Elevation of TP53 Autoantibody Before CA125 in Preclinical Invasive Epithelial Ovarian Cancer

Ovarian cancer ranks fifth in cancer-related deaths among women in 2016, and is the leading cause of death from gynecologic malignancies (1). If detected although still confined to the ovaries (Stage I) or to the pelvis (Stage II), 5-year survival rates are 90% and 70%, respectively. Once the disease has spread to the peritoneal cavity (Stage III) or beyond (Stage IV), the 5-year survival decreases sharply to less than 20% (2–4). Computer simulation suggests that earlier detection could reduce mortality by as much as 43% (5). Recent results from the United Kingdom Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) document that screening for ovarian cancer using a multimodal (two-stage) strategy can lead to a stage shift and an estimated 20% reduction in mortality (6).

In the multimodal (two-stage) screening strategy used both in UKCTOCS (6) and in the Normal Risk Ovarian Screening Study (NROSS; ref. 7), the level of the serum biomarker CA125 (MUC16) was determined annually in postmenopausal women at average (general population) risk of developing ovarian cancer. A Bayesian statistical method, the Risk of Ovarian Cancer Algorithm (ROCA), was used to calculate each woman's risk of having ovarian cancer based on age-specific incidence, dates of blood draws, and longitudinal CA125 levels after every new CA125 test. ROCA assesses whether CA125 has significantly increased above an individual's CA125 baseline levels (8, 9). If risk has not changed, women return in one year for the next CA125 determination. If the ROCA is strongly elevated, transvaginal ultrasound (TVS) is obtained and, if this suggests malignancy, an operation is performed. If a modest increase in risk is observed, CA125 is repeated in 3 months. This two-stage strategy (ROCA + TVS), where two independent tests are both positive, improves the limited specificity of CA125 using a single threshold (e.g., 30 or 35 U/mL) as well as the limited specificity of TVS and results in only 3 to 4 operations (40% positive predictive value) for each case of ovarian cancer detected (6, 7). It also increases the sensitivity for earlier detection of ovarian cancer to around 85% (10, 11). A recent report suggesting that this approach may reduce disease specific mortality (6), adds impetus to the search for new markers that can complement CA125, especially in women who are not detected using CA125 alone.

About 80% of ovarian cancers express significant amounts of CA125 at a tissue level (12). However, only 50% to 60% of early-stage cancers have elevated CA125 serum levels at presentation (13). This suggests that a panel of biomarkers will almost certainly be required to detect all early-stage ovarian cancers. During the last two decades, more than 110 different markers have been studied to increase sensitivity of CA125 for detecting ovarian cancer. HE4 and CA72.4 can detect a fraction of cases missed by CA125 (14, 15), but in studies with preclinical samples to date, no biomarker has been consistently elevated before CA125.

The sensitivity of tumor-derived biomarkers for early detection is limited by the expression of a biomarker within a cancer, the rate of biomarker-shedding and the volume of early-stage cancers. Small cancers may not shed sufficient amounts of biomarker(s) to raise serum levels, but could evoke an immune response (16). Small volumes of cancer with mutated, overexpressed or aberrantly compartmentalized tumor associated proteins could evoke a humoral immune response with readily detectable titers of specific autoantibody (17). Therefore, detecting an autologous immune response to tumor associated antigens might improve lead time and detect cancers that are too small to be detected by antigen assays.

The TP53 tumor-suppressor gene is frequently mutated in a variety of human cancers. TP53 is mutated in virtually all high-grade serous ovarian cancers (18). Approximately two thirds of TP53 mutations are missense that stabilize TP53 protein and increase TP53 accumulation (19). Autoantibody reactive with wild-type TP53 have been reported in sera from approximately 15% of women with ovarian cancers, but most reports have studied only a limited number of cases at the time of symptomatic clinical diagnosis (20–36).

