Exploring Vitamin D Metabolism in Gynaecological Malignancies: Redox-Dependent and Redox-Independent Anticancer Mechanisms From Preclinical Models to Clinical Evidence.

Hoi Ching Cherry Chan: methodology, writing – review and editing, writing – original draft. Sachin Bhoora: writing – original draft, writing – review and editing. Esther Zhou: writing – original draft, writing – review and editing. Sumari Marais: conceptualization, methodology, supervision, writing – original draft, writing – review and editing. Rivak Punchoo: conceptualization, methodology, writing – original draft, writing – review and editing, supervision, funding acquisition.

The authors have nothing to report.

This research was funded by the South African Medical Research Council (SAMRC) under a Self‐Initiated Research Grant.

This review adopts a narrative approach, with the literature search designed to be representative and critically evaluative rather than comprehensive. PubMed/MEDLINE, ClinicalTrials.gov and the Surveillance, Epidemiology and End Results (SEER) databases were searched for records available up to June 2026; with no pre‐specified cut‐off date used for systematic reviews and meta‐analyses. Additionally, trial registry status update was conducted up to August 2026. Database searches were supplemented by manual screening of reference lists and citation tracking of key mechanisms in the preclinical and clinical papers identified in the literature search. Search terms covered four conceptual domains using Boolean operators and, where available, Medical Subject Headings. The vitamin D domain included ‘vitamin D’, ‘calcitriol’, ‘1,25‐dihydroxyvitamin D 3 ’, ‘25‐hydroxyvitamin D’, ‘calcifediol’, ‘cholecalciferol’, ‘vitamin D receptor’, ‘VDR’, ‘CYP27B1’ and ‘CYP24A1’. The redox domain included ‘reactive oxygen species’, ‘ROS’, ‘oxidative stress’, ‘mitochondrial membrane potential’, ‘NADPH oxidase’, ‘thioredoxin’ and ‘TXNIP’. Apoptosis and related pathways were searched using ‘apoptosis’, ‘autophagy’, ‘paraptosis’, ‘cell cycle arrest’, ‘caspase’, ‘Bcl‐2’ and ‘Bax’. Gynaecological terms included ‘ovarian cancer’, ‘cervical cancer’, ‘endometrial cancer’, ‘cervical intraepithelial neoplasia’, ‘endometrial hyperplasia’ and ‘endometriosis’. A representative Boolean combination was (‘vitamin D’ OR ‘calcitriol’ OR ‘VDR’) AND (‘reactive oxygen species’ OR ‘oxidative stress’) AND (‘apoptosis’ OR ‘autophagy’) AND (‘ovarian cancer’ OR ‘cervical cancer’ OR ‘endometrial cancer’). ClinicalTrials.gov was searched by pairing ‘vitamin D’ with each cancer type and its principal precursor lesion. SEER and Global Burden of Disease Study outputs were consulted directly for the incidence and survival statistics cited in the Introduction section of this article. Consistent with the narrative, non‐systematic design of this review, study selection did not follow a pre‐registered or reproducible protocol. Records were not screened in duplicate, formal screening and inclusion/exclusion counts were not maintained, and a PRISMA‐type flow diagram was not generated. Non‐gynaecological evidence, including breast‐cancer models and the VITAL trial, was selected purposively when it provided mechanistic rationale or dosing and scheduling information unavailable in the gynaecological literature. Such evidence is identified as extrapolated at each point of use. Accordingly, this review should be interpreted as a critical narrative synthesis rather than an exhaustive or reproducible systematic search. We prioritised preclinical studies conducted in vitro or in vivo; clinical studies providing individual‐level data, including case–control, cohort and trial designs; and ecological or modelling analyses providing population‐level or projected estimates. Eligible evidence addressed vitamin D metabolites, analogues or VDR signalling in relation to ROS, apoptosis, autophagy, cell‐cycle regulation or clinical outcomes in ovarian, cervical or endometrial cancer or their principal precursor lesions. Evidence from non‐gynaecological cancers and general cancer‐mortality trials (e.g., VITAL) was included selectively and is explicitly identified as extrapolated. It was used to provide mechanistic precedent or dosing and scheduling information not otherwise available within the gynaecological literature. Given the heterogeneity of study designs, vitamin D compounds, dosing regimens and outcome measures, quantitative synthesis was not undertaken. Findings are instead synthesised narratively by cancer type, beginning with preclinical mechanistic evidence and followed by individual‐level clinical observational evidence and ecological or modelling data, with study design identified where relevant. Cross‐cutting themes, including CYP24A1‐mediated resistance, dose–response hormesis and VDR‐dependent ROS signalling, are considered across all three malignancies and are summarised with the main translational barriers and research priorities in Tables  1 , 2 , 3 , 4 . Summary of experimental evidence for vitamin D compound‐induced reactive oxygen species (ROS) and apoptosis across ovarian, cervical, endometrial and breast cancer models, indicating whether antioxidant rescue and VDR dependency were tested, and whether findings are direct or extrapolated. Calcifediol: 10 μM Calcipotriol: 250 μM Calcitriol: 50 nM, 200 nM Perifosine: 10 μM, 30 μM Abbreviations: ER, endoplasmic reticulum; HGSOC, high‐grade serous ovarian carcinoma; HPV, human papillomavirus; LPS, lipopolysaccharide; NOX4, NADPH oxidase 4; p53/TP53, tumour protein 53 (gene); ROS, reactive oxygen species; VDR, vitamin D receptor.

Two case–control studies have examined vitamin D in populations of African ancestry. These populations are of particular interest because high ambient ultraviolet‐B (UVB) exposure can coexist with a high prevalence of vitamin D deficiency, influenced by skin pigmentation, dietary patterns and inequities in healthcare access. A population‐based case–control study of African‐American women reported no independent association between dietary vitamin D intake and ovarian cancer risk, whereas calcium intake was inversely associated with risk [ 32 ] (Table  3 ). Longer summer sun exposure was also associated with lower risk, a finding consistent with a possible role for endogenous vitamin D synthesis. However, sun exposure also reflects behaviours such as physical activity, so a vitamin D‐specific mechanism cannot be inferred from this association. The study also illustrates the methodological importance of distinguishing dietary vitamin D intake from total vitamin D exposure when evaluating ovarian cancer risk in populations with a high prevalence of vitamin D deficiency. Clinical trial, individual‐level observational (cohort and case–control) and ecological/modelling evidence from studies of vitamin D and gynaecological cancer. Note: ClinicalTrials.gov registry status for registered trials was rechecked on 22 August 2026. Abbreviations: 25(OH)D, 25‐hydroxyvitamin D; BMI, body mass index; BRCA1/2, breast cancer susceptibility gene 1/2; CI, confidence interval; CIN, cervical intraepithelial neoplasia; CRP, C‐reactive protein; FFQ, food frequency questionnaire; HIV, human immunodeficiency virus; HPV, human papillomavirus; HR, hazard ratio; IU, international units; LSIL, low‐grade squamous intraepithelial lesion; NMSC, non‐melanoma skin cancer; OR, odds ratio; ORR, objective response rate; PRIMMO, Pembrolizumab plus Radiotherapy In Mismatch‐repair deficient or Microsatellite‐instable tumours; Q, quartile; RCT, randomised controlled trial; RR, relative risk; SEER, Surveillance, Epidemiology and End Results (Program); UVB, ultraviolet B radiation; UVR, ultraviolet radiation; VDR, vitamin D receptor. A case–control study in Lagos, Nigeria extended this question to a sub‐Saharan African population by measuring serum 25(OH)D directly [ 33 ] (Table  3 ). The association lost statistical significance after adjustment, illustrating the susceptibility of vitamin D‐cancer associations to residual confounding by factors such as age, adiposity, menopausal status, family history and co‐morbidity. Population‐specific VDR haplotype frequencies in African‐ancestry populations remain poorly characterised and may modify dose–response relationships. Their characterisation is therefore an important priority for genetic epidemiology in underrepresented populations. These two case–control studies do not provide consistent evidence that vitamin D protects against epithelial ovarian cancer in African‐ancestry populations. They do, however, highlight two methodological priorities for future studies: distinguishing dietary intake from total vitamin D exposure and accounting carefully for factors that confound associations with vitamin D status. Characterisation of VDR polymorphisms may also help account for genetic heterogeneity and remains an important research priority. Prospective cohort studies strengthen temporal assessment and reduce some sources of bias relative to case–control designs, although they do not establish causality. They have been central to evaluating whether vitamin D status at diagnosis or during treatment is associated with survival in ovarian cancer. A prospective analysis from the Australian Ovarian Cancer Study (AOCS) found that higher serum 25(OH)D at diagnosis was associated with improved overall survival [ 34 ] (Table  3 ), raising the possibility that 25(OH)D may have prognostic value. Residual confounding by performance status, co‐morbidity, physical activity and diet cannot be excluded. In addition, the association was observed at diagnosis rather than after treatment, leaving the biologically relevant exposure window unresolved. By contrast, a large nationwide prospective cohort found no association between 25(OH)D and ovarian cancer‐specific survival [ 35 ] (Table  3 ). The modest difference between sampling time points could reflect seasonal variation, reduced outdoor activity during treatment, regression to the mean or treatment‐related effects on vitamin D metabolism. The cohort was predominantly vitamin D replete, with relatively few deficient participants, which may have limited power to detect effects within the biologically relevant deficient range. The null