Sex differences in non-reproductive cancers: mechanistic roles of sex-hormone signaling

Abstract

Sex differences in the incidence and outcomes of non-reproductive cancers persist across many tumor types, even after adjustment for major exposure- and care-related factors. This review examines how sex-hormone signaling may contribute to these patterns through tumor-intrinsic mechanisms and regulation of the tumor microenvironment. In tumor cells, ER, AR and PR mediate classical nuclear transcriptional programs, whereas membrane-associated or cytoplasmic receptor pools, together with GPER, support rapid non-genomic signaling through PI3K/AKT, MAPK/ERK and related kinase or second-messenger pathways. Intratumoral steroid handling can create local ligand conditions that differ from circulating hormone levels and modify context-specific receptor activity. Hormonal context may also influence vascular, stromal and immune phenotypes. Clinically, sex-hormone-related tumor states may be better captured by integrated activity-based readouts, including receptor status, pathway activation, local steroid availability and immune context, rather than receptor immunostaining alone. Overall, sex-hormone signaling offers a hypothesis-generating framework for understanding sex-biased tumor biology beyond traditional hormone-driven cancers, but its clinical relevance requires further mechanistic and prospective validation.

Introduction

Cancer remains a leading cause of morbidity and mortality globally, with an estimated 20 million new cases and 9.7 million deaths in 2022 [1]. While sex specificity is intrinsic to cancers of the reproductive organs, sex differences extend beyond these malignancies. Notably, consistent sex disparities are observed across a broad spectrum of non-reproductive cancers [2–4]. Multiple studies indicate that, across various regions and age groups, men consistently exhibit higher incidence and mortality rates than women in non-reproductive cancers such as liver, esophageal, gastric, bladder, and renal cancers. Conversely, women tend to have higher rates of certain cancers, including thyroid and gallbladder [1, 5–7]. These patterns are broadly consistent across regions and age groups, indicating that sex disparities in non-reproductive cancers are unlikely to be explained by measured exposures or clinical covariates alone. Existing research indicates that cancer disparities between males and females are shaped by both biological sex-related mechanisms and sociocultural gender-related factors [4, 8]. This review acknowledges the latter but focuses primarily on the biological mechanisms underlying sex differences in non-reproductive cancers. Figure 1 provides a conceptual overview of the major biological and gender-related factors that may shape these disparities, while highlighting the biological focus of the review. In summary, these observations suggest that sex-related cancer disparities arise from interacting biological and exposure-related factors, with downstream effects on tumor biology and treatment response [9, 10].

Fig. 1.

A framework linking sex and gender to hormone decoding and sex-biased cancer phenotypes. Global sex disparities in non-reproductive cancers arise from the combined impact of biological sex (chromosomes, gonads, life-course hormonal milieu) and gender- and environment-related factors (behaviors, exposures, healthcare access). These inputs converge at a central hub involving sex hormone–receptor interactions, where both systemic and local ligand availability are decoded by tumor and host compartments. Downstream, hormone-conditioned states influence tumor-intrinsic biology (e.g., genome maintenance, oxidative stress, checkpoint control, oncogenic insults) and tumor immune microenvironment (TME) architecture, as revealed by sex-stratified multi-omics. Ultimately, these factors affect clinical outcomes, including differential responses and toxicities to endocrine, immune, and targeted therapies. Abbreviations: E2, estradiol; T, testosterone; DHT, dihydrotestosterone; P4, progesterone; ER, estrogen receptor; AR, androgen receptor; PR, progesterone receptor; GPER, G protein–coupled estrogen receptor; TME, tumor microenvironment

Building on these epidemiological observations, the concept of “sex as a biological variable” provides a useful framework for considering why sex differences in cancer incidence and survival may persist after adjustment for lifestyle and clinical factors [7, 9, 11, 12]. These differences may arise from both sex-chromosome complement and the sex-hormone milieu, because sex differences are mainly determined by chromosomal and hormonal factors [13, 14]. Sex chromosomes may influence tumor biology through gene-dosage effects, escape from X-chromosome inactivation, Y-chromosome loss, and chromatin or immune regulatory programs [15, 16]. These mechanisms may operate independently of circulating hormones or interact with endocrine signals in a tissue-specific manner. At the same time, sex hormones shape receptor-dependent programs throughout development and adulthood [17–19]. Because a full review of sex-chromosome biology is beyond the scope of this article, we focus here on sex-hormone signaling while recognizing it as one component of a broader framework of sex-biased cancer biology. Within this broader framework, sex-hormone signaling offers a focused mechanistic entry point because it connects systemic endocrine exposure, local receptor activity, intratumoral steroid metabolism, and tumor–microenvironmental regulation across diverse tumor sites [17]. Sex hormones include estrogens, androgens, and progesterone, with estradiol (E2), testosterone (T), and dihydrotestosterone (DHT) representing major bioactive ligands. These hormones exert their effects by binding to receptors and activating downstream signaling pathways. Steroid hormone receptors, including estrogen receptors (ERα and ERβ), androgen receptor (AR), and progesterone receptor (PR), belong to the nuclear receptor superfamily and classically function as ligand-regulated transcription factors. However, their activity is not confined to the nucleus. Membrane-associated or cytoplasmic pools of steroid hormone receptors can also mediate rapid kinase-coupled signaling, as can the G protein-coupled estrogen receptor (GPER) [18, 19].

Sex-hormone signaling is frequently cited as a key factor in cancer biology, but it is not routinely considered an explicit, cross-cancer mechanism [20]. This clinical relevance is clearest in classical hormone-responsive malignancies, including ER-positive breast cancer, androgen-dependent prostate cancer and selected endometrial cancers, where steroid-receptor signaling is an established therapeutic target [21–23]. However, it is becoming increasingly clear that these steroid pathways extend beyond these canonical cancer sites. They intersect with fundamental carcinogenic processes such as oxidative stress and genomic instability, and in virally driven cancers, they may influence both viral oncogenesis and host responses to environmental carcinogens [24–26]. Another layer of complexity arises from immune modulation. Experimental and translational research suggests that hormonal exposure can alter myeloid cell states and cytokine signaling [27], as well as influence checkpoint regulation and broader effects on the tumor immune microenvironment [28]. These findings highlight that steroid hormone signaling can shape both tumor-cell programs and the immune microenvironment beyond classical hormone-dependent cancers.

Although epidemiological, multi-omics, and clinical studies have increasingly documented sex-related differences in cancer incidence, outcomes, and tumor biology [4, 29, 30], the evidence remains fragmented across tumor types and biological compartments. Many studies focus on individual hormones, receptors, immune features, or molecular programs [31, 32], whereas fewer integrate systemic endocrine exposure, local steroid metabolism, receptor activity, tumor-intrinsic signaling, and immune regulation into a cross-cancer framework. Recent studies have further extended these observations to therapeutic and immune regulation, including the role of sex hormones in antitumor immunity and immune checkpoint inhibitor (ICI) responses [28], as well as AR pathways in male-predominant tumors [33, 34]. However, sex-hormone signaling is still commonly framed as a context-specific modifier rather than a unified cross-cancer mechanism. Consequently, current mechanistic models do not fully account for the epidemiological patterns of sex-biased cancer risk and outcomes [27]. Bridging this gap is critical for improving risk stratification and therapy selection. Table 1 therefore summarizes key knowledge gaps and research priorities for sex-hormone signaling across selected non-reproductive cancers.

Table 1.

