P53 AUTOANTIBODIES AS POTENTIAL DETECTION AND PROGNOSTIC BIOMARKERS IN SEROUS OVARIAN CANCER

Abstract

Background:

This study examined the value of serum p53 autoantibodies (p53-AAb) as detection and prognostic biomarkers in ovarian cancer.

Methods:

p53-AAb were detected by ELISA in sera obtained pre-operatively from women undergoing surgery for a pelvic mass. This group included women subsequently diagnosed with invasive serous ovarian cancer (N=60), non-serous ovarian cancers (N=30), and women with benign disease (N=30). Age-matched controls were selected from the general population (N=120). Receiver operating characteristic curves were constructed to compare the values of p53-AAb, CA 125, and HE4 as a screening biomarker. Kaplan-Meier curves and Cox proportional hazards modeling were used to assess its prognostic value on survival.

Results:

p53-AAb were detected in 25/60 (41.7%) of serous cases, 4/30 (13.3%) non-serous cases, 3/30 (10%) benign disease cases, and 10/120 (8.3%) controls (combined p=0.0002). p53-AAb did not significantly improve detection of cases (AUC=0.69) or discrimination of benign versus malignant disease (AUC=0.64) compared with CA 125 (AUC=0.99) or HE4 (AUC=0.98). In multivariate analysis among cases, p53-AAb correlated only with a family history of breast cancer (p=0.01). Detectable p53 antibodies in pre-treatment sera were correlated with improved overall survival (p=0.04, HR=0.57 (95% CI=0.33, 0.97)) in serous ovarian cancer.

Conclusions:

Antibodies to p53 are detected in the sera of 42% of patients with advanced serous ovarian cancer.

Impact:

Although their utility as a pre-operative diagnostic biomarker, beyond CA 125 and HE4, is limited, they are prognostic for improved overall survival.

INTRODUCTION

Changes in tumor antigens, whether due to overexpression, mutation, or altered degradation, can lead to the development of autoantibodies(1). Tumor antigen-specific autoantibodies have been identified in the sera of patients with solid tumors, with antibody levels generally increasing with tumor burden(2–8). The long half-life and in vitro stability of these antibodies make them potential biomarkers for the early detection and/or prognosis of cancer.

As an autoantibody biomarker, p53-AAb are attractive because p53 is mutated in a variety of cancers(9). The development of p53-AAb is associated with tumor p53 mutations that lead to decreased protein degradation(9, 10), and reflect p53-dependent changes in tumor biology. p53-AAb are detected in the sera of 6–7% of patients with limited-stage and 19–30% of patients with late-stage ovarian cancer(11, 12), suggesting that p53-AAb would have limited application as a diagnostic biomarker. Evidence for the utility of p53-AAb as a prognostic biomarker in ovarian cancer is mixed. Goodell et al. found a correlation with improved overall survival(11), but two other groups found no correlation with disease-specific survival(12, 13). These differences may reflect patient selection or differences in epitope detection in the assays. It is not known if cancer autoantibodies are involved in active immunologic surveillance, or if they are simply byproducts of altered protein structure found in cancer cells.

We have developed a custom ELISA assay, termed Rapid Antigenic Protein In situ Display (RAPID), for the detection of antibodies to tumor antigens in patient sera(14). RAPID ELISA eliminates the need for a priori protein purification and minimizes the risk of immunogenic co-purified bacterial antigens as cross-reacting serologic targets in ELISA. p53-AAb are detected on RAPID ELISA with comparable specificities and limits of detection as standard recombinant protein ELISA for p53 antigen, with a linear range of detection covering three orders of magnitude(8). In this study, we investigated the utility of p53-AAb as biomarkers of diagnosis and prognosis for serous ovarian cancer by itself and compared to the current best ovarian cancer biomarkers, CA125 and HE4 (15).