finding may therefore not generalise to populations such as sub‐Saharan Africa or South Asia, where severe deficiency is more prevalent. A 2025 meta‐analysis reported a statistically significant reduction in ovarian cancer incidence and susceptibility outcomes per unit increase in circulating 25(OH)D [ 36 ] (Table  3 ). This estimate should be interpreted cautiously because assay heterogeneity and seasonal variation in blood sampling may attenuate pooled dose–response estimates, whilst inconsistent definitions of deficiency limit threshold‐based analyses. Greater standardisation of assay platforms and sampling protocols would strengthen future meta‐analyses. Despite the volume of observational evidence, prospective interventional studies specifically evaluating vitamin D supplementation in patients with ovarian cancer remain limited. Two registered ClinicalTrials.gov studies are particularly relevant (Table  3 ). NCT01744821 [ 37 ] (Table  3 ) was one of the few prospective studies designed to test vitamin D chemoprevention in a genetically high‐risk population. Its termination because of insufficient recruitment illustrates a recurring challenge in prevention trials that enrol asymptomatic individuals and should inform the design of future studies in high‐risk populations. ClinicalTrials.gov reports that seven participants were enrolled and that the planned biomarker outcomes were not analysed because of insufficient accrual, although limited participant‐flow and safety data are posted. NCT03028519 [ 38 ] (Table  3 ) is a prospective, non‐randomised, uncontrolled interventional cohort study in newly diagnosed ovarian cancer patients undergoing systematic correction of vitamin D deficiency with participants assigned to two cohorts based on baseline vitamin D status and no prospective control arm. Outcomes include treatment‐related morbidity and biological markers without alteration of standard therapy. As of 22 August 2026, the registry listed the study as completed, but no results had been posted. Because the study lacks a concurrent randomised control group, the design cannot isolate the effect of vitamin D with the rigour of a randomised trial; any reported findings should therefore be considered hypothesis‐generating. In addition, no ovarian cancer trial has reported results for survival, treatment‐response or chemotherapy‐toxicity endpoints, an important translational gap summarised in Table  4 . Key translational barriers and research priorities for vitamin D research by gynaecological cancer type. Only two registered interventional trials exist for vitamin D in ovarian cancer and neither has produced usable efficacy data NCT01744821 , a chemoprevention trial in genetically high‐risk women (including BRCA1/2 carriers), was terminated due to failure to recruit sufficient participants, a recurring obstacle in prevention trials enrolling asymptomatic individuals NCT03028519 , a deficiency‐repletion trial in newly diagnosed patients, has completed but its results remain unreported; its non‐randomised, uncontrolled design will in any case limit interpretability once published No trial has reported results for survival, treatment‐response or chemotherapy‐toxicity endpoints in ovarian cancer patients Redesigned chemoprevention trials in high‐risk (BRCA1/2) populations with recruitment strategies that directly address the barriers behind NCT01744821 's termination Timely reporting of NCT03028519 outcome data Adequately powered, concurrently controlled (non‐historical) RCTs of deficiency repletion, examining treatment‐related morbidity and biological markers alongside standard chemotherapy A first RCT powered to detect a survival or treatment‐response endpoint in ovarian cancer patients Adequately powered RCTs with standardised 25(OH)D assay platforms and sampling protocols All cervical‐specific RCT evidence comes from two small trials from the same research group ( n  = 29/arm), using an intermittent high‐dose (bolus) regimen. A 14‐trial individual‐patient‐data meta‐analysis in broader cancer populations subsequently found daily dosing, but not bolus dosing, to be associated with lower cancer mortality ([ 39 , 40 ]; [ 44 ]). No RCT has been conducted in invasive cervical cancer; interventional evidence is confined to precancerous CIN. The largest vitamin D RCT to date (VITAL, n  = 25,871) was not cervical‐cancer‐specific and found no reduction in overall cancer incidence; its secondary mortality signal reached significance only in sensitivity analyses excluding early follow‐up [ 41 ]. Existing trial‐level meta‐analyses ([ 42 ]; [ 43 ]) pool heterogeneous cancer types, doses and regimens rather than isolating cervical cancer. The only HPV‐clearance intervention study was non‐randomised, allocated participants according to prior supplementation history and used a 6‐month follow‐up, shorter than the 12–24 months typically required for HPV clearance [ 52 ]. No trial has used HPV clearance as a prespecified primary or co‐primary endpoint with adequate follow‐up and no Phase II/III trial has tested vitamin D as an adjunct to platinum‐based chemoradiation in locally advanced cervical cancer. Large, adequately powered RCTs in cervical cancer specifically, using daily (not bolus) dosing regimens with prespecified target serum 25(OH)D concentrations CIN regression and HPV clearance as co‐primary endpoints, with a minimum 24‐month follow‐up True randomisation (rather than history‐based allocation) in future HPV‐clearance intervention studies Phase II trials of vitamin D as an adjunct to platinum‐based chemoradiation in locally advanced cervical cancer, with translational sub‐studies of VDR expression and immune microenvironment Stratified trial analyses by VDR genotype to identify subgroups most likely to benefit RCTs specifically enrolling invasive cervical cancer populations, not only precancerous CIN The Phase II PRIMMO trial is the sole interventional dataset in histologically confirmed endometrial carcinoma [ 55 ] PRIMMO's five‐drug immunomodulatory combination (cyclophosphamide, aspirin, lansoprazole, vitamin D, curcumin, plus pembrolizumab and radiotherapy) precludes isolating any effect attributable to vitamin D specifically PRIMMO was single‐arm with no control group and a small sample size ( n  = 25); its primary efficacy endpoint was not met No RCT has ever tested vitamin D supplementation against histological or molecular endpoints in endometrial hyperplasia, the principal precursor lesion No RCT has evaluated vitamin D as a chemopreventive strategy in high‐risk endometrial cancer populations Epidemiological studies are consistently null but confounded by BMI/obesity, precluding causal inference No adequately powered RCTs against histological/molecular endpoints; PRIMMO trial confounded by multi‐drug design VDR polymorphism studies underpowered and unreplicated in GWAS RCTs designed to isolate vitamin D's independent contribution (single‐agent or placebo‐controlled), adequately powered for cancer or hyperplasia endpoints Trials specifically enrolling endometrial hyperplasia populations, using histological regression as a primary endpoint Trial stratification by baseline 25(OH)D (particularly ≤ 25 nmol/L) and by VDR expression/polymorphism status Incorporation of histological/molecular endpoints (Ki‐67 proliferation index, apoptotic fraction, CYP24A1 expression) rather than biomarker measures alone Abbreviations: 25(OH)D, 25‐hydroxyvitamin D; BMI, body mass index; BRCA1/2, breast cancer susceptibility gene 1/2; CIN, cervical intraepithelial neoplasia; CYP24A1, 24‐hydroxylase; GWAS, genome‐wide association study; HPV, human papillomavirus; Ki‐67, proliferation marker encoded by MKI67; PRIMMO, Pembrolizumab plus Radiotherapy in Mismatch‐repair deficient or Microsatellite‐instable tumours; RCT, randomised controlled trial; VDR, vitamin D receptor; VITAL, Vitamin D and Omega‐3 Trial. Narrative and systematic reviews have synthesised the mechanistic and epidemiological relevance of vitamin D to cervical carcinogenesis and human papillomavirus (HPV)‐related inflammation [ 93 , 94 , 95 ]. Clinical studies in cervical intraepithelial neoplasia (CIN) suggest a potential role for vitamin D in lesion regression or recurrence and report changes in HPV clearance and inflammatory biomarkers. However, the available trials are small and use heterogeneous supplementation protocols, limiting inference across studies without formal meta‐analysis. Evidence in invasive cervical cancer is substantially more limited, with most data derived from preclinical models or studies of precursor lesions. Cervical tumourigenesis typically progresses from dysplasia to invasive carcinoma and is driven primarily by persistent high‐risk HPV infection. The most direct randomised controlled trial (RCT) evidence comes from two double‐blind, placebo‐controlled trials conducted by the same research group in CIN1 and CIN2/3 cohorts, respectively. Both reported benefit from vitamin D 3 supplementation for lesion regression or recurrence [ 39 ] (Table  3 ). In the CIN2/3 cohort, a follow‐up trial reported lower combined CIN1/2/3 recurrence with vitamin D supplementation than with placebo (18.5% vs. 48.1%, p  = 0.02) [ 40 ]. However, when recurrence was restricted to CIN2/3, the target population of the trial, the difference was no longer significant. The reported benefit was therefore driven mainly by lower‐grade CIN1 recurrence rather than a demonstrated reduction in recurrence of the higher‐grade lesions of greatest clinical concern. Both trials were small and therefore had limited statistical power. As summarised in Table  4 , they constitute the cervical‐specific RCT evidence base and neither enrolled patients with invasive disease. The supplementation regimen may also influence efficacy, although direct evidence comes from non‐cervical cancer populations. Cross‐cancer meta‐analyses have reported lower overall cancer mortality with vitamin D supplementation [ 42 , 43 ] (Table  3 ). An individual‐patient‐data meta‐analysis, also not specific to cervical cancer, found a mortality benefit with daily dosing but not with bolus regimens [ 44 ] (Table  3 ). The general‐population VITAL trial found no reduction in overall cancer