Principal knowledge gaps and research priorities for sex-hormone signaling in selected non-reproductive cancers

Cancer type	Reported sex-bias pattern	Hormone focus	Principal unresolved question/ research priority	Evidence status	Representative references (PMID)
Hepatocellular carcinoma	Male-predominant incidence and mortality in many populations	AR; estrogen-related signaling; local steroid metabolism	Clarify whether hormone-related signaling contributes directly to sex disparity or mainly reflects underlying inflammatory and metabolic liver injury; determine why estrogen-related effects appear protective in some models but not others.	●	32,290,381; 36,563,929; 40,846,185
Bladder cancer	Strongly male-predominant	AR; ER signaling; local steroid milieu	Determine whether AR-associated findings represent true pathway dependence; define why local androgen availability may differ substantially from circulating levels and reconcile mixed estrogen-related findings.	●	32,967,818; 37,666,817; 39,342,807
Thyroid cancer	Female-predominant, with age-related variation	ERα/ERβ expression or signaling balance	Establish whether altered ERα/ERβ balance contributes to female predominance, disease progression, or both, and determine how menopausal or age-related hormonal changes modify these associations.	◐	28,124,274; 40,137,860; 40,716,901
Gallbladder	Gallbladder cancer is female-skewed in many populations; patterns differ across biliary sites	Hormone exposure; gallstone-related inflammation	Determine whether hormone-related associations contribute to sex disparity or mainly reflect associated biliary disease, gallstone risk, menopausal hormone exposure and local inflammatory states.	◐	24,634,588; 34,766,362; 41,791,643
Colorectal cancer	Historically male-predominant overall; age- and cohort-specific patterns are changing	ERβ; estrogen exposure and metabolism	Define whether estrogen-related mechanisms contribute differently across age groups and disease settings, particularly in view of changing early-onset incidence patterns and limited causal evidence for circulating estradiol.	◐	39,030,619; 39,232,785; 41,675,768
Lung cancer	Sex differences vary by histology, smoking status, and molecular subtype	ER signaling; aromatase; hormone–immune associations	Identify which tumor subsets show functional estrogen-related activity and determine whether observed links with immune features reflect biological or therapeutic relevance rather than model-specific associations.	○	37,370,722; 37,285,115; 39,754,298

This table presents selected knowledge gaps and research priorities for sex-hormone signaling in non-reproductive cancers. It is intended to orient readers to representative unresolved questions and is not a definitive causal or quantitative comparison across cancer types. Evidence status reflects a qualitative judgment based on the reported sex-bias pattern, biological plausibility, availability of hormone-related evidence, and remaining mechanistic uncertainty. Representative references support the inclusion of each cancer type and research focus but do not imply causal proof

Evidence status: ● = relatively well-studied sex-bias pattern with substantial but incompletely resolved mechanistic evidence; ◐ = established relevance with dispersed or inconsistent evidence; ○ = preliminary evidence with mechanisms requiring further validation

In summary, epidemiological, mechanistic, and multi-omics studies suggest that sex hormones may contribute to cancer risk, tumor biology, and treatment response beyond reproductive cancers. This review synthesizes current evidence on sex-hormone signaling in non-reproductive cancers, focusing on tumor-intrinsic receptor programs, local steroid metabolism, non-genomic signaling, immune and microenvironmental regulation, and the implications for biomarker development and clinical trial design.

Tumor-intrinsic mechanisms of sex-hormone signaling

Sex-hormone signaling in tumor cells operates through two primary mechanisms: classical nuclear receptor pathways and non-classical membrane-initiated pathways [19]. The classical pathway involves nuclear hormone receptors—ERα, ERβ, AR, and PR—which, upon ligand binding, act as transcription factors to modulate enhancer–promoter interactions and reprogram cellular states [35]. Concurrently, membrane-associated receptors, including both the same nuclear receptors localized to the plasma membrane and the specialized GPER, mediate rapid non-genomic signaling. These pathways transduce hormonal signals to swiftly activate PI3K/AKT and MAPK/ERK kinase cascades, modulating cell proliferation and survival [36, 37]. The relative contribution of each pathway is influenced by intratumoral steroidogenesis and tissue-specific cofactor availability. These integrated mechanisms may contribute to sex-biased transcriptional and signaling outputs, even in non-reproductive tissues. Figure 2 provides an overview of these nuclear and membrane-initiated pathways.

Fig. 2.

Tumor-intrinsic decoding of sex-hormone signals integrates genomic receptor programs with rapid kinase-coupled signaling. Sex hormones engage classical (genomic) nuclear receptor signaling, in which ligand-bound ER/AR/PR translocate to the nucleus and activate context-dependent transcription via hormone response elements. These elements are shaped by chromatin state, cofactors, and tissue-specific receptor isoforms/variants. Simultaneously, non-classical rapid signaling from membrane/cytosolic receptor pools (including GPER) activates PI3K/AKT, MAPK/ERK, SRC/PKC, and second-messenger pathways, enabling rapid state shifts. Local steroidogenesis and ligand conversion (e.g., CYP19A1 aromatase, SRD5A1/2 5α-reductases) can decouple intratumoral exposure from circulating hormones, stabilizing sex- and tissue-specific signaling states. Bidirectional pathway crosstalk (notably AR↔PI3K/AKT and ER↔MAPK/ERK) supports compensatory activation and plasticity, while kinase-driven chromatin remodeling can lock in sex-skewed transcriptional programs. Abbreviations: AR, androgen receptor; cAMP, cyclic AMP; CYP19A1, aromatase; ER, estrogen receptor; ERK, extracellular signal-regulated kinase; GPER, G protein-coupled estrogen receptor; HRE, hormone-response element; MAPK, mitogen-activated protein kinase; PI3K, phosphoinositide 3-kinase; PKC, protein kinase C; PR, progesterone receptor; SRC, SRC proto-oncogene; SRD5A1/2, 5α-reductases

Classical receptor mechanisms

Steroid hormone signaling in tumors predominantly occurs through the classical nuclear receptor axis. ERα/ERβ, AR, and PR act as ligand-activated transcription factors that convert steroid inputs into sustained transcriptional programs [38]. ERα and ERβ bind estrogens but differ in tissue distribution, cofactor preference, and transcriptional output. As a result, the same estrogenic exposure can produce distinct effects across organs. For example, ERβ is the predominant ER subtype in normal colonic epithelium and is frequently reduced during colorectal tumorigenesis, whereas ERα has been linked to estrogen-driven proliferative responses in breast cancer [39, 40]. In differentiated thyroid cancer, both ERα and ERβ have been detected, and an ERα-positive/ERβ-negative pattern has been associated with more aggressive clinicopathological features in some studies [41]. Therefore, the balance between ERα and ERβ may be more informative than the expression of either receptor alone, particularly when interpreting sex-biased or estrogen-responsive tumor states. AR is the primary mediator of androgen signaling, and its effects vary depending on the availability of specific transcriptional partners. AR can either reinforce lineage programs or support stress-adaptation states [42]. PR exists as two isoforms, PRA and PRB, and shifts in their balance—along with the local co-regulator environment—can redirect downstream gene programs even when total PR levels remain similar [43]. Upon ligand binding, these receptors undergo conformational changes, dimerize, and translocate to the nucleus, where they bind hormone response elements and recruit co-activators or co-repressors to shape enhancer–promoter communication. This ultimately regulates gene networks that control proliferation, survival, metabolic adaptation, and lineage specification [35, 44]. The magnitude, direction, and tissue specificity of nuclear-receptor outputs are thus influenced by tissue-restricted isoforms, dynamic co-regulator availability, and chromatin accessibility, which determine which regulatory elements are competent for receptor binding and transcriptional control [35]. Sex-linked variables further modulate these processes. Systemic hormone milieus change over the life course, and intratumoral steroidogenesis can locally synthesize or convert ligands within tumor–stroma compartments, adjusting receptor activation in a tissue-dependent manner [45–47]. These factors may contribute to sex-dimorphic programs even in non-reproductive tissues and may help explain sex differences in susceptibility and progression in selected tumor types.

