observed exclusively in female patients and was entirely absent in male patients. These findings provide a mechanistic basis for the superior overall survival observed in female melanoma patients with higher KDM6A expression levels [52].

KDM5C

KDM5C encodes lysine-specific demethylase 5C, a nuclear chromatin regulator that removes methyl groups from histone H3 lysine 4 (H3K4) [38]. Through its H3K4me2/3 demethylase activity, KDM5C participates in extensive epigenetic regulatory networks across physiological and pathological contexts [53]. KDM5C possesses a Y-linked gametolog, KDM5D, and typically escapes XCI [10,38].

KDM5C is implicated in various malignancies, including acute myeloid leukemia (AML), glioblastoma multiforme (GBM), nasopharyngeal carcinoma, breast cancer, pancreatic cancer, hepatocellular carcinoma, colorectal cancer, and renal cell carcinoma. Somatic mutations have been identified in renal cell carcinoma, pancreatic cancer, and pediatric acute leukemia, whereas KDM5C overexpression is associated with colorectal, breast, and ovarian cancers. Conversely, downregulation of KDM5C correlates with genomic instability in clear cell renal cell carcinoma (ccRCC) [53,54].

Males exhibit an approximately two-fold higher risk of developing ccRCC compared to females. KDM5C mutations are detected in approximately 20% of ccRCC patients, showing a particularly strong bias toward males [55]. In males, the Y-linked gametolog KDM5D partially overlaps with KDM5C in function, though their functional roles are not entirely identical. Whereas KDM5C mutations reduce metastatic potential, KDM5D mutations suppress tumor growth. However, concurrent mutations in both genes restore the tumor growth that had been suppressed by the KDM5D mutation alone. Conversely, phenotypic changes driven by KDM5C mutations are not rescued by KDM5D mutations. This indicates that even if KDM5C escapes XCI to maintain equal gene dosage between sexes, KDM5C mutations counteract the tumor-suppressive effect of KDM5D mutations, highlighting how functional disparities between these two gametologs contribute to sex-specific oncogenesis [55].

GBM is pathologically characterized by tumor hypoxia. KDM5C functions as an oxygen sensor and is co-expressed with HIF1A, p75, and BIRC5 (survivin) under hypoxic conditions [54]. Higher expression levels of KDM5C positively correlate with elevated expression of HIF1A that encodes hypoxia-inducible factor 1-alpha. Because HIF1A overexpression is linked to poor clinical outcomes, this suggests that KDM5C may similarly contribute to unfavorable prognoses in GBM. Hypoxia-induced stabilization of HIF-1α is mediated by p75 upregulation, which enhances GBM aggressiveness. Furthermore, HIF-1α regulates the expression of BIRC5, a protein key to tumor progression whose elevated expression is significantly associated with adverse survival outcomes in patients with GBM. The p75–HIF1A–KDM5C–BIRC5 signaling axis represents a crucial hypoxia-mediated oncogenic cascade with significant therapeutic potential [54].

KDM5C deficiency demonstrates striking sex disparities in the repression of germline genes. Female epiblast-like cells (EpiLCs) are more severely impacted by KDM5C loss than male EpiLCs, which is attributed to higher baseline KDM5C expression in females resulting from XCI escape. Although heterozygous females harboring a single mutant KDM5C allele would theoretically express levels similar to wild-type males, and KDM5C-deficient males lack KDM5C expression entirely, one might expect males to exhibit more differentially expressed genes (DEGs) in the germline network. Surprisingly, heterozygous females exhibited more than twice as many germline DEGs compared to KDM5C-deficient males [56].

Thyroid cancer occurs more frequently in females than in males; however, male patients often present with more aggressive tumors and higher mortality rates. The KDM family of genes is known to partially mediate these sex differences. KDM5C and KDM6A mRNA levels were found to be elevated in female thyroid cancer tissues relative to normal thyroid tissues. In normal thyroid tissues, female KDM5C mRNA expression was 2.67-fold higher than in males [57]. Although KDM5C was detected in both normal and cancerous female thyroid tissue samples, nuclear loss of KDM5C was frequently observed in thyroid cancer tissues. Real-time PCR analysis of normal thyroid tissues similarly confirmed significantly higher KDM5C expression in females. Taken together, these findings suggest that elevated KDM5C expression in females plays a protective role against thyroid cancer development [57].