incidence with daily vitamin D3; a secondary mortality benefit emerged only in a sensitivity analysis excluding early follow‐up [ 41 ] (Table  3 ). None of these data derive from cervical cancer or CIN cohorts and they should not be interpreted as demonstrating a dosing effect within cervical disease. Observational evidence in Black women has reported relationships among human immunodeficiency virus (HIV) status, serum 25(OH)D concentrations and high‐grade dysplasia [ 45 ] (Table  3 ). These findings raise the possibility of immune‐inflammatory interactions and support further investigation of whether vitamin D‐mediated immunomodulation and redox biology influence HPV persistence, clearance or progression. Two ecological analyses report associations in the same direction. SEER‐18 registry data show an inverse association between ambient ultraviolet radiation (UVR) exposure and cervical cancer incidence [ 46 ] (Table  3 ), whilst a diagnosis of non‐melanoma skin cancer (NMSC), used as a proxy for cumulative UVB exposure, is associated with lower subsequent cervical cancer risk [ 47 ] (Table  3 ). The latter association may be influenced by more intensive medical surveillance and local immunosuppression in patients with NMSC, neither of which is captured adequately by the proxy‐based design. More broadly, ecological studies rely on aggregate or proxy exposure measures and are vulnerable to aggregation bias and unmeasured confounding by outdoor activity, skin pigmentation and access to screening. The ecological findings are therefore compatible with, but do not establish, a protective role for vitamin D in cervical carcinogenesis. A separate ecological modelling analysis projected substantially lower population‐level cancer incidence at higher population‐level 25(OH)D concentrations [ 51 ] (Table  3 ). Because this projection is derived from ecological correlations, it remains vulnerable to confounding and aggregation bias. It cannot establish individual‐level causation and requires prospective validation before it can inform public‐health recommendations. At the individual level, a Japanese case–control study reported a dose‐dependent reduction in invasive cervical carcinoma risk with higher dietary calcium and vitamin D intake, but no association with CIN3 [ 48 ] (Table  3 ). The absence of an association with CIN3, despite lower risk of invasive carcinoma, is difficult to reconcile with a model in which vitamin D acts predominantly during the earliest stages of carcinogenesis. One explanation is measurement‐related: dietary vitamin D intake is a weaker indicator of vitamin D status than directly measured serum 25(OH)D, and an effect at the CIN3 stage may therefore be difficult to detect using dietary assessment. Alternatively, vitamin D may exert a greater effect during later progression from in situ to invasive disease through processes such as angiogenesis, stromal invasion or immune evasion; this remains a hypothesis rather than an established mechanism. The findings support prospective studies using serum‐based measures of vitamin D status. Although dose–response trends across intake quartiles strengthen the internal consistency of the study, interpretation remains limited by dietary recall, residual confounding and uncertain generalisability beyond the Japanese population. Two narrative reviews similarly report inverse associations between circulating 25(OH)D and cervical neoplasia or low‐grade squamous intraepithelial lesion (LSIL) persistence [ 49 ] (Table  3 ), and between vitamin D status and persistent HPV infection or progression [ 50 ] (Table  3 ). Because these are narrative rather than systematic or individual‐level analyses, they inherit the limitations of the underlying primary studies and should not be treated as independent confirmatory evidence. Overall, the observational literature provides moderate epidemiological support for a protective association, particularly during later stages of cervical carcinogenesis. Prospective cohorts with individual‐level serum measurements, standardised HPV genotyping and adequate adjustment for confounding are needed for confirmation. Other evidence is methodologically more limited. A hospital‐based, non‐randomised comparison of HPV‐positive women found no significant effect of vitamin D on HPV regression, although supplemented women had improved metabolic and inflammatory markers [ 52 ] (Table  3 ). This is biologically plausible because chronic inflammation can facilitate HPV persistence, but the study cannot establish causality. Allocation was based on supplementation history rather than randomisation, creating substantial potential for selection bias, and the six‐month follow‐up was shorter than the 12–24 months often required to observe HPV clearance. Replication in randomised, adequately powered trials with at least 24 months of follow‐up is required before clinical conclusions can be drawn. An important unresolved question is whether vitamin D‐mediated immunomodulation directly facilitates HPV clearance or acts mainly by reducing the chronic inflammatory state that promotes viral persistence. This distinction requires adequately powered trials with at least 24 months of follow‐up and prespecified viral‐clearance endpoints. The interaction between vitamin D status and the cervicovaginal microbiome also warrants investigation. Vitamin D status, the cervicovaginal microbiome and local mucosal immunity may represent an interacting axis in HPV‐driven carcinogenesis, but prospective evidence remains limited. Future research should include large, adequately powered RCTs using daily vitamin D supplementation with prespecified, mechanistically justified target serum 25(OH)D concentrations. At a minimum, the regimen should correct deficiency aligned to clinical guidelines, with exploratory upper targets informed by preclinical and observational data. CIN regression and HPV clearance should be considered co‐primary endpoints, with at least 24 months of follow‐up. Additional priorities include Phase II trials of vitamin D as an adjunct to platinum‐based chemoradiation in locally advanced cervical cancer, with translational sub‐studies assessing VDR expression and the tumour immune microenvironment. Stratified analyses by VDR genotype and prospective evaluation of the vitamin D‐microbiome‐mucosal‐immunity axis are also needed (Table  4 ). At present, vitamin D supplementation is not recommended as a therapeutic or adjunct intervention for cervical cancer or CIN outside a clinical trial. Its immunomodulatory, anti‐inflammatory and antiproliferative properties provide a rationale for further study but not for a change in clinical practice. This question should remain separate from correction of vitamin D deficiency according to general clinical guidelines for bone and metabolic health. Despite mechanistic evidence implicating the vitamin D‐VDR axis in endometrial tumourigenesis, large epidemiological studies have not demonstrated a consistent statistically significant association between circulating 25(OH)D concentrations and endometrial cancer incidence. This largely null pattern is observed across several exposure measures, including dietary intake, predicted biomarkers, directly measured serum concentrations and genetic proxies, making a simple measurement artefact an unlikely explanation. The Vitamin D Pooling Project of Rarer Cancers provides supporting evidence from a nested case–control analysis of seven international cohorts [ 53 ] (Table  3 ). A marginal case–control difference was fully attenuated after adjustment for body mass index (BMI), and no dose–response relationship was observed across 25(OH)D categories. The multicentre design and centralised biomarker measurement make this one of the more methodologically rigorous evaluations in the field. Nevertheless, the small proportion of severely deficient participants may have limited power to detect subgroup‐specific effects, supporting the need for larger, genotype‐stratified cohorts with mechanistic endpoints. More recent retrospective data from a multicentre analysis of patients with uterine or ovarian tumours found lower vitamin D concentrations in malignant than in benign pathology [ 54 ] (Table  3 ). The cross‐sectional design, however, cannot establish whether hypovitaminosis D preceded or followed malignancy. Advanced cancer can itself lower circulating 25(OH)D through reduced sun exposure, impaired hepatic hydroxylation and inflammation, whilst the absence of information on supplementation, co‐morbidity and treatment status leaves substantial potential for residual confounding. These findings should therefore be regarded as hypothesis‐generating rather than causal evidence. Variation in VDR function represents an additional factor not captured by serum biomarker studies alone. Canonical VDR polymorphisms, including FokI (rs2228570), BsmI (rs1544410), TaqI (rs731236) and ApaI (rs7975232), have been examined in relation to endometrial cancer risk in candidate‐gene studies and meta‐analyses. Evidence remains heterogeneous, with no single variant showing a consistent, replicated association across ethnically diverse populations. This inconsistency partly reflects limitations of the candidate‐gene literature: many studies predate genome‐wide association methods, were underpowered to detect small genetic effects and did not correct adequately for multiple comparisons across correlated variants. Population stratification and between‐study differences in ethnicity and genotyping platforms add further heterogeneity. The absence of replication of VDR variants as endometrial cancer risk loci in large genome‐wide association study (GWAS) datasets further weakens the case for any single polymorphism as a clinically meaningful modifier. Haplotype‐level analyses in adequately powered, ethnically stratified cohorts remain appropriate for further investigation. VDR status nevertheless remains mechanistically relevant because tumours with low VDR expression are unresponsive to 1,25‐dihydroxyvitamin D 3 in vitro. The relationship between circulating vitamin D and endometrial cancer risk may therefore depend partly on receptor‐level competency, which population‐level serum biomarker