Non-classical pathways

Beyond genomic regulation, sex hormones also elicit rapid, non-genomic signaling via membrane-proximal and cytosolic receptor pools. A small fraction of ERα/ERβ, AR, and PR localize to the plasma membrane, with GPER providing an additional membrane sensor. These receptors can transduce hormonal signals within minutes to kinase and second-messenger hubs, providing a plausible route by which hormone-related signaling may occur even when nuclear receptor immunostaining is modest [48]. However, low nuclear receptor immunostaining should not be taken as evidence that membrane-associated receptor pools dominate the hormone response. In most non-reproductive cancers, the relative contribution of membrane-associated and nuclear receptor pools has not been quantitatively defined in fresh human tumor tissue. Although initiated at the membrane, these signals can feedback to transcriptional regulators and chromatin programs.

Mechanistically, membrane-initiated signaling often converges on phosphoinositide 3-kinase/protein kinase B (PI3K/AKT) and mitogen-activated protein kinase/extracellular signal-regulated kinase (MAPK/ERK) pathways, frequently involving Src kinase/PKC modules and Ca2+/cAMP second messengers. These rapid outputs modulate apoptosis thresholds, mitochondrial flux, cytoskeletal organization, and motility without direct DNA binding, thereby influencing survival and invasiveness on a short timescale [49, 50].

In male-predominant cancers, androgen-associated signaling is potentially relevant under stress-permissive conditions, where membrane or cytosolic receptor pools quickly couple to PI3K/AKT and MAPK/ERK. Hepatocellular carcinoma (HCC) may provide an example of this stress-coupled hormone-associated pathway: under inflammatory or metabolic stress, AR-related signaling has been associated with PI3K/AKT, while hormone-associated signaling may also involve MAPK/ERK, and intratumoral steroid availability may be influenced by local aromatase (CYP19A1). Oxidative estrogen metabolites may further contribute to ROS generation and DNA adduct formation, potentially reinforcing stress-adapted tumor states [51–54]. A complementary example is urothelial cancer, where the membrane-associated aspects of AR biology may contribute to tumor-related signaling. Membrane-associated AR signaling may interact with SRC–PI3K signaling [55, 56], potentially contributing to epithelial-to-mesenchymal transition (EMT)-like programs, enhanced motility, and plasticity, even in the absence of strong nuclear AR signaling [57]. Related stress-adaptation mechanisms have been investigated in other male-predominant cancers. In renal cancer, GPER-mediated PI3K/AKT signaling has been associated with migratory and invasive phenotypes, while hypoxia-associated alterations in AR signaling may further contribute to stress-adapted tumor states [58–60]. In head-and-neck squamous carcinomas, estrogen receptor–EGFR crosstalk can enhance MAPK activation, while AR-related EGFR/Akt signaling has been linked to migratory phenotypes, supporting a potential role for rapid hormone-associated signaling in tumor invasion [61–63].

In cancers with a female-skewed incidence, estrogen-centered membrane signaling is often emphasized, particularly in contexts where ER immunoreactivity is low but estrogen responsiveness remains. In thyroid cancer, membrane-bound ER and GPER have been linked to MAPK/ERK and PI3K/AKT signaling, both associated with proliferation and stress resilience, with intrathyroidal steroidogenesis potentially enriching local estrogen exposure in certain subsets [64]. In biliary tract cancers, particularly cholangiocarcinoma, cholestatic inflammation may provide a microenvironment conducive to membrane-initiated estrogen signaling. Estrogen-related ERK1/2 and PI3K/AKT signaling has been associated with proliferation and invasion, while local aromatase activity may further modify estrogen-related signaling in association with GPER/GPR30 expression [65–68]. These female-leaning examples suggest that estrogen sensing can remain biologically relevant via kinase coupling, even when canonical nuclear ER signaling is limited.

Despite organ-specific variations, non-genomic hormone signaling consistently maps onto two key mechanisms that influence downstream microenvironmental and immune phenotypes. The first is reciprocal crosstalk with dominant oncogenic pathways, such as AR with PI3K/AKT and ER or GPER with MAPK/ERK. This interconnection facilitates rapid compensation under stress and can maintain a phenotype even when one pathway is inhibited [48]. The magnitude and persistence of these responses are further regulated by tissue stress integrators, including YAP/TAZ for mechanical cues, WNT or β-catenin for lineage programs, and NF-κB for inflammatory tone, rendering signaling direction and durability highly context-dependent [69–71]. Equally significant is the availability of local ligands at the tumor–stroma interface. Intratumoral steroidogenesis and steroid conversion, involving enzymes like CYP19A1, SRD5A1/2, and related factors, can decouple serum hormone measurements from intratumoral hormone levels [46]. In bladder cancer, this divergence has been directly demonstrated by evidence showing substantial differences between intratumoral and circulating steroid measurements [47]. These enzymes may be produced by both tumor cells and stromal or immune lineages, providing a direct link through which endocrine context becomes integrated into the microenvironment. Sustained kinase activity, in turn, feeds back onto chromatin regulators and pioneer factors, stabilizing transcriptional states that differ in prevalence between sexes in certain tissues [31].

This interpretation should remain cautious, because the evidence supporting GPER-mediated signaling in non-reproductive cancers is still strongest in established cell-line systems. Pharmacological studies using GPER agonists or antagonists, including G-1 and G-15, have helped define rapid estrogen-responsive pathways, but such responses may vary by model and may not always establish receptor-specific effects. In this context, low nuclear receptor immunostaining should be viewed as a reason to examine non-classical signaling rather than as evidence that membrane-associated receptor pools dominate the hormone response. More direct validation in primary tumor tissue, organoid models and in vivo systems, ideally combined with GPER knockdown, knockout, rescue and knock-in approaches, will be important for defining the true contribution of GPER across different tumor types and model systems.

Immunity and the tumor microenvironment

Hormonal signals can affect tumors at multiple levels, extending beyond cancer cells to host physiology and tumor-associated compartments. By influencing metabolism, perfusion, and immune tone, sex hormones may contribute to changes in endothelial function, vascular organization, stromal architecture, metabolic states, and local steroidogenic activity. These changes establish gradients of perfusion, hypoxia, and chemokines, which in turn regulate immune cell entry, positioning, and functional programming within the tumor. Additionally, hormone receptors expressed by immune lineages provide a direct pathway for modulating T-cell and myeloid programs, contributing to sex-biased immune responses across non-reproductive cancers. Therefore, systemic hormonal context and local tissue remodeling shape tumor microenvironment architecture and the quality of antitumor immunity (Fig. 3).

Fig. 3.

Sex hormones shape the TME and antitumor immunity by setting “entry conditions” for immune infiltration and function. a Systemic hormones and intratumoral steroidogenesis increase local ligand levels in response to inflammation and hypoxia, partially decoupling the TME from circulating hormone levels. b Hormone-conditioned programs remodel key TME modules—vasculature, metabolism, and stroma/ECM mechanics—that collectively define perfusion, hypoxia, and stress landscapes. c These features create chemokine and nutrient/oxygen gradients, influencing immune entry along a spectrum from myeloid-recruiting niches (e.g., CCL2/IL-10) to IFN–CXCL9/10-driven CXCR3 + CD8+ T cell recruitment, with dose/receptor balance influencing Treg-supporting signaling. d Direct receptor signaling in immune lineages fine-tunes effector and suppressive functions across T cells, dendritic cells, macrophages, MDSCs, NK cells, and γδ T cells. e Tumor-cell hormone loops can modulate immune visibility (e.g., PD-L1 upregulation; reduced MHC-I/antigen processing via non-genomic cascades), altering ICI sensitivity or myeloid-dominant resistance pathways. Abbreviations: CCL2, C–C motif chemokine ligand 2; CXCR3, C–X–C chemokine receptor 3; DC, dendritic cell; ECM, extracellular matrix; IFN, interferon; IL-10, interleukin 10; ICI, immune checkpoint inhibitor; MDSC, myeloid-derived suppressor cell; MHC-I, major histocompatibility complex class I; NK, natural killer; PD-L1, programmed death-ligand 1; ROS, reactive oxygen species; STS, steroid sulfatase; TAM, tumor-associated macrophage; TME, tumor microenvironment; Treg, regulatory T cell; UGT, UDP-glucuronosyltransferase; HSD, hydroxysteroid dehydrogenase

