|Home | About | Journals | Submit | Contact Us | Français|
To determine the epidemiology of CA-125 in women without ovarian cancer.
We analyzed demographic, medical and lifestyle characteristics related to CA-125, measured using the Centocor CA-125II RIA assay, among 25,608 multi-ethnic U.S. women aged 55–74 years enrolled in a cancer screening trial and found to have no evidence of ovarian cancer.
Mean CA-125 level was 11.9 U/ml (SD 8.3); median 10.0 U/ml, interquartile range 8.0–14.0. High levels, using the clinical cut point of ≥ 35 U/ml, were associated with increased age (p<0.001) and former smoking (p<0.021), while hysterectomy and obesity were protective (p<0.001). Mean levels were higher with increasing age (p<0.001), ever use of hormone therapy (p<0.001), former smoking (p<0.017) and history of breast cancer (p<0.002), but lower (p<0.001) with non-White status, previous hysterectomy, current smoking, and obesity. Current hormone therapy use was not associated with CA-125 in women without a uterus.
In post-menopausal women without ovarian cancer, CA-125 level is influenced by a number of factors, including race/ethnicity, age, hysterectomy, smoking history and obesity.
CA-125 is a tumor antigen that is elevated in the majority of ovarian cancers and has been shown to be increased in some women with early stage disease. Although a single determination of CA-125 does not have the sensitivity, specificity and positive predictive value to be used as a stand-alone screen, CA-125 measurement may be shown to be a valuable part of a multi-modal and/or longitudinal screening algorithm.[2–8] If this approach proves feasible, knowledge of the usual pattern of CA-125 levels among populations of women without ovarian cancer will be a prerequisite.
CA-125 is an antigenic determinant on a high molecular weight glycoprotein encoded by the MUC16 gene, [10;11] that was first recognized by a monoclonal antibody, OC125, and subsequently a series of other antibodies. CA-125 has been shown to be elevated in women with a number of physiologic and pathologic processes including cirrhosis and congestive heart failure and has been reported to vary by age, race, and the presence of benign gynecologic conditions (endometriosis, hysterectomy, hormone therapy use) and other malignancies besides ovarian cancer (breast, colon, pancreatic, lung, gastric, liver cancer).[1;9;13;14]
Most studies of CA-125 have been conducted among cancer patients. While some have studied this marker among general populations, often through analyzing women enrolled in screening studies, many analyses have had relatively small sample sizes and few have been geographically and ethnically diverse. We had available baseline measurements of CA-125 in an exceptionally large, racially and ethnically diverse population of post-menopausal women, residing in multiple locations across the United States. In this population, CA-125 is currently being tested as one modality in a screening trial for ovarian cancer. Our primary goal was to evaluate the epidemiology of CA-125 at baseline measurement in a population examined and found to be without evidence of ovarian cancer, with a focus on the effects of personal characteristics that can readily be obtained clinically including race/ethnicity, age, personal and family medical history, smoking behavior and body mass index.
The population under study was selected from the Prostate, Lung, Colorectal and Ovarian Cancer (PLCO) Cancer Screening Trial. From 1993 through 2001, the PLCO Trial recruited over 150,000 men and women to a randomized controlled trial of screening methods for four cancers and has been described in detail elsewhere. Eligible subjects were from 55 to 74 years of age and not diagnosed previously with prostate, lung, colorectal, or ovarian cancer. Criteria for exclusion included current treatment for cancer other than non-melanoma skin cancer and enrollment in another cancer screening or prevention trial. Beginning on April 15, 1995, individuals who had received a colonoscopy, sigmoidoscopy, or barium enema in the past 3 years were also excluded. Initially, women who by self-report had undergone oophorectomy were ineligible, but in 1996 this restriction was lifted because low accrual of women threatened to jeopardize screening endpoints for lung and colon cancer.
During recruitment, 78,237 women were enrolled and 39,115 were randomized to the intervention arm. Women in this arm, unless they reported a history of oophorectomy, received a baseline CA-125 measurement and transvaginal ultrasound. During the screening process, CA-125 results ≥ 35 U/ml were classified as abnormal (positive) and generated letters to the subject and her physician urging follow-up. Procedures and any diagnoses resulting from abnormal screening results were ascertained by study staff and the entire population was surveyed annually for the occurrence of any cancer diagnoses.
Serum frozen locally was shipped to the Immunogenetics Laboratory at UCLA for testing using the Centocor CA-125II RIA assay, a heterologous, double determinant immunoassay. The capture antibody is the M11 murine monoclonal antibody, and the tracer antibody is the OC 125 monoclonal antibody originally generated by immunization of BALB/c mice with the OVCA 433 ovarian cancer cell line. Quality assurance was done in accordance with the manufacturer’s suggested protocol. The coefficients of variation were found to be 4.07% at the lower concentration of 52.7 U/ml and 3.78% at the higher concentration of 106.5 U/ml. The corresponding 95% confidence intervals (CI) were 3.92%–4.22% and 3.64%–3.92%, respectively. These results are in good agreement with those reported by the manufacturer.
