Article Text
Abstract
Objective Triaging patients with presumptive ovarian cancer to the appropriate specialist may improve survival. Therefore, there is increasing interest in complementary diagnostic markers to the standard serum CA125. In patients with pelvic masses, we examined the ability of epidemiologic variables and preoperative differential blood counts to improve detection of ovarian cancer over CA125 alone.
Methods From pathology reports, patients were classified as having: epithelial ovarian cancer (n=743), including fallopian tube and primary peritoneal cancer, non-epithelial ovarian cancers (n=46), non-ovarian cancers (n=122), or benign disease (1,129). From women with epithelial ovarian cancer, we excluded those who received prior neoadjuvant chemotherapy (n=19). Women were also excluded if they did not have a serum CA125 or complete blood count measured within 180 days prior to surgery (n=1099) or did not have both tests within 90 days of each other (n=13). Categorizing patients by menopausal status, we calculated Pearson correlations between differential counts or ratios and CA125, and used t tests to identity univariate predictors of malignancy and stepwise logistic regression and likelihood ratio tests to create models best distinguishing epithelial ovarian cancer from benign disease.
Results 337 women with epithelial ovarian cancer and 365 with benign disease were included in the analysis. Compared with cancers, women with benign disease had lower average: age, 52.5 versus 58.4 years (p<0.0001); serum CA125, 20 versus 239 U/mL (p<0.0001), neutrophil-to-lymphocyte ratio, 2.4 versus 3.5 (p<0.0001); and platelet-to-lymphocyte ratio, 158 versus 222 (p<0.0001); but greater average body mass index, 28.5 versus 26.8 kg/m2 (p=0.004), and lymphocyte-to-monocyte ratio, 5.6 versus 3.9 (p<0.0001). Correlations between counts and ratios and serum CA125 were seen in both epithelial ovarian cancer and benign disease groups and differed by menopausal status. In premenopausal women, a multivariate model including serum CA125, smoking, family history, lymphocytes, and monocytes performed similarly to the model with lymphocyte-to-monocyte ratio replacing counts. In postmenopausal women, a model including body mass index, parity, monocytes, and basophils performed similarly to the model replacing counts with platelet-to-lymphocyte ratio and lymphocyte-to-monocyte ratio. Models including epidemiologic variables and either counts or ratios were better at fitting data than models with serum CA125 and menopausal status alone. A single model applying to all women overstated performance for premenopausal women and understated performance for postmenopausal women.
Conclusions Epidemiologic variables and differential counts or ratios better distinguished between benign and malignant disease when compared with serum CA125 alone using separate models for pre- and postmenopausal women.
- ovarian cancer
- ovarian cysts
Statistics from Altmetric.com
Footnotes
Contributors DWC was responsible for data collection and writing; WJB was responsible for data analysis and writing; AFV was responsible for database construction, data analysis, and writing; RB was responsible for subject recruitment and writing; AKG was responsible for subject recruitment and writing; and UM was responsible for subject recruitment and writing.
Funding Supported by NIH grant No R35CA197605.
Disclaimer The funding source had no role in the design, analysis, writing, or submission of this work.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval The study protocol was approved by the Human Subject Review Boards of Brigham and Women’s Hospital, Dana Farber Cancer Institute, and the Massachusetts General Hospital.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request. To protect sensitive data and participant identities, we will restrict access to the data and strip datasets of any variables that could be used to identify individual participants. In accordance with the NIH policy on Data Sharing, investigators will make any data resulting from these projects available to researchers upon request after it has been submitted for publication. At all times, human subject’s guidelines for protection of participant privacy will be honored.