MATERIALS AND METHODS

Patient Sera

Sera used in these analyses were obtained from the Brigham and Women’s Hospital and the Dana-Farber Cancer Institute with support from the NCI Early Detection Research Network. Sera derived from ovarian cancer patients were obtained at the time of presentation prior to surgery, and patients received routine post-operative therapy. The non-serous cases were derived from 10 patients with endometrioid cancer, 10 patients with clear cell carcinoma, and 10 patients with mucinous carcinoma. The benign disease samples were derived from 19 patients with serous cystadenomas and 11 patients with mucinous cystadenomas. Sera from age-matched general population control women were obtained from the Brigham and Women’s Hospital using a standardized serum collection protocol and stored at −80°C until use. Cases and matched controls were processed simultaneously. Women with a personal history of cancer (other than non-melanoma skin cancer) were excluded as controls. Written consent was obtained from all subjects under institutional review board approval. For the 60 serous cases included in the survival analysis, medical records were reviewed and details related to presentation and treatment abstracted.

RAPID ELISA for p53 Antibodies

Detection of p53 antibodies in patient sera using Rapid Antigenic Protein In situ Display (RAPID) ELISA was performed as described(14). Briefly, 96-well detection plates coated with anti-GST antibody (GE Healthcare, Piscataway, NJ) were blocked overnight at 4°C with blocking buffer (5% milk/0.2% Tween20 in phosphate buffered saline (PBS-T)). On the next day, full-length cDNA’s in a pCITE vector optimized for in vitro expression (from Harvard Institute of Proteomics, Cambridge, MA) encoding wild-type p53-GST, p21-GST, EBNA-1-GST, and/or pCITE control vector were expressed using the TNT T7 Coupled Reticulocyte Lysate System (IVTT, Promega, Madison, WI) per manufacturer’s recommendations. To each tube of lysate was added: 16 μl reaction buffer, 8 μl polymerase, 4 μl minus-methionine, 4 μl minus-leucine, 8μl RNaseOUT (Invitrogen, Carlsbad, CA), 500 ng DNA, to a final volume of 400 μl. The DNA-lysate mixture was incubated at 30°C for 90 minutes. After incubation, PBS was added to the mixture and 50 μl transferred to each well. The plate was incubated for 2 hours at room temperature (rt.) while shaking at 800 rpm, and then washed 5 times with blocking buffer. Human serum samples were diluted 1:300 in blocking buffer and 100 μl added to each well. To test gene expression, mouse anti-GST monoclonal antibody (Cell Signaling Technology, Danvers, MA) was diluted 1:1000 in SuperBlock blocking buffer (Pierce, Rockford, IL) and 100 μl added to each well. Secondary antibodies were diluted at 1:1000 and were HRP-conjugated goat anti-human IgG antibody (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) or anti-mouse IgG secondary antibody (GE Healthcare, Piscataway, NJ). Luminescence detection was performed using the SuperSignal ELISA Femto Max Sensitivity (Pierce, Rockford, IL) and a Wallac plate reader (Perkin Elmer, Waltham, MA). Study specimens were tested in duplicate on the same plate.

Detection of CA 125 and HE4 in sera

The CA 125 assay is a double antibody chemiluminescence immunoassay based on two monoclonal antibodies [M 11 and OC 125, Fujirebio Diagnostics]. The Roche CA125II test has been standardized against the Enzymun-Test CA 125 II method that in turn was standardized against the CA 125 II RIA from Fujirebio Diagnostics. The analytical sensitivity of the test is 0.6 U/mL. The dynamic range of the assay is 0.60 to 5,000 U/mL and the reportable range extends up to 25,000 U/mL using a 5-fold dilution. Study specimens were tested in singlet. The imprecision [% CV] of the test, based on two concentrations of quality control materials [35.5 and 104.8 U/mL] was 4.8 % to 5.4 %. The upper 95^th percentile cutoff for healthy pre- and post-menopausal women is 35 U/mL.

