Chronic Medical Conditions and CA125 Levels among Women without Ovarian Cancer

An estimated 22,240 new cases and 14,070 deaths from ovarian cancer occur annually in the United States (1), making it the third most common gynecologic cancer in the United States and the leading cause of deaths associated with cancers of the female genital tract. This high mortality largely reflects its tendency to be diagnosed at later stages, which is associated with poorer survival (2–4). Development of a reliable screening biomarker would allow earlier detection and intervention, resulting in reduced morbidity and mortality.

Cancer antigen (CA) 125 is a high-molecular weight glycoprotein and is recognized by the mAb OC125 (5). CA125 is used clinically to monitor cancer progression and is a prime candidate for a screening biomarker (6). However, randomized screening trials of CA125 alone, or in combination with transvaginal ultrasound, have not provided clear mortality benefit (4, 7). In addition to searching for more sensitive and specific biomarkers, ovarian cancer screening might be improved by taking into consideration cofactors that influence CA125 levels independent of ovarian cancer.

Because MUC16 (the gene that encodes CA125) is expressed in a variety of tissues, including those lining the gastrointestinal, respiratory, reproductive tracts, and body cavities, CA125 levels can vary with a range of exposures and conditions affecting those tissues. These include hysterectomy, parity, oral contraceptive or menopausal hormone therapy use, and history of endometriosis, as well as demographic factors such as race, age, body mass index (BMI; refs. 8–14). Medical conditions associated with ascites such as heart failure, tuberculosis, and chronic liver disease may also raise CA125 to levels similar to those detected in women affected by ovarian cancer (11, 15–22). In this study, we examined a list of common medical conditions self-reported by women without ovarian cancer to determine associations with CA125 levels.

This analysis was based on women without ovarian cancer (controls) who participated in the New England Case Control (NEC) study, a population-based study that enrolled newly diagnosed ovarian cancer cases and controls from eastern Massachusetts and New Hampshire between 1992 and 2008. A detailed account of the study design has been described previously (23). Briefly, the controls were identified through random digit dialing, driver's license lists, and town-resident lists. A total of 2,100 controls (54%) were eligible and enrolled. Nearly all participants (>95%) provided blood specimens at the time of enrollment, and this study included 2,004 women who had a banked blood sample available for CA125 measurement.

The venous blood samples were collected by trained phlebotomists during or shortly after the in-person interview in the participants’ homes or other convenient locations. The samples were transported to the processing laboratory, usually within 24 hours, where samples were centrifuged and separated into serum, plasma, red blood cells, and buffy coat. Samples were stored at −80°C. CA125 was measured on plasma samples of 100 μL each using the CA125II assay at the Clinical and Epidemiology Research Laboratory (CERLab) at Boston Children's Hospital. The reportable range of the assay was 0.6 to 500 IU/L, with a normal reference interval for females of less than 35 U/mL. We assessed the reproducibility of the assay by including five blinded aliquots of quality control pool in each of the 46 batches. The between batch coefficient of variation (CV) was 1%.

All control participants completed an in-person interview. We asked about sociodemographic variables, reproductive and family history, detailed medical history including disease diagnoses, hospitalization, and current medication use. We selected medical conditions based on their known or hypothesized relationship with ovarian cancer or CA125. In addition, we only included medical conditions with at least 5% prevalence. The questions were asked in the following format: “Did you have any of the following conditions or surgeries before (a reference date of one year before study enrollment)?” and “before (the reference date), did you have any other medical conditions that lasted 6 months or longer, or any condition you understand to have a genetic basis?” Coronary artery disease was defined as myocardial infarction and or angina. Inflammatory bowel disease included ulcerative colitis and Crohn's disease.

Thyroid disorders were divided into underactive thyroid (hypothyroidism, Hashimoto's thyroiditis, or thyroidectomy), overactive thyroid (hyperthyroidism, Grave's disease), others which included thyroid cysts, unknown over or under thyroid, thyroid nodule, and thyroid cancer. Gall bladder problems include cholecystitis, cholelithiasis/gall stones, and gall bladder surgery. Psychologic disorders included anxiety and depression. Obesity was defined as BMI ≥ 30 kg/m². Other medical conditions considered include allergies, appendectomy, asthma/bronchitis, bone fracture, non-ovarian cancers, colon polyps, endometriosis, gastroesophageal reflux/ulcers, headaches/migraines, hypertension, hypercholesterolemia/atherosclerosis, hysterectomy, osteoarthritis, osteoporosis, and uterine fibroids.