Women completing a baseline screen including a questionnaire and a CA-125 test were potentially eligible for these analyses. Analyses were conducted on data available as of July 2004. From the initial women enrolled in the intervention arm, the following were excluded: women without ovaries who were not screened or who were screened inadvertently and those not receiving the CA-125 test/assay, women with a diagnosis of ovarian cancer at baseline or within 2 years of screening or with lack of post-baseline follow-up of up to two years (therefore ovarian cancer status was unknown at 2 yrs post baseline), and women with missing information on one or more of the baseline variables included in the analyses.
Univariate and multivariate logistic and linear regression techniques were used to evaluate whether the study variables were associated with CA-125 level. Distributions of each variable were assessed for outliers. Models were constructed with CA-125 as a binary outcome (≥35 U/ml as a cut point) and as a log-transformed continuous variable. For the log-transformed regression models, parameter estimates are reported along with their exponentiated values, since the latter estimates the ratio between mean CA-125 levels for subjects with a factor versus those without. All model assumptions were checked to assure they were not violated and the data were examined for collinearity among explanatory variables.
The associations between race/ethnicity (White, Black, Hispanic, Asian, Pacific Islander/Native American) and age at screening (55–59, 60–64, 65–69, 70–74 years) and CA-125 were examined as well as other variables including age at last menstrual period as a marker for age at menopause (<40, 40–44, 45–49, 50–54, 55+ years), a personal history of breast cancer, a history of ovarian cancer in first degree relatives, a history of breast cancer in first degree relatives, previous hysterectomy, former or current use of hormone therapy (HT) (yes/no) and body mass index (BMI) (weight/height2 with normal <25, overweight 25–29, obese 30+). The model also included other variables considered to be potentially associated with CA-125 including history of endometriosis (yes/no), uterine fibroids (yes/no), benign ovarian tumors or cysts (yes/no), partial oophorectomy (yes/no), as well as cigarette smoking status (never/former/current). The analyses were repeated separately for women with and without a hysterectomy as the endometrium is a principal source of CA-125and therefore factors associated with this marker might vary depending on the presence or absence of the uterus.
All aspects of the PLCO study were approved by the Human Rights Committees of each institution and the National Cancer Institute, and written informed consent was obtained from each subject.
There were 34,288 women who were eligible for a baseline CA-125 measurement. Of these, 6,107 did not receive a CA-125 or were ineligible for these analyses due to an ovarian cancer diagnosis or lack of follow-up. Of the remaining subjects, the 2,573 women with CA-125 testing but incomplete baseline survey data did not differ from the final 25,608 subjects in the analyses with regard to mean CA-125. However there were statistically significantly higher proportions of Black, older and less than high school educated women among those with missing data. Table 1 shows the distribution of the 25,608 women analyzed by race/ethnicity, age group and other variables by ranges of CA-125 levels along with means and medians. Women were predominantly white (90%) and between the ages of 55–64 years (66%), with 26% reporting a hysterectomy, 65% ever using hormone therapy, and 24% with a BMI ≥ 30 (obese). Although the majority was white, minority representation was substantial with 1,132 Black and 900 Asian participants.
The mean CA-125 level was 11.9 (1 SD 8.3) and the median was 10.0 with an interquartile range of 8.0–14.0 U/ml (Table 1). Just over 1.6% of all participants had CA-125 levels of 35 U/ml or greater. An additional 1.1% and 7.4% had levels between 30-<35 and 20-<30 U/ml, respectively. This left nearly 90% of the population having values between 0 and 20 U/ml, with an even distribution in the 0-<10 (44.7%) and 10-<20 (45.2%) groups. Black women had the lowest mean and median CA-125. Mean CA-125 increased with age category, with a decline of percentage of women in the lowest CA-125 category (0-<10 U/ml) as age increased corresponding to increasing percentages in the highest CA-125 category. The earlier the age at menopause, the lower the mean CA-125, and the more likely a woman would fall in the lowest CA-125 category.
Compared to White women, minority women were at lower risk for an elevated CA-125 (35 U/ml or greater), with the adjusted odds ratio (aOR) for Asians (aOR=0.53, 95% confidence interval (CI) 0.28–1.00) of borderline statistical significance (p<0.051) (Table 2). Increasing age groups were each associated with a statistically increased risk for a high CA-125. Hysterectomy (aOR = 0.58; CI 0.41–0.81, p<.001) and obesity (aOR=0.53; CI 0.39–0.71, p<0.001) were associated with lower risk for elevated CA-125. Although early age at menopause was protective in the univariate analyses, there was no evidence of an association of menopausal age with high CA-125 in the multivariate analyses. There was no association with reported history of endometriosis. Women reporting a history of uterine fibroids had an increased aOR of 1.26, although the confidence intervals, 0.97–1.64, included 1.0.