The method used to measure [HE4] in serum is a research-only, double monoclonal ELISA [RK®] obtained from Fujirebio Diagnostics, Inc. [New Jersey]. The assay was performed according to the manufacturer’s recommendations and results are reported as pM based on the manufacturer’s calibration. There is no internationally recognized HE4 reference standard available at this time. The lowest calibrator concentration is 29 pM. The dynamic range of the assay is 15 to 900 pM and the reportable range can be extended by dilution with manufacturer-provided diluent. Study specimens were tested in duplicate, on separate plates [each plate contains a calibration curve]. The imprecision [% CV] of the test, based on two concentrations of manufacturer-supplied quality control materials averaging 47.1 and 390.4 pM, was 3.8% and 3.4 % [n = 36 assays], respectively. The imprecision [% CV] of the test determined by measurement of an in-house serum pool averaging 32.4 pM was 8.7% [n = 36 assays].

Immunohistochemistry for p53 and tumor necrosis

A monoclonal antibody to p53 was used to localize p53 protein (OP43, Oncogene Science, Cambridge, MA), as previously described(16). A positive score required strong immunostaining obscuring nuclear detail. The percentage of cells staining positive was estimated and recorded for each case.

Statistical Analysis

RAPID ELISA analyses were performed in duplicate and average values were used in analyses. Differences between cases and controls were assessed by chi-square tests. To assess the added value of p53-antibodies along with CA 125 and HE4 to discriminate cases from benign disease and healthy controls, we constructed ROC curves and calculated areas under the curve. Associations between clinical characteristics and p53 antibody detection among cases were tested using logistic regression adjusted for age and Jewish ethnicity. Overall survival (OS) was determined from the date of surgery to the date of death or the date last seen in the medical record. We used Kaplan-Meier plots and Cox proportional hazards models to test the associations between marker levels and mortality. All p-values were two-sided. Statistical analyses were performed using SPSS 14.0 software (SPSS Inc, Chicago, IL) and SAS (SAS Institute Inc., Cary, NC).

RESULTS

Detection of high-titer p53-specific autoantibodies by RAPID ELISA

Antibodies to wild-type p53-GST fusion proteins are detected in the sera of a subset of breast cancer patients by RAPID ELISA(8). In ovarian cancer, p53 antigen is specifically detected by RAPID ELISA at serum titers ranging from 1:600–1:48,600, compared to antibodies to the control antigen EBNA-1-GST and vector control (Supplemental Figure 1A, control curves superimposed, and data not shown). Serum was used at a dilution of 1:300 for all subsequent ELISA analyses. Inter-assay coefficients of variation of RAPID ELISA is 3.9%(14). Both p21-GST and p53-GST are readily expressed and detected by immunoblotting with anti-GST Mab (Supplemental Figure 1B, left), but only p53-GST is detected with the ovarian cancer patient serum (Supplemental Figure 1B, right).

Detection of p53 antibodies in benign and malignant ovarian disease

Sera were collected preoperatively from women undergoing surgery for a pelvic mass. Control sera and questionnaires were collected from healthy women in the Boston area with no history of cancer. The serous cases were primarily stage III/IV (95%). An initial 30 cases of invasive serous cancer and their matched controls were randomly selected as a training set for the detection of p53-AAb by RAPID ELISA, and a cut-off value of 13.1 × 10⁶ was established (mean of the controls + 2 standard deviations (S.D.)). Using this cut-off, p53-AAb were detected in 13 of 30 cases (43%) and 0 of 30 control sera (p<0.001) (Figure 1A). Table 1 shows the age distribution, menopausal status, and sample collection details of the cases and controls selected for these studies. Cases and controls did not differ in age, race, menopausal status, year of blood collection, or length of storage. When the p53-AAb cut-off value was applied to the entire case and control groups, antibodies were seen in 25/60 (41.7%) of serous cases, 4/30 (13.3%) of non-serous cases, 3/30 (10%) of benign disease cases, and 10/120 (8.3%) of controls (combined p=0.0002) (Table 1). We observed a lower frequency of p53-AAb in non-serous ovarian cancer (41.7% in serous carcinoma vs. 13.3% in non-serous cases), consistent with a lower frequency of p53 mutations(17). 3/30 (10%) of benign ovarian disease controls were positive for p53-AAb. This is similar to detection of p53 protein overexpression in 7% of benign ovarian tumors(18). Within the cohort of serous carcinomas, all three cases of early-stage serous carcinoma (one case of stage I and two cases of stage II disease) were positive for p53-AAb.