Data on covariates known to influence CA125, including age, race, menopausal status/age at menopause, BMI (kg/m²), oral contraceptives use, hysterectomy, hormone replacement therapy, smoking status, and parity (8, 13, 14), were also collected during the interview. Women were classified as premenopausal if they reported that their periods were still occurring, with or without birth control or menstrual-regulatory hormones. Women were classified as postmenopausal if they reported their periods had stopped or were only occurring because of menopausal hormones. Women who reported no periods because of a hysterectomy or a medical condition/treatment were classified as premenopausal if they were age <50 and postmenopausal if age ≥50 years. The study was conducted in accordance with the guiding principles of human research by the Declaration of Helsinki and the U.S. Common Rule. The ethical principles of respect for persons, beneficence, and justice were observed. Informed written consent was obtained from all participants before recruitment into the study.

The study was approved by the institutional review boards of Brigham and Women's Hospital (Boston, MA) and Dartmouth College (Hanover, NH).

We identified two outliers of CA125 defined as greater than two SDs from the mean, but those observations were included in the analysis given the accuracy of the assay. CA125 was log-transformed to achieve normal distribution. We assessed geometric mean levels of CA125 associated with medical conditions according to the woman's menopausal status (pre- vs. postmenopausal) at the time of her blood sampling, as differences in CA125 by menopausal status are well established (13).

All models were adjusted for age, race, parity, and smoking. Models conducted in postmenopausal women were additionally adjusted for hormone replacement therapy (HRT). All statistical analyses were performed using SAS 9.2 (SAS Institute).

The average age of participants was 53 years, and 60% were over age 50 (Table 1). Most participants were white (97%), 14% were current smokers, and 39% former smokers. The most common medical conditions in this study population were anxiety/depression (24%), history of bone fracture (39%), hypercholesterolemia (15%), hypertension (18%), and obesity (18%).

Among the premenopausal women (Table 2), CA125 was lower in those with a history of colon polyps (−31%; P = 0.06) and in those with hysterectomy (−23%; P = 0.01). CA125 levels were higher in women with endometriosis (14%, P = 0.05) and coronary artery disease (221%, P < 0.01), although the later was based on a small number of exposed women (n = 2). Among postmenopausal women, CA125 was higher in women with a history of breast cancer (12%; P = 0.05) and inflammatory bowel disease (14%; P = 0.04). CA125 levels were significantly lower in women reporting osteoporosis (−12%, P = 0.03), hypercholesterolemia/atherosclerosis (−7%, P = 0.02), and osteoarthritis (−9%, P = 0.01). Results were similar when the two outliers were excluded.

We analyzed the association between the number of medical conditions and the percent change in CA125 levels (Supplementary Table S1) and observed no consistent patterns of change in CA125 levels with the number of medical conditions in pre- or postmenopausal women. In addition, we used a multivariable model to simultaneously adjust for all medical conditions that were associated with CA125 in the univariate analyses (Table 3). In premenopausal women, all conditions that were associated with a percent change in CA125 when assessed singly in univariate models (colon polyps, coronary artery disease, endometriosis, and hysterectomy) remained associated after mutual adjustment. With the exception of endometriosis, for which the percent increase was attenuated, the percent change in CA125 after adjustment for multiple medical conditions was similar to that seen in the univariate models. In postmenopausal women, only osteoporosis remained significantly associated with CA125 after adjustment for medical conditions, and the percent decrease in CA125 was similar to that observed for osteoporosis in the univariate model.

We additionally assessed specific medical conditions singly with adjustment for the medication used to treat that condition. For each medical condition, these models produced results similar to those seen in the univariate models (Supplementary Table S2). For example, in postmenopausal women, we assessed medical conditions in relation to CA125 after additional adjustment for medications such as statins for hypercholesterolemia, bisphosphonates for osteoporosis, aminosalicylates (sulfasalazine) and aspirin for inflammatory bowel disease and nonsteroidal anti-inflammatory drugs (NSAIDs) for osteoarthritis; adjusting for these medications generally did not affect the percent of change in CA125 for any condition. We did note, however, an attenuation in the percent of increase associated with breast cancer when adjusted for tamoxifen and arimidex. We also assessed the association between colon polyps and percent change in CA125 among postmenopausal with and without HRT use and found no difference in the association (Supplementary Table S3).