studies cannot capture. Ecological studies using geographic latitude as a proxy for ultraviolet‐B exposure and cutaneous vitamin D synthesis have produced mixed findings for endometrial cancer, unlike the more consistent inverse gradients reported for colorectal and breast cancers. As in the cervical cancer literature, these analyses are vulnerable to aggregation bias because population‐level associations cannot establish individual‐level causation. Geographic patterns also co‐vary with socioeconomic status, diet, healthcare access, obesity prevalence, physical activity and hormone use, all of which may influence endometrial cancer risk independently. Ecological evidence is therefore insufficient to resolve causality and should be regarded as hypothesis‐generating. The absence of a consistent geographic association is nevertheless notable because such a pattern might be expected if vitamin D were a dominant aetiological determinant of endometrial cancer. A further gap is the limited evidence on vitamin D status in endometrial hyperplasia, the principal morphological precursor of endometrioid endometrial carcinoma. Hyperplasia represents an earlier stage of disease, is responsive to progestogen therapy or surgery, and commonly occurs in populations characterised by obesity and metabolic syndrome, both of which are independently associated with vitamin D insufficiency. This setting therefore provides an important opportunity for prevention‐oriented mechanistic research. To date, no adequately powered study has determined whether correction of vitamin D deficiency alters progression of hyperplasia or promotes histological regression, a gap identified as a research priority in Table  4 . In advanced disease, the Phase II PRIMMO trial is the only prospective interventional dataset in which vitamin D was administered to patients with histologically confirmed endometrial carcinoma, as part of a five‐drug immunomodulatory combination with pembrolizumab and radiotherapy [ 55 ] (Table  3 ). The primary endpoint was not met. Because the study used a multi‐component, single‐arm, uncontrolled design, the contribution of vitamin D cannot be isolated. The trial is therefore best interpreted as evidence of the feasibility and safety of the combination strategy rather than evidence for a specific vitamin D effect. No RCT has evaluated vitamin D as a single agent in endometrial carcinoma (Table  4 ). Across study designs, the endometrial cancer literature shows a clear disparity between mechanistic plausibility and largely null epidemiological and clinical findings. Prospective cohorts, pooled multicentre case–control studies, Mendelian randomisation analyses and ecological investigations generally show a null or negligible association between vitamin D status and endometrial cancer incidence. Cross‐sectional associations between low 25(OH)D and cancer diagnosis may reflect reverse causation and residual confounding, particularly from obesity, insulin resistance and systemic inflammation. These factors can lower circulating vitamin D whilst independently promoting endometrial carcinogenesis through hormonal and metabolic pathways. Statistical adjustment may not fully address this confounding structure. Mendelian randomisation therefore provides useful complementary evidence, and its null findings further weaken the case for vitamin D deficiency as a primary aetiological factor in endometrial cancer. A major evidence gap is the absence of rigorously designed RCTs examining vitamin D supplementation against histological endpoints in endometrial hyperplasia or as a chemopreventive strategy in populations at high risk of endometrial cancer. Existing interventional studies are dominated by multi‐drug combinations and surrogate biomarker outcomes and cannot isolate the effect of vitamin D. Future trials should stratify participants by baseline 25(OH)D to examine potential threshold or dose–response effects, particularly in severely deficient individuals (≤ 25 nmol/L). They should incorporate histological or molecular endpoints, such as the Ki‐67 proliferation index, apoptotic fraction and CYP24A1 expression, rather than relying only on circulating biomarkers. Follow‐up should be long enough to capture incident cancer or regression of hyperplasia, and VDR expression and VDR variant status should be assessed prospectively as potential predictive biomarkers (Table  4 ). Without trials designed to isolate the independent contribution of vitamin D and adequately powered for clinically relevant endpoints, the evidence will remain inconclusive. Current evidence does not support vitamin D supplementation as a strategy to prevent or treat endometrial cancer. This conclusion should be interpreted in the context of the limited trial evidence: absence of demonstrated benefit is not equivalent to definitive proof of inefficacy, and supplementation in severely deficient, high‐risk women remains insufficiently studied. When vitamin D deficiency is identified in gynaecological oncology patients, it should be corrected according to established general clinical guidelines for bone and metabolic health rather than through cancer‐specific protocols. Any associated improvements in immune function, musculoskeletal health, cardiovascular risk or perioperative recovery should be regarded as supportive health benefits rather than direct antitumour effects. This review does not provide evidence to support routine oncology‐driven screening of all patients with gynaecological cancer for vitamin D deficiency. Where screening is undertaken, it should follow standard clinical indications independent of the cancer diagnosis or treatment plan.

The evidence base for vitamin D in gynaecological oncology is markedly asymmetric: mechanistic characterisation is extensive, whereas clinical evaluation is sparse and predominantly non‐interventional. Across ovarian, cervical and endometrial cancers, preclinical studies identify vitamin D and its analogues as plausible adjunctive agents that can engage ROS‐mediated mitochondrial apoptosis, modulate autophagic flux, induce cell‐cycle arrest and attenuate oncogenic signalling pathways including NF‐κB, PI3K/AKT/mTOR and c‐Myc. In cervical neoplasia, the strongest clinical evidence is limited to CIN regression or recurrence and HPV‐related outcomes and cannot be extrapolated to therapeutic efficacy in invasive cervical cancer because RCT data are absent. In ovarian cancer, inverse associations between 25(OH)D and mortality remain vulnerable to confounding, and interventional evidence is limited. In endometrial cancer, cohort, case–control and Mendelian randomisation studies generally converge on a null association with incidence. These inconsistent clinical findings reflect important limitations of the current evidence base, including small samples, non‐randomised designs and multi‐agent interventions that cannot isolate the contribution of vitamin D. They should not be interpreted as definitive proof that vitamin D has no clinical effect. Addressing this gap requires the integrated preclinical and clinical priorities outlined in Tables  2 and 4 , including stronger disease‐relevant models, and adequately powered and appropriately controlled trials. Until such evidence is available, it will remain difficult to determine whether optimising vitamin D status provides antitumour benefit in gynaecological malignancies. Overall, the evidence supports a clear distinction between correction of vitamin D deficiency according to established general clinical guidelines, for bone and metabolic health, irrespective of cancer diagnosis, and the use of vitamin D as an anticancer intervention, for which evidence remains insufficient. Routine screening of all gynaecological oncology patients for vitamin D deficiency solely for oncological purposes is not supported, nor is supplementation intended to modify cancer outcomes outside a clinical trial. These are separate clinical questions, and the evidence reviewed here does not justify a change in current oncological practice.

Ovarian cancer remains the most lethal gynaecological malignancy, with a five‐year relative survival rate of approximately 52% for 2016–2022 [ 56 ]. First‐line treatment is based on surgery and platinum‐based chemotherapy [ 57 ]. Vitamin D metabolism is dysregulated in ovarian tumours, although the direction and extent of this dysregulation vary across studies. Expression of the activating enzyme CYP27B1 (1α‐hydroxylase) remains debated. One study detected CYP27B1 in ovarian cancer cell lines, consistent with the capacity for intratumoural bioactivation [ 58 ]. By contrast, subsequent tissue‐based studies reported significantly lower CYP27B1 expression in ovarian cancers than in normal ovarian epithelium, with the lowest expression in poorly differentiated tumours and metastases and a positive association with overall survival [ 59 ]. A systematic review also noted upregulation of vitamin D‐metabolising enzymes in ovarian cancer cells, although several subsequent studies did not confirm this pattern [ 60 ]. Differences in experimental model, tumour grade and histological type may contribute to these discrepant findings and argue against a uniform model of autocrine vitamin D bioactivation in ovarian cancer. The evidence for the catabolic enzyme CYP24A1 is more consistent: its overexpression correlates independently with poor prognosis and may counteract retained CYP27B1 activity, thereby reducing net intratumoural calcitriol availability [ 61 ]. This should nevertheless be regarded as a cross‐tumour hypothesis rather than an established general mechanism, because the ovarian evidence derives largely from separate, non‐comparable study designs and CYP24A1 has not been tested directly as a unifying resistance marker. Among the structural analogues tested, PRI‐5202 showed the greatest potency for inducing CYP24A1 mRNA expression in both patient‐derived high‐grade serous ovarian carcinoma (HGSOC) lines. Its EC 50 was lower in the CYP24A1‐high line (14,433: 0.92 ± 0.20 nmol/L) than in the CYP24A1‐low line (13,781: 2.98 ± 1.10 nmol/L) [ 62 ]. Paradoxically, the CYP24A1‐low line was more sensitive overall, with greater reductions in viability and