Sex hormones and the tumor microenvironment (TME)

Sex hormones shape the tumor microenvironment both through direct signaling in cancer cells and by conditioning the vascular, metabolic, and stromal landscape that immune cells encounter. In many contexts, rapid membrane-associated signaling through ERs, AR, and GPER modulates endothelial tone and permeability. These vascular effects can contribute to heterogeneous perfusion and hypoxic niches. However, glycolytic reprogramming in cancer is not simply a consequence of hypoxia, as tumor cells may preferentially use aerobic glycolysis even under oxygen-replete conditions. Hypoxia and altered perfusion can further amplify lactate accumulation, ROS production, angiogenic signaling, and metabolic stress [72–74]. These metabolic shifts may also amplify hormone signaling by altering local steroid metabolism. Evidence from classical hormone-responsive tumor contexts provides a mechanistic precedent for this possibility. Inflammation and hypoxia at the tumor–stroma interface can upregulate aromatase in breast adipose stroma, increasing local estradiol production. In androgen-deprived or androgen-independent prostate cancer, increased expression of enzymes involved in adrenal androgen conversion and intratumoral DHT synthesis can maintain local androgen availability. These examples illustrate how local tissue stress or therapeutic pressure can weaken the correspondence between circulating hormone levels and intratumoral steroid exposure, although the extent to which analogous mechanisms operate across non-reproductive cancers remains to be directly established [75–77].

Across tissues, vascular stress, local steroidogenesis, and stromal mechanics emerge as recurring modules that are reorganized into distinct immune states. Table 2 maps representative sex-biased configurations in non-reproductive cancers and links each pattern to hormone-associated mechanisms and dominant tissue stressors. The underlying logic is that endocrine inputs are filtered by local injury and tissue architecture, which sets a baseline immune state before therapeutic intervention. In the liver, chronic viral or metabolic injury can create a stress-permissive environment in which AR-related and hypoxia-associated immune programs may be more evident in male-predominant disease, consistent with the higher incidence and more aggressive clinical behavior of HCC [5, 33, 78]. In the urothelium, carcinogen exposure can coincide with androgen-associated kinase activity and stromal remodeling; in biliary tract cancers, local aromatase activity and GPER/GPR30 expression have been reported, potentially aligning with female-skewed disease patterns [66]. In these examples, local steroid production, perfusion stress, and stromal organization form a coupled axis that translates hormonal context into tissue-specific, sex-biased immune ecosystems.

Table 2.

Representative sex-biased immune and TME features across selected non-reproductive cancers

Cancer type	Sex bias	Representative immune/ TME features	Candidate hormone–TME links	TME stressors
Hepatocellular carcinoma	Male	Low CD8⁺, weak IFN, myeloid-rich	AR signaling, local androgen synthesis	Viral/ metabolic injury
Bladder cancer	Male	Immune-cold, EMT-like, myeloid dominance	Membrane AR–SRC–PI3K	Carcinogens, hypoxia
Thyroid cancer	Female	IFN-tonic, CD8⁺ enriched	Estrogen–GPER–MAPK	Oxidative stress
Biliary tract cancer	Female	Treg-favoring or IFN-modulated	Aromatase, GPER signaling	Cholestasis
Lung adenocarcinoma	Variable	Context-dependent	Local estrogen synthesis	Smoking, hypoxia

Once established, stromal architecture propagates these sex-dependent cues into the immune landscape. Fibroblast and matrix programs (e.g., TGF-β activity, collagen crosslinking, and extracellular matrix stiffness) are shaped by ER/AR crosstalk with mechanosensing regulators such as YAP/TAZ and developmental pathways like WNT/β-catenin. These changes alter interstitial pressure and the physical “tracks” available to both cancer and immune cells [79–81]. As these physical constraints evolve, chemokine and cytokine gradients are also remodeled. In several settings, androgen-associated programs have been linked to CCL2/IL-10-rich, myeloid-recruiting niches, while estrogenic cues can enhance CXCL9/10–IFN circuits or, depending on receptor balance and hormone dose, support regulatory T-cell programs [27]. Within this framework, sex-discordant immune “baselines” may emerge as downstream effects of hormone-conditioned vascular, metabolic, and stromal states—for example, hypoxia- and myeloid-skewed niches in male-predominant liver and bladder cancers versus interferon-tonic or bile-acid–modulated milieus in female-predominant thyroid and biliary cancers, although direct comparative evidence remains limited across tumor types [78, 82].

Sex hormones and antitumor immunity

Building on these hormone-conditioned gradients and niches, sex hormones further modulate antitumor immunity at the cellular level by reshaping differentiation, trafficking, and checkpoint programs across lymphoid and myeloid lineages. These effects, however, should not be reduced to a simple androgen-suppressive versus estrogen-protective model, because hormone–immune interactions vary with cancer type, immune-cell subset, receptor subtype, hormone concentration, timing of exposure, and the local inflammatory or stromal state. Accordingly, ER, AR, PR, and GPER are expressed in a tissue- and activation-dependent manner across CD8⁺ T cells, CD4⁺ T cells, regulatory T cells, dendritic cells, macrophages, neutrophils, and natural killer (NK) cells, providing a receptor basis for ligand-dependent regulation of these immune programs. In T cells, androgen signaling has been linked to skewed differentiation toward attenuated effector maturation and reduced acquisition of cytotoxic programs, including progenitor-like exhausted or stem-like states in which AR activity influences TCF1/TCF7-associated programs, and may modulate responses to immunotherapy [83, 84]. These findings support a role for AR in shaping T-cell states, although the extent to which this axis operates across non-reproductive cancers remains less well defined, given that part of the evidence, including the study by Guan et al., comes mainly from prostate cancer models. Estrogen signaling, particularly through rapid, non-genomic mechanisms, can regulate dendritic cell differentiation, maturation, and cross-presentation under selected conditions, rather than uniformly enhancing these functions. These effects depend on receptor signaling, dose, timing, dendritic-cell subset and inflammatory state [85–87]. However, higher or chronic estrogen exposure can, in certain contexts, favor the expansion of regulatory T cells [88]. Although estrogen can enhance selected systemic immune responses, this does not necessarily translate into effective anti-tumor immunity within the TME. Tumor-local conditions such as hypoxia, metabolic stress, stromal remodeling and chronic inflammation may redirect estrogen-related effects toward Treg expansion, myeloid suppression, angiogenesis, PD-L1 expression or immune exclusion [89]. Consistent with this framework, male-predominant hepatocellular and urothelial carcinomas have been shown to exhibit lower intratumoral CD8⁺ T cell infiltration and reduced interferon signatures compared to their female counterparts, even after adjusting for key covariates [31, 90]. Furthermore, local steroid metabolism may reshape intratumoral hormone exposure and thereby influence immune states, although its direct contribution to T-cell differentiation programs in non-reproductive cancers remains to be established [47, 91]. Intratumoral AR has been shown to repress interferon gamma (IFNG) and granzyme B (GZMB) in CD8⁺ T cells via chromatin-level mechanisms, pushing T cells toward progenitor-like exhausted states and weakening cytotoxic function [83, 84, 92]. Sex steroid receptors may also influence anti-tumor immunity through immune checkpoint-related and immune-evasion pathways. In melanoma, androgen receptor signaling has been linked to MICA shedding, escape from natural killer-cell-mediated cytotoxicity, and reduced responsiveness to immune checkpoint inhibitors, while ERβ-related pathways have also been proposed as relevant to melanoma biology and therapeutic response [93, 94]. Beyond immune-intrinsic signaling, in non-small-cell lung cancer, tumor-cell aromatase has been reported to support an autocrine E2–ER loop that upregulates programmed death-ligand 1 (PD-L1), biasing the tumor microenvironment toward immune exclusion. In preclinical and early translational models, aromatase inhibition has been shown to partially reverse these features, increasing γδ T cells, NK cells, and activated CD8⁺ T cells [95, 96].

