Role of androgen hormones and 11 oxyandrogen metabolites in endometrial and ovarian cancers

Gjorgoska, Marija (author), Lanišnik Rižner, Tea (mentor)

Background: Endometrial cancer (EC) and ovarian cancer (OC) are the sixth and seventh most diagnosed cancers in women worldwide, with incidence rates rising due to demographic changes. Both cancers mainly affect postmenopausal women, with a median diagnosis age in the early to mid-sixties. EC generally has better survival rates, whereas OC, especially its high-grade serous subtype (HGSOC), is often diagnosed at an advanced stage, resulting in poor survival and frequent chemoresistance. Currently, there is an unmet need for accurate, cost-effective triage tools, less invasive diagnostics, reliable prognostic biomarkers, and new therapeutic targets for both cancers. Steroid hormones have been poorly studied for their role in the pathophysiology of both EC and OC. They are also gaining new attention in cancer research for their role in modulating anti-tumor immunity and chemoresistance. Others and our group have shown that both ECs and HGSOCs express steroid-metabolizing enzymes and receptors, supporting intra-tumoral steroid interconversion that may influence local hormone signaling and thereby impact tumor cell behavior. In this doctoral dissertation, we focused on a specific class of steroid hormones, namely androgens and their 11-oxygenated derivatives, i.e., 11-oxyandrogens. We systematically investigated: (i) the local metabolism of classic androgen and 11-oxyandrogen precursors in well-characterized cell models; (ii) the transcriptomic and metabolomic effects of potent androgens and 11-oxyandrogens on cell models, along with the prognostic relevance of steroid-metabolizing enzymes and steroid receptors in tumor tissues; and (iii) the diagnostic potential of these steroids in distinguishing EC and OC from benign uterine conditions and non-malignant adnexal masses, respectively. Methods: We used four EC and six HGSOC cell lines, along with immortalized control cell lines derived from normal endometrial and ovarian surface epithelium. To investigate local androgen metabolism, we performed quantitative gene expression analysis of key enzymes in androgen and 11-oxyandrogen metabolism and incubated cell lines with physiologically relevant concentrations of classical androgen precursors (dehydroepiandrosterone sulphate (DHEAS), dehydroepiandrosterone sulphate (DHEAS), dehydroepiandrosterone (DHEA) and androstenedione (A4)) as well as 11-oxyandrogen precursors (11ß-hydroxyandrostenedione (11OHA4) and 11-ketoandrostenedione (11KA4)). The resulting steroid metabolites were measured using liquid chromatography-tandem mass spectrometry (LC-MS/MS). Quantitative gene expression analysis of key enzymes involved in classic and 11-oxyandrogen metabolism was performed using qPCR, and the metabolic profiles were interpreted in the context of enzyme expression levels. To assess the functional effects of androgens, selected EC and HGSOC cell lines were treated with potent classic and 11-oxyandrogens, followed by untargeted transcriptomic analysis via RNA sequencing and metabolomic profiling using nuclear magnetic resonance (NMR). Additionally, publicly available clinical datasets were analyzed to evaluate the expression of key steroid-metabolizing enzymes and steroid receptors in EC and HGSOC tumors, along with their associations with patient survival. Finally, two prospective clinical studies were conducted to evaluate the diagnostic potential of circulating steroids in EC and OC. Serum steroid levels were quantified by LC-MS/MS, and diagnostic accuracy was assessed using machine learning. Results: (i) In both EC and HGSOC cell models, classic androgen precursors did not support local production of 11-oxyandrogens, attributable to the observed absence of CYP11B1 expression across all cell lines. Furthermore, DHEAS and DHEA served as limited sources of bioactive androgens, testosterone (T) and 5?-dihydrotestosterone (DHT), primarily due to low HSD3B1/2 expression, likely diverting DHEA toward alternative metabolites such as 5?-androstenediol via reductive HSD17B enzymes or oxygenated derivatives through CYP3B isoforms. Among EC cell lines, the low-grade model RL95-2 exhibited the highest capacity to metabolize DHEAS and DHEA into T. In HGSOC, the primary tumor cell line Caov-3 and the non-cancerous control line HIO-80 showed the greatest T production from these precursors. Similarly, A4 was a poor substrate for bioactive androgen synthesis in both cancer types, suggesting that it is preferentially converted to 5?