greater nuclear VDR accumulation despite its weaker transcriptional response. CYP24A1 induction potency therefore did not predict antiproliferative efficacy. Whether this discrepancy reflects CYP24A1‐mediated catabolism, downstream signalling uncoupling or other inter‐line differences requires direct pharmacokinetic and functional investigation. Calcitriol exerts antiproliferative and pro‐apoptotic effects across ovarian cancer cell lines, although the pharmacological relevance of these findings varies substantially by experimental context. Several concentrations used, particularly in MDAH‐2774 and SKOV‐3 cells [ 15 , 16 ] (Table  1 ), substantially exceed pharmacologically achievable serum levels. These findings therefore primarily demonstrate mechanistic pathway engagement, including VDR/p53 upregulation, Bax induction and Bcl‐2 suppression. OVCAR4 data are more encouraging, showing growth suppression through c‐Myc downregulation and p27 upregulation across a broader and more physiologically plausible concentration range [ 63 ]. Combination studies suggest that calcitriol may augment established cytotoxic regimens, although it remains unclear whether these effects are more than additive. In SKOV‐3 cells, co‐administration with cisplatin produced greater apoptosis and VEGF suppression than either agent alone, but formal interaction analysis, such as a combination index, was not performed [ 16 ]. In OVCAR4 cells, calcitriol combined with docosahexaenoic acid (DHA) produced greater growth inhibition than monotherapy. The effect was stronger at lower calcitriol concentrations during extended treatment (12 days) than during short‐term treatment (3 days), supporting further investigation of sustained low‐dose schedules [ 63 ]. A distinct transcriptional mechanism involves calcitriol‐mediated suppression of CCAT2 , a lncRNA overexpressed in ovarian cancer. Calcitriol reduced TCF7L2 binding at the c‐Myc promoter, decreased c‐Myc protein abundance and attenuated proliferation, migration and invasion in vitro [ 64 ]. Whether this axis operates independently of or converges with canonical apoptotic pathways remains unresolved because c‐Myc downregulation influences both proliferative and apoptotic thresholds, and direct EMT marker quantification was not reported. Validation in three‐dimensional culture systems or in vivo models is required before mechanistic independence can be established. VDR agonist derivatives may also influence the tumour microenvironment. In a 2025 study using patient‐derived primary cells and a mouse model, these derivatives inhibited TGF‐β1‐induced mesenchymal transition of peritoneal mesothelial cells and reduced ovarian cancer adhesion to the peritoneum [ 65 ]. These findings suggest a potential role for VDR signalling in peritoneal dissemination, a process central to ovarian cancer progression that remains underexplored in vitamin D research. Evidence also links VDR activation with ROS generation and mitochondrial apoptosis in ovarian cancer. In ovarian cancer cells, calcitriol enhanced radiation‐induced apoptosis through NOX4‐derived ROS. NOX4 and p22phox were associated with mitochondrial membrane depolarisation, cytochrome c release and effector caspase activation, whilst antioxidant reversal supported ROS as an essential mediator [ 14 ] (Table  1 ). These findings raise the possibility that vitamin D may be more effective as a priming intervention before genotoxic stress than as a standalone cytotoxic agent, although this scheduling hypothesis has not been tested directly. The pro‐oxidant response observed in tumour cells contrasts with the cytoprotective role of VDR in normal ovarian physiology. VDR loss is associated with oxidative stress and folliculogenesis defects [ 66 ], whilst vitamin D 3 stabilises mitochondrial function through MAPK/ERK1/2 signalling in PCOS granulosa cells [ 67 ]. The basis for these context‐dependent ROS effects remains uncertain. Differential co‐activator recruitment and transcriptional reprogramming within malignant cells may contribute, but this proposed mechanism requires direct evaluation before therapeutic selectivity can be assumed. Breast cancer models provide a relevant but extrapolated mechanistic parallel. In MCF‐7 cells, calcitriol promotes Bax translocation to mitochondria, cytochrome c release and ROS production. These events define a caspase‐independent intrinsic apoptotic pathway in which commitment to cell death precedes and does not require effector caspase activation [ 29 ] (Table  1 ). A parallel calcium/calpain pathway, in which calpain cleaves procaspase‐12, has been demonstrated in rodent models [ 30 , 31 , 68 ] (Table  1 ). However, caspase‐12 is a non‐functional pseudogene in most humans [ 69 ], limiting the direct relevance of this cleavage mechanism to human gynaecological cancer. In human systems, the calpain‐dependent pathway may instead act on substrates such as calpastatin and cytoskeletal proteins without obligatory caspase‐12 involvement. The mitochondrial‐ROS and calcium‐calpain pathways are not mutually exclusive and may converge on shared apoptotic effectors. Direct quantitative comparisons of VDR expression and ROS‐mitochondrial biology between MCF‐7 and ovarian cancer lines remain limited, so this framework requires validation in ovarian‐specific models. ROS induction is an established therapeutic mechanism across several drug classes in ovarian cancer and provides useful context for evaluating the potential adjuvant role of vitamin D, although it is not direct evidence for vitamin D itself [ 70 ]. Cisplatin induces a mitochondrial ROS response that augments cytotoxicity caused by nuclear DNA damage. Mitochondrial content and ROS‐generating capacity have been identified as important determinants of sensitivity to cisplatin‐induced apoptosis in HGSOC cell lines [ 71 , 72 ]. PARP inhibitors act through impaired base‐excision repair and can independently increase ROS through upregulation of NADPH oxidases 1 and 4; antioxidant rescue experiments indicate that this ROS amplification contributes to growth inhibition beyond impaired DNA repair alone [ 73 ]. Other ROS‐inducing agents provide further mechanistic precedent. CYT‐Rx20 induces apoptosis and DNA damage in ovarian cancer models, with these effects reversed by thiol antioxidants [ 74 ]. Berberine induces oxidative DNA damage and suppresses homologous recombination repair, increasing sensitivity to the PARP inhibitor niraparib in vitro and in xenograft models [ 75 ]. Gedunin similarly induces marked ROS generation, mitochondrial membrane depolarisation, cytochrome c release and intrinsic apoptosis, with these effects reversed by N‐acetylcysteine [ 76 ]. Vitamin D and its analogues have mechanistic features consistent with engagement of redox‐sensitive pathways through VDR‐mediated effects on mitochondrial function and oxidative signalling. In breast cancer models, vitamin D enhanced mitochondrial membrane depolarisation and cytochrome c release [ 77 , 78 ]. These observations are consistent with but do not establish a role for calcitriol as an adjuvant to established ROS‐inducing regimens at doses below those associated with hypercalcaemia. Direct ovarian‐specific dose‐ranging and combination studies are therefore required. Evidence for VDR‐dependent NOX4 activation by calcitriol in ovarian cancer remains limited to a single study. Broader mechanistic inference relies on related cancer models and on the NOX1/NOX4 upregulation observed with PARP inhibition in ovarian cancer [ 73 ], an associative rather than directly demonstrated link to calcitriol. A central unresolved question is whether gain‐of‐function mutant TP53 , present in approximately 96% of HGSOC [ 79 ], can reprogramme VDR‐dependent ROS signalling from a pro‐apoptotic to an anti‐apoptotic response. This concern derives from a single mechanistic study outside the gynaecological cancer literature. ChIP‐on‐chip analysis showed that mutant p53 physically and functionally interacts with VDR, is recruited to VDR‐regulated gene promoters and can reprogramme calcitriol into an anti‐apoptotic agent in tumour cells carrying gain‐of‐function p53 mutations [ 17 ] (Table  1 ). Whether this effect occurs in ovarian cancer has not been tested and requires direct evaluation in HGSOC‐specific models before its clinical significance can be determined (Table  2 ). Key translational barriers and corresponding research priorities for advancing Vitamin D‐based therapeutic strategies in ovarian, cervical and endometrial cancers. Uncharacterised interaction between gain‐of‐function TP53 mutation (present in ~96% of HGSOC) and VDR‐dependent ROS/apoptosis signalling; mutant p53 may reprogramme calcitriol into an anti‐apoptotic agent, but this is untested in ovarian models No pharmacokinetic‐pharmacodynamic model links achievable intratumoural calcitriol levels to the NOX4‐ROS threshold required for apoptotic commitment Reliance on supraphysiological concentrations in key cell‐line studies, limiting clinical extrapolation Formal synergy testing with cisplatin/DHA not performed VDR's role in peritoneal mesothelial‐mesenchymal transition and dissemination remains underexplored VDR haplotype frequencies uncharacterised in African‐ancestry populations TP53‐stratified xenograft and patient‐derived organoid models to resolve the mutant‐p53‐VDR interaction Dose‐ranging/PK‐PD studies defining intratumoural calcitriol exposure relative to ROS thresholds Standardised combination/synergy studies with platinum agents and PARP inhibitors In vivo and 3D models evaluating VDR's role in peritoneal tumour microenvironment modulation Genetic epidemiology characterising VDR haplotypes in underrepresented populations Absence of antioxidant‐rescue experiments, so ROS is associated with, but not proven necessary for, apoptosis In vitro evidence confined almost exclusively to HPV‐positive lines (SiHa, CaSki), with no HPV‐negative or organoid comparators Radiosensitisation mechanism inferred from proteomic data rather than direct DNA damage Functional ROS‐rescue studies to establish causality in cervical cancer apoptosis Expansion of models to HPV‐negative and patient‐derived organoid systems with VDR stratification Direct DNA