Anti-MUC1 Antibodies and Ovarian Cancer Risk: Prospective Data from the Nurses' Health Studies

The human mucin family member MUC1 is a heavily glycosylated protein expressed in low levels by many normal epithelial cells, whereas a poorly glycosylated form is overexpressed by many adenocarcinomas, including breast and ovarian cancers (1). Circulating anti-MUC1 antibodies have been identified in patients at the time of cancer diagnosis and generally predict better survival (2-4). Antibodies also have been found in healthy individuals, especially in women who are pregnant or lactating (4-7), leading to the hypothesis that natural immunity against MUC1 might develop and account for the long-term protective effects of pregnancy and breastfeeding on breast cancer risk (8). In addition, several other factors that reduce risk of ovarian cancer, including oral contraceptive use, tubal ligation, and mastitis during breastfeeding, have been associated with higher antibody levels (9). These findings led to the hypothesis that immunity developed against MUC1 could account, at least in part, for the long-term protective effect of certain reproductive and hormonal factors on ovarian cancer risk (9).

Because ovarian tumors can express MUC1 and lead to antibody development, it is important to evaluate antibody levels well before diagnosis in a prospective study design. Thus, we measured anti-MUC1 antibodies in plasma samples collected from Nurses' Health Study (NHS) participants in a nested case-control study of ovarian cancer, in which cases provided blood samples at least 3 years before a diagnosis, to evaluate the relationship between circulating anti-MUC1 antibodies and risk of ovarian cancer.

The NHS was established in 1976 among 121,700 U.S. female registered nurses ages 30 to 55 years at entry, and the NHSII in 1989 among 116,430 U.S. female registered nurses ages 25 to 42 years at enrollment (10, 11). Information on potential risk factors for ovarian cancer, including reproductive history, was obtained through baseline and follow-up questionnaires sent every 2 years. Heparinized blood specimens were provided by 32,826 NHS participants between 1989 and 1990 (12), and by 29,611 NHSII participants between 1996 and 1999 (13). Samples were shipped through overnight courier to the laboratory, where they were separated into plasma, RBC, and WBC components, and stored in liquid nitrogen freezers. Follow-up of NHS and NHSII blood cohorts was 98% in 2004 and 2003, respectively. NHS and NHSII protocols have been reviewed and approved by the Committee on the Use of Human Subjects in Research at the Brigham and Women's Hospital.

Incident cases of epithelial ovarian cancer diagnosed at least 3 years after blood collection but before June 1, 2004 (NHS) or June 1, 2003 (NHSII) were identified. A total of 117 cases (111 from NHS, 6 from NHSII) were confirmed by medical record review and matched to 339 controls (∼3 per case) on age (±1 y), menopausal status at blood collection and diagnosis, postmenopausal hormone use at blood collection, and month (±1 mo), time of the day (±2 h), and fasting status at the blood draw. All controls had intact ovaries at the time of case diagnosis. Except for nonmelanoma skin cancer, cases and controls had no history of cancer before blood collection.

Anti-MUC1 antibody assays were conducted at the Department of Immunology, University of Pittsburgh. Antibodies were measured against a synthetic 100-mer MUC1 peptide corresponding to five tandem repeats of the MUC1 polypeptide core repeat region (9, 14). Briefly, MUC1-coated Immulon wells (Dynax) and peptide-negative plates were incubated overnight and washed thrice with PBS before the addition of 100 μL of 2.5% bovine serum albumin in PBS. Serially diluted plasma (1:40-1:80 in PBS) was added to MUC1-coated plates and incubated at room temperature. Plates were washed 5 times with 100 μL PBS and 0.1% Tween 20 detergent. Alkaline phosphatase–labeled goat anti-human polyvalent IgM, IgG, and IgA (50 μL; Sigma-Aldrich) diluted 1:1,000 were added before plates were again washed 5 times with PBS-Tween. Alkaline phosphatase substrate pNPP (100 μL; Sigma-Aldrich) was added. Plates were incubated before the stop solution (0.5 mol/L NaOH) was added. We used the MRX Revelation plate reader (Thermo Labsystems) to read absorbance values at 405 to 410 nm, which were subtracted from the absorbance values obtained from antigen-negative plates to account for nonspecific binding.