The goals of this study are (i) to develop and validate a new and more sensitive immunoassay for autoantibodies against wild-type TP53 protein, (ii) to determine the fraction of ovarian cancer patients with early-stage disease who have elevated levels of TP53 autoantibody, (iii) to discover whether TP53 autoantibody can provide lead time over CA125 and detect cases that do not have elevated CA125, and (iv) to test whether autoantibodies against patient specific mutations of TP53 provide more sensitive assays. Availability of serial pre-clinical specimens from the UKCTOCS has permitted measurement of TP53 autoantibody in large numbers of women who went on to develop ovarian cancer. We have tested for the first time whether TP53 autoantibody is elevated before CA125. In earlier studies, autoantibody has been detected to wild-type TP53 protein. With access to serum specimens obtained at clinical diagnosis from the Australian Ovarian Cancer Study (AOCS) for which TP53 sequences are known, we have tested for the first time whether greater sensitivity could be attained by detecting autoantibody to the patient's specific mutant TP53 protein rather than to the wild-type TP53 protein.

The MagPlex/xMAP technology (Luminex Corp.) was used to develop an indirect serological assay for detecting human TP53 autoantibody. Coupling recombinant wild-type TP53 antigen to microspheres, confirmation of effective coupling and calibration of the assay were accomplished using protocols modified from the Luminex xMAP Cookbook. In brief, 1 × 10⁶ microspheres were coupled with recombinant human wild-type TP53 protein using an xMAP antibody coupling kit. Coupling was completed by a carbodiimide reaction linking the primary amino groups on TP53 protein and the carboxyl groups on the microsphere surface. The antigen conjugation procedure was performed according to the manufacturer's instructions. Biotin-conjugated anti-human TP53 antibody was used to validate coupling efficiency. A TP53 autoantibody calibrator (10 U/mL) was used to establish a standard curve for quantitating the titer of TP53 autoantibody in human serum samples. The standard operating protocol for performing the assay was established during assay development.

A new xMAP bead-based immunoassay for detecting TP53 autoantibody in human serum samples was performed by an established standard operating protocol. In brief, a suspension of TP53 antigen-microspheres was prepared by diluting the coupled microsphere stocks (1 × 10⁶ beads/mL) to a final concentration of 50 beads/μL in PBS buffer. Aliquots of microsphere suspension were placed in each well of a 96-well polystyrene microplate. Standard curves for quantitating TP53 autoantibody were plotted from triplicate assays of half-log dilution series of the TP53 autoantibody calibrator with concentrations ranging from 0.031 to 0.5 U/mL. Positive and negative controls for TP53 autoantibody were prepared in triplicate according to the manufacturer's instructions. Serum samples for assay (2 μL) were diluted with serum matrix to mimic the native analyte environment. Serum samples were added to 96-well polystyrene microplates and incubated with TP53-coupled microspheres for 1 hour at room temperature with gentle shaking. To wash the beads, each plate was clipped onto a magnetic plate separator for 1 minute and the liquid was discarded by inverting the plate. Following washing with PBS buffer, microspheres were incubated with detection antibody biotinylated goat anti-human IgG. The plates were covered to protect them from light for 30 minutes at room temperature with gentle shaking. Plates were washed two times with PBS buffer and microspheres were incubated with fluorescence reporter Streptavidin-R-phycoerythrin (SAPE). The plates were covered to protect them from light and incubated for 30 minutes at room temperature with gentle shaking. After a final wash, the beads were re-suspended in PBS buffer and fluorescence measured on the MAGPIX system (Luminex Corp.) with at least 50 beads per well. The data were acquired and analyzed by xPONENT software version 4.2 (Luminex Corp.).

Details are available in the Supplementary Materials and Methods.