damage/cell‐cycle endpoint studies combining calcitriol and radiation in vivo Hormetic (non‐linear) dose–response of vitamin D analogues uncharacterised across histological types and VDR expression levels Much mechanistic evidence derives from non‐malignant/bovine models rather than endometrial carcinoma itself Proposed Trx/TXNIP selective‐cytotoxicity mechanism and paraptosis‐like death pathway lack direct functional validation In vivo evidence limited to a single immunocompromised xenograft study using a non‐VDR‐dependent analogue Concurrent tumour upregulation of activating hydroxylases and CYP24A1 creates an unresolved resistance mechanism Patient‐derived organoid/primary cell models to validate CYP24A1 resistance, dose–response and combined calcitriol‐progesterone findings Site‐specific syngeneic or humanised mouse models reproducing tumour microenvironment and hormonal context Functional TXNIP knockdown/overexpression studies to test the proposed selectivity mechanism Biomarker co‐development (VDR expression, CYP24A1 activity, progesterone receptor status) for early‐phase trial stratification Mendelian‐randomisation and genotype‐stratified cohorts to resolve BMI‐related confounding RCTs in endometrial hyperplasia using histological/molecular endpoints (Ki‐67, apoptotic fraction, CYP24A1 expression) with adequate duration Trials isolating vitamin D's independent contribution, free of combination‐drug confounding Abbreviations: BMI, body mass index; CYP24A1, 24‐hydroxylase; DHA, docosahexaenoic acid; DNA, deoxyribonucleic acid; HGSOC, high‐grade serous ovarian carcinoma; HPV, human papillomavirus; Ki‐67, proliferation marker encoded by MKI67; NOX4, NADPH oxidase 4; p53/TP53, tumour protein 53 (gene); PARP, poly (ADP‐ribose) polymerase; PK‐PD, pharmacokinetic‐pharmacodynamic; RCT, randomised controlled trials; ROS, reactive oxygen species; Trx, thioredoxin; TXNIP, thioredoxin‐interacting protein; VDR, vitamin D receptor. Preclinical models of cervical cancer consistently demonstrate that vitamin D metabolites exert concentration‐dependent antiproliferative and pro‐apoptotic effects through mechanistically convergent but distinct pathways, including VDR‐autocrine signalling, ROS‐mediated mitochondrial apoptosis and cell cycle arrest. A defining feature of cervical cancer biology is the expression of a functional vitamin D‐metabolising system comprising CYP27B1, CYP24A1 and VDR, which permits local calcitriol signalling. Calcifediol treatment in SiHa and HeLa cells reduced viability and induced apoptosis together with coordinated upregulation of CYP27B1 , CYP24A1 and VDR [ 18 , 19 ] (Table  1 ). This pattern is consistent with autocrine bioactivation and negative‐feedback induction, but it does not establish intratumoural bioactivation as the operative anticancer mechanism. Treatment of SiHa cells with the upstream precursor cholecalciferol produced a different enzyme response, with CYP27B1 downregulated rather than upregulated [ 23 ]. Concurrent upregulation of CYP2R1 , the primary hepatic and extrahepatic 25‐hydroxylase, together with VDR and CYP24A1, suggests autocrine conversion of cholecalciferol to calcifediol as an initiating step, followed by VDR activation, CYP24A1 induction and CYP27B1 suppression through negative feedback. Further validation, including receptor‐binding and functional rescue experiments, is required. Across HeLa and CaSki cell studies, calcitriol, calcifediol and the synthetic analogue calcipotriol induced ROS generation, mitochondrial membrane depolarisation, caspase 3/7 activation and phosphatidylserine externalisation at supraphysiological concentrations, a pattern consistent with intrinsic apoptosis [ 19 , 24 , 80 ]. None of the cited studies performed antioxidant rescue experiments (Table  2 ). The observed dose pattern, with growth inhibition at nanomolar concentrations but significant apoptosis only at micromolar concentrations, indicates that clinical translation will require either targeted intracellular delivery or analogues with substantially greater potency. Calcitriol also engages cell‐cycle checkpoints through a distinct pathway. In HeLa S3 cells, calcitriol downregulated the oncogene HCCR‐1 and upregulated the cyclin‐dependent kinase inhibitor p21. Rescue of HCCR‐1 reversed G1 arrest, supporting a causal role for HCCR‐1 suppression [ 81 ]. Whether this pathway operates independently of ROS‐driven apoptosis has not been tested. Radioresistance remains an important determinant of treatment failure in locally advanced cervical cancer and contributes to persistent local recurrence despite concurrent chemoradiotherapy. This creates a rationale for molecularly targeted adjuncts. The radiosensitising effects of agents such as cisplatin or gemcitabine are largely attributed to impaired DNA double‐strand break (DSB) repair and redistribution of the cell cycle towards radiosensitive phases. By contrast, preclinical evidence suggests that calcitriol acts through VDR‐dependent reprogramming of the autophagy‐apoptosis pathway. This mechanism does not appear to involve DSB amplification or cell‐cycle perturbation and may therefore complement existing approaches, although such complementarity has not been tested directly in combination studies. In vitro radiosensitisation data derive from two sequential studies using SiHa and CaSki cells, both HPV‐16‐positive, VDR‐expressing squamous cervical cancer lines. Calcitriol reduced clonogenic survival relative to irradiation alone, increased apoptosis and decreased autophagy. Proteomic analysis showed upregulation of caspase‐8 in the combined‐treatment group [ 20 ] (Table  1 ), and a subsequent study confirmed the radiosensitisation signal [ 21 ] (Table  1 ). VDR dependence is also consistent with broader evidence on vitamin D‐mediated autophagy modulation. In an extrapolated model of anti‐oestrogen‐resistant breast cancer, VDR activation suppressed pro‐survival autophagy; VDR silencing abolished the sensitising effect of calcitriol and restored survival autophagy through the IRE1α‐JNK pathway [ 82 ]. The proposed mechanism involves calcitriol‐mediated degradation of Ambra1, a positive regulator of autophagy initiation, together with increased caspase‐8 activation [ 21 ] (Table  1 ). Rescue experiments support a causal role: pharmacological caspase inhibition and Ambra1 overexpression each restored tumour‐cell proliferation during combined calcitriol and radiation treatment. Whether Ambra1 depletion reflects transcriptional regulation or occurs downstream of caspase‐mediated cleavage remains unresolved. Neither study linked the effect to canonical radiation‐induced pathways such as double‐strand break signalling or cell‐cycle redistribution; the mechanistic conclusions therefore rest primarily on proteomic inference rather than direct interrogation of DNA damage. A separate study showed that pharmacological autophagy inhibition enhanced radiation‐induced cytotoxicity and apoptosis in cervical cancer cells through Akt/mTOR/P70S6K downregulation [ 83 ], supporting the broader principle that suppression of pro‐survival autophagy can enhance radiosensitivity. In vivo, oral calcitriol monotherapy reduced tumour volume in HPV‐16‐positive TC‐1‐bearing mice relative to placebo, whilst calcitriol combined with cisplatin improved both tumour‐volume and survival outcomes compared with cisplatin alone [ 22 ]. This protocol did not include a radiation arm; the survival benefit therefore reflects the calcitriol‐cisplatin combination only. Formal interaction or combination‐index analysis was not performed, and the HPV‐16 E6/E7‐driven syngeneic TC‐1 model may not capture the heterogeneous mutational landscape of clinical cervical cancer (Table  2 ). Several methodological limitations constrain clinical translation. The available in vitro radiosensitisation data are derived from VDR‐expressing HPV‐positive lines (SiHa and CaSki), with no published comparison using HPV‐negative models or patient‐derived organoids. This is relevant because HPV status may influence autophagy‐radiation interactions and radiosensitivity. A recent study reported distinct autophagy‐receptor expression and radiation‐response profiles in HPV‐positive cervical cancer cells compared with HPV‐negative C33A cells; autophagy inhibition enhanced radiation‐induced cytotoxicity selectively in the HPV‐positive cells [ 84 ]. The calcitriol radiosensitisation phenotype may therefore reflect VDR expression, HPV‐associated cellular context or both. Current data cannot distinguish the relative contribution of these factors, and HPV dependence should be treated as a hypothesis requiring direct testing in matched HPV‐positive and HPV‐negative models. Transcriptomic profiling further indicates that calcitriol has broad effects in cervical cancer cells. In CaSki cells, more than 1000 protein‐coding genes and approximately 400 non‐coding RNAs were altered, including significant repression of migration‐associated matrix metalloproteinases [ 80 ]. These effects extend beyond autophagy suppression, whilst their implications for normal cervical epithelium remain uncharacterised. Overall, the preclinical evidence in Table  1 is promising but remains insufficient to justify unselected clinical evaluation until the dosing, scheduling and model‐system limitations summarised in Table  2 are addressed. Vitamin D signalling, ROS and cellular fate have been examined in endometrial models in relation to both endometriosis and malignant transformation. A study using bovine and human endometrial cell models reported that calcitriol reduced oxidative‐stress markers and attenuated lipopolysaccharide (LPS)‐induced apoptosis through modulation of the PI3K/AKT/mTOR pathway ( p  < 0.05) [ 25 ]. The cytoprotective effect did not appear to result primarily from direct antioxidant neutralisation; enhanced autophagic flux was implicated as a candidate mechanism. Upregulation of Beclin‐1, ATG5 and ATG7 accompanied the protective response, suggesting that autophagy may contribute to cytoprotection. This finding derives from a single study, however, and requires independent replication before autophagy can be considered the dominant mechanism. A second line of evidence concerns inflammatory