Laboratory personnel were blinded to case/control status of the samples. Blood specimens from cases and their matched controls were assayed on the same plate, and each plate included two to eight quality control samples. The within-plate coefficient of variation was 9%, showing excellent reproducibility, and the between-plate coefficient of variation was 23% for absorbance values at the 1:40 dilution, and 9% and 24% at the 1:80 dilution. The Spearman rank correlation between the two dilutions was high (ρ = 0.93; P < 0.001); readings at the 1:40 dilution were used in this analysis. Cases and their matched controls were assayed together and on the same plate.

The distribution of anti-MUC1 antibody levels among subjects was skewed right; therefore, values were natural-log transformed for all analyses. No antibody values were identified as outliers (15). Generalized linear models were used to compare mean anti-MUC1 antibody levels across categories of factors possibly associated with antibodies, adjusting for age and assay plate, among controls and characteristics of the cancer among cases. The estimated number of ovulatory cycles for each woman was calculated by subtracting age at menarche, 1 year for each term pregnancy, duration of oral contraceptive use, and breastfeeding from age at menopause (or age if premenopausal), and multiplying by 12, as previously described (16).

We used conditional logistic regression to estimate relative risks (RR) and 95% confidence intervals (95% CI) in comparing quartiles (based on the control distribution) of anti-MUC1 antibody levels. To examine whether the association between antibody levels and ovarian cancer risk differed according to categories of age at blood collection, we compared the likelihood of models with and without an interaction term between quartiles of antibody levels weighted by median antibody levels (modeled continuously) and age (in years). All P-values were two-sided. Analyses were conducted on SAS 9.1 version (SAS Institute).

The average age at blood collection was 57 years for both cases and controls, and the average age at ovarian cancer diagnosis for cases was 65 years (Table 1). Compared with controls, women who developed ovarian cancer were more likely to be nulliparous, have a family history of ovarian cancer, and have relatively shorter durations of oral contraceptive use.

Among controls, women in the top quartile of age at blood collection (i.e., ≥64 y) had significantly lower levels of antibodies compared with younger women (P = 0.03; Table 2). Antibody levels also decreased with increasing number of estimated ovulatory cycles (P-trend = 0.04) but were higher among controls with a history of tubal ligation (P = 0.03). Antibody levels were slightly but not significantly lower among controls whose age at menopause was ≥51 years or had regularly used talc in genital hygiene.

Among cases, women whose age at blood collection was ≥64 years had a nonsignificantly higher level of antibodies than those whose age at collection was younger (P = 0.28; Table 3). No significant differences in antibodies were observed between invasive or borderline tumors, or by histology. Women who went on to develop late-stage versus early-stage disease tended to have higher antibody levels (P = 0.04), and women who became cases >11 years after their blood was drawn had higher antibody levels than those diagnosed earlier in the follow-up period (P = 0.01). As might be expected, only 5 (17%) of the 29 women with the shortest interval to diagnosis (and lowest anti-MUC1 antibody levels) were diagnosed after age 65 years compared with 20 (67%) of the 30 women with the longest interval to diagnosis (and highest antibody levels; P < 0.0001). None of the factors found to affect antibody levels in controls were significant in the cases (data not shown).

Among all women, those with antibody levels in the top quartile had a nonsignificant 28% lower risk of developing ovarian cancer compared with those in the lowest quartile (95% CI, 0.39-1.31); there was no significant trend across quartiles (P-trend = 0.30; Table 4). However, there was a statistically significant difference in the association by age at blood draw (P-interaction = 0.005). Because the RRs in quartiles 2 through 4 were similar, we collapsed these quartiles to increase power in the analysis stratified by age. Among those <64 years at blood collection (n = 90 cases and 257 controls), the RR for ovarian cancer comparing quartiles 2 to 4 with quartile 1 was 0.53 (95% CI, 0.31-0.93); P-trend = 0.03). The comparable RR in women ≥64 years at blood draw (n = 27 cases and 82 controls) was 2.11 (95% CI, 0.73-6.04; P-trend = 0.05). Neither adjusting for known risk factors of ovarian cancer (e.g., oral contraceptive use, parity, and tubal ligation) nor excluding NHSII participants (6 cases and 8 controls) altered the results (data not shown).

