Differences in Prognostic Molecular Markers between Women Over and Under 45 Years of Age with Advanced Ovarian Cancer

Invasive epithelial ovarian cancer is typically a disease of postmenopausal women, who account for 80% of such patients. This disease, particularly advanced-stage disease, is uncommon in women in the reproductive-age group (≤45 years). Furthermore, young age at diagnosis is reported to be a favorable prognostic factor in women with advanced epithelial ovarian cancers (1, 2). In our previous study, we compared the survival rates of reproductive-age women with a control group of older patients with stage III and IV epithelial ovarian cancers (3). The two groups were well balanced with respect to extent of primary cytoreductive surgery, stage of disease, tumor histology, number of cycles of chemotherapy, and family history of cancer. The 5-year survival rate of 48% and median survival duration of 54 months in the reproductive-age group was significantly better compared with those of 22% and 34 months in the older group (P = 0.003; ref. 3).

It is unclear why there are marked differences in prognosis for different age categories. Age-associated clinical determinants such as a better performance status and a tendency for clinicians to treat younger patients more intensively may contribute to this difference but do not completely explain it (3). There are limited studies that have determined the prognostic significance of biological tumor markers in reproductive- versus nonreproductive-age women with advanced ovarian cancers (4). Accordingly, we analyzed the expression of CD34, p53, and HER2 in our study group to assess this possibility.

CD34 is an antigen present on endothelial cells. It is a reliable marker of angiogenesis and predicts for disease-free survival in ovarian cancer patients (5). As in many solid tumors, including breast, lung, and head and neck tumors, angiogenesis has been demonstrated to be an important prognostic factor in epithelial ovarian carcinoma (6, 7, 8, 9, 10, 11, 12, 13, 14). P53 is a tumor suppressor gene product that responds to cellular damage by inhibiting cell cycle progression and inducing cell death. Mutations in p53 lead to inhibition of apoptosis and uncontrolled cell growth. Altered p53 function has been associated with poor outcomes in many tumors such as breast, lung, hematopoietic, and gynecological malignancies including ovarian cancers (15, 16, 17, 18). HER2 is an oncogene product related to the epidermal growth factor receptor. This tyrosine kinase receptor is involved in cellular proliferation. Overexpression of HER2 has been described as a factor associated with poor prognosis in ovarian carcinomas and other cancers (19, 20, 21). The aim of this study was to evaluate the relative frequency of microvessels and p53 and HER2 expression in women over and under 45 years of age to determine whether variations in these age groups might influence prognosis.

All women (≤45 years) with stage III and IV epithelial ovarian carcinoma diagnosed between 1984 and 2001 (n = 52) were identified from tumor registry databases at two hospitals. Those with borderline and mixed tumors were excluded (3). A group of 52 controls (>45 years) who were well balanced for extent of primary cytoreductive surgery, stage of disease, tumor histology, number of cycles of chemotherapy, and family history of cancer were selected. Data were collected from hospital charts, clinic records, and tumor registry databases at the University of California, Irvine Medical Center (Orange, CA) and Memorial Medical Center (Long Beach, CA). Institutional review board approvals were obtained from both institutions. After an earlier study (3) found an improved prognosis in the ≤45 year age group compared with the >45 year age group, we retrospectively obtained paraffin blocks from these patients. Of the 46 blocks available for analysis, 24 were from the ≤45 year age group, and 22 were from the >45 year age group. Tissue blocks from 58 patients were either not available (n = 50) or not adequate (n = 8) for analysis.

Immunohistochemical studies were performed using formalin-fixed, paraffin-embedded tissues on a Ventana automated immunohistochemistry system (Ventana Medical System, Tucson, AZ). Four-micrometer–thick histologic sections were prepared for immunostaining and hematoxylin and eosin staining. Prediluted (Ventana) monoclonal antibodies anti-CD34, anti-p53, and anti-HER2neu were used, and the avidin-biotin complex peroxidase method was used per standard protocol. Positive and negative control sections were included in each staining run.