Figure 1. p53 autoantibodies are highly specific biomarkers in serous ovarian cancer.

A. Sera derived from 30 ovarian cancer patients and 30 age-matched healthy women were tested for p53-specific antibodies by RAPID ELISA. The cut-off value (mean signal + 2 S.D. of the controls) is shown as a dotted line (13.1 × 10⁶). B. Distribution of p53 autoantibodies in serous cases and controls. The signal intensity for serous cases (n=60) and all controls (n=120) are shown as a percentage of the total sera. The distribution of p53-AAb signal intensity for controls (open bars) is a unimodal distribution and the cases show a bimodal distribution (filled bars).

Table 1.

Characteristics of cases and controls.

	Serous cases (N=60)	Non-serous cases (N=30)	Benign disease (N=30)	Controls (N=120)	p-value
	N (%)	N (%)	N (%)	N (%)
Age
42–56	20 (33.3)	24 (80.0)	17 (56.7)	62 (51.7)
57–63	17 (28.3)	3 (10.0)	8 (26.7)	27 (22.5)
63–71	23 (38.3)	3 (10.0)	5 (16.7)	31 (25.8)	0.99*
Menopausal status
Pre	8 (13.3)	14 (48.3)	10 (33.3)	40 (33.3)
Post	52 (86.7)	15 (51.7)	20 (66.7)	80 (67.7)	0.28*
Race
White	58 (98.3)	25 (86.2)	27 (90.0)	118 (98.3)
Non-white	1 (1.7)	4 (13.8)	3 (10.0)	2 (1.7)	0.05*
Year of specimen collection
2001–2002	16 (26.7)	8 (26.7)	10 (33.3)	34 (28.6)
2003–2004	24 (40.0)	12 (40.0)	7 (23.3)	41 (34.4)
2005–2008	20 (33.3)	10 (33.3)	13 (43.3)	44 (37.0)	0.97*
Length of storage
<4 years	20 (33.3)	10 (33.3)	13 (43.3)	44 (36.7)
4–5.7 years	24 (40.0)	12 (40.0)	7 (23.3)	36 (30.0)
>5.7 years	16 (26.7)	8 (26.7)	10 (33.3)	40 (33.3)	0.57*
p53 status
Negative	35 (58.3)	26 (86.7)	27 (90.0)	110 (91.7)
Positive	25 (41.7)	4 (13.3)	3 (10.0)	10 (8.3)	<0.0001†
CA 125 (mean, 95% CI)	775 (500, 1200)	84 (50, 140)‡	22 (13, 35)	12 (11, 14)	<0.0001§
HE4 (mean, 95% CI)	503 (379, 669)	--	60 (52, 69)	61 (58, 65)	<0.0001§

^*Chi-square test comparing all disease groups to controls.

^†Chi-square test comparing serous cases to controls. p=0.40 and 0.77 comparing non-serous cases and benign disease subjects to controls, respectively.

^‡CA 125 values from medical records.

^§t-test comparing serous cases to controls.

p<0.0001 and 0.02 comparing non-serous cases and benign disease subjects to controls, respectively. CA 125 and HE4 levels were not measured for the 30 non-serous cases and their matched controls.

The distribution of p53-AAb for invasive serous cases (n=60) compared to all controls (n=120) is shown in Figure 1B and reveals that the distribution is unimodal in controls, with a peak at 9.08 × 10⁶ units. Cases, in contrast, have a bimodal distribution, with a similar baseline peak at 9.08 × 10⁶ units, and a second peak at 30.3 × 10⁶ units. A more stringent cut-off of 16.4 × 10⁶ units (mean normals + 3 S.D.) increases specificity from 91.7 to 96.7 but decreases sensitivity from 41.7 to 33.3.