The MDACC-NROSS samples included 50 preoperative sera from patients with stage III–IV invasive epithelial ovarian/tubal/peritoneal cancer in the MD Anderson Cancer Center Gynecologic Cancer Tissue Bank and 216 sera from healthy controls who did not develop ovarian cancer on the NROSS coordinated by the MDACC Ovarian SPORE (7). The AOCS sample set included preoperative sera from 108 clinically diagnosed patients with invasive epithelial ovarian/tubal/peritoneal cancer, preoperative sera from 109 patients with benign ovarian tumors and age-matched sera from 464 healthy Australian controls. The UKCTOCS sample set was obtained from participants of this population-based, multicenter randomized controlled trial of ovarian cancer screening in the United Kingdom (10, 37). The sample set included 164 screen-detected and 56 screen-negative women with 1,053 preclinical serial samples predating diagnosis of invasive epithelial ovarian/tubal/peritoneal cancer up to 5 years and 619 age-matched controls (3,069 serial samples) who did not develop any type of cancer during follow-up. The combined sample sets are generally representative of the performance of multimodal screening in the UKCTOCS, but have been enriched for screen negative cases. Ethical approval was obtained from these studies from the appropriate IRB/ethical committees. All participants had provided consent for use of samples in ethically approved secondary studies.

We chose our cutoff value to achieve high specificity in multiple datasets as described below. We applied bootstrapping to construct 95% confidence intervals for the TP53 autoantibody cutoff values associated with 98% specificity in the MDACC, AOCS, and UKCTOCS cohorts. A cutoff value of 78 U/mL, which falls within all three of the above 95% confidence intervals (CI) and provides an overall specificity of 97.4%, was chosen as the common threshold for detecting ovarian cancer.

Line plots and dot plots were drawn to compare differences in early detection times between TP53 autoantibody, CA125, and ROCA in the longitudinal samples of UKCTOCS cohort. Wilcoxon signed rank tests were performed to assess whether the additional early detection time provided by TP53 autoantibody improved upon CA125 or ROCA when both tests indicated a case. In addition, Fisher exact tests were used to assess the independence of TP53 autoantibody and ROCA in the three sample cohorts. Bootstrapping was applied to determine whether TP53 autoantibody could detect ROCA negative cases in the UCKTOCS cohort.

Receiver operating characteristic (ROC) analysis adapted for early-detection-times was performed to compare performance between single and combination biomarker panels in the UKCTOCS cohort. The R package partial ROC (pROC) was used for ROC computations. Longitudinal sets of values were reduced to single values as follows. First, 35 and 78 U/mL were specified as cutoff values for CA125 and TP53 autoantibody, respectively. If a participant, either a case or a control, was never detected by one of the biomarkers, the maximum value of that marker was assigned to that participant. If a marker ever exceeded its cutoff, the value seen at the first time that the marker exceeded its cutoff was recorded as the summary value. If a case was detected by both CA125 and TP53 autoantibody assays, and the difference in detection times was larger than a certain threshold, such as 3, 6, or 12 months, a zero for the biomarker that detected the cancer later was recorded, as an indication that the latter marker had “failed” by not catching the tumor as early as it could have been caught. After this adjustment for early-detection-time, ROC analyses were performed both for single biomarkers and biomarker combinations. For the biomarker combination, a logistic regression model was built with the summary values as inputs and case/control status as targets, and the predicted logistic regression values were applied as the biomarker combination “score.” ROC curves were plotted both for the single biomarker and for the biomarker combination. Area under the curve (AUC), partial AUC (pAUC), sensitivity, and accuracy were computed at 98% specificity. Next, bootstrapping was used to compute 95% CIs for the AUC, as well as P values for the differences in AUC and pAUC values between single biomarkers and the biomarker combination.

We have developed a sensitive, rapid and high-throughput MagPlex microsphere-based immunoassay for quantitating TP53-specific autoantibody in small volumes (2 μL) of serum. The procedure for immunoassay development, validation, and assessment of performance in nested case–control studies has been outlined in Supplementary Fig. S1. After establishing and optimizing of the assay method, a validation study was performed to demonstrate the accuracy, reproducibility and reliability of the TP53 autoantibody immunoassay. Serum samples with three different TP53 autoantibody titers (low, medium, and high range of each sample) were measured in triplicate. The intra-assay coefficient of variation was 3.5% to 4.6% and the inter-assay coefficient of variation was 6.6% on three consecutive days. The linearity (R²) was 0.9936 and the linear range was between 5.83 and 250 U/mL. The Limit of Detection (LoD) value was 5.83 U/mL and the maximum detectable TP53 autoantibody titer was 250 U/mL for this immunoassay.

