Risk prediction for development of benign ovarian tumors in postmenopause

Full Text

Introduction

Ovarian neoplasms represent a common pathology in women of late reproductive and postmenopausal age. According to literature data, up to 25 % of all female reproductive system neoplasms are ovarian tumors, with 45 % of ovarian tumors being more frequently detected in peri- and postmenopausal women [1; 2]. Given the increasing life expectancy, high prevalence of benign ovarian tumors (BOT) in postmenopause, and low adherence to gynecologist visits in this age group, risk factor stratification for BOT development in these patients is particularly relevant.

Current scientific literature actively investigates the etiology and pathogenesis of BOT in postmenopause, debating the roles of environmental factors, age, obstetric-gynecological and somatic history, metabolic processes, micronutrient/macronutrient deficiencies, hormonal imbalances, etc. [2; 3]. Researchers are evaluating various imaging modalities for uterine appendage tumor prognosis: ultrasound (utilizing IOTA criteria – International Ovarian Tumor Analysis; O-RADS – Ovarian-Adnexal Reporting and Data System; ADNEX model – Assessment of Different Neoplasias in the Adnex), computed tomography, and magnetic resonance imaging [3–5]. Serological markers are being extensively studied for ovarian tumor prediction, including cancer marker 125 (Cancer Antigen 125, CA-125), insulin-like growth factor 1 (IGF-1), leptin, zinc, adipokines, etc. Literature data indicate that patient age, low vitamin D levels, elevated leptin, IGF-1, and sex hormone-binding globulin (SHBG), along with decreased blood zinc levels, may contribute to various reproductive system neoplasms, including ovarian tumors [6–10]. However, despite widespread implementation of comprehensive ovarian tumor screening programs, prediction remains an unsolved challenge – no single diagnostic marker used in isolation can effectively predict BOT development risk.

The objective of the study was to develop a mathematical model for predicting the risk of benign ovarian tumors (BOT) in postmenopause.

Materials and Methods

A prospective study was conducted involving patients admitted to the gynecological department of Perm Regional Clinical Hospital. The patients were divided into two groups: the main group consisted of postmenopausal women with BOT (n = 40), and the comparison group included postmenopausal women without masses of the uterine appendages (n = 20). Clinical and anamnestic data were analyzed, and laboratory tests were performed to measure levels of vitamin D, insulin-like growth factor 1 (IGF-1), leptin, zinc, estradiol, testosterone, sex hormone-binding globulin (SHBG), and cancer marker 125 (CA-125). Results of instrumental examinations (pelvic ultrasound), surgical treatment protocols, and histological reports were evaluated. Data were processed using Microsoft Excel (2010), StatSoft Statistica 6.0 (StatSoft, USA), and SPSS Statistics, version 22.0 (IBM Microsoft, USA). Statistical hypothesis testing was performed using the chi-square test (c²), p-values, odds ratio (OR) with 95 % confidence interval (CI), and risk ratio (RR) with 95 % CI. Intergroup comparisons of quantitative variables were conducted using the Mann–Whitney U-test. ROC curve analysis was employed to assess the relationship between sensitivity and specificity of the studied parameters.

Results and Discussion

The distribution of patients in the main group by duration of postmenopause according to the STRAW +10 (Stages of Reproductive Aging Workshop) scale was as follows: 77.5 % (n = 31) of patients were at STRAW +2 stage (c² = 47.4; p < 0.001), 20 % (n = 8) at STRAW +1c stage (c² = 6.8; p = 0.009), and 2.5 % (n = 1) at STRAW +1b stage (c² = 0; p = 1.0). The comparison group’s postmenopausal age characteristics were comparable to those of the main group. Thus, BOT were more frequently detected in patients with late postmenopause. Notably, the majority of postmenopausal women with BOT had no history of menstrual dysfunction (92.7 %; p < 0.001), and over 90 % had fulfilled their reproductive desires (p < 0.001).