Each slide was evaluated for expression of tumor markers by investigators blinded to the patients’ ages and clinical outcomes. Microvessel density (MVD) was assessed by light microscopy according to the original method by Weidner et al. (6). One block of tumor per case was stained. We chose the blocks by selecting the one that contained the most amount of viable ovarian carcinoma. The area of highest neovascularization was found by scanning the entire slide at low magnification (×40 and ×100). We then selected stromal areas within the tumor or adjacent to its advancing margin for highest areas of neovascularization (so-called “hot spot”). After the hot spot was identified, individual microvessels were counted on a ×400 field. Any brown-stained endothelial cell or cluster of cells clearly separate from an adjacent microvessel was considered a countable structure. The presence of a lumen was not required to identify a brown-stained image as a microvessel. Each count was reported as the highest number of microvessels identified in a single field on ×400 magnification. To achieve assay reproducibility, we only evaluated slides that were adequately stained based on comparisons with positive controls. To reduce subjectivity and interobserver variability in the evaluation of MVD, p53, and HER2 expression, an expert pathologist and two gynecologic oncologists using a discussion microscope performed the analyses simultaneously, and all had to agree on the microvessel count and grade of staining in each specimen. For both p53 and HER2, we evaluated the entire slide and selected all areas of invasive carcinoma to perform a semiquantitative analysis to determine the percentage of tumor cells staining positive for p53 and HER2. Anti-p53 was used to identify an accumulation of p53 gene product. Only nuclear staining was reported as a positive result. We attributed a grade of p53 expression (range, 1–4) based on the percentage of tumor cells stained compared with the total number of tumor cells. We defined grade 1 as 0% to 25% of tumor cells stained, grade 2 as 26% to 50% of tumor cells stained, grade 3 as 51% to 75% of tumor cells stained, and grade 4 as 76% to 100% of tumor cells stained. We defined low and high expression of p53 as grade 1–2 and grade 3–4, respectively. We graded the transmembrane protein expression of HER2 from 0 to 3, with grade 0 defined as the absence of staining, and grade 3 defined as 100% staining of cancer cells. We defined negative and positive expression of HER2 as grade 0–1 and grade 2–3, respectively.

Kaplan-Meier survival and Cox Proportional Hazard analyses were used to determine survival differences and identify predictors for outcome. A two-sided P < 0.05 was considered significant.

From an earlier study of 104 women with stage III or IV ovarian carcinoma, 46 tissue blocks were available for molecular marker analysis [24 patients in the ≤45 year age group and 22 patients in the >45 year age group (3)]. The mean age was 41 years (range, 22–45 years) in the younger group compared with 61 years (range, 47–85 years) in the older cohort. The patients’ clinical profiles are shown in Table 1. The younger women were not significantly different from the older patients with respect to stage of disease, extent of initial surgical treatment, number of cycles of chemotherapy treatment, and histologic cell type. They were also comparable with respect to race and family history of breast and gynecologic cancer. Eastern Cooperative Oncology Group performance status was significantly different between the two groups: 87.5% (21 of 24) of the younger women had a performance status of 0 compared with only 36.4% (8 of 22) in the older age group (P < 0.0005). The distribution of tumor grade based on patient age revealed significantly more grade 1 and 2 tumors versus grade 3 tumors in the younger group (P = 0.04). Despite this difference in distribution of cases based on tumor grade, tumor grade was not an important prognostic factor in multivariate analysis in this patient population (hazard ratio adjusted for age and MVD = 1.6; P = 0.3). The 5-year survival rate and median survival were 23.1% and 28.5 months in older patients compared with 36.8% and 44.4 months in the younger cohort, respectively (P = 0.3). Although our previous larger study showed a difference in survival between these two age groups (3), this subset did not maintain that relationship. The median follow-up after primary surgery was 37 months (range, 3–115 months). The mean ± SE MVD per ×400 field in the women over 45 years of age was 10.3 ± 1.5 microvessels compared with 16.1 ± 2.1 microvessels in the younger cohorts (P = 0.03). However, the expression of p53 and HER2 was not significantly different between the two age groups (P = 0.13, p53; P = 0.49, HER2; Table 2).