Hormone responsiveness also extends to myeloid compartments, which translate endocrine signals into suppressive niches and secondary barriers to T-cell access. AR activity has been associated with macrophage differentiation-associated and M2-like phenotypes, accompanied by increased IL-10 expression and other immunosuppressive features in selected tumor contexts [97]. Estrogenic signals can enhance dendritic cell interferon programs and modulate NK-cell activity, yet in some contexts, may also promote the accumulation and immunosuppressive function of myeloid-derived suppressor cells (MDSCs) through ERα/STAT3-linked pathways [98]. As myeloid cells play a critical role in vessel normalization and matrix remodeling, these hormone-dependent programs may secondarily influence tissue perfusion and the physical accessibility of effector T cells to tumors, thereby contributing to sex differences in effective immune surveillance [99].

Local hormone availability functions as an amplifier, reinforcing immune baselines. Intratumoral steroidogenesis—mediated by enzymes such as CYP19A1, SRD5A1/2, steroid sulfatase (STS), UDP-glucuronosyltransferases (UGTs), and hydroxysteroid dehydrogenases (HSDs) at the tumor–stroma interface—creates autocrine and paracrine pools of bioactive ligands that sustain immune set-points despite similar systemic hormone levels. Inflammatory signals can regulate components of these pathways in chronically injured liver tissue, while intratumoral steroid profiles in bladder cancer can differ substantially from circulating measurements. Together, these observations suggest that local steroid handling may connect site-specific stressors with sex-linked immune profiles [47, 91, 100]. Mechanistically, non-genomic hormone signaling can converge on immune gene regulation: kinase cascades downstream of ER/AR/GPER phosphorylate chromatin remodelers and pioneer factors in tumor and immune cells, reshaping enhancer accessibility at antigen-presentation and interferon-response loci. Hormone-related signaling may also intersect with tumor-cell plasticity and immune recognition. In preclinical lung cancer models, ER blockade with fulvestrant reduced mesenchymal features and sensitized tumor cells to antigen-specific T-cell- and NK-cell-mediated cytotoxicity [101]. At the population level, sex-stratified datasets suggest that antigen-presentation–deficient or myeloid-dominant states may be enriched in some male-predominant cancers, such as hepatocellular and bladder carcinoma, whereas female-predominant sites more often display interferon-tonic or Treg-favoring environments shaped by estrogen balance and tissue physiology [90, 99].

Several preclinical studies highlight the therapeutic potential of hormone–immune coupling. Estrogen signaling in tumor-associated macrophages (TAMs) has been linked to M2-like polarization and ICI resistance; macrophage-specific ERα ablation or selective ER degrader (SERD) treatment (e.g., fulvestrant) can restore T-cell–inflamed features and improve ICI efficacy in preclinical models [102]. In liver metastases, estrogen-dependent MDSC recruitment via TNF-α-associated pathways has been implicated in primary resistance to PD-L1 blockade, with ovariectomy or tamoxifen shown to re-establish inflamed TME traits in experimental models [103, 104].

This context dependence is also relevant to female-predominant cancers such as thyroid and gallbladder cancer, where estrogen-related pathways may contribute to tumor development through epithelial, stromal, angiogenic, inflammatory, or oxidative-metabolic mechanisms, even when some estrogen-dependent immune effects may be protective in other settings.

In summary, sex hormones shape antitumor immunity in a context-dependent manner. Androgen-associated pathways are often linked to altered CTL activity and suppressive myeloid niches, while estrogen signaling can enhance antigen presentation and IFN–CXCL9/10 chemokine activity in some settings, but may also support Treg-promoting, stromal, angiogenic, or myeloid-suppressive programs depending on dose, receptor balance, tissue state, and tumor microenvironment. Local ligand availability and tissue-specific stressors—such as viral or lipotoxic liver injury, urinary carcinogens, hypoxia, and bile-acid–induced cholestasis—may amplify these patterns, providing a plausible framework for interpreting sex-biased immune ecosystems observed across non-reproductive cancers [28, 105].

Clinical implications

Sex hormones influence cancer risk and progression by modulating receptor-driven programs and local ligand availability. From a translational standpoint, the key question is whether a hormone-associated program represents a therapeutically tractable dependency in a given tumor, particularly outside of classical hormone-driven cancers. Recent clinical and translational research across various tumor types highlights a consistent message: candidate therapeutic signals may lie in pathway activity and its downstream phenotypic effects, while receptor immunostaining alone may fail to capture hormonally wired biology. Concurrently, sex-stratified multi-omics datasets have facilitated the interpretation of hormone-response programs in conjunction with immune visibility, stress-adapted states, and kinase networks, positioning “hormone biology” as a measurable tumor state rather than a mere background factor. Consequently, clinical translation is most compelling when it links tangible therapeutic targets to operational biomarkers and pharmacodynamic readouts, which can validate, monitor, and refine interventions. The following section outlines the strongest clinical therapeutic levers targeting hormone axes, followed by an examination of how hormone-axis interventions interact with dominant oncogenic and immune programs in biomarker-defined settings. Figure 4 provides an overview of therapeutic hormone axes, activity-based biomarkers, and rational combination strategies.

Fig. 4.

Therapeutic levers and activity-first clinical translation of sex-hormone signaling. Candidate hormone-axis states may extend beyond classical endocrine tumors by targeting the ER, AR, PR, and GPER axes, along with local intratumoral ligand supply (steroidogenesis and conversion). Representative clinical contexts and agents illustrate axis-directed strategies, such as oral SERDs for ER programs, emerging AR-directed approaches in AR-active non-prostate cancers, PR modulation in selected gynecological contexts, and early-phase targeting of rapid GPER signaling. An activity-first biomarker framework emphasizes hormone-response gene signatures, kinase activity readouts, steroidogenic programs, and immune-context features to identify functional dependencies and guide rational combinations. This model supports the evaluation of rational combinations with CDK4/6 inhibitors, kinase inhibitors, and PD-1/PD-L1 blockade, and encourages sex-aware trial designs that stratify by pathway activity and hormone-response states, with longitudinal multi-omics and immune profiling. Abbreviations: ADT, androgen deprivation therapy; AI, aromatase inhibitor; AR, androgen receptor; CDK4/6, cyclin-dependent kinase 4/6; ER, estrogen receptor; GPER, G protein-coupled estrogen receptor; HCC, hepatocellular carcinoma; HER2, human epidermal growth factor receptor 2; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; PR, progesterone receptor; SERD, selective estrogen receptor degrader

Hormone-axis levers

Table 3 summarizes candidate hormone-axis mechanisms and the clinical readouts used to operationalize pathway dependency in non-endocrine tumors.

Table 3.