Comparison of p53 antibodies with CA 125 and HE4 biomarkers

Serum CA 125 and HE4 levels were obtained for the case and controls groups, and are shown in Table 1. Not surprisingly, CA 125 levels were markedly elevated in these (advanced) serous cases vs. controls (mean 775 vs. 12, p<0.0001). Similarly, HE4 levels were elevated in serous cases vs. controls (mean 503 vs. 61, p<0.0001). Non-serous cases had lower levels of CA 125 (mean 84), as expected. To determine if p53-AAb have additive value as a screening biomarker for serous carcinoma beyond CA 125 and HE4, ROC curves were constructed comparing all three biomarkers. In this dataset, both CA 125 (AUC=0.99) and HE4 (AUC=0.98) were superior biomarkers at discriminating cases vs. controls, in comparison to p53-AAb (AUC=0.69), and p53-AAb added very little to CA 125 and HE4 in combination (Figure 2A). Nor did we find evidence that p53-AAb (AUC=0.64) added to CA 125 (AUC=0.94) or HE4 (AUC=0.97) in predicting benign disease from invasive serous cases (Figure 2B).

Figure 2. Comparison of ROC curves of p53 autoantibodies, CA 125, and HE4.

A. Sera derived from 60 serous ovarian cancer patients and 90 age-matched healthy women were tested for CA 125 (blue), HE4 (red), and p53-specific antibodies (green), and ROC curves were calculated. The combination of all three biomarkers is shown in orange. B. Sera derived from 60 serous ovarian cancer patients and 30 women with benign ovarian disease were tested for CA 125 (blue), HE4 (red), and p53-specific antibodies (green), and ROC curves were calculated.

Detection of p53 antibodies in the setting of low CA 125

In this cohort, CA 125 is elevated in over 95% of cases, due to selection of patients with serous carcinomas undergoing surgery. To determine if p53-AAb have potential additive benefit beyond CA 125 for the detection of serous carcinomas, 20 sera were identified from patients with serous carcinoma who had low CA 125 levels (median 40, range 15–76.7). These cases were matched by age and stage with 20 sera with high CA 125 levels (median 2116, range 718–23,010). p53-AAb were detected in 6/20 sera (30%) in the low CA 125 cohort, and 11/20 (55%) in the high CA 125 cohort. Of the 6 sera with p53-AAb in the low CA 125 cohort, three had stage I/II and three had late stage III/IV serous carcinoma, with a median CA 125 level median of 32.4 (range 23–76.7).

p53 antibodies correlate with improved survival in serous ovarian cancer

Table 2 compares the clinical characteristics of serous cases who did not have p53-AAb (n=35) compared to those who had p53-AAb (n=25). No recognized risk factors for ovarian cancer such as pregnancies, oral contraceptive use, or talc use were significantly different between the groups. All patients received standard surgical and adjuvant therapy for advanced serous carcinoma. HE4 and CA 125 levels were not associated with the detection of p53-AAb. There was a tendency for more recent cases to have p53-AAb (p=0.05). However, no significant correlation was found between level of antibodies and length of storage (r = −0.14, p = 0.28) for serous cases nor for the other groups shown in Table 1 (data not shown). Serous cases with p53-AAb were more likely to have a family history of breast cancer (p=0.01) than those who did not. This may reflect the increased frequency of p53 mutations in the tumors of BRCA-mutation carriers(19), but BRCA1 and BRCA2 mutation testing is not available for these cohorts. In 8 cases where we had assessment of p53 protein overexpression by IHC, p53-AAb were detected in 2 of 5 cases whose tumors expressed p53 and none of the 3 cases who were negative by IHC, consistent with prior observations that development of p53-AAb in the sera are strongly correlated with p53 antigen overexpression(9). We saw no difference in association of p53-AAb and tumor necrosis (data not shown).

Table 2.