-reduced metabolites via SRD5A isoforms, which were expressed by both EC and HGSOC cells. In contrast, 11-oxyandrogen precursors, 11OHA4 and 11KA4, were efficiently converted into the potent androgen receptor (AR) agonist 11-ketotestosterone (11KT) in selected cell lines. This conversion, catalyzed by the AKR1C3 enzyme, was most pronounced in low-grade EC models (Ishikawa, HEC-1-A, RL95-2) and chemo-sensitive HGSOC lines (OVSAHO, Kuramochi). Notably, the amount of 11KT formed from 11-oxyandrogen precursors was several-fold higher than T produced from classic precursors, underscoring intra-tumoral 11-oxyandrogen metabolism as significant sources of androgen receptor (AR)-activating metabolites in low-grade ECs and chemo-sensitive HGSOCs. (ii) In EC cell models, treatment with DHT and 11-keto-dihydrotestosterone (11KDHT) induced modest transcriptomic changes after 48 hours of incubation. In AR-expressing Ishikawa cells, androgen exposure upregulated genes such as MYO1D and LAMC3, whereas no significant transcriptomic effects were observed in RL95-2 cells, which express low AR levels. In contrast, the AR-positive HGSOC model OVSAHO exhibited pronounced transcriptomic responses to potent androgens (T, DHT) and 11-oxyandrogens (11KT, 11KDHT) following 72 hours of incubation. These responses included activation of stress response and cell cycle-related pathways. At the metabolomic level, both EC and HGSOC cell models exhibited mild intracellular alterations following 72 hours of incubation with potent androgens and 11-oxyandrogens. Specifically, in the low-grade EC model Ishikawa, DHT and 11KDHT reduced intracellular levels of amino acids and lactic acid, while in the HGSOC model OVSAHO, 11KDHT lowered intracellular amino acid and glutathione levels. Analysis of tumor tissues revealed that expression patterns of AR and steroid-metabolizing enzymes varied by tumor type and clinical characteristics. In EC, higher intra-tumoral expression of HSD11B2, SRD5A2, and AR was associated with improved survival, whereas elevated SRD5A1 expression correlated with poorer outcomes. Similarly, in HGSOC, increased expression of HSD11B2, HSD17B2, and AR was linked to better survival, while higher levels of PAPSS1, H6PD, and HSD17B4 were associated with worse prognosis. (iii) In patients with EC (n = 62), preoperative serum levels of classic androgens (A4, T), 11-oxyandrogens (11OHA4, 11OHT), and glucocorticoids (17?-hydroxyprogesterone, 11-deoxycortisol) were elevated compared to controls with benign uterine conditions (n = 70). Conversely, patients with primary or recurrent OC (n = 43) showed reduced preoperative serum levels of T and 11-oxyandrogens (11OHT, 11KT) relative to controls with non-malignant adnexal masses (n = 56). Individually, steroid hormones had limited diagnostic accuracy for both EC and OC (AUC < 0.7). However, in EC, a logistic regression model combining the androgen pool (DHEA, A4, T), tumor biomarkers (CA125, HE4), and clinical parameters (BMI, parity) achieved an AUC of 0.87 (sensitivity 74.7%, specificity 79.1%), significantly outperforming CA125, HE4, or their combination alone in differentiating EC from benign uterine conditions. For OC, two models integrating 11-oxyandrogens, age, and either CA125 or HE4 reached an AUC of 0.91, with sensitivities of 88.9% and 94.4% and specificities of 82.0% and 77.3%, respectively, surpassing CA125, HE4, and the ROMA index in differentiating OC from non-malignant adnexal masses. These findings emphasize the added diagnostic value of steroid hormone profiling in the preoperative evaluation of gynecologic tumors, however validation in independent cohorts is needed. Conclusions: This doctoral dissertation provides new insights into (i) local steroid metabolism in EC and HGSOC cell models; (ii) transcriptomic and metabolomic effects of androgen signaling in EC and HGSOC cell models; and (iii) diagnostic potential of circulating steroids in distinguishing EC and OC from their respective control counterparts. We identified distinct patterns of 11-oxyandrogen metabolism across tumor types and clinical subgroups, between low- and high-grade EC, between EC tumors and normal endometrium, and between chemo-sensitive and chemo-resistant HGSOC models, as well as between HGSOC tumors and normal ovarian epithelium. Functionally, androgen and 11-oxyandrogen exposure triggered significant transcriptomic changes in AR-expressing HGSOC cells, particularly in stress response and cell cycle pathways. In tumor tissues, we identified prognostic biomarker candidates, including steroid-metabolizing enzymes and AR, that could improve risk stratification and support personalized clinical management. Importantly, we also uncovered, for the first time, systemic alterations in circulating 11-oxyandrogens in patients with EC and OC compared to their respective control groups. By integrating steroid profiles with traditional tumor markers and clinical parameters we developed novel diagnostic models for EC and OC, which significantly outperformed tumor biomarkers CA-125 and HE4. To our knowledge, these are the first steroid-protein-based models for these cancers. If validated in larger, independent cohorts, they hold promise to improve cancer detection and ultimately improve clinical outcomes.