We determined the prognostic significance of tumor markers independent of patient age. In the overall study group, the median number of microvessels per ×400 field was 11 (range, 0–43 microvessels per ×400 field). The median survival of patients with MVD ≤ 11 microvessels (n = 22) compared with MVD > 11 microvessels (n = 24) was 23.9 and 59.7 months, respectively (P = 0.001; Fig. 1). When we analyzed the survival of the two age groups with respect to MVD count, MVD ≤ 11 microvessels was associated with a decreased survival in women >45 years of age [23.9 months (n = 14) versus 58.8 months (n = 8); P = 0.03]. In younger patients, the median survival of those with MVD ≤ 11 microvessels was 21.3 months (n = 7) versus 54.6 months (n = 17) in those with MVD > 11 microvessels (P = 0.09). In the total study group, patients with tumors that had a high expression of p53 had a median survival of 28.6 months compared with 51.7 months in those with low expression of p53 (P = 0.09; Fig. 2). Lastly, the median survival of women with tumors that have a positive expression of HER2 was 23.9 versus 38.6 months in those who have a negative expression of HER2 receptors (P = 0.15; Fig. 3).

To determine whether various tumor histologies may have influenced our results, we evaluated the prognostic significance of MVD in patients with serous tumors only. Lower MVD remained as a poor prognostic factor in overall stratified Kaplan-Meier analysis in those with serous histology (P = 0.012). Moreover, positive HER2 expression was also a poor prognosticator in women with serous tumors (P = 0.011). However, the prognostic significance of p53 expression remained unchanged after excluding the nonserous tumors (P = 0.391). In the younger patients with serous tumors, MVD ≤ 11 microvessels and positive HER2 expression were associated with poorer outcomes (MVD, P = 0.05; HER2, P = 0.02). On multivariate analysis, women with lower MVD had significantly worse survival after adjusting for age, stage, and grade of disease (adjusted hazard ratio = 3.4; 95% confidence interval, 1.4–8.1).

Our previous report confirmed the observation that age is a significant independent prognostic factor for survival in women with advanced epithelial ovarian cancer (3). To determine the outcomes of reproductive-age versus nonreproductive-age women with advanced ovarian cancer, we selected 45 years as a demarcation because the majority of women over 45 years old do not have reproductive function. The objective of this study was to test the hypothesis that differences in molecular markers might account for some of the improved survival seen in younger patients with advanced ovarian cancer. We approached this hypothesis by analyzing biological markers of angiogenesis (CD34, endothelial cell antigen), cell growth regulation (p53, tumor suppressor gene product), and cellular proliferation (HER2, oncogene product) in our previously described study group.

Our initial analysis of 104 patients showed a median survival advantage of 20 months for the women ≤45 years of age versus the women > 45 years of age (54 versus 34 months; ref. 3). In the subgroup of 46 patients with available pathological specimens, however, the difference in survival between the younger and older cohorts no longer reached significance, likely due to the small sample size. Nevertheless, even in this group of 46 patients, the younger cohort had a survival advantage of >15 months over the older group. Although the survival difference was not statistically significant, it is clinically relevant, suggesting that these findings may be extrapolated to a larger group. Moreover, acquiring the genetic status of these women (such as BRCA mutations) may have strengthened our study. However, this information was not available because this is a retrospective study, and genetic testing was not yet considered standard of care.

Investigators have studied the prognostic significance of various biological tumor markers, including angiogenesis (5, 6, 7, 9), p53 (15, 16, 17, 18), and HER2 expression (19, 20, 21). Weidner et al. (6) have shown that increased angiogenesis is associated with increased risk of metastatic disease in breast cancer patients. Furthermore, Gasparini et al. (8) demonstrated that tumor vessel density significantly impacted progression-free and overall survival in breast cancer patients. Because most patients with ovarian cancers die of complications from local relapse instead of blood-borne metastasis, whether observations on breast cancer patients are applicable to ovarian cancers is uncertain. Data supporting the hypothesis that angiogenesis is a predictor of prognosis in these patients include those of Schoell et al. (7), who demonstrated that ovarian cancers of patients who died of cancer had markedly more angiogenesis than tumors of survivors. Other investigators have also found a relationship between increased angiogenesis and poor survival (10).