To determine a common cutoff value for the TP53 autoantibody immunoassay, we first used a bootstrap technique to construct 95% CIs at the 98th percentile of TP53 autoantibody values for healthy controls across three datasets (Supplementary Fig. S2). For the MDACC-NROSS dataset, the 98th percentile for controls was 100.9 (95% CI, 32.4–148.3) U/mL. For the AOCS dataset, the 98th percentile for healthy controls was 48.5 (95% CI, 35.0–84.4) U/mL. For the UKCTOCS dataset, where serial samples were available from each healthy control, we chose the single highest value from each participant to calculate the 98^th percentile of 94.6 (95% CI, 65.6–128.0). A common cutoff value of 78 U/mL was chosen for TP53 autoantibody levels. This value was included within the 95% CIs for the 98^th percentile across all datasets (Supplementary Fig. S2). Using all controls, this provided a specificity of 97.4%. TP53 autoantibody levels above 78 U/mL were considered positive.

We first analyzed the MDACC-NROSS set of specimens to validate the newly developed immunoassay for detecting TP53 autoantibody in human serum samples. Patient characteristics and TP53 autoantibody positivity by stage or histology are described in Supplementary Table S1. Results are summarized in Fig. 1. With a 78 U/mL cutoff value for TP53 autoantibody, 15 of 50 sera (30%) from cancer cases were positive, whereas 5 of 216 sera (2.3%) of healthy controls exhibited TP53 autoantibody (Fig. 1A).

We also compared CA125 values (the cutoff value = 35 U/mL) with TP53 autoantibody levels in each cancer case (Fig. 1A). CA125 was positive in 45 of 50 cases (90%) and 4 of 216 controls (1.9%). The box plot showed that both TP53 autoantibody and CA125 values were significantly elevated in cases when compared with controls. Remarkably, 12 of 15 (80%) ovarian cancer patients had extremely high titers of TP53 autoantibody (>250 U/mL; Fig. 1B). Elevation of both CA125 and TP53 autoantibody was observed in 13 of 50 (26%). We further evaluated whether TP53 autoantibody correlated with ovarian cancer histology. All TP53 autoantibody positive cases were serous or serous mixed with endometrioid histotype (Supplementary Table S1).

Next, we analyzed sera from the AOCS study. Patient characteristics and TP53 autoantibody positivity by stage or histology are described in Supplementary Table S2. TP53 autoantibody values were significantly elevated in ovarian cancer cases when compared with patients with benign ovarian neoplasms or healthy controls. With a 78 U/mL threshold, sera from 23 of 108 invasive epithelial cancer cases (21.3%) had elevated TP53 autoantibody levels (Fig. 2). By contrast, 3 of 109 sera (2.8%) from patients with benign ovarian neoplasms and 6 of 464 sera (1.3%) from healthy controls had elevated TP53 autoantibody levels (Fig. 2). We further evaluated whether TP53 autoantibody correlated with ovarian cancer stage or histology. Elevated TP53 autoantibody levels were found in 2 of the 12 early-stage (I/II; 16.7%) and 19 of 90 late-stage (III/IV; 21.1%) patients. All TP53 autoantibody–positive cases exhibited a serous histotype (Supplementary Table S2).