Candidate hormone-axis mechanisms and proposed clinical readouts

Hormone axis	Receptor mode	Representative settings	Tumor-intrinsic effects	TME/ immune effects	Proposed clinical readouts
Estrogen (E2)	Nuclear ERα/ERβ	Thyroid, biliary, lung	Transcriptional reprogramming, lineage bias	IFN-tonic niches, DC activation/ Treg expansion	ER activity score (target-gene programs) ± ERα/ERβ balance/isoform proxies; avoid ER IHC alone
	Membrane ER/ GPER	Thyroid, gallbladder	PI3K–AKT, MAPK–ERK, stress adaptation	Endothelial permeability, immune entry	p-ERK/ p-AKT ± immediate-early gene response; GPER expression as supportive (not sole) marker
Androgen (T/DHT)	Nuclear AR	Liver, bladder	Lineage reinforcement, survival	Myeloid-skewed niches	Androgen-response signature/ AR activity module ± downstream pathway wiring; AR IHC as supportive only
	Membrane AR	Bladder, HCC	SRC–PI3K signaling, EMT-like states	Hypoxia, immune exclusion	p-SRC/ p-AKT ± EMT/plasticity score; motility/adhesion modules
Progesterone (P4)	Nuclear PR-A/PR-B	Endometrial-like subsets	Isoform-dependent transcription	Context-dependent immune modulation	PR program score ± PR-A: PR-B isoform balance proxies; interpret with tumor state/context
Local steroidogenesis	Ligand axis	Liver, bladder, biliary	Local ligand supply	Immune set-point stabilization	Enzyme program score ± intratumoral steroid profiling

This table summarizes candidate hormone-axis mechanisms and activity-based readouts that may support patient stratification and pharmacodynamic assessment in future studies. These readouts are intended to complement receptor immunostaining by capturing pathway activity, local ligand availability, downstream signaling, and immune-state changes. Their predictive performance in non-reproductive cancer cohorts will require prospective validation

ER axis

The clearest clinical precedent for therapeutically targeting estrogen signaling comes from ER-driven breast cancer, where next-generation ER degradation has shown measurable benefits in molecularly selected settings. In the phase III EMERALD trial, the oral SERD elacestrant improved progression-free survival compared to standard endocrine therapy in ER⁺/HER2⁻ metastatic breast cancer, reinforcing the concept that functional ER dependence can be clinically identified, particularly in tumors with acquired ESR1 mutations [106]. This activity-focused perspective is further supported by developments in oral SERDs, where outcomes are more closely linked to evidence of pathway activation than to ER immunostaining alone. In the phase II SERENA-2 study, camizestrant demonstrated dose-dependent antitumor activity relative to fulvestrant, highlighting the potential for deeper ER suppression in advanced ER⁺/HER2⁻ disease [107]. In the phase II acelERA trial, giredestrant was compared with physician’s choice endocrine therapy, highlighting both the promise and limitations of extending ER-axis therapy in tumors with heterogeneous dependency and reinforcing the importance of biomarker-defined ER circuitry [108]. These positive or mixed results should also be interpreted alongside negative evidence. For example, the oral SERD amcenestrant failed to improve progression-free survival in AMEERA-3 and was later discontinued after the phase III AMEERA-5 trial was stopped for futility, indicating that ER-axis targeting is not uniformly successful even in hormone-dependent breast cancer [109, 110]. For non-reproductive tumors where estrogen-response programs intersect with kinase signaling (e.g., certain biliary or thyroid cancers), a viable translational strategy involves using ER-axis agents as hypothesis-testing tools in cohorts enriched for measurable estrogen-response activity and defined downstream signaling states, rather than directly applying breast-cancer endocrine paradigms [108].

AR axis

AR signaling provides an illustrative steroid-hormone axis for therapeutic exploration. Prostate cancer established the core translational principle that sustained disease control is achievable when a tumor is genuinely dependent on a ligand–receptor transcriptional program and therapy aligns with that functional dependence. This principle extends beyond reproductive tissues. Salivary duct carcinoma (SDC) provides a non-classical clinical example: AR pathway activity is prevalent and can be used prospectively for patient selection. In the phase II Alliance A091404 study, enzalutamide demonstrated limited single-agent activity and did not meet protocol-defined success, highlighting that AR positivity alone may be insufficient and motivating intensified/combined AR-axis blockade and refined biomarkers [111]. The phase II YATAGARASU study further evaluated deeper AR-axis suppression with apalutamide plus goserelin, suggesting that intensified endocrine blockade is feasible in selected patients, but this should be viewed as investigational rather than as proof that combination therapy can overcome enzalutamide monotherapy failure [112].

These salivary-gland findings highlight a practical takeaway for sex-biased cancers: AR targeting is most rational when AR activity is regarded as a measurable state rather than a binary IHC label. In urothelial cancer, molecular androgen-response scores have been examined in relation to clinical outcomes, supporting the evaluation of AR-associated tumor states beyond receptor expression alone [105, 113]. This suggests that AR-axis studies in this context may benefit from prioritizing androgen-response signatures over receptor presence alone. To date, AR-targeted strategies in the non-prostatic settings discussed above remain investigational or have produced limited or inconclusive evidence, particularly in unselected or weakly stratified cohorts. This highlights the need to distinguish AR-associated biology from proven AR dependency before broader clinical adoption.

PR axis

Clinical PR targeting remains most established in gynecologic oncology, where progesterone-based strategies are routinely employed in PR-expressing endometrioid diseases and fertility-sparing settings. Contemporary guidance emphasizes biomarker-aware selection and monitoring over PR staining as a stand-alone decision tool [114, 115]. Systematic syntheses further indicate that clinical benefit from progestin-centered approaches is heterogeneous, with relapse being common. This reinforces the view that PR should be interpreted as a dynamic pathway whose direction and magnitude depend on tissue context and regulatory state, not merely as a static immunophenotype [22, 116]. Recent translational work has extended PR antagonism beyond histology-specific assumptions by enrolling PR-positive solid tumors in biomarker-defined designs. Onapristone extended release was evaluated across PR-positive recurrent adult granulosa cell tumors, low-grade serous ovarian cancer, and endometrioid endometrial cancer in a phase II basket study, providing prospective evidence that PR-directed therapy can be tested across tumor types when a coherent PR program is present [117]. These findings support a clinically practical view of PR as both a lineage marker and a context-dependent therapeutic vulnerability, motivating a sex-aware interpretation. Progesterone exposure, receptor isoform balance, and co-regulator environments can vary by sex and tissue. Evidence from non-small-cell lung cancer further suggests that PR-related tumor states may have sex- and compartment-dependent clinical relevance, as PR positivity has been reported more frequently in female patients, while epithelial and stromal PR expression show distinct sex-associated prognostic patterns [118, 119].

GPER axis

GPER represents a potentially relevant estrogen-sensing pathway with properties distinct from classical nuclear ER signaling, particularly in settings in which estrogen-related responses persist despite modest nuclear ER immunoreactivity. At present, however, this interpretation should be viewed as provisional. The oral GPER agonist LNS8801 has shown pharmacological feasibility and evidence consistent with target engagement in GPER-positive models, but it remains in early-phase development, and its clinical efficacy in non-reproductive cancers has not yet been established [120]. These data support further biomarker-enriched investigation rather than broad extrapolation across tumor types. A germline coding variant in human GPER has also been reported to attenuate responsiveness to LNS8801, suggesting that GPER genotype may become relevant for future patient stratification, although this possibility still requires prospective validation [121]. Complementary mechanistic data in uveal melanoma indicate that GPER agonism can alter tumor-relevant signaling and immune-related phenotypes, providing a rationale for further study while remaining hypothesis-generating at this stage [120].

Local steroidogenesis axis

A recurring theme across sex-biased, non-reproductive cancers is that tumors and stroma can locally generate or convert steroid ligands, weakening the link between serum hormone levels and intratumoral exposure, thus enabling sustained signaling in a tissue-specific manner. This concept has clinical precedent in endocrine oncology, where inhibiting ligand supply (e.g., suppressing estrogen synthesis as part of endocrine therapy strategies) is a well-established method for controlling receptor-driven programs, showing that ligand availability can be therapeutically relevant when tumor cells are functionally dependent on this pathway [21]. Beyond classical endocrine settings, translational syntheses emphasize that intratumoral steroidogenic and steroid-converting enzymes can maintain local pools of bioactive E2/T/DHT, potentially sustaining hormone-related signaling even when nuclear receptor staining is modest. This provides a concrete rationale for measuring steroidogenic programs as part of clinical selection [47, 122, 123]. The clinical implication is that “local ligand supply” should be treated as a distinct axis, with its own biomarkers and pharmacodynamic readouts. Tumors can be functionally hormone-enabled through either high receptor activity, high intratumoral ligand production, or both, so the most interpretable trials will define dependency using convergent evidence rather than a single marker [124, 125]. Given that intratumoral ligand production can sustain receptor activation despite systemic castration, axis blockade alone may be insufficient, thus providing a mechanistic rationale for combination therapies with either ligand-synthesis inhibitors or downstream effector inhibitors.