Association between markers, epidemiologic variables, survival and p53-Ab status among serous cases.

	p53 Negative (N=35)	p53 Positive (N=25)	p-value*
	N (%)	N (%)
CA 125 (mean (95% CI))	761 (432, 1342)	795 (379, 1667)	0.90†
HE4 (mean (95% CI))	466 (315, 690)	560 (362, 868)	0.42†
Year of diagnosis
2001–2002	12 (34.3)	4 (16.0)
2003–2004	14 (40.0)	10 (40.0)	0.17
2005–2008	9 (25.7)	11 (44.0)	0.05
Age
42–56	11 (31.4)	9 (36.0)
57–63	8 (22.9)	9 (36.0)	0.55
63–71	16 (45.7)	7 (28.0)	0.35
OC use
Ever	15 (44.1)	15 (60.0)
Never	19 (55.9)	10 (40.0)	0.15
Parity
Nulliparous	4 (11.8)	4 (16.0)
Parous	30 (88.2)	21 (84.0)	0.84
Genital talc use >5 yrs
No	26 (76.5)	19 (76.0)
Yes	8 (23.5)	6 (24.0)	0.92
Jewish ethnicity
No	31 (88.6)	19 (76.0)
Yes	4 (11.4)	6 (24.0)	0.40
Smoking
Never	15 (44.1)	16 (64.0)
Ever	19 (55.9)	9 (36.0)	0.19
BMI
<20	1 (2.9)	3 (12.5)
20–24.9	17 (50.0)	9 (37.5)
25–29.9	9 (26.5)	8 (33.3)
≥30	7 (20.6)	4 (16.7)	0.77
Tubal ligation
No	26 (81.2)	21 (84.0)
Yes	6 (18.8)	4 (16.0)	0.46
Personal history of cancer
No	31 (91.2)	18 (72.0)
Yes	3 (8.8)	7 (28.0)	0.10
Family history of breast cancer
No	31 (96.9)	16 (64.0)
Yes	1 (3.1)	9 (36.0)	0.01
Immunohistochemistry
Negative	3 (50.0)	0 (0.0)
Positive	3 (50.0)	2 (100.0)	0.46
Survival
Alive	13 (37.1)	17 (68.0)
Dead	22 (62.9)	8 (32.0)	0.03

^*Adjusted for age, Jewish ethnicity, and batch number.

^†p-value for test of trend.

We assessed the association between the presence of p53-AAb and survival. To fully adjust this association for other variables related to survival we conducted a medical record review of the 60 serous cases and obtained information on clinical and treatment characteristics. All cases were grade 3, and 58 (97%) were stage III/IV. Six patients had residual disease >1 cm and 36 had ≤1 cm, but residual disease information was unavailable for the remaining 18 cases. Amount of residual disease was highly correlated with stage. Fifty-nine of the 60 cases received chemotherapy, which was adjuvant platinum with or without paclitaxel in 39 (66%) patients. 37 (63%) received at least 6 cycles of adjuvant chemotherapy. In a model adjusted for age, year of diagnosis, platinum-containing chemotherapy and number of cycles of chemotherapy, stage, and laboratory batch number, we found that p53-AAb were associated with significantly improved survival (HR=0.57, 95% CI (0.33, 0.97), p=0.04) (Table 3). The association was similar when additionally adjusted for CA125 and HE4 (data not shown).

Table 3.

Hazard ratios and 95% CIs for mortality serous among cases

	HR (95% CI)	p-value
Log transformed CA 125	1.11 (0.80, 1.53)	0.53
Log transformed HE4	0.81 (0.49, 1.34)	0.41
Log transformed continuous measure for p53	0.56 (0.38, 0.93)	0.03
Age	1.00 (0.95, 1.06)	0.99
Year of diagnosis	1.07 (0.83, 1.39)	0.60

	HR (95% CI)	p-value
Log transformed continuous measure for p53	0.57 (0.33, 0.97)	0.04
Age	0.97 (0.92, 1.03)	0.36
Year of diagnosis	1.06 (0.80, 1.41)	0.67
Adjuvant platinum with or without paclitaxel	0.81 (0.32, 2.07)	0.66
<6 cycles of first chemo regimen	1.59 (0.44, 5.72)	0.48
Unknown number of cycles of first chemo regimen	2.53 (0.81, 7.87)	0.11
Stage (<4 vs. 4)	2.73 (1.10, 6.78)	0.03

^*Adjusted for batch number.