Availability of sera from the UKCTOCS trial permitted detection of TP53 autoantibody in preclinical sera from a large number of women who subsequently developed invasive epithelial ovarian cancer. Patient characteristics and TP53 autoantibody positivity by stage or histology are shown in Table 1. Elevated TP53 autoantibody levels were found in sera from 43 of 220 patients (19.5%) with invasive epithelial ovarian cancer and in sera from 17 of 619 subjects (2.7%) without cancer detected during the study (Fig. 3A). The TP53 autoantibody levels were also compared to ROCA/CA125 values in each of the ovarian cancer cases. In the UKCTOCS dataset, ROCA could detect 164 of 220 (74.5%) cases and 34 cases were detected by both ROCA and TP53 autoantibody assays. In box plots, both values were significantly elevated in cases compared to controls (Fig. 3A). In addition, scatter plot analysis suggested that CA125 is, as anticipated, a good biomarker to separate cases from controls (Fig. 3B). We further evaluated whether TP53 autoantibody correlated with ovarian cancer stage or histology. Elevated TP53 autoantibody levels were found in 11 of the 81 early-stage (I/II; 13.6%) and 32 of 139 late-stage (III/IV; 23.0%) patients (Supplementary Table S3). Some 76.7% of cancer cases with elevated TP53 autoantibody levels were of serous histotype, but 1 to 2 cases with endometrioid, clear cell and mucinous histotype were also associated with elevated TP53 autoantibody (Table 1).

In the UKCTOCS dataset, among the 220 women who went on to develop ovarian cancer, 164 cases (74.5%) were detected with rising CA125 using the ROCA and TVS (screen positives) and 56 cases (25.5%) were not detected by ROCA (screen negative). Thirty-four of 164 (20.7%) screen-positive cases and 9 of 56 (16.1%) screen-negative cases had elevated levels of TP53 autoantibody (Fig. 3A). Thus, in retrospect, TP53 autoantibody could identify 16% of ovarian cancer cases missed by the ROCA.

To determine whether TP53 autoantibody titers might increase before elevation of CA125 above the 35 U/mL threshold or to an indication of elevated risk determined by the ROCA, we measured TP53 autoantibody in serial serum specimens from ovarian cancer cases in the UKCTOCS dataset. Serial values for CA125 and TP53 autoantibody from two representative early-stage cancer cases are plotted in Supplementary Fig. S3A and S3B. TP53 autoantibody titers were markedly elevated (>250 U/mL) 8 and 7 months before cancer diagnosis, whereas CA125 levels remained within the normal range (<35 U/mL) during the entire period. In addition, the TP53 autoantibody titers were elevated in cancer cases at late stage (Supplementary Fig. S3C and S3D). Although CA125 remained within normal range, autoantibody titers increased dramatically (190∼250 U/mL) at 4 years 4 months and 1 year 6 months before cancer diagnosis.

Among 43 cases with autoantibody elevation, TP53 autoantibody was detected 11.8 months (mean) before an elevation of the ROCA and 11.0 months before CA125 elevation (≥35 U/mL; Fig. 4A). In 34 cases with both autoantibody elevation and ROCA positivity, TP53 autoantibody were detected 9.2 months before ROCA positivity, and 8.1 months before CA125 elevation (Fig. 4B). In 9 cases with autoantibody positivity that were not detected by the ROCA, TP53 autoantibody titers rose 22.9 months (mean) before cancer diagnosis (Fig. 4C). Thus, TP53 autoantibody levels provide a promising biomarker with clinically significant lead time over either elevation of CA125 levels alone or elevation of risk using the ROCA algorithm to analyze CA125 levels.

To determine whether TP53 autoantibody levels enhanced the ability of CA125 to detect ovarian cancer, we generated ROC curves to discriminate cancer cases from controls for the UKCTOCS. The values first reaching the TP53 autoantibody cutoff (78 U/mL) and CA125 cutoff values (35 U/mL) were chosen for ROC analysis. The AUC for TP53 autoantibody, CA125 and the combination of these two markers have been calculated. In the UKCTOCS dataset (Fig. 5A), the AUC of TP53 autoantibody was 0.699. Adding TP53 autoantibody values to the panel can significantly increase the AUC (CA125, 0.838 vs. TP53 AAb + CA125, 0.867, P = 0.007).