• Language: English
• Keywords: Endometrial cancer, Ovarian cancer, HGSOC, Steroid hormones, Intracrinology, Biomarker discovery, Liquid chromatography-tandem mass spectrometry, Transcriptomics, Metabolomics, Machine learning
• Work type: Doctoral dissertation
• Organization: MF - Faculty of Medicine
• Year: 2025
• PID: 20.500.12556/RUL-176297
• Publication date in RUL: 27.11.2025

Secondary language

• Language: Slovenian
• Title: Vloga androgenih hormonov in 11 oksiandrogenih metabolitov pri raku endometrija in raku jajčnikov

Abstract

Ozadje: Rak endometrija (EC) in rak jajčnikov (OC) sta šesti oziroma sedmi najpogosteje diagnosticirani obliki raka pri ženskah po svetu, pri čemer incidenca narašča zaradi demografskih sprememb. Oba raka najpogosteje prizadeneta ženske po menopavzi, v zgodnjih do srednjih šestdesetih letih. EC praviloma omogoča boljše preživetje, medtem ko OC, zlasti serozni karcinom visokega gradusa (HGSOC), pogosto diagnosticirajo v napredovalem stadiju, kar vodi v slabši izid in pogosto kemorezistenco. Trenutno obstaja velika potreba po natančnejših, cenovno sprejemljivih orodjih triaže, zgodnjih in manj invazivnih diagnostičnih metodah, zanesljivih prognostičnih označevalcih ter novih terapevtskih tarčah za obe vrsti raka. Vloga steroidnih hormonov v patofiziologiji EC in OC je bila doslej slabo raziskana, vendar steroidni hormoni pridobivajo vse več pozornosti zaradi vpliva na protitumorsko imunost in kemorezistenco. Naša in druge raziskovalne skupine so pokazale, da tako EC kot HGSOC izražata gene encimov, vključenih v sintezo in metabolizem steroidov, in njihove receptorje, kar omogoča sintezo in delovanje v tumorskih celicah, ki lahko vplivata na lokalni prenos signala hormonov in posledično na signalne poti in odziv tumorskih celic. V doktorski disertaciji smo se osredotočili na skupino steroidnih hormonov – androgene in njihove 11-oksigenirane derivate, tj. 11-oksiandrogene. Sistematično smo preučili: (i) lokalni metabolizem prekurzorjev androgenov in 11-oksiandrogenov v dobro okarakteriziranih celičnih modelih; (ii) transkriptomske in metabolomske učinke aktivnih androgenov in 11-oksiandrogenov v celičnih modelih, skupaj z napovedno vrednostjo encimov, vključenih v metabolizem steroidov, in steroidnih receptorjev v tumorskem tkivu; ter (iii) diagnostični potencial teh steroidov pri razlikovanju EC in OC od benignih patologij endometrija in jajčnikov. Metode: Uporabili smo štiri celične linije EC in šest celičnih linij HGSOC ter kontrolne linije, pridobljene iz normalnega endometrijskega epitela in površinskega epitela jajčnikov. Za preučevanje lokalnega metabolizma androgenov smo izvedli kvantitativno analizo izraženosti genov ključnih encimov, vključenih v metabolizem klasičnih androgenov, in 11-oksiandrogenov, ter inkubirali celične linije s fiziološko relevantnimi koncentracijami prekurzorjev androgenov (dehidroepiandrosteron sulfat (DHEAS), dehidroepiandrosteron (DHEA) in androstendion (A4)) ter prekurzorjev 11-oksiandrogenov (11β-hidroksiandrostendion (11OHA4) in 11-ketoandrostendion (11KA4)). Nastale steroide smo identificirali in kvantificirali s tekočinsko kromatografijo sklopljeno z masno spektrometrijo (LC-MS/MS). Za funkcionalno oceno učinkov androgenov smo izbrane celične linije EC in HGSOC inkubirali z biološko aktivnimi klasičnimi in 11-oksiandrogeni, nato pa izvedli netarčno transkriptomsko