ATRX

ATRX encodes a chromatin remodeler that utilizes ATP-dependent activity to regulate chromatin organization and gene expression [38]. Belonging to the SNF2/RAD54 helicase family, ATRX induces H3.3 histone deposition at telomeres, pericentric heterochromatin, and imprinted loci via the ATRX–DAXX complex [38]. Failure of H3.3 deposition leads to global alterations in chromatin structure and the continuous activation of chronic DNA damage signaling cascades [58]. Furthermore, ATRX directly binds to the repeat A region of Xist lncRNA to recruit the PRC2 complex, which subsequently trimethylates H3K27 to silence gene transcription [59,60]. ATRX mutations are frequently identified in various malignancies, including neuroblastoma, pancreatic neuroendocrine tumors, and osteosarcoma [61].

In certain brain tumors, such as isocitrate dehydrogenase (IDH)-mutant astrocytomas which exhibit a higher incidence in males, ATRX mutations frequently occur, upregulating the alternative lengthening of telomeres (ALT) pathway in tumor cells and thereby accelerating disease progression. This suggests that biallelic ATRX expression in female cells confers enhanced protection against oncogenesis [62]. ATRX mutations almost exclusively co-occur with IDH mutations and are rarely observed as solitary events [63]. In most glioblastomas, the combination of ATRX loss and IDH mutation leads to immunosuppression; however, IDH/ATRX co-mutated astrocytomas exhibit greater immune cell infiltration than oligodendrogliomas harboring IDH mutations with wild-type ATRX. These opposing immunological effects of ATRX loss depend on the sequence and timing of ATRX mutations relative to IDH mutations, which is attributed to intrinsic epigenetic alterations within the chromatin [61].

ATRX mutations profoundly impact tumor biology and prognosis in neuroblastoma. ATRX deficiency predominantly manifests in patients aged 5 years or older, presenting with indolent disease progression yet yielding a low overall survival rate [64]. These mutations induce defects in telomeric H3.3 deposition, leading to telomere maintenance via the ALT pathway while simultaneously repressing genes governing neuronal differentiation, thereby creating a microenvironment that sustains continuous cell proliferation. Moreover, ATRX alterations elevate DNA replication stress and genomic instability, particularly in combination with 11q deletions, allowing tumor cells to evade DNA damage-induced apoptosis. ATRX mutations are mutually exclusive with MYCN amplification, as concurrent alterations induce synthetic lethality that impairs cell viability [61].

ATRX mutations also influence gastric cancer biology. Female gastric cancer patients harboring ATRX mutations exhibited enhanced anti-tumor immunity compared to those without mutations, whereas ATRX mutation status had no significant impact on anti-tumor immunity in male patients. Crucially, this enhanced anti-tumor immunity in female patients was observed independently of sex-based immune heterogeneity [65].

PHF6

PHF6 encodes plant homeodomain (PHD) finger protein 6, a PHD-like zinc finger protein involved in transcriptional regulation [38]. PHF6 is linked to chromatin organization and regulatory pathways, including transcription control by RNA polymerase II. Mutations in PHF6 have been identified in T-cell acute lymphoblastic leukemia (T-ALL), B-cell acute lymphoblastic leukemia (B-ALL), AML, and myelodysplastic syndromes (MDS) [37]. Meanwhile, PHF6 is overexpressed in various cancers, including breast, colorectal, gastric, and glioblastoma, whereas it is downregulated in esophageal cancer [66]. Immune cell infiltration patterns also vary according to PHF6 expression levels; elevated PHF6 expression is generally accompanied by an increase in immune cells associated with tumor suppression. However, no significant correlation was observed between PHF6 expression and tumor immune cell infiltration in esophageal and uterine cancers. These functional characteristics stem from direct binding of PHF6 to chromatin to regulate gene transcription, and this chromatin–transcriptional regulatory interaction is reported to influence tumor initiation and development [67].

T-ALL exhibits a higher incidence in males. In 2025, an estimated 6,100 cases of acute lymphoblastic leukemia occurred in the United States, comprising 3,450 males and 2,650 females. Mortality was likewise higher in males (720 deaths) than in females (680 deaths) [68]. Studies have shown that PHF6 mutations are identified in T-ALL but not in B-ALL, indicating that PHF6 mutations are restricted to T-ALL. In the same study, biallelic expression was absent in cells from female T-ALL patients, implying that PHF6 rarely escapes XCI. Notably, PHF6 mutations were observed exclusively in male T-ALL cells [36]. However, no significant sex differences were identified in the transcriptional expression levels of PHF6 [69].

AML patients harboring PHF6 mutations exhibit lower overall survival rates compared to non-mutated patients, a trend that is particularly pronounced in males. In contrast, PHF6 mutations alone do not significantly alter survival rates in female AML patients [37]. This suggests that the overall reduction in survival among AML patients with PHF6 mutations may largely stem from the baseline male predominance among AML cases. Indeed, in the United States in 2025, newly diagnosed male AML cases numbered 12,060 compared to 9,950 female cases, with male deaths (6,130) exceeding female deaths (4,960) by approximately 23.6% [68]. While PHF6 mutations alone have little impact on female survival, co-occurrence with RUNX1 mutations leads to worsened survival outcomes, an effect similarly observed in males [37].