In this study, we assessed the association between several commonly diagnosed medical conditions and CA125 levels in women without an ovarian malignancy. This is the first study to address in an extensive fashion the role of medical conditions on CA125 in the general population using a single assay. Our observations suggest that pathologic processes in several organ systems including genitourinary, bone, breast, cardiovascular, and gastrointestinal could affect CA125 levels in women with no evidence of ovarian cancer. In addition, the importance of these systems in influencing CA125 levels varied between premenopausal and postmenopausal women.

We observed increased levels of CA125 in women with a history of coronary artery disease (CAD) with significant association being limited to premenopausal women. However, this observation was based on a small number of cases. Postmenopausal women with CAD also had higher CA125 levels although the difference was not statistically significant. These observations are consistent with previous reports (24, 25). Previous studies suggest pericardial, pleura, or peritoneal irritations or reactions from effusions as the potential sources of elevated CA125 in heart diseases including heart failure (26, 27). In addition, heart disease results in overexpression of inflammatory cytokines, especially IL6, which may raise CA125 levels (28–30). Interestingly, we observed lower CA125 levels in postmenopausal women with a history of hypercholesterolemia/atherosclerosis. Our findings are similar to those of Joo and colleagues who reported a significant inverse correlation of CA125 with metabolic syndrome consisting of dyslipidemia, higher total cholesterol and elevated triglycerides (31).

In our study, the association between endometriosis and elevated CA125 was restricted to premenopausal women. Previous studies have observed higher CA125 levels in women with active endometriosis and proposed CA125 as a candidate biomarker for detecting endometriosis (17, 32). We did not observe an association between history of endometriosis and CA125 in postmenopausal women. Interestingly, in the PLCO study, a history of endometriosis was associated with significantly lower CA125 in postmenopausal women (7). Because endometriosis is thought to derive from cells exfoliated from the uterine lining, our observation of no association between endometriosis and CA125 among postmenopausal women may simply reflect atrophy of the endometrium and endometriosis that occurs after menopause. Alternatively, the lack of association may reflect prior hormone suppressive therapy or hysterectomy for treatment of endometriosis. Notably, adjusting for these factors in the PLCO study attenuated the inverse association between CA125 and prior endometriosis in that study (8). These results suggested that there is a complex association of endometriosis, its treatment, and menopausal status with CA125 levels.

In our study, hysterectomy was associated with lower CA125 levels in premenopausal women, but not in postmenopausal women. Two studies addressing hysterectomy and CA125 in postmenopausal women showed similar relationships (8, 14). Furthermore, we previously observed lower CA125 levels in premenopausal and postmenopausal women who had had a hysterectomy in the EPIC cohort, although the association was statistically significant only in postmenopausal women (13). Whether or not postmenopausal women with hysterectomy were taking menopausal hormonal therapy may also be a factor because exogenous hormones likely increase CA125 levels (8, 33, 34). In our study, the average CA125 in hysterectomized postmenopausal women known to have taken hormone therapy was a little higher (12.3 units/mL) compared with those not known to have used HRT (11.9 units/mL; P = 0.28). The two observations that endometriosis raises CA125 and hysterectomy lowers it in premenopausal women suggest the importance of the uterus (or perhaps more precisely the uterine lining) in the physiology of CA125. CA125 is indeed expressed in the endometrium, and in naturally cycling women, CA125 is highest during the menstrual phase (35, 36).

We found two medical conditions related to the colon suggestively associated with CA125 levels. CA125 is expressed in gastrointestinal tract (37–39) and some reports have previously described the association between colorectal cancer, gastrointestinal tumors, and CA125 (40–42). In our study, colon polyps were associated with significantly lower CA125 levels among premenopausal women and suggestively lower levels in postmenopausal women, although we were not able to examine the relationship with prior colon cancer due to the small number of women with this condition. Because colon polyps may be precursors to colon cancer, one may have expected to see higher CA125 levels in women with polyps. However, removal of these polyps could have lowered the CA125 levels. It is also possible that conditions known to be associated with colon polyps, including smoking and obesity, could underlie the association because these exposures can lower CA125 (8). While we adjusted for these potential confounders in our study, the possibility of residual confounding remains. In contrast, the association we observed between higher CA125 and inflammatory bowel diseases among postmenopausal women may be due to the inflammatory nature of these diseases as CA125 is known to be elevated in some inflammatory conditions of the gastrointestinal tract (37, 43).