analizo (RNA sekvenciranje) ter metabolomsko analizo z jedrsko magnetno resonanco (NMR). Poleg tega smo analizirali javno dostopne baze podatkov, da bi ocenili izražanje genov ključnih encimov, vključenih v metabolizem steroidov, in steroidnih receptorjev v tumorjih EC in HGSOC ter njihovo povezavo s preživetjem bolnic. Na koncu smo izvedli dve prospektivni klinični študiji za oceno diagnostičnega potenciala sistemskih steroidov pri EC in OC. Ravni serumskih steroidov smo določili z LC-MS/MS, njihov diagnostični potencial pa ocenili z metodami strojnega učenja. Rezultati: (i) Pri obeh vrstah raka, tako EC kot HGSOC, prekurzorji androgenov niso omogočali tvorbe 11-oksiandrogenov zaradi odsotnosti izražanja gena CYP11B1 v vseh celičnih linijah. Poleg tega sta bila DHEAS in DHEA omejena vira bioaktivnih androgenov, testosterona (T) in 5α-dihidrotestosterona (DHT), in sicer predvsem zaradi nizke izraženosti genov HSD3B1/2, kar je najverjetneje povzročilo pretvorbo DHEA v alternativne metabolite, kot je 5α-androstendiol (prek reduktivnih HSD17B encimov), ali oksigenirane derivate (prek CYP3B izooblik). Med celičnimi linijami je model EC nizkega gradusa RL95-2 izkazoval najvišjo sposobnost pretvorbe DHEAS in DHEA v T. Pri HGSOC sta celična linija primarnega HGSOC Caov-3 in kontrolna linija HIO-80 tvorili največ T iz teh prekurzorjev. Tudi A4 je bil slab substrat za sintezo bioaktivnih androgenov pri obeh vrstah raka, kar nakazuje, da se verjetno pretvarja v alternativne metabolite, vključno s 5α-reduciranimi metaboliti preko izooblik SRD5A, ki so jih izražale tako celice EC kot HGSOC. Nasprotno so se prekurzorji 11-oksiandrogenov (11OHA4 in 11KA4) v izbranih celičnih linijah učinkovito pretvorili v 11-ketotestosteron (11KT), ki je pomemben agonist receptorja za androgene (AR). Ta pretvorba je bila najizrazitejša v modelih EC nizkega gradusa (Ishikawa, HEC-1-A, RL95-2) in kemosenzitivnih linijah HGSOC (OVSAHO, Kuramochi). Pomembno je, da je bila količina 11KT, ki je nastal iz prekurzorjev 11-oksiandrogenov, pri modelih EC nizkega gradusa in kemosenzitivnega HGSOC večkrat višja kot količina T iz klasičnih prekurzorjev, kar poudarja pomen sinteze 11-oksiandrogenov v tkivu tumorja kot ključnega vira agonistov AR v teh tumorskih modelih. (ii) V celičnih modelih EC je 48-urna inkubacija z DHT in 11-keto-DHT (11KDHT) na ravni transkriptoma povzročila zmerne spremembe. V celicah Ishikawa, ki izražajo AR, je izpostavljenost androgenom povečala izražanje genov, kot je MYO1D, medtem ko pri RL95-2 celicah z nizko izraženim AR niso bile zaznane statistično značilne spremembe. Model HGSOC OVSAHO, ki prav tako izraža AR, je po 72-urni inkubaciji na ravni transkriptoma izkazal izrazite odzive na androgene (T, DHT) in 11-oksiandrogene (11KT, 11KDHT), vključno z aktivacijo poti, vpletenih v odziv na stres in uravnavanje celičnega cikla. Na metabolomski ravni sta oba modela EC in HGSOC pokazala manjše spremembe po inkubaciji z androgeni. V modelu EC nizkega gradusa Ishikawa sta DHT in 11KDHT zmanjšala raven aminokislin in mlečne kisline, medtem ko je pri modelu HGSOC OVSAHO 11KDHT znižal raven aminokislin in glutationa. Analiza tumorskih tkiv je pokazala, da se izražanje gena AR in genov, ki kodirajo encime, vključenih v metabolizem steroidov, razlikuje glede na tip tumorja in klinične značilnosti. Pri EC je bila višja izraženost genov HSD11B2, SRD5A2 in AR povezana