In this first-of-its-kind study, we prospectively evaluated the relationship between circulating anti-MUC1 antibody levels and risk of ovarian cancer. We observed a statistically significant interaction by age for the association between anti-MUC1 antibodies and risk of ovarian cancer, with an inverse association among women <64 years at assessment (i.e., the controls had higher antibody levels on average than the cases) and a positive association in those ≥64 years (i.e., the controls had lower antibody levels on average than the cases). A decline in anti-MUC1 antibodies in healthy individuals with increasing age has been observed previously (4, 9). This may be due to a mechanism called immunosenescence, which reflects a declining immune response toward many common antigens with increasing age and time since antigen presentation, such as after immunization (17, 18). Many of the exposures, that we have postulated could lead to protective antibodies, occur during the reproductive years, and their effect could wane with an increasing time interval from those events. This interpretation would be compatible with data from published epidemiologic studies indicating that the protective effect of oral contraceptive use (19, 20) and, possibly, with breastfeeding (21) and tubal ligation (22), attenuates over time. In addition, the drop in antibody levels among controls matches age-specific increases in risk for ovarian cancer in the population; that is, about a 2-fold change between women <64 years compared with those ≥64 years (27 versus 56 cases per 100,000 in NHS data).

The events associated with antibody levels among controls found in our study include known risk factors for ovarian cancer. Notably, women who had undergone a tubal ligation had ∼20% higher antibody levels compared with those who had not had one; this is similar to a prior finding among 705 healthy women (9). Explanations for the protective effect of tubal ligation on ovarian cancer risk have focused on blockage of potentially harmful contaminants from reaching the ovaries; however, our results suggest the injury associated with tubal ligation may trigger the release of MUC1 into the circulation and the development of protective anti-MUC1 antibodies.

Increasing number of ovulatory cycles, which has been linked with higher risk for ovarian cancer, was associated with lower antibody levels in this study and in one prior study examining this association (23). The ability of uninterrupted ovulations to lower anti-MUC1 antibody levels may be due to an absence of reproductive events, such as oral contraceptive (OC) use and breastfeeding, which may have otherwise elicited antibody production. In addition, MUC1 has higher expression in secretory endometrium (24), and circulating levels are highest during the luteal phase (25). Thus, it is possible that repeated cyclic variation of MUC1 in blood over many years could lead to an immune tolerance that ultimately lowers anti-MUC1 antibody levels.

A prior study of 705 women without ovarian cancer noted associations between anti-MUC1 antibody levels and other factors, including OC use, hysterectomy, breastfeeding, bone fracture, and talc use (9). Although we did not observe statistically significant associations for these exposures in this study (possibly due to the smaller number of controls), the direction of these relationships were similar to the prior study. For example, talc use was associated with lower anti-MUC1 antibody levels in both studies, fitting with its role as a factor that has been shown to increase risk for ovarian cancer (26-28). Like ovulatory cycles, regular talc use may represent an exposure that can trigger a chronic inflammatory response, lead to MUC1 antigenemia, and result in MUC1-specific immune tolerance. Clearly, additional research is necessary to fully understand how “acute” events translate into protection and “chronic” events into increased risk, and how these may relate to the difference in the association by age. We believe this will require an understanding of the effects of the events and age on cellular immune reactions that may be most critical for cancer elimination or escape.