Considering mean levels of CA-125 as an outcome (Table 3), race/ethnicity analyses again using Whites as a reference and adjusting for other variables revealed statistically significant (p<0.001) lower mean CA-125 levels for each of the other race/ethnic groups. There was a statistically significant association between increasing age category and higher mean CA-125 levels for each age group (p<0.0001). Those women with hysterectomy had lower CA-125 (p<0.001). Women with a history of breast cancer, late menopause, former and current hormone therapy use, and former smokers had statistically significantly higher mean CA-125 levels. Obesity and current smoking were associated with statistically significantly lower mean CA-125 levels. Adjusting for other variables abrogated univariate inverse associations between low education, early menopause, endometriosis, uterine fibroids, ovarian cysts and partial oophorectomy as related to CA-125 as a continuous variable.
Repeating these analyses and considering only the 18,955 women who still had their uterus, both former (p<0.013) and current (p<0.0001) hormone therapy use was associated with increased CA-125 level. Restricting the analyses to the 6,653 women with a previous hysterectomy, only former use of hormone therapy (versus no history of use) was associated with an increased CA-125 as a continuous variable (p<0.01).
In this study of CA-125 in post-menopausal women with no evidence of ovarian cancer, mean CA-125 level was 11.9, with a very low percentage of women, 1.6%, having values greater than the standard clinical threshold of 35 U/ml. The most striking associations, based on the multivariate parameter estimates (Table 3), were for race/ethnicity, corresponding to 19%, 12%, 8% and 11% lower mean CA-125 levels in Black, Hispanic, Asian and Pacific Islander/Native American women relative to White women, hysterectomy (9% lower mean), current smoking (9% lower mean) and to a much lesser degree obesity (3% lower mean). The higher mean CA-125 levels in women who were older (7%), had a history of breast cancer (5%), late menopause (4%), former smokers (1.4%) and users of hormone therapy (3–4%), were modest. Interestingly, the demographic patterns are complementary to incidence patterns for ovarian cancer, with higher incidence in White and older women.
The most comparable study we could find was data from 18,748 post-menopausal women in a United Kingdom screening trial. However, the focus of their analyses was to develop a parsimonious predictive model rather than our approach of including all variables of interest in our models with the purpose of calculating risk estimates for individual variables. While Pauler et al. included women from 40 to 60 years of age, our study subjects were older, with a range from 55–74 years.
Prior studies generally report a decrease in CA-125 levels with increasing age.[9;13;19–21] It has been consistently demonstrated that CA-125 is higher in pre- versus post-menopausal women, so it is important to consider that our population was all post-menopausal. Our mean and median values for Whites were identical to those in a reference value study of 938 Dutch post-menopausal women with a mean of 12 U/ml and median of 10 U/ml. These authors found a slight decrease in age using categories from <45 years to >65 years, but did not consider any other variables. Another study from the Netherlands indicated a dramatic drop in CA-125 between age categories from 60–70 years in a healthy control group of 370 women, again a univariate analysis. However, Grover et al found no age association in post-menopausal women, and an increase in CA-125 with increasing age in a pre-menopausal and peri-menopausal sample. Pauler et al. demonstrated a slight reduction in CA-125 levels with increasing age adjusted for some of the same variables we used. They reported that the decrease by age was attenuated in women with a previous history of cancer. Our results indicating an increase in CA-125 by increasing post- menopausal age category were based on a very large, diverse, and older sample of women and allowed simultaneous adjustment for numerous collected potentially important variables. If valid, this association could possibly be a consequence of aging processes at the cellular and immunological level. Interestingly, when we analyzed CA-125 using results from the original Centocor assay available for 5371 initially enrolled women aged 60 years and older, the values were lower (age 60–64, mean 9.2 and median 8.0; age 65–69, mean 9.2, median 7.7; age 70–74, mean 9.9, median 8.0), and there was no statistical difference by age categories, suggesting that perhaps the assay used could affect results.
While endometriosis has consistently been found to be associated with CA-125 level, which is expected based on the underlying biology, we did not find this association. This may reflect the limitation that a history of endometriosis was based solely on self report. More importantly, endometriosis often resolves after menopause and therefore a post-menopausal CA-125 might not be expected to be related to a historical diagnosis of this condition.