As the regular ROC analysis may not reflect elevation of TP53 autoantibody before CA125, we performed new ROC analyses incorporating lead time. ROC analysis was performed to compare performance between single and combination marker panels in the UKCTOCS cohort corrected for different lead times as described in the Methods section. ROC curves for TP53 autoantibody and CA125 in the UKCTOCS study were calculated using 3, 6, and 12 months as the cutoffs to define “significant early detection.” With 3 months lead time (Fig. 5B), the AUC of TP53 autoantibody was 0.636. Adding TP53 autoantibody to the panel significantly increased the AUC (CA125, 0.751 vs. TP53 AAb + CA125, 0.861, P = 0.000). With 6 and 12 month lead times a combination of CA125 and TP53 autoantibody was not superior to CA125 alone (Supplementary Fig. S4A and S4B). Thus, a combination of CA125 and TP53 autoantibody significantly enhanced the ROC curve relative to either biomarker alone.

At a set cutoff value (78 U/mL) for TP53 autoantibody, our immunoassay using wild-type TP53 detected autoantibody in 21%∼30% of invasive epithelial ovarian cancer patients in three datasets. As virtually all high-grade serous cancers, which comprise approximately 60% of all ovarian cancers, have TP53 mutations at a variety of sites, we may have missed autoantibody against epitopes expressed by these mutant TP53 proteins, but not by wild-type TP53. Indeed, our previous screening result in the AOCS trial showed that cancer cases who harbored a TP53 mutant gene have higher TP53 autoantibody titers compared with TP53 wild-type cancer cases (Supplementary Fig. S5), consistent with previous reports (34). Therefore, we asked whether assays incorporating the specific mutant TP53 proteins might detect antibody in a larger fraction of patients or at higher titer.

At present, 95 different TP53 mutations have been identified in high-grade serous ovarian cancers (18). As the TP53 gene has been sequenced for each case in the AOCS trial, we could express the specific mutant proteins from the ovarian cancers of these patients using a mammalian protein expression system. We then established immunoassays and compared autoantibody titers against wild-type TP53 with titers against each specific mutant TP53 at 98% specificity in AOCS controls. We selected 15 TP53 point mutation candidates for multiplex immunoassay development based on previous screening results in the AOCS trial (Supplementary Fig. S6). Mutant TP53 proteins were chosen where mutations at the same locus were associated with anti-TP53 antibody in some patients, but not in others or where mutations were associated with intermediate TP53 wild-type autoantibody titers.

When sera from 29 patients in the AOCS study were assayed against 15 corresponding mutant TP53 proteins and the results compared with those from assays against wild-type TP53 protein, increased sensitivity was not observed (Supplementary Table S4). Patients with undetectable TP53 wild-type autoantibody had undetectable TP53 mutant autoantibody titers. Differences in titers, when present, did not exceed thresholds for either TP53 wild-type or TP53 mutant for positivity at 98% specificity. Subsequently, we analyzed sera from the AOCS trial against TP53 wild-type and 9 mutant protein candidates. When compared to titers against TP53 wild-type protein, most of the TP53 mutant proteins detected the same number or fewer cases (19.4%–22.2%) with the thresholds set at 98% specificity (Supplementary Table S5). In general, they detected exactly the same cases as TP53 wild-type protein. Only TP53 R248W, R273H and R273L mutant proteins detected an additional 2–3 more cases (24.0%–25.0%) compared with TP53 wild-type protein (Supplementary Table S5). Consequently, it appears that the predominant epitope(s) recognized by the autologous humoral response is likely to be expressed on the wild-type TP53 protein.