Figure 3 illustrates the Kaplan-Meier curves for the two groups. Overall median survival time was 36.8 months (25^th percentile: 22.2 months; 75^th percentile: 68.6 months). Thirty-one patients have died (median survival: 28.5 months) and 29 were alive at last follow-up (median survival 67.2 months).

Figure 3. Overall survival is increased in patients with p53 antibodies.

Kaplan-Meier curves comparing overall survival between p53 antibody-positive (red) and antibody-negative (blue) patients.

DISCUSSION

The early diagnosis of ovarian cancer is associated with lower morbidity and longer overall survival(20), but the majority of ovarian cancers are detected at advanced stages. Serum CA 125 is used as a biomarker for screening high-risk women, but only 50% of early-stage ovarian cancer is detected(21, 22). There remains an urgent need for biomarkers for early detection of ovarian cancers, as well as prognostic biomarkers to help guide therapeutic decisions. Several antigens, including MMP1(23), cytokines(24, 25), plasminogen activator receptor(26), osteopontin(6), MUC16, MMP7(27), B7-H4(28), HE4, and kallikreins(28, 29) have been examined as potential biomarkers for the early detection and prognosis of ovarian cancer(30). Of these, HE4 has shown promise in addition to CA 125(15).

p53 antibodies as biomarkers for early detection

We observed that 41.7% of patients with serous ovarian cancer have p53-specific antibodies. This frequency is higher than published reports of ovarian cancers (16–30% frequency(11, 12)). This is likely due to patient selection, as our results pertain largely to late stage serous carcinomas, which have a high rate (70–84%) of p53 mutations and overexpression(31). Because this assay uses mammalian IVTT-generated recombinant protein, there may also be differences in antigenic structure between antibody assays. In contrast to serous ovarian cancer, other ovarian cancer subtypes have a lower frequency of p53 mutations, including endometrioid (28%), mucinous (16%), and clear cell carcinomas (10–16%) (17, 19). We confirmed that the frequency of p53-AAb is lower (13.3%) in other ovarian cancer subtypes. No significant differences between cases and controls were observed for endometrioid (p=0.39), mucinous (p=0.53), and clear cell carcinomas (p=0.91, data not shown). Similarly, we confirmed that p53-autoantibodies are detected at a low frequency in patients with benign ovarian disease (10%). However, we determined that there was limited additive benefit of p53-AAb to the biomarkers CA 125 and HE4 for discrimination of ovarian cancer vs. healthy normal women, or of cancer vs. benign ovarian disease in this cohort. Because the majority of specimens in this study were from women with high grade, advanced invasive serous cancers, the very high frequency of elevated CA 125 and HE4 levels is expected, so that an additive benefit of any diagnostic biomarker in this setting would be minimal. However, p53-AAb were also detected in 30% of patients selected with false-negative CA 125 levels, suggesting that it may still be of benefit as a biomarker for detection and monitoring for this subset of patients.

p53 antibodies as biomarkers for improved prognosis

Several prior studies have examined the role of p53-AAb as prognostic biomarkers in ovarian cancer (11–13, 32–35). Early studies showed no correlation with overall survival(33, 34), or association with poor prognosis(36, 37). More recently, in a study of pre-surgical serum from 104 women with ovarian cancer, p53-AAb were associated with a favorable overall survival (51 months vs. 24 months)(11). In contrast, in an extensive study of 233 patients, 15.9% of patients had p53-AAb, which was not associated with any difference in disease-specific survival, even when selected for advanced stage patients(12). A recent study of 130 patients showed an association between p53-AAb and p53 mutation, but not survival(13). Our data agree with the results of Goodell et al., that p53-AAb are associated with a modest favorable prognosis for serous ovarian cancers. We have also shown in multivariate analysis that detection of p53-AAb is independent of other clinicopathologic risk factors for ovarian cancer, as well as independent of the biomarkers CA 125 and HE4. Although Table 2 suggested the possibility that p53-AAb might be affected by length of storage of the specimen, antibody levels did not correlate significantly with duration of storage. Nevertheless we included year of diagnosis as an adjustment variable in our model (Table 3); and the association between presence of antibodies and survival remained. We also adjusted for other treatment related variables, such as clinical stage, platinum-containing chemotherapy, and number of cycles of adjuvant chemotherapy, and found that the association between antibodies and survival remained.