Another striking finding from these PLCO CA-125 analyses is the differences by racial/ethnic group. These results adjusted for education level as a surrogate marker for socioeconomic status and confirm the prior UK study addressing race/ethnicity as a possible factor in CA-125 levels in which only 80 Asian and 89 African women out of a total of 18,748 subjects were evaluated. Even so, race was a statistically significant variable with mean CA-125 levels slightly lower (by 1.2 U/ml) in Asians and markedly lower (by 5.2 U/ml) in African relative to White women. In the UK study, the Asian group is most probably predominantly from the Indian subcontinent, whereas in the PLCO study the Asian group is largely comprised of women with Japanese, Chinese, and Filipino heritage.
In our data, hysterectomy was associated with decreased mean CA-125, confirming other work.[9;22] The finding that current hormone therapy use increased CA-125 only in women with a uterus comports with the premise that hormone therapy resulted in stimulation of CA-125 in the endometrium, which is a known source of CA-125 in healthy women. Although some women with hysterectomy who reported having their ovaries had likely also had oophorectomy, the concentration of CA-125 in the healthy ovary is small compared to the endometrium and oophorectomy had no significant impact on CA-125 levels in other studies. [9;23] CA-125 levels vary during the menstrual cycle, suggesting an influence of ovarian steroid hormones. Kurihara et al. demonstrated in a small study that CA-125 was higher among healthy post-menopausal women using hormone therapy than among non-users. Two clinical trials examined CA-125 levels in response to initiation of hormone therapy.[25;26] In both studies there was no effect of current estrogen therapy on CA-125 levels in women with hysterectomies, which corroborates our data. However, for women with a uterus, Karabacak reported that 100 ug/day transdermal estradiol was associated with a significant increase in CA-125, again consistent with our results. In contrast, Cengiz et al. reported that current use of combination of estrogen and progestin resulted in lower CA-125 levels in women with a uterus, while Okon et al. saw no change in their 12 month follow-up study.[26;27] The type of hormone therapy used was not determined in our study, but it is likely that estrogen-only therapy would predominate in the women with hysterectomy and estrogen-progestogen therapy would predominate in women with a uterus.
It is not obvious why current smoking and obesity are associated with a lower CA-125. It is intriguing that Pauler et al. also found a protective effect of current smoking, which they considered was most likely a fluke or possibly due to an effect on liver enzymes and enhanced metabolic degradation of CA-125. Perhaps a higher plasma volume, associated with obesity,[28;29] dilutes the level of serum CA-125. Why former smoking or hormone therapy use may be associated with increased CA-125 is even less clear. A limitation of the baseline questionnaire used in this study is that “former” was not linked to dates or duration, so we cannot reconstruct histories linking the timing of hysterectomy, smoking, and timing and duration of hormone therapy use as related to the date of CA-125 measurement.
Another limitation is that while this population is unusually large and represents geographic and racial/ethnic diversity, it is comprised of women agreeing to participate in a long term cancer prevention screening trial. Potential subjects were recruited using a multitude of approaches, with random mailings accounting for the majority of enrollees. It has been demonstrated that the PLCO population has lower mortality rates than the general population, suggesting they are healthier in general. Analyses of data from one site comparing those enrolled to those invited from within a health care system population demonstrated that 11% of those asked joined the study and suggested that those who were White, in their sixties, with higher income and with fewer co-morbidities were somewhat more likely to participate. Our analyses are therefore likely subject to some degree of selection (volunteer) bias. Further, although the trial deployed numerous strategies to recruit minority subjects , the percentages of enrolled women in these groups was lower than the percentages these groups represent in the United States, suggesting that volunteer bias and representativeness may be even more of an issue for non-Whites. This is noteworthy since an important result of our analyses is the lower CA-125 values associated with minorities. Additionally, there is always the possibility that our results could be affected by unmeasured confounders such as subclinical chronic disease processes or genetic variation and that characteristics such as age and race are markers for some other factor directly affecting CA-125. Finally, it is possible that a small number of women with sub-clinical ovarian cancer remained in the study population.
In summary, CA-125 levels were found to be associated with a number of demographic and medical factors. A notable finding was substantially lower mean CA-125 levels in minority women relative to White women, with levels 8–19% lower. If CA-125 is incorporated in a screening algorithm, these variables may prove to be important in clinical evaluations. Future analyses will consider changes in CA-125 over time using the baseline and measurements available from five subsequent annual screening exams among women without ovarian cancer in this population.
The authors gratefully acknowledge the contribution of the study staff at each of the ten screening centers; Westat, Inc.; Information Management Services, Inc.; the central Immunogenetics Laboratory at UCLA; and the study investigators and staff at the National Cancer Institute.
Funding Sources: This work was supported by individual contracts from the National Cancer Institute to each of the ten screening centers and to the coordinating center.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of Interest Statement:
None of the authors has a conflict of interest.