Analysis of laboratory results demonstrated that CA-125 was elevated in only one patient from the main group with histologically verified endometrioma; in all other cases, CA-125 levels were within reference ranges in both study groups. In the main group, estradiol levels were normal in 92.5 % (p< 0.001), with a mean value of 15.5 pg/mL. FSH levels exceeded the age-specific reference range in 17.5 % of BOT patients (p=0.018), averaging 57.5 mIU/mL. SHBG levels were elevated in 5 % of the main group (mean 61.1 nmol/L). Normal zinc levels were recorded in 85 % of patients (p< 0.001), with a mean of 13.2 μmol/L. Leptin levels did not exceed the norm in 75 % of the main group (p< 0.001), averaging 16.3 ng/mL. Vitamin D deficiency was detected in 77.5 % of BOT patients (mean 22.1 ng/mL). In the comparison group, laboratory markers showed: vitamin D deficiency in 66.7 % (p< 0.001; mean 36.4 ng/mL); elevated FSH in 60 % (p=0.001; mean 45.4 mIU/mL); normal estradiol in 80% (p< 0.001; mean 11 pg/mL); elevated leptin in 53.3 % (mean 9 ng/mL); SHBG above reference values in 20 % (mean 96 nmol/L); and normal zinc levels in 93 % (mean 15.7 μmol/L). Thus, the study revealed that isolated assessment of the studied laboratory markers lacks statistical significance for determining BOT development risk in postmenopause.

The study focused on predicting BOT development in postmenopausal patients through analysis of a representative sample. The research objective was formulated as follows: to identify combinations of independent variables that optimally predict the dependent variable. To test the hypothesis, integrated ROC curves were constructed for these parameters. A combined model was developed to improve performance, determining BOT presence in postmenopausal women based on the studied variables (Table). The model achieved 92.59 % accuracy (sensitivity 95.0 %, specificity 85.71 %). The area under the ROC curve was 0.95, demonstrating high model efficacy despite the small sample size. Model coefficients were derived, followed by analysis of sensitivity and specificity for the resulting multiparametric diagnostic index (p).

Characteristics of diagnostic parameter quality

The mathematical model for predicting the risk of benign ovarian tumors (BOT) in postmenopause incorporates eight parameters:

where: p = risk of benign ovarian tumor (BOT) development in a woman; e = base of natural logarithm; x₁ = age (years); x₂ = CA-125 level (U/mL); x₃ = vitamin D level (ng/mL); x₄ = SHBG level (nmol/L); x₅ = IGF-1 level (ng/mL); x₆ = FSH level (mIU/mL); x₇ = estradiol level (pg/mL); x₈ = zinc level (μmol/L).

A p-value of 0.55 or higher predicts high risk of benign ovarian tumor (BOT) development in postmenopausal women, while p < 0.55 indicates low risk of BOT in postmenopausal patients.

Thus, the efficacy of the integrated index calculated using Formula reached 92.59 % (sensitivity 95.0 %, specificity 85.71 %), outperforming individual parameters. The area under the integrated ROC curve was 0.95, confirming the high diagnostic quality of the composite index (Figure).

Fig. ROC curve for the model identifying benign ovarian tumors (BOT) based on patient age and the integrated index: for Formula (1) AUC = 0.95; for the “age” parameter AUC = 0.69

Thus, the area under the ROC curve (AUC) of 0.95 demonstrates the high diagnostic quality of the integrated index and characterizes the developed model as a highly effective method for predicting benign ovarian tumor (BOT) risk in postmenopause.

Conclusions

The mathematical model for predicting benign ovarian tumor (BOT) development in postmenopause, developed through this study, demonstrated improved efficacy in risk prognosis for postmenopausal patients. The results confirm the relevance of a multifactorial approach–assessing factor combinations–as no single marker in isolation provides sufficient screening sensitivity for postmenopausal BOT risk. Based on the prognosis, high-risk patients can be identified and actively monitored in outpatient settings.

About the authors

Yu. A. Shashurina

Author for correspondence.
Email: jusya15@yandex.ru

Degree Candidate of the Department of Obstetrics and Gynecology no. 1, Obstetrician-gynecologist of the Gynecology Department

References

Supplementary files

Supplementary Files

Action

1. JATS XML

Download

2. Fig. ROC curve for the model identifying benign ovarian tumors (BOT) based on patient age and the integrated index: for Formula (1) AUC = 0.95; for the “age” parameter AUC = 0.69

Download (122KB)

Indexing metadata