Combination strategies and sex-aware evaluation

ER-driven breast cancer exemplifies how hormone-axis therapy gains clinical relevance. Endocrine treatment is rarely employed merely as a “hormone blocker” because ER output is closely linked to cell-cycle execution and can sustain growth programs even under therapeutic constraints. Across multiple randomized phase III trials, the addition of cyclin-dependent kinase 4/6 (CDK4/6) inhibitors to endocrine therapy has extended disease control, consistent with the inhibition of the proliferative machinery that converts ER signaling into a growth advantage [126–128]. The broader implication is practical: combination therapy is more rational to evaluate, and its clinical activity more readily interpretable, when it is matched to an active hormone-driven program and the downstream pathways through which that program supports tumor growth, rather than selected solely on the basis of receptor staining.

This activity-oriented approach is now being tested in “non-classical” histologies where hormone wiring is not assumed a priori but may be sustained by membrane or cytosolic receptor pools, kinase coupling, or local ligand availability. SDC provides a useful model for AR targeting because AR pathway activity is frequent and has been prospectively evaluated. In molecularly selected SDC, AR-directed strategies become more interpretable when enrollment and response assessment are tied to evidence of pathway activity and whether the disease state shifts as expected on therapy, rather than relying solely on AR immunostaining positivity [111, 112]. A similar approach has been applied to PR, where a phase II study of onapristone extended release prospectively evaluated PR-positive tumors and observed clinical benefit in some patients, although no objective responses were observed and the primary endpoint was not met [117]. This experience supports the feasibility of biomarker-informed testing, while also indicating that receptor positivity alone does not establish a therapeutically actionable hormone dependency.

The interface with immunotherapy is most informative when hormone-response states and immune context are evaluated together in real patient cohorts, rather than relying solely on mechanistic plausibility. In urothelial cancer, analyses of neoadjuvant pembrolizumab-treated cohorts have examined androgen/estrogen-response programs alongside immune features and outcomes, providing a concrete framework for interpreting “hormone + immunity” as a coupled tumor state that can be measured and related to clinical outcomes [113]. More broadly, sex-stratified multi-omics resources consistently reveal male–female differences in immune context across tumor types, including antigen-presentation modules and myeloid–T-cell balance. This provides a reference backdrop for evaluating whether hormone-modulatory interventions shift immune visibility or reinforce suppression [90, 99, 129]. Whether sex-hormone modulation causally reshapes immune context or merely correlates with pre-existing immune differences remains to be tested in interventional trials that incorporate sequential biopsies and immune profiling. In this context, sex-aware evaluation becomes a data-supported layer of interpretation—linking hormone programs, immune state, and clinical outcomes—rather than an abstract modifier.

These translational opportunities should nevertheless be interpreted cautiously. Several hormone-axis interventions have shown limited, negative or inconclusive results in unselected or weakly stratified cohorts, including failed SERD development with amcenestrant and uncertain AR-directed activity in urothelial or hepatocellular carcinoma. In salivary duct carcinoma, combined AR-axis suppression should be viewed as investigational rather than as proof that combination therapy can overcome enzalutamide monotherapy failure. Moreover, activity-based biomarkers, gene signatures, kinase readouts and local steroid measurements remain promising but insufficiently validated tools, with limited data on sensitivity, specificity and prospective predictive value in non-reproductive cancer cohorts. These gaps highlight the need to distinguish mechanistic plausibility from clinically demonstrated dependency.

Discussion

Sex differences in cancer burden are persistent, reproducible and highly site-specific, particularly in non-reproductive malignancies, where a male excess in incidence and mortality is common but a female excess is observed in certain sites. A central message from this review is that sex-hormone signaling should be viewed as an observable, context-dependent tumor state rather than as a uniform determinant of cancer behavior. This state may arise through layered interactions between classical nuclear receptor transcriptional programs, membrane-initiated kinase signaling, intratumoral steroid processing and hormone-conditioned vascular, stromal and immune microenvironments. This framework reconciles consistent population-level sex disparities with variable hormone-associated effects across tumor types and disease stages. The key question is therefore not whether estrogen or androgen is universally protective or tumor-promoting, but when a hormone-linked program becomes a functional dependency in a specific tissue and disease state.

Sex-hormone signaling should also be considered alongside sex-chromosome-linked mechanisms. X-linked tumor suppressor genes that escape X-chromosome inactivation may provide an additional protective layer in females, whereas males may be more vulnerable to loss-of-function events because they carry only one X chromosome [15]. Loss of the Y chromosome in male cells has also been linked to immune evasion and poorer cancer outcomes [16]. Four Core Genotypes models further indicate that sex chromosomes and gonadal hormones can make independent and context-dependent contributions to disease phenotypes [13]. These findings do not conflict with a hormone-focused framework. Rather, they indicate that sex-biased cancer biology may involve chromosomal, hormonal, immune and tissue-specific mechanisms. Thus, the hormone-focused perspective developed here should be interpreted as part of a broader model that also includes sex-chromosome-linked mechanisms.

Sex-hormone effects also cannot be reduced to static comparisons of “higher versus lower” circulating hormone levels. Blood hormones are upstream signals; the downstream output depends on how tissues “decode” this input. Receptor repertoires, isoform composition, co-regulator abundance, chromatin accessibility and enhancer landscapes vary across organs and tumor states, allowing the same hormonal stimulus to produce distinct transcriptional outputs. Thus, ERα/ERβ balance, PR isoform composition and AR cofactor dependence are better interpreted as context-specific determinants of receptor output than as fixed indicators of hormone responsiveness. This helps explain why sex bias is often tumor-type and sometimes subtype-specific, and why simplified assertions like “estrogen is protective” or “androgen is pro-tumor” rarely apply across non-reproductive sites. A similar distinction applies to immunity, where systemic hormone-associated immune effects may be redirected by local TME conditions such as chronic inflammation, hypoxia, stromal barriers and tumor-derived suppressive cues. Cross-cancer analyses further support this notion, showing that hormone-related programs can appear as coherent molecular states aligned with stress-adapted phenotypes and characteristic pathway wiring in subsets of non-reproductive tumors [31, 32, 99].

Within this pathway wiring, non-genomic, kinase-coupled signaling may connect receptor biology with tumor behavior, but its contribution should be interpreted cautiously. Membrane-associated steroid receptors and GPER can engage PI3K/AKT, MAPK/ERK and related pathways, yet the relative contribution of membrane versus nuclear receptor pools to hormone-responsive signaling remains insufficiently defined in fresh human tumor tissues, particularly in non-reproductive cancers. Apparent discrepancies across studies may reflect differences in dose, timing, receptor composition, tissue stress, hypoxia and inflammation [48, 130–132]. The role of GPER also appears to be context-dependent rather than uniformly tumor-promoting or therapeutically exploitable. For example, GPER deficiency accelerated DEN-induced hepatocarcinogenesis in a liver cancer model and was associated with enhanced inflammation and fibrosis, suggesting that GPER signaling may have protective effects in some inflammatory tissue contexts [132]. This evidence cautions against viewing GPER activation as a universal strategy and supports evaluating GPER-related interventions by tissue type, disease stage, inflammatory state, cellular compartment and genetic background.