We initially observed higher CA125 levels in postmenopausal women with prior breast cancer but the association became attenuated after additionally adjusting for some medications used in treating breast cancer including tamoxifen and arimidex. CA125 may be elevated at breast cancer diagnosis and may be used in the management of breast cancer, especially for the advanced metastatic disease (44–46). Pauler and colleagues reported significantly higher CA125 among women with any prior (non-ovarian) cancers in the UK screening study (14).

That both osteoporosis and osteoarthritis were related to lower CA125 among the postmenopausal women is consistent with a correlation previously reported between the degree of bone mineral density loss and CA125 (47). This may reflect declining CA125 level with an increasing hypoestrogenic state. Similarly, an increasing hypoestrogenic state in postmenopausal women may in part underlie osteoarthritis (46) and could explain our observation that postmenopausal women with osteoarthritis have lower levels of CA125. As previously pointed out, menopausal hormone replacement can raise CA125 levels (33), and may also increase bone mineral density and reduce the risk for osteoarthritis of the hip (33, 48).

We did not find consistent patterns in CA125 levels associated with the number of medical conditions. This is not surprising, as some medical conditions measured here were associated with an increase of the percent change in CA125, while others were associated with a decrease. Consequently, the net change in CA125 corresponding to a given number of comorbidities likely would reflect the commonness of a morbidity, the frequency with which comorbidities migrate together, and the strength and direction of their impact on CA125 levels. Interestingly, however, our data indicated that, in premenopausal women, the influence of individual medical conditions on CA125 levels was similar when adjusted for multiple comorbidities. After adjusting for comorbidities in postmenopausal women, only osteoporosis remained associated with a percent change in CA125 levels, and the observed decrease was comparable with that seen in the univariate model. We found no evidence that treatments associated with various medical conditions measured here changed the association between the medical condition and the percent change in CA125 levels.

Together, our results suggest that several common medical conditions may influence CA125 levels in the general population and are worth consideration in the context of measuring CA125 as a screening biomarker. As is known, CA125 levels vary not only due to the influence of medical conditions but also with technical and biologic functions as well as normal interindividual and within subject biologic variations. In randomized screening trials to date, a single threshold of 35 U/mL is used to classify women as having abnormally high levels that warrant additional screening. However, adjusting this threshold for abnormal CA125 levels based on personal characteristics may yield a more sensitive and specific test. Several studies have evaluated nonmedical factors on a much larger scale (8, 13, 14). However, our study adds many medical conditions that may be used to further adjust the CA125 screening threshold. The limitations of our study include lack of comprehensive inclusion of all medical conditions that may influence CA125, small sample size, self-reported exposures, lack of validation in medical records, a cross-sectional design that prevents repeated measures of CA125 and precludes assessment of temporality, and lack of racial diversity in the New England population.

Future work should assess the relationship of these and other medical conditions on CA125 in a more diverse population. If confirmed, our findings may help identify factors that could influence false positive and false negative readings for CA125, and ultimately help enhance prevention or detection of ovarian cancer at an early stage with significant prognosis.

No potential conflicts of interest were disclosed.

Conception and design: B.O. Akinwunmi, D.W. Cramer, L. Titus, S.S. Tworoger, K.L. Terry

Development of methodology: B.O. Akinwunmi, L. Titus, K.L. Terry

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.F. Vitonis, L. Titus, K.L. Terry

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B.O. Akinwunmi, A. Babic, A.F. Vitonis, S.S. Tworoger, K.L. Terry

Writing, review, and/or revision of the manuscript: B.O. Akinwunmi, A. Babic, A.F. Vitonis, D.W. Cramer, L. Titus, S.S. Tworoger, K.L. Terry

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B.O. Akinwunmi, D.W. Cramer

Study supervision: L. Titus, K.L. Terry

Other (PI of the grant that leads the work): D.W. Cramer

This work was supported by NIH grants R01 CA193965, R01 CA054419, and R35 CA197605.

The authors gratefully acknowledge the contributions of Ajay Singh, Finnian McCausland, and the entire Master of Medical Sciences in Clinical Investigation (MMSCI) team, Harvard Medical School. We are also grateful for the contributions of all members of the OB/GYN Epidemiology Center and NEC study participants.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.