z boljšim preživetjem, medtem ko je bila višja izraženost gena SRD5A1 povezana s slabšim izidom. Podobno je bilo pri HGSOC večje izražanje genov HSD11B2, HSD17B2 in AR povezano z boljšim preživetjem, medtem ko je bila višja izraženost genov PAPSS1, H6PD in HSD17B4 povezana s slabšo prognozo. (iii) Pri bolnicah z EC (n = 62) so bile predoperativne ravni klasičnih androgenov (A4, T), 11-oksiandrogenov (11OHA4, 11OHT) in glukokortikoidov (17α-hidroksiprogesteron, 11-deoksikortizol) povišane v primerjavi z bolnicami z benignimi boleznimi maternice (n = 70). Nasprotno pa so imele bolnice s primarnim ali ponavljajočim se OC (n = 43) znižane predoperativne serumske ravni T in 11-oksiandrogenov (11OHT, 11KT) v primerjavi z bolnicami z benignimi adneksalnimi masami (n = 56). Posamično so imeli steroidni hormoni omejeno diagnostično moč za EC in OC (AUC < 0,7). Vendar pa je za EC logistični regresijski model, ki je združeval androgene (DHEA, A4, T), tumorske biooznačevalce (CA-125, HE4) in klinične parametre (indeks telesne mase (ITM), pariteto), dosegel AUC 0,87 (občutljivost 74,7 %, specifičnost 79,1 %) in s tem pomembno presegel diagnostično moč CA-125, HE4 ali njune kombinacije. Za OC sta dva modela, ki sta vključevala 11-oksiandrogene, starost in bodisi CA-125 bodisi HE4, dosegla AUC 0,91, z občutljivostmi 88,9 % in 94,4 % ter specifičnostmi 82,0 % in 77,3 %, kar je preseglo diagnostično učinkovitost CA-125, HE4 in ROMA indeksa pri razlikovanju OC od benignih adneksalnih mas. Naši rezultati poudarjajo dodatno diagnostično vrednost profiliranja steroidnih hormonov pri predoperativni oceni ginekoloških tumorjev, vendar je potrebna validacija na neodvisnih kohortah. Zaključki: Doktorska disertacija prinaša nova spoznanja o (i) lokalnem metabolizmu steroidov v celičnih modelih EC in HGSOC; (ii) vplivu androgenov na transkriptom in metabolom; ter (iii) diagnostičnem potencialu sistemskih steroidov pri razlikovanju EC in OC od kontrolnih skupin. Ugotovili smo različne vzorce metabolizma 11-oksiandrogenov med modeli EC nizkega in visokega gradusa, med modeli EC in normalnim endometrijskim epitelijem, med kemosenzitivnimi in kemorezistentnimi modeli HGSOC ter med modeli HGSOC in normalnim ovarijskim epitelijem. Funkcionalno je izpostavljenost androgenom in 11-oksiandrogenom sprožila pomembne spremembe na ravni transkriptoma v AR-pozitivnem modelu HGSOC, zlasti v signalnih poteh odziva na stres in celičnega cikla. V tumorskem tkivu smo identificirali kandidate za prognostične označevalce, vključno z encimi za sintezo in metabolizem steroidov, ter AR, ki bi lahko podprli personalizirano klinično obravnavo. Pomembno je, da smo prvi ugotovili sistemske spremembe v 11-oksiandrogenih pri bolnicah z EC in OC v primerjavi z bolnicami z benignimi patologijami. Z integracijo steroidnih profilov s tradicionalnimi tumorskimi biooznačevalci in kliničnimi parametri smo razvili nove modele za neinvazivno diagnostiko EC in OC, ki pomembno presegajo trenutno uporabljene serumske biooznačevalce, CA-125 in HE4. Če bodo validirani v večjih, neodvisnih kohortah, imajo ti modeli potencial za izboljšano in neinvazivno odkrivanje teh rakov ter posledično izboljšanje kliničnih izidov.

• Keywords: Rak endometrija, rak jajčnikov, HGSOC, steroidni hormoni, intrakrinologija, tekočinska kromatografija sklopljena z masno spektrometrijo, transkriptomika, metabolomika, strojno učenje