Predicting circulating CA125 levels among healthy premenopausal women

Naoko Sasamoto, Ana Babic, Bernard A. Rosner, Renee T. Fortner, Allison F. Vitonis, Hidemi Yamamoto, Raina N. Fichorova, Anne Tjønneland, Louise Hansen, Kim Overvad, Marina Kvaskoff, Agnes Fournier, Francesca Romana Mancini, Heiner Boeing, Antonia Trichopoulou, Eleni Peppa, Anna Karakatsani, Domenico Palli, Valeria Pala, Amalia MattielloRosario Tumino, Chiara C. Grasso, N. Charlotte Onland-Moret, Elisabete Weiderpass, J. Ramon Quiros, Leila Lujan-Barroso, Miguel Rodríguez-Barranco, Sandra Colorado-Yohar, Aurelio Barricarte, Miren Dorronsoro, Annika Idahl, Eva Lundin, Hanna Sartor, Kay Tee Khaw, Timothy J. Key, David Muller, Elio Riboli, Marc J. Gunter, Laure Dossus, Rudolf Kaaks, Daniel W. Cramer, Shelley S. Tworoger, Kathryn L. Terry

Research output: Contribution to journalArticlepeer-review


Background: Cancer antigen 125 (CA125) is the most promising ovarian cancer screening biomarker to date. Multiple studies reported CA125 levels vary by personal characteristics, which could inform personalized CA125 thresholds. However, this has not been well described in premenopausal women. Methods: We evaluated predictors of CA125 levels among 815 premenopausal women from the New England Case Control Study (NEC). We developed linear and dichotomous (-35 U/mL) CA125 prediction models and externally validated an abridged model restricting to available predictors among 473 premenopausal women in the European Prospective Investigation into Cancer and Nutrition Study (EPIC). Results: The final linear CA125 prediction model included age, race, tubal ligation, endometriosis, menstrual phase at blood draw, and fibroids, which explained 7% of the total variance of CA125. The correlation between observed and predicted CA125 levels based on the abridged model (including age, race, and menstrual phase at blood draw) had similar correlation coefficients in NEC (r = 0.22) and in EPIC (r= 0.22). The dichotomous CA125 prediction model included age, tubal ligation, endometriosis, prior personal cancer diagnosis, family history of ovarian cancer, number of miscarriages, menstrual phase at blood draw, and smoking status with AUC of 0.83. The abridged dichotomous model (including age, number of miscarriages, menstrual phase at blood draw, and smoking status) showed similar AUCs in NEC (0.73) and in EPIC (0.78). Conclusions: We identified a combination of factors associated with CA125 levels in premenopausal women. Impact: Our model could be valuable in identifying healthy women likely to have elevated CA125 and consequently improve its specificity for ovarian cancer screening.

Original languageEnglish
Pages (from-to)1076-1085
Number of pages10
JournalCancer Epidemiology Biomarkers and Prevention
Issue number6
Publication statusPublished - Jun 1 2019

ASJC Scopus subject areas

  • Epidemiology
  • Oncology


Dive into the research topics of 'Predicting circulating CA125 levels among healthy premenopausal women'. Together they form a unique fingerprint.

Cite this