signalling. Calcitriol suppresses LPS‐mediated NF‐κB activation and attenuates NLRP3 inflammasome signalling in ectopic endometrial stromal cells and the Ishikawa endometrial carcinoma line, with supporting in vivo data from a rodent endometriosis model [ 85 ]. NF‐κB and NLRP3 contribute to the pro‐inflammatory microenvironment that supports ectopic endometrial survival and immune evasion. These findings are consistent with interacting autophagic and anti‐inflammatory effects, although this integrated model has not been tested as a unified pathway. An additional complication is the non‐linear dose–response of vitamin D compounds. Comparative analysis of the synthetic analogue alfacalcidol and calcitriol in HEC‐1A endometrial adenocarcinoma cells showed a hormetic pattern: low‐dose exposure suppressed proliferation and migration, whereas supraphysiological concentrations increased both outcomes [ 25 , 26 ] (Table  1 ). This opposing concentration dependence has not been systematically characterised across histological types of endometrial cancer or correlated with baseline VDR expression (Table  2 ). Differences in the timing and dynamics of the two compounds further suggest that pharmacokinetic, as well as pharmacodynamic, factors may contribute to the divergent responses. This distinction remains unconfirmed. Vitamin D may also have context‐dependent effects on redox homeostasis in the endometrium, with cytoprotective effects reported in non‐malignant stromal and epithelial cells and pro‐oxidant, cytotoxic effects in malignant cells. The thioredoxin (Trx) system provides one possible molecular explanation. Trx is an intracellular ROS‐buffering protein inhibited by thioredoxin‐interacting protein (TXNIP), which was first identified as a vitamin D 3 ‐induced protein (VDUP1) in haematopoietic cells [ 86 ]. Calcitriol‐induced TXNIP expression in endometrial cancer cells appears to be context‐ and glucose‐dependent and is not reproducible across all cancer cell lines; it should therefore be regarded as provisional rather than established [ 27 ]. By shifting the Trx/TXNIP balance towards TXNIP, calcitriol may reduce thioredoxin‐mediated ROS buffering, increase intracellular oxidative stress, induce cell‐cycle arrest and trigger apoptosis in susceptible malignant cells. This remains a proposed model because functional TXNIP knockdown has not been reported in this context. Immunohistochemical analysis shows markedly higher Trx expression in endometrial carcinoma than in normal endometrium, with correspondingly diminished VDR‐mediated TXNIP induction, consistent with tumour‐associated suppression of a pro‐apoptotic axis. Elevated baseline Trx and reduced TXNIP may make cancer cells more susceptible to vitamin D‐induced redox imbalance, whereas the lower baseline ROS burden of non‐malignant cells may remain below the apoptotic threshold after TXNIP upregulation alone. This proposed selectivity mechanism requires direct functional testing, including TXNIP knockdown or overexpression with ROS and apoptosis readouts in matched malignant and non‐malignant endometrial cell lines before it can be considered established. Vitamin D‐associated cell death may extend beyond classical intrinsic apoptosis. In HEC‐1A cells, calcitriol combined with the AKT inhibitor perifosine triggered caspase‐mediated apoptosis together with paraptosis‐like, autophagic and necrotic death. Electron microscopy showed morphologically distinct death patterns, although these have not been functionally validated and mixed phenotypes arising from incomplete apoptosis cannot be excluded [ 28 ] (Table  1 ). Engagement of multiple death programmes may be relevant in advanced endometrial carcinoma, where BCL‐2 overexpression, p53 mutation and PI3K/AKT hyperactivation contribute to resistance to intrinsic apoptosis and limit chemotherapy efficacy. ROS‐amplification strategies that exploit elevated basal oxidative stress are biologically plausible, but resistance and toxicity remain unresolved concerns. The cell‐cycle effects of VDR activation are consistent with patterns reported in other malignancies. VDR‐mediated transcription predominantly induces G0/G1 arrest through p21 and p27 upregulation, suppression of CDK4/CDK6‐cyclin D and CDK2‐cyclin E complexes, reduced c‐MYC expression and accumulation of hypophosphorylated retinoblastoma protein. These changes maintain E2F transcriptional repression and prevent S‐phase entry [ 87 ]. This pathway has been demonstrated in endometrial cancer cell lines treated with calcitriol and progesterone, where G0/G1 arrest was accompanied by downregulation of cyclins D1 and D3 and upregulation of p27 [ 88 ]. VDRE‐driven regulation of p21 represents a direct genomic action of the VDR‐RXR heterodimer, whereas c‐MYC suppression is more likely to reflect indirect transcriptional repression. Because indirect targets depend on downstream transcriptional cascades rather than direct VDRE binding, their response may be more sensitive to VDR expression and co‐activator availability. This may help explain variation in the extent of cell‐cycle arrest across models. The therapeutic potential of vitamin D in endometrial cancer must also be considered within the hormonal context of endometrial biology. Available evidence supports a multi‐level interaction between progesterone and vitamin D involving receptor cross‐talk, transcriptional effects and enzymatic regulation of calcitriol bioavailability. Each mechanism may contribute independently, although the model has not been tested as an integrated system within a single study. At the receptor level, progesterone upregulates VDR expression in endometrial cancer cell lines and enhances the biological effects of calcitriol. Combined calcitriol‐progesterone treatment produces greater growth inhibition than either agent alone, accompanied by caspase‐3 activation, G0/G1 cell‐cycle arrest and a validated shift in the BAX:BCL‐2 ratio [ 88 ]. Proteomic analysis identified histone H1.4 (HIST1H1E), histidine triad nucleotide‐binding protein 2 (HINT2), IFN‐induced double‐stranded RNA‐activated protein kinase (EIF2AK2) and Bcl‐2‐associated X protein (BAX) as candidate effectors of the combined treatment. Each showed lower abundance in cancer cells than in immortalised endometrial epithelial cells and increased expression after calcitriol or progesterone exposure. The BAX‐mediated mechanism is consistent with the capacity of VDR activation to repress anti‐apoptotic BCL‐2 family members and enhance DNA‐damage‐response signalling and may lower the apoptotic threshold in cells with BCL‐2‐associated chemoresistance. For most of the identified proteins, however, the proteomic associations are not causal. Gain‐ and loss‐of‐function validation has been reported only for the BAX pathway and the remaining candidate proteins require independent functional confirmation. Local vitamin D metabolism provides a further level of hormonal interaction within the tumour microenvironment. Immunohistochemical analysis of endometrial carcinoma tissue microarrays ( n  = 157 EC; n  = 60 normal endometrium) showed lower VDR and CYP24A1 expression in carcinoma than in normal endometrium. In contrast, the activating hydroxylases CYP27A1 and CYP2R1 were upregulated ( p  = 0.0002 and p  = 0.03, respectively) and correlated positively with nuclear VDR and inversely with Ki‐67 [ 89 ]. Functional studies showed reduced viability and colony formation after vitamin D exposure, together with intracellular 25(OH)D accumulation and nuclear VDR translocation. These findings suggest that circulating 25(OH)D, rather than systemic calcitriol, may provide the substrate for local antitumour activity in some endometrial carcinomas and support investigation of the vitamin D axis independently of circulating calcitriol. Whether the observed hydroxylase upregulation represents compensation for impaired downstream signalling remains unresolved. The potential relevance of activating‐enzyme upregulation is complicated by CYP24A1, the principal calcitriol‐catabolising enzyme, which is overexpressed in advanced‐stage tumours and cancer cell lines relative to normal tissue [ 90 ], although these observations derive from separate cohorts rather than a single integrated dataset. This pattern may create a resistance mechanism in which tumours both generate and degrade the active ligand. Two lines of evidence support a functional role for CYP24A1 : progestins suppress calcitriol‐induced CYP24A1 in a progesterone‐receptor‐dependent manner, prolonging calcitriol activity and enhancing growth inhibition; and siRNA‐mediated CYP24A1 knockdown independently sensitises cells to calcitriol. The finding that actinomycin D, but not cycloheximide, blocked calcitriol‐induced CYP24A1 splicing indicates a transcriptionally driven mechanism that does not require new protein synthesis. Together, these data provide a mechanistic rationale for progesterone‐calcitriol combination therapy, although the combination has not been compared directly with either agent alone in vivo. In vivo evidence remains a major translational gap. The principal proof‐of‐concept data derive from xenografts treated with PT19c, a non‐hypercalcaemic vitamin D 2 derivative that reduced tumour volume without inducing hypercalcaemia. The effect was independent of VDR binding and involved disruption of actin‐filament organisation and focal‐adhesion formation [ 91 ]. This VDR‐independent mechanism is relevant given the reduced VDR expression reported in advanced disease and suggests that VDR‐low tumours might still respond to cytoskeletally active analogues. However, the evidence is limited to a single study in an immunocompromised model without site‐specific implantation to assess local invasion or peritoneal spread (Table  2 ). The preclinical studies supporting the endometrial cancer findings, including the PT19c xenograft data, are summarised alongside the ovarian and cervical evidence in Table  1 . Addressing the identified resistance, validation and model‐system limitations will require moving beyond immortalised cell lines and single xenograft studies towards models that better reproduce the tumour microenvironment and hormonal context, as outlined in Table  2 [ 92 ].