In this study, we have developed a new, high-throughput xMAP bead-based immunoassay for quantitating TP53-specific autoantibody in ultra-low volumes of human serum. This assay was used to analyze sera from patients with invasive ovarian, tubal and primary peritoneal cancer from three large datasets to test whether TP53 autoantibody might serve as a biomarker for early detection of ovarian cancer. We set a cutoff value of TP53 autoantibody titers (78 U/mL) for each study, reflecting a specificity of 97.4%. Our assay detected elevated levels of TP53 autoantibody in 21% to 30% of cases at the time of clinical diagnosis, whereas previous studies had found TP53 autoantibody in around 15% of cases (20–36). In the UKCTOCS trial, TP53 autoantibody could detect 16% of ovarian cancer patients who were not detected using the CA125 based multimodal strategy. Remarkably, in the UKCTOCS screen detected cases, TP53 autoantibody levels rose a mean of 11.0 months earlier than did CA125 (>35 U/mL) and 11.8 months earlier than the ROCA. In cases that were not screen detected using the CA125-based ROCA screening strategy, TP53 autoantibody titers rose an average of 22.9 months before cancer diagnosis. Among more than 110 ovarian cancer biomarkers identified to date, this is the first promising candidate that is able to detect invasive epithelial ovarian cancer at an earlier time than CA125 (38). Recently, Russell and colleagues (39) identified new biomarkers Protein Z, Fibronectin and C-reactive protein by the proteomic platform and those biomarkers showed a lead time on CA125 in small UKCTOCS sample sets (40).

The TP53 gene is frequently mutated in a variety of human cancers and plays a crucial role in their development, continued growth and genetic instability. Our immunoassay for detecting TP53 autoantibody may have potential applications for early detection of other types of cancer. Indeed, detection of TP53 autoantibody has also been reported in other cancers such as esophageal cancer (41, 42), head and neck cancer (43, 44), colorectal cancer (45–47), hepatocellular carcinoma (48, 49), lung cancer (50–52), and breast cancer (53, 54) etc. If TP53 autoantibody is used for screening in a general population, more prevalent cancers may also be detected. This could be an asset rather than a limitation, if a cost-effective diagnostic algorithm were developed to differentiate other types of cancer in addition to ovarian cancer in women with an elevated TP53 autoantibody titer.

We also evaluated the potential benefit in combining CA125 with TP53 autoantibody for detecting ovarian cancer. ROC curve analysis for the UKCTOCS dataset showed that combination of CA125 with TP53 autoantibody can significantly increase the AUC (Fig. 5A). To further confirm those results, we performed new ROC analyses incorporating 3, 6, and 12 months lead time of CA125 and TP53 autoantibody (Fig. 5B and Supplementary Fig. S4A and S4B). Combination of CA125 with TP53 autoantibody can significantly increase the AUC (Fig. 5B; CA125, 0.751 vs. TP53 AAb + CA125, 0.861, P = 0.000). Our data suggest that the addition of TP53 autoantibody may improve upon CA125 alone or the ROCA for identifying patients with early-stage ovarian cancer. Further large-scale clinical studies will be required to evaluate the clinical efficacy of a combination of CA125 with TP53 autoantibody assays for detecting ovarian cancer.

We tested whether using assays that incorporated the mutant TP53 protein corresponding to that in the ovarian cancer from each patient might provide a more sensitive assay than wild-type TP53. Similar sensitivity was obtained with specific mutant and wild-type TP53. Given current technology, mutant proteins and peptides could be incorporated in assays for early detection, if there were an advantage to this approach. However, we did not find such an advantage in this study. Consequently, this was an important issue to resolve. This also suggests that the autoantibody might be directed against wild-typeTP53 and the amount of non-degraded TP53 in ovarian cancer cells might be an important factor in developing autoantibody.

To our knowledge, this study is the first to assay TP53 autoantibody in large numbers of preclinical sera. As only 20% to 25% of patients with invasive epithelial ovarian cancer have elevated levels of autoantibody against TP53, this biomarker may have only a limited, but potentially significant, impact on the ability to detect early stage disease. Our study does demonstrate that TP53 autoantibody levels: (i) can complement CA125 at the time of diagnosis; (ii) provide substantial lead time over CA125 in a fraction of cases; and (iii) could contribute one important member of a panel of biomarkers that would improve substantially upon the performance of CA125. Given the proof of concept that TP53 autoantibody can be elevated in advance of CA125, we are identifying other autoantibodies that would complement TP53 autoantibody levels and create a panel of tests that would detect CA125-negative cases and provide lead time in a greater fraction of cases.