The development of p53-AAb is strongly associated with overexpression of p53 protein and high-grade tumors(10, 12, 35, 38). In two small studies in breast cancer, p53-AAb more commonly developed in the setting of p53 point mutations (n=13(39)), and with mutations that were associated with p53/HSP70 complexes(40). Our data are consistent with the finding that p53-AAb are restricted to patients with tumors that express high levels of p53. The impact of p53 overexpression on prognosis is mixed, but has been associated with reduced progression free survival (PFS) and overall survival (OS) in ovarian cancer (41–43). In a large study of 783 ovarian carcinomas, p53 overexpression (53%) was associated with shorter overall survival in univariate analysis, but was only an independent factor for low-intermediate grade cancers(44). Similarly, in a large multicenter study of 476 patients, p53 expression was not an independent prognostic factor of PFS or OS(45). Our data suggest that p53-AAb identify a cohort of patients with a more favorable prognosis within this high-risk cohort. A prospective study that incorporates full p53 mutational analysis, IHC protein expression, and serum p53-Ab detection is needed to confirm these findings.

p53 antibodies as biomarkers of anti-tumor immunity

It is not known whether p53-AAb indicate functional immunologic control of tumor, or whether they are surrogate but inert biomarkers for differences in tumor biology. p53-AAb may induce amplification of p53-specific T cell immunity, which has been detected in tumor-infiltrating lymphocytes in ovarian cancer(46). Since infiltrating T cells are correlated with improved prognosis in ovarian cancer(3), and cell-mediated immunity is associated with improved tumor responses to chemotherapeutic agents(47), it is likely that p53-AAb may help localize cytotoxic T cells to sites of minimal residual disease after chemotherapy.

The development of p53-AAb could also reflect host differences in the induction of antibody responses in general. However, we see no evidence of impaired immune competence to the infectious antigen EBNA-1 in patients who don’t make antibodies to p53 (data not shown and(8)). Several functional polymorphisms are associated with enhancement of antibody-dependent cytotoxicity or complement function and autoimmunity. The FCGR2A H131R SNP(48, 49) regulates Fc binding, and patients who are homozygous for FCGR2A H/H have increased ADCC and rituximab responses(50). Similarly, the CFH Y402H SNP regulates complement activation(51), and patients with CFH Y/Y have increased trastuzumab responses(52). In a subset analysis of 16 patients, no association between p53-AAb and the H131R or the CFH Y402H SNP was found (data not shown).

Future applications of p53-Ab biomarkers

The potential clinical application of p53-AAb as a biomarker for the early detection of ovarian cancer is limited by the lower sensitivity in early stage disease(8, 11) and the non-specific induction of p53-AAb in different p53-overexpressing cancers. However, recent evidence points to early mutations in p53 in the distal fallopian tubes as candidate precursor lesions for ovarian carcinogenesis(16), suggesting that p53-specific biomarkers may still have utility for ovarian cancer diagnosis in the subgroup of BRCA patients who appear to be susceptible to tumors originating in the Fallopian tubes. We did detect p53-AAb in stage I as well as stage II (n=3) serous ovarian cancers, but testing p53-AAb in defined pre-diagnostic serum collections is required to determine whether p53-AAb have utility prior to clinical diagnosis. To increase both sensitivity and specificity, antibodies to panels of tumor antigens that include p53 have been proposed(3, 4). Recent developments in proteomics technologies, including spotted-protein microarrays(5), reversed-phase protein arrays(53), phage display(6, 54), and programmable protein arrays(8) hold great promise for the future development of autoantibody biomarkers for cancer detection.