Alongside receptor localization and downstream pathway wiring, local ligand availability adds a further layer of complexity. Intratumoral steroid handling—encompassing uptake, conversion, and inactivation by both tumor and stromal cells through enzymes such as aromatase, 5α-reductases, steroid sulfatase and hydroxysteroid dehydrogenases—may decouple local hormone exposure from serum levels in non-reproductive cancers, although direct evidence is currently strongest in selected settings such as bladder cancer. This is particularly relevant for sex differences, as hormone signaling is not solely dictated by the host’s endocrine state but is shaped by the tumor’s microenvironment. In stressed tissues, these steroidogenic and steroid-converting programs can maintain local estrogenic or androgenic activity and may contribute to both transcriptional and kinase-coupled signaling outputs [47, 91]. Consistent with this possibility, intratumoral steroid profiling in bladder cancer reveals significant divergence between the ligand landscape within tumors and systemic hormone levels [47]. However, whether inflammation- or hypoxia-linked changes in local steroid availability contribute to persistent sex bias across non-reproductive cancers requires direct testing [75, 91]. This distinction between local and systemic hormone biology is important when interpreting epidemiological and genetic evidence. For example, Larsson et al. used Mendelian randomization to examine genetically predicted serum estradiol and reported causal support mainly for estrogen receptor-positive breast cancer and endometrioid endometrial cancer, but not for most other site-specific cancers in women [133]. This finding should not be interpreted as excluding a role for hormone signaling in non-reproductive tumors. Rather, it highlights that circulating estradiol is only one component of a broader steroid network and may not capture intratumoral steroid conversion, receptor localization, receptor isoform balance, membrane-associated signaling, or downstream pathway activity.

Thus, the hormone sensed within a tumor may not simply reflect the dominant systemic hormone associated with the patient’s sex. Estrogen signaling in prostate cancer and androgen signaling in tumors arising in women illustrate that “opposite-sex” hormones can be biologically relevant when receptor expression, local ligand availability and downstream pathway activity permit pathway engagement [134, 135]. This supports a more flexible model beyond a simple male-androgen versus female-estrogen framework. Overall, these observations support a context-dependent, hypothesis-generating interpretation of sex-hormone signaling in non-reproductive cancers, with further validation needed in tumor tissues and functional models.

These tumor-intrinsic and ligand-level mechanisms ultimately converge on the TME, where vascular, metabolic, stromal and immune states determine whether hormone-linked programs become clinically meaningful. Sex-stratified datasets support this view by showing differences in antigen-presentation modules, myeloid–T-cell balance and broader immune contexture between male and female tumors [27]. Overall, sex differences in cancer should be interpreted as emergent properties of interacting biological layers, including context-dependent hormone decoding, local steroid handling, and microenvironmental and immune reprogramming, rather than as simple reflections of systemic hormone levels alone. This perspective helps explain why the same hormonal axis may associate with opposing clinical outcomes across tumor types or even across molecular subtypes within the same tumor type, as downstream programs are shaped by tissue-specific receptor architecture, cofactor availability, chromatin state, local ligand supply and stress-adapted tumor states [52, 136]. A practical implication is that translational analyses should prioritize functional circuit activity by integrating hormone-response programs with immune context, intratumoral steroid context and treatment outcomes, rather than relying on static receptor expression alone.

Several limitations temper the interpretation of the current evidence. A major source of uncertainty is that many clinical datasets do not adequately capture the life-course and exposure-related variables that shape hormone biology, including menopausal status, menstrual-cycle phase, exogenous hormone use, including oral contraceptives, hormone replacement therapy and selective estrogen receptor modulators used outside oncological treatment contexts, metabolic and inflammatory comorbidities, and sex-specific pharmacokinetics [137–140]. Female hormone exposure should therefore not be treated as a single biological state, because menstrual cycling, perimenopausal transition, postmenopausal decline and exogenous hormone use may each shape tumor, immune and microenvironmental phenotypes. Age is also important, as some sex differences, including the female predominance in thyroid cancer, change across the life course and may narrow after menopause [141]. The recent rise in early-onset colorectal cancer among young women in some populations further shows that cancer sex disparities are dynamic rather than fixed [142, 143]. This trend should be interpreted cautiously and may reflect interacting cohort, lifestyle, reproductive and biological factors rather than a direct hormone-driven effect. Transgender and gender-diverse populations may help separate exogenous hormone exposure from chromosomal and developmental background, although current cancer outcome data remain limited; for example, de Blok et al. reported altered breast cancer risk among transgender people receiving gender-affirming hormone treatment [144]. Hormone metabolism, clearance and tissue uptake may further shape systemic and intratumoral steroid availability and should be considered when interpreting sex-aware cancer datasets [139].

These exposure-related uncertainties are compounded by limitations in pathway measurement and causal inference. Thresholds for defining pathway “activity” remain non-standardized. IHC-based assessment of receptor expression or pathway activation remains valuable in clinical settings [145, 146], but static receptor staining may underestimate functional signaling, particularly when membrane-initiated pathways or local ligand supply contribute substantially to pathway activity. Intratumoral steroid quantification by LC-MS/MS provides a complementary approach for measuring local hormone exposure. In bladder cancer, this approach has demonstrated substantial divergence between intratumoral and circulating steroid measurements, although its broader clinical use still requires standardization, adequate tissue handling, and prospective validation [47]. Much of the current evidence is cross-sectional, making it difficult to determine whether hormone signaling contributes to ecosystem states or whether evolving tumor states reshape steroid handling and receptor wiring. These gaps are not merely technical but will influence the confidence with which sex-aware hormone biology can be translated into reproducible risk stratification and clinically testable hypotheses.

Looking ahead, study designs that address context dependence and test causality will be crucial. Longitudinal cohorts and interventional trials will be particularly important. Ideally, these studies should incorporate on-treatment biopsies, paired multi-omic profiling, immune spatial measurements, and pharmacodynamic readouts to test whether hormone-axis modulation shifts tumor state in the predicted direction. Standardized activity signatures for ER, AR, PR, and GPER programs should be benchmarked against downstream kinase activity and, where feasible, intratumoral steroid measurements. This approach would enable dependency to be defined by convergent evidence, rather than relying on any single marker. Aging and the life course should also be given focused attention, as systemic hormone milieus and local steroid handling evolve over time, potentially explaining age-stratified shifts in sex bias. More broadly, it remains unclear whether shared cross-tumor principles of sex-dimorphic regulation exist or if most mechanisms are tissue-restricted. Sex-chromosome effects and their interaction with endocrine signaling likely contribute to sex-biased phenotypes beyond circulating hormone levels and should be integrated with hormone-centered models, rather than treated as competing explanations. Continued efforts to incorporate sex as a biological variable in oncology trials, along with improved sex reporting in foundational experimental systems, will be essential for translating mechanistic insights into reproducible, sex-aware precision strategies.

Conclusion

Stable and significant differences in cancer incidence and outcomes between men and women are well-documented. Evidence from population studies, animal models, and cell experiments suggests that sex hormones and their receptor signaling pathways play a role in shaping these differences in a variety of non-reproductive tumors. However, the direction and intensity of these effects are dependent on cancer type, tissue, and microenvironment, and some inconsistencies and controversies remain. Mechanistically, sex hormone signaling can reprogram transcription and cell states through both classical genomic pathways and rapid non-genomic pathways, activating kinase networks via membrane-associated receptor pools. This signaling influences tumor proliferation, stress responses, migration, invasion, metabolic adaptation, and treatment response. These intrinsic tumor processes further interact with local steroidogenesis, the matrix, and the immune microenvironment to create sex-biased tumor ecosystems. With advancements in molecular profiling and accurate diagnostics, more verifiable risk assessments and individualized treatment strategies can be developed for tumors exhibiting sex differences.

Authors’ contributions

Jingsheng Xu: conceptual design, full text writing, original draft preparation; Xinyu Zhang: document retrieval, chart making; Zhenyu Li: professional interpretation, structural optimization; Siqi Li: format review, academic guidance; Anhui Ning: format review, literature screening; Dingding Li: review and editing; Minjie Chu: academic guidance, resource support, supervision, review and editing. Haiyan Gong: academic guidance, funding acquisition, supervision, review and editing. All authors reviewed the manuscript.

Funding

The research project described in this paper was supported by grants to M.J.C and H.Y.G including the National Natural Science Foundation of China (No. 82273715) and the scientific research projects of Jiangsu commission of health (M2024094). This work was also supported by the Postgraduate Research & Practice Innovation Program of Jiangsu Province (No. KYCX25-3832).

Data availability

Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.