I. Jacobs is an employee of Abcodia Ltd. and Women's Health Specialists Ltd., holds ownership interest (including patents) in ROCA Algorithm, and is a consultant/advisory board member for Abcodia Ltd. S.J. Skates is a consultant/advisory board member for Abcodia Ltd. W. He holds ownership interest (including patents) in OriGene. U. Menon is listed as a coinventor on a patent on implications of RANKL inhibition in prevention of hormone driven breast cancer, which is owned by Institute of Molecular Biotechnology GmbH. R.C. Bast Jr receives royalties for CA125 from Fujirebio Diagnostics Inc. No potential conflicts of interest were disclosed by the other authors.

Conception and design: W.-L. Yang, A. Simmons, K.H. Lu, I. Jacobs, S.J. Skates, W.-W. He, R.C. Bast Jr

Development of methodology: W.-L. Yang, A. Gentry-Maharaj, A. Simmons, Y. Zhao, W.-W. He, R.C. Bast Jr

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): W.-L. Yang, A. Gentry-Maharaj, A. Simmons, A. Ryan, E.O. Fourkala, K.H. Lu, D. Bowtell, U. Menon

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W.-L. Yang, A. Gentry-Maharaj, A. Ryan, K.A. Baggerly, Y. Zhao, R.C. Bast Jr

Writing, review, and/or revision of the manuscript: W.-L. Yang, A. Gentry-Maharaj, A. Simmons, A. Ryan, E.O. Fourkala, Y. Zhao, K.H. Lu, D. Bowtell, I. Jacobs, S.J. Skates, U. Menon, R.C. Bast Jr

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W.-L. Yang, A. Simmons, E.O. Fourkala, K.H. Lu, W.-W. He, R.C. Bast Jr

Study supervision: Z. Lu, R.C. Bast Jr

We are grateful to the UKCTOCS participants who donated their samples for use in secondary studies. The authors are solely responsible for the design of the study, the analysis and interpretation of the data, the writing of the article, and the decision to submit the article for publication. We also acknowledge the contribution of the AOCS nurses and research assistants and would like to thank all of the women who participated in the study.

This work was supported by funds from the Early Detection Research Network (5 U01 CA200462-02) and the MD Anderson Ovarian SPORE (P50 CA83639), National Cancer Institute, Department of Health and Human Services; the Cancer Prevention Research Institute of Texas (RP160145); Golfer's Against Cancer, the Mossy Foundation, the Roberson Endowment, National Foundation for Cancer Research; UT MD Anderson Women's Moon Shot; and generous donations from Stuart and Gaye Lynn Zarrow. The UT MD Anderson Cancer Center Odyssey Program provided support (to W.-L. Yang), the Theodore N. Law Endowment for Scientific Achievement (to W.-L. Yang); the Clyde H. Wright Memorial Fund (to W.-L. Yang) and Bristol-Myers Squibb Award in Clinical Research (to W.-L. Yang). S.J. Skates received additional support from the NCI Early Detection Research Network (CA152990). UKCTOCS was core funded by the Medical Research Council, Cancer Research UK, and the Department of Health with additional support from the Eve Appeal, Special Trustees of Bart's and the London, and Special Trustees of UCLH and supported by researchers at the National Institute for Health Research University College London Hospitals Biomedical Research Centre. AOCS was supported by the U.S. Army Medical Research and Materiel Command under DAMD17-01-1-0729, The Cancer Council Victoria, Queensland Cancer Fund, The Cancer Council New South Wales, The Cancer Council South Australia, The Cancer Foundation of Western Australia, The Cancer Council Tasmania, and the National Health and Medical Research Council of Australia (NHMRC; ID400413, ID400281). The AOCS gratefully acknowledges additional support from S. Boldeman, the Agar family, the Peter MacCallum Cancer Centre Foundation, Ovarian Cancer Australia and Ovarian Cancer Action (UK).

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.