MAGEC3

MAGEC3 is a member of the melanoma antigen family C (MAGEC) group that is not expressed in normal adult tissues except for the testes, but is aberrantly expressed across various histopathological cancer types [38]. Genes of the MAGE family are implicated in diverse biological processes, including DNA repair, cell cycle regulation, and protein-protein interactions [70]. MAGEC3 escapes XCI and lacks a functional gametolog on the Y chromosome. Expression levels of MAGEC3 vary depending on the specific cancer type [71]. Significantly elevated expression has been observed in GBM, HNSC, and thyroid carcinoma (THCA). Notably, its expression was most prominent in GBM, whereas moderate expression levels were generally observed in other malignant tumors [71]. In colorectal cancer, MAGEC3 expression was also significantly elevated, which correlated with worse patient prognoses compared to the other MAGE family genes [71,72].

Overexpression of MAGEC3 is similarly observed in esophageal squamous cell carcinoma [72]. MAGEC3 promotes tumor progression by activating epithelial-to-mesenchymal transition (EMT), leading to an increased risk of lymphatic metastasis and reduced overall survival rates. Furthermore, MAGEC3 has been shown to modulate cytokine secretion in T cells to suppress anti-tumor immunity, suggesting that it exerts immunosuppressive functions within the tumor microenvironment [72].

Meanwhile, a large log-fold change in TSIX expression was observed in female patients expressing the MAGEC3 protein, implying a potential association between MAGEC3 protein expression and the XCI process [70]. In ovarian cancer, elevated MAGEC3 expression was associated with favorable prognoses, whereas lower expression showed no such correlation. MAGEC3 expression is lower in ovarian cancer tissues compared to normal ovarian tissue, and high MAGEC3 expression is associated with poor prognosis. Given that tumor expression levels were not significantly elevated relative to normal tissue, these findings were interpreted not as evidence of a direct tumor-promoting function of MAGEC3, but rather as reflecting a distinct disease subtype unrelated to MAGE-driven mechanisms. In addition, tumors exhibiting high MAGEC3 expression without MAGE induction may retain potential DNA repair functions associated with MAGE proteins. The clustering of differentially expressed genes related to MAGEC3 at specific genomic loci supports the hypothesis that MAGEC3 influences chromatin accessibility through stress response programs tied to cell cycle control and DNA repair [70].

Sex-specific differences in MAGEC3 alteration patterns have also been confirmed. Across 21 cancer types, MAGEC3 mutations occurred more frequently in male patients; LoF variants were similarly observed at a higher frequency in males [10].

CNKSR2

CNKSR2 encodes a multidomain scaffold protein, Connector Enhancer of Kinase Suppressor of Ras 2, that mediates downstream Ras signaling. The expression of this protein is induced by vitamin D and inhibits apoptosis in specific cancer cells [38]. Additionally, it functions as a molecular adaptor in signal transduction cascades and participates in postsynaptic organization and specialization assembly [38]. Downregulation of CNKSR2 mRNA expression has been documented in pancreatic cancer, prostate cancer, rectal cancer, melanoma, gastric cancer, and cervical squamous cell carcinoma [73].

LoF mutations in CNKSR2 are observed more frequently in male cancer patients. Unlike EXITS genes with a Y-linked gametolog, CNKSR2 lacks a comparably direct reference for assessing dosage compensation; nevertheless, current evidence for its XCI-escape status remains insufficient and may be tissue-dependent rather than definitively absent. Analysis of CNKSR2 expression levels in non-mutated tumor tissues revealed elevated expression in female LUAD cases, suggesting that CNKSR2 may escape XCI in a tissue-specific manner in LUAD [10].

During tumorigenesis and malignant progression, immunosuppressive molecules disrupt immune synapse formation, thereby repressing the activation of tumor-reactive T cells [74]. Conversely, natural killer (NK) cells activate dendritic cell precursors to facilitate immune synapse formation and promote T-cell activation [75]. CNKSR2 functions in regulating synaptic assembly, and in cervical cancer with high CNKSR2 expression, extensive infiltration of diverse immune cell populations has been observed. This indicates that CNKSR2 may act to promote anti-tumor immune responses [73].