We suggested that the decline in anti-MUC1 antibodies in controls reflected immunosenescene. Clearly, the higher antibody levels observed among cases who were ≥age 64 years at blood draw (Table 3) requires a different interpretation. First, we point out that the level of antibodies in the oldest age group was about the same as in the younger controls. This may indicate that there is a subset of women who do not have the expected decline in antibody levels as they age, perhaps due to chronic antigen challenge provided by an ongoing chronic inflammatory condition or a developing cancer. In addition, we note that cases who were ≥64 years at blood draw had an age at diagnosis of 75 years and overlapped with the group in Table 3, which had the greatest lag between age at blood draw and cancer diagnosis, and the highest antibody levels. Unfortunately, our data cannot clarify whether the higher antibody levels in older cases represent a persistence of high antibodies that may delay the onset of disease or new appearance of antibodies reflecting a chronic inflammatory response. Sorting out the possibilities will require larger studies with serial measurements of anti-MUC1 antibody levels across the important mid-60s age period. Besides anti-MUC1 antibodies, it may be necessary to measure free MUC1 antigen and immune complexes involving the antigen and antibodies. These have been described and could interfere with assays for either free antigen or free antibody (29). A comprehensive study might also establish whether the appearance (or persistence) of high levels of anti-MUC1 antibodies in later life might represent an early detection signal many years in advance of ovarian cancer.

In the current study, we focused our attention on MUC1 because of the wealth of data that exists on its expression in normal tissues versus cancer, inflammation, and other pathologic conditions (1, 30). We measured antibodies against the unglycosylated protein core of MUC1, commonly found in tumors. However, there are other aberrantly glycosylated MUC1 epitopes closely associated with tumor MUC1 that might be even more immunogenic, eliciting additional protective immunity (31, 32). It is also possible that additional ovarian cancer antigens can be presented to the immune system through other nonmalignant events, and lead to the creation of antibodies and cellular immunity that would also be associated with a reduced risk for ovarian cancer. MUC1 thus represents one of many targets of immunosurveillance that can collectively elicit protective immunity and lower cancer risk. The development of new high-throughput technologies may allow identification of antibodies that are present in circulation before ovarian cancer diagnosis, and the exposures or events that are associated with increased antibody levels and define serologic signatures of cancer risk reduction. Further, understanding how antibodies against MUC1 or other antigens are formed during exposures or events that predate ovarian cancer development may suggest ways to duplicate or amplify these protective effects through vaccination.

A limitation of this study is that we were unable to measure MUC1-specific memory T cells due to the lack of archived viable white cells. However, T-cell activation is indirectly measured through the presence of various isotypes of anti-MUC1 antibodies, all of which require antigen-specific helper T cells for their generation. We also lacked information on certain exposures that have previously been associated with anti-MUC1 antibody levels, such as mastitis, mumps parotitis, and endometriosis. The relatively small number of cases in this study clearly limited our power to examine the effect of antibodies by histologic type of ovarian cancer or by other potential effect modifiers, and the relatively high between-plate coefficients of variation, often seen in immune-based assays like the one used here, also contributed to a lack of precision.

Although ours is the first prospective study investigating anti-MUC1 antibodies in relation to ovarian cancer, one study examined humoral immune responses to MUC1 among BRCA carriers and related it to subsequent breast cancer, another MUC1-expressing cancer (33). Women who carried a mutation had significantly lower anti-MUC1 immunoglobulin antibodies compared with controls, and carriers who went on to develop breast cancer had lower (although not significantly so) antibody levels than carriers not found to have cancer at prophylactic mastectomy. Thus, this study fits with our observations in suggesting that anti-MUC1 antibodies correlate with a known risk factor for the cancer and risk for developing disease.

In conclusion, our prospective study provides important evidence that circulating anti-MUC1 antibody, either as an effector mechanism or as a marker of a more comprehensive immune response, may be associated with a lower risk of ovarian cancer in women <64 years and offers an explanation for the mechanism of action of several factors known to alter risk of ovarian cancer. These data further suggest a new paradigm that immunosurveillance may play an important role in ovarian carcinogenesis. Larger studies are necessary to clarify the role of MUC1 immunity in ovarian cancer, particularly in relation to the potential differential association by age at antibody assessment. Ultimately, this avenue of research could lead to new methods for ovarian cancer prevention or early detection.

No potential conflicts of interest were disclosed.

We thank the participants in the Nurses' Health Studies for their continued participation and dedication.

Grant Support: NIH grants [grant numbers CA105009 and CA123170 (D.W. Cramer), CA49449 and CA67262 (SEH), and P01 CA73743 (O.J. Finn).

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.