Gynaecological malignancies, encompassing ovarian, cervical and endometrial cancers, contribute substantially to cancer‐related morbidity and mortality in women worldwide. According to the Global Burden of Disease Study 2021, these three cancers accounted for over 1.4 million incident cases in 2021. Their burden differs markedly across socio‐demographic settings: cervical cancer disproportionately affects women in low‐ and middle‐income countries, whereas ovarian and endometrial cancers carry the greatest mortality burden in high‐income countries, where treatment‐resistant recurrent disease contributes substantially to adverse outcomes [ 1 , 2 ]. Despite advances in surgical techniques, cytotoxic regimens and targeted biological agents, treatment varies considerably by stage and histology. Surgery remains the primary, and often sole, curative modality for early‐stage disease, particularly endometrial cancer, in which most patients present with Stage I disease treated by hysterectomy alone. Platinum‐based chemotherapy, with or without radiotherapy, is used for advanced, high‐risk or recurrent disease across the three malignancies: as first‐line therapy in advanced ovarian cancer, as concurrent chemo‐radiotherapy in locally advanced cervical cancer and as adjuvant or combined therapy, increasingly with immune checkpoint inhibitors, in high‐risk or advanced endometrial cancer. In these settings, long‐term efficacy remains constrained by intrinsic and acquired resistance. Recurrent ovarian cancer is characterised by the eventual emergence of platinum resistance, which markedly worsens prognosis and restricts therapeutic options [ 3 ]. These limitations support the search for adjunctive strategies that can complement established regimens through distinct mechanisms. Vitamin D exists in two major forms: cholecalciferol (vitamin D 3 ), synthesised cutaneously following ultraviolet B irradiation and ergocalciferol (vitamin D2), derived from dietary sources [ 4 ]. Both precursors undergo hepatic hydroxylation by 25‐hydroxylases to yield 25‐hydroxyvitamin D [25(OH)D], the principal circulating form and standard clinical measure of vitamin D status. Subsequent renal hydroxylation by 1α‐hydroxylase (CYP27B1) generates 1,25‐dihydroxyvitamin D [1,25(OH) 2 D 3 ] or calcitriol, the hormonally active metabolite responsible for most physiological effects of vitamin D. Bioavailability is tightly regulated by 24‐hydroxylase (CYP24A1)‐mediated catabolism, which converts 1,25(OH)2D 3 to less active derivatives such as 1,24,25(OH) 3 D 3 and thereby limits hypercalcaemic toxicity. At the cellular level, 1,25(OH) 2 D₃ acts predominantly through genomic transcriptional mechanisms, although non‐genomic actions are also described. Ligand‐bound VDR heterodimerises with the retinoid X receptor (RXR). The resulting VDR‐RXR complex binds vitamin D response elements (VDREs) within target gene promoters and modulates transcription of genes involved in calcium transport, immune function and cell‐cycle regulation (Figure  1 ). Vitamin D synthesis, VDR‐RXR genomic signalling pathway, and autocrine and systemic biological actions. Created in BioRender. Punchoo, R. (2026) https://BioRender.com/skfx3td . In this review, the compound tested is named explicitly rather than referred to generically as ‘vitamin D’. Cholecalciferol and ergocalciferol are inactive precursors; calcifediol/25(OH)D is the major circulating precursor hormone and standard clinical biomarker of vitamin D reserve status; calcitriol/1,25(OH) 2 D 3 is the hormonally active endogenous VDR ligand; and synthetic analogues such as alfacalcidol and calcipotriol have distinct pharmacological properties. Alfacalcidol is a prodrug that requires hepatic activation, whereas calcipotriol is a low‐calcaemic VDR agonist developed for topical use. These compounds differ in receptor affinity, potency, calcaemic liability and clinical availability and are therefore not treated as pharmacologically interchangeable. Concentrations reported across the cited literature are similarly heterogeneous and are described in three broad tiers. Physiological concentrations refer to endogenous circulating ranges; calcitriol is approximately 75–200 pmol L −1 , whilst serum 25(OH)D concentrations typically range from 25 to 125 nmol L −1 in winter and 50–300 nmol L −1 in summer [ 5 ]. Pharmacological concentrations are those achievable with clinically tolerated dosing without inducing hypercalcaemia. Supraphysiological concentrations exceed levels attainable in patients and predominate in much of the in vitro literature discussed below; these are identified explicitly where relevant. Calcitriol is the high‐affinity ligand of the vitamin D receptor (VDR), a nuclear transcription factor whose genomic footprint spans more than 1000 target genes involved in cellular proliferation, differentiation, immune regulation and apoptosis across a broad range of tissues [ 6 ]. VDR is widely distributed, and several extra‐renal cell types, including gynaecological epithelia, can locally bioactivate vitamin D precursors through autocrine CYP27B1 activity. Calcitriol therefore has paracrine and intracrine functions that are directly relevant to tumour biology and extend beyond its classical endocrine role [ 7 ]. Vitamin D signalling intersects with cancer biology through its regulation of reactive oxygen species (ROS). ROS comprise a chemically heterogeneous group of oxygen‐containing molecules, including superoxide anions (O 2 • − ), hydrogen peroxide (H 2 O 2 ), hydroxyl radicals (•OH) and singlet oxygen ( 1 O 2 ). They are generated continuously as by‐products of aerobic metabolism and enzymatic activity and can also arise from exogenous insults such as ionising radiation, environmental toxins and chemotherapeutic agents [ 8 ]. Although once viewed primarily as mediators of cellular damage, ROS are now recognised as important second messengers in proliferative, inflammatory and stress‐response signalling. The redox imbalance that supports tumour progression may therefore also create a therapeutically exploitable vulnerability. The biological effects of ROS are strongly concentration‐dependent and follow a non‐monotonic hormetic dose–response relationship (Figure  2 ). At sub‐physiological concentrations, insufficient ROS, termed reductive stress, impairs oxidative post‐translational modifications required for normal redox signalling and disrupts processes such as protein folding, enzyme activation and transcription‐factor regulation. At the opposite extreme, supraphysiological ROS accumulation overwhelms cellular antioxidant capacity and produces oxidative stress, characterised by lipid peroxidation, protein carbonylation, nucleic acid oxidation, mitochondrial membrane depolarisation and activation of apoptotic and necrotic cell‐death pathways. Between these extremes lies a hormetic zone in which moderate ROS concentrations function as mitogenic and adaptive signals. These signals can increase antioxidant enzyme activity, including superoxide dismutase and catalase, activate stress‐responsive transcription factors and reinforce mitochondrial quality‐control mechanisms. The threshold at which adaptive signalling transitions to pathological damage is cell‐type‐ and context‐specific and remains an active area of investigation [ 9 ]. A conceptual illustration of the hormetic dose–response relationship between ROS concentration and cellular outcome, showing the reductive‐stress, adaptive and oxidative‐stress zones. Created in BioRender. Punchoo, R. (2026) https://BioRender.com/7fk0bo0 . At low to moderate levels, ROS function as pro‐tumourigenic second messengers that activate oncogenic pathways including NF‐κB, PI3K/Akt and NRF2, thereby supporting cancer‐cell proliferation, survival and drug resistance [ 10 , 11 ]. When ROS accumulation exceeds cellular antioxidant buffering capacity, oxidative damage to mitochondrial membranes, proteins and nucleic acids can instead commit the cell to apoptosis or other oxidative cell‐death pathways [ 12 ]. Transformed cells generally maintain higher basal ROS levels than their normal counterparts. Although this adaptation supports oncogenic signalling, it also narrows the oxidative reserve before cytotoxicity is reached. This elevated steady‐state oxidative burden is therefore a potential therapeutic vulnerability: pro‐oxidant interventions that raise ROS beyond the lethal threshold in cancer cells may achieve selective cytotoxicity without equivalent injury to normal tissue [ 13 ]. Several established anticancer agents, including cisplatin, exploit redox‐dependent mechanisms, providing a rationale for adjunctive strategies that engage the same apoptotic axis through complementary upstream pathways. These observations provide a rationale for evaluating vitamin D in gynaecological oncology. Although experimental and epidemiological evidence increasingly implicates vitamin D signalling in tumour‐cell biology, no comprehensive review has examined mechanistic and translational evidence across ovarian, cervical and endometrial cancers specifically through the lens of ROS‐mediated apoptosis or assessed how the preclinical findings align with clinical evidence. The evidence base is markedly imbalanced: preclinical studies are comparatively extensive, whereas clinical evaluation remains limited to a small number of underpowered or non‐interventional studies. This narrative review therefore synthesises preclinical and clinical evidence on vitamin D and ROS‐mediated apoptosis across the three major gynaecological malignancies, identifies the principal translational limitations and outlines priorities for further clinical development. ROS‐mediated mitochondrial apoptosis is an important anti‐cancer mechanism associated with vitamin D, but it does not operate in isolation. Autophagy, cell‐cycle arrest, paraptosis‐like death, immunomodulation and anti‐inflammatory signalling are also reported as parallel or interacting pathways in the preclinical literature. They are included here where they intersect with or occur upstream of, redox‐dependent cell death. Non‐malignant gynaecological conditions, principally endometriosis and cervical intraepithelial neoplasia, are discussed where they model VDR‐mediated pathways implicated in malignant transformation or represent precursor lesions in which the chemopreventive potential of vitamin D has been tested directly. Evidence from non‐gynaecological cancers and general cancer‐mortality trials (e.g., the Vitamin D and Omega‐3 Trial [VITAL]) is included selectively to provide mechanistic or pharmacological context when gynaecological‐specific evidence is absent or underpowered. Such evidence is not used as direct evidence of efficacy for a specific gynaecological malignancy.

The authors declare no conflicts of interest.