CNKSR2 is also frequently overexpressed in papillary THCA (PTC), where its expression is associated with downregulation of an oxidative-phosphorylation gene set; more broadly, the related KSR/CNKSR scaffold family (KSR1, KSR2, and CNKSR1) is coordinately linked to Notch-pathway activation, and functional knockdown of KSR1 confirmed that this family is required for BRAFV600E-induced extracellular signal-regulated kinase (ERK) phosphorylation and NOTCH1 signaling in PTC cells [76]. This supports a role for CNKSR2 in promoting tumor progression in PTC as well [73].

CONCLUSION AND FUTURE PERSPECTIVES

The male bias in incidence and aggressiveness observed across most non-reproductive cancers extends beyond simple lifestyle differences or sex hormone variations; it is deeply rooted in genetic and epigenetic imbalances of the sex chromosomes [77]. These genetic and epigenetic mechanisms are not mutually exclusive with the hormone-based hypotheses discussed in the introduction, but rather function complementarily to shape sex differences in cancer. In this review, we focused on the dosage-protection mechanism conferred by seven X-linked tumor suppressor genes with varying degrees of XCI escape—from established EXITS genes to genes with limited or undetermined escape status (KDM6A, KDM5C, DDX3X, ATRX, PHF6, MAGEC3, and CNKSR2) [10]. While female cells reduce their susceptibility to the "two-hit" model by compensating LoF mutations on one X chromosome with normal allele expression from the Xi, males are exposed to a "single-hit vulnerability" wherein a single mutation suffices to disrupt the tumor suppressor network [30].

However, among the 23 protein-coding X-linked tumor suppressor genes cataloged in the TSGene 2.0 database, only three (DDX3X, KDM6A, and PHF6) were among the seven genes analyzed above; the remaining twenty genes (including BTK, DMD, DUSP9, FHL1, FLNA, GPC3, FOXO4, RBBP7, RPL10, ZNF185, SRPX, TREX2, RBMX, RPS6KA6, FOXP3, TCEAL7, EDA2R, BCORL1, BEX2, AMER1) remain critical research gaps that have not yet been fully elucidated in sex-difference cancer studies [11]. The four other genes discussed above—KDM5C, ATRX, MAGEC3, and CNKSR2—were identified through an independent literature search rather than from TSGene 2.0's own X-linked tumor-suppressor list, and are therefore not counted among these twenty. The primary reasons for this knowledge gap include:

Facultative and tissue-specific XCI escape

Beyond the twenty genes listed above, other X-linked tumor suppressors show the same pattern; for example, STAG2 (a component of the cohesin complex) exhibits partial escape depending on specific tissue types or cellular differentiation stages. Consequently, standard bulk genomic data analyses have failed to effectively demonstrate the female-specific dosage protection effect [78].

Limitations in statistical power of cohorts

Likewise, mutations in MED12 and BCOR are concentrated in specific cancer types (e.g., AML and brain tumors), resulting in insufficient sample sizes within large-scale cancer genomic cohorts (such as TCGA) to establish statistical significance across sex subgroups [79].

Inadequate application of single-cell multi-omics

High-dimensional analytical approaches capable of measuring XCI erosion (incomplete inactivation of Xi) driven by epigenetic instability or allele-specific expression (ASE) at the individual cell level have not yet been actively implemented [31].

Molecular elucidation of X-linked tumor suppressor genes and XCI escape provides a decisive opportunity for the realization of sex-specific cancer prevention and precision oncology [10]. For high-risk male populations harboring mutations in the genes reviewed herein (KDM6A, KDM5C, DDX3X, ATRX, PHF6, MAGEC3, and CNKSR2), where a single mutation dismantles the tumor suppressor network, establishing sex-specific cancer risk profiling and screening guidelines with shortened examination intervals is an urgent priority. Conversely, in female cancer patients, personalized therapeutic strategies leveraging epigenetic modulators, such as DNA methyltransferase (DNMT)/histone deacetylase (HDAC) inhibitors, can be explored to induce epigenetic reactivation (X-reactivation) of the Xi or to sustain demethylase activity [31].

For high-risk male patients, incorporating EXITS-gene mutation status into risk-stratified screening algorithms could enable earlier detection, following the precedent of BRCA1/2-based screening models already established for breast and ovarian cancer [80]. Tumors arising from loss of EXITS-gene dosage protection may also represent candidates for synthetic-lethality-based therapeutic strategies that exploit their heightened dependence on compensatory pathways [81]. Furthermore, single-cell multi-omics-based follow-up studies should be prioritized for the remaining twenty uncharacterized TSGs. Ultimately, incorporating the gene dosage imbalance of X-linked TSGs—including these twenty candidates—as an essential variable in drug development and clinical trial design for preventive interventions will overcome sex disparities in cancer development and progression, thereby establishing true sex-specific precision medicine [77,79].

Funding Statement

FUNDING This Research was supported by Duksung Women's University Research Grants 2025.