On the contrary, studies have also revealed that increased angiogenesis is an important predictor for improved prognosis in ovarian cancer. For example, in clear cell cancers of the ovary, higher MVD was independently correlated with improved progression-free survival (11). Likewise, others have found that increased MVD was associated with an increased response to chemotherapy in patients with ovarian carcinomas (9). Similarly, Gadducci et al. (22) also showed that patients with advanced ovarian cancers with an elevated MVD had an improved response to paclitaxel/platinum-based chemotherapy compared with those with lower MVD. In their study, elevated MVD was an independent prognostic variable for improved progression-free and overall survival (22). These observations may be explained by the fact that higher tumor vascularity might permit better drug and oxygen availability, resulting in a therapeutic benefit, whereas tumor hypoxia leads to a poor response to chemotherapy (12). On the other hand, some investigators have found that MVD has no prognostic significance in ovarian cancer patients (13, 14, 23), possibly reflecting the hypothesis that new vessels formed in association with angiogenesis have structural and functional abnormalities with chaotic flow and intermittent stagnation, making it difficult to deliver drugs uniformly throughout the tumor (24). Thus, the role of angiogenesis in ovarian cancer prognosis and response to therapy remains unclear.

In this current study, MVD (per ×400 field) of ≤11 microvessels predicted for worse survival, independent of patient age. In fact, the median survival of patients with MVD ≤ 11 microvessels was 36 months less than those with MVD > 11 microvessels (P = 0.001). Thus, the prognostic disadvantage in the patients over 45 years of age compared with the younger ones appeared to be a function of decreased angiogenesis in the older group. Previous studies have identified a significant correlation between low MVD and tumor hypoxia (25). Hypoxia in ovarian and other carcinomas is associated with a poor prognosis (26, 27, 28, 29). Thus, the poor outcome of patients with tumors with low MVD in this report may be associated with tumor hypoxia.

There are several methods used in the detection of tumor angiogenesis, including immunohistochemical analysis of tumor neovessels, molecular expression of angiogenic cytokines, measurement of circulating cytokines, and functional expression of proteins involved in neovessel formation. The data generated from these techniques are correlative and provide an insight into the process of angiogenesis but are not considered definitive. The numerous techniques used in the assessment of angiogenesis may partly explain the varying results demonstrated by different investigators on the prognostic significance of MVD. We selected the technique first introduced by Weidner et al. (6). Most investigators use the vascular hot spot tumor regions that contain the greatest number of vessels for analysis and have demonstrated this to be a reliable and reproducible technique. Furthermore, Hollingsworth et al. (5) compared various antibodies including Ulex, vWF, and CD34 and found that the CD34 antibody stained most consistently and reproducibly among this group, making it the preferred marker for analyzing angiogenesis. Accordingly, we elected to use CD34 as a marker for our study.

In our study, women with tumors that have p53 mutations had a median survival of 23 months less than those who did not have this gene mutation. Furthermore, women with tumors that overexpressed HER2 in our study had a median survival of 15 months less than women with tumors that underexpressed this receptor. Although the prognostic value of both tumor markers did not reach statistical significance, likely due to the small sample size, the differences in median survival between those with and without the expression of p53 and HER2 are clinically relevant and are in agreement with previously published reports (15, 16, 17, 18, 19, 20, 21).

Recent results from a large phase III trial on metastatic colorectal cancer showed that patients who received chemotherapy with an antiangiogenic agent, bevacizumab, had a significantly longer survival than those treated with chemotherapy alone (30). The results from our study have strong clinical implications toward the treatment of advanced ovarian carcinoma with antiangiogenic agents, particularly in women over 45 years of age. Because patients over 45 years of age had a decreased survival with significantly lower MVD compared with younger women in our study, antiangiogenic therapies may have a variable effect in older women with advanced ovarian carcinoma. Therefore, it is important to develop reliable surrogate markers to determine the efficacy of angiogenic inhibitors in specific subsets of patients. In this manner, antiangiogenic therapies can be individualized toward patients who can benefit most from this form of targeted biological therapy.