Importance of spondin 1 and cellular retinoic acid binding protein 1 in the clinical diagnosis of ovarian cancer

Abstract

Background: Diagnosis of ovarian cancer is often delayed because of subtle symptoms and a lack of a specific, sensitive test useful for the general population. The majority of cases are diagnosed at late stages, after the tumor has metastasized and implanted on many other abdominal organs and cavity surfaces. A paucity of prognostic markers makes it difficult to define which tumors will act aggressively and shorten survival. Hence, it is imperative to have new screening tests for diagnosis of ovarian cancer at earlier stages, prior to metastatic progression. Diagnosis at these early stages will dramatically increase the overall survival of women with ovarian cancer. Material and Methods: Based on previously published literature on proposed molecular cell markers in ovarian carcinoma, we sought to validate the overexpression of two genes (cellular retinoic acid Binding Protein, CRABP-1, and spondin 1) through immunohistochemistry. Results: We verified the overexpression of spondin 1 in ovarian cancer. Expression of cellular retinoic acid Binding Protein, CRABP-1 in whole ovarian cancer tissue sections was higher than in the TMA tissue cores. Conclusion: Our results thus demonstrate that spondin 1 is a useful marker for ovarian cancer; additionally, the high percentages of tumors that are positive for spondin 1 make it an ideal target for therapy. CRABP-1 was not expressed at high levels in any subtype of ovarian cancer, making it a poor marker.

Introduction

Ovarian cancer is the most lethal gynecological cancer, with an estimated 204,499 new cases and 124,860 deaths world-wide in 2002 [1]. Ovarian cancer accounts for 3% of all the female cancers diagnosed, but 5% of the female cancer deaths [2]. Although the death rate has decreased slightly over the years (9.51 deaths per 100,000 women in 1991 to 8.62 deaths per 100,000 women in 2005), the five year survival rate for ovarian cancer patients is still dismal [2]. When detected while still localized, the five year survival is 93%; however this drops to 71% for patients with regional metastasis, and to 31% for patients with distant metastasis [2].

Unfortunately, fewer than 20% of cases are detected when the tumor is still localized; over 65% are detected after distant metastasis has occurred [2]. Because of the late diagnosis in many cases, the overall five year survival is only 46%, compared to an 89% five year survival for breast cancer [2].

Currently there are no diagnostic markers that are both specific and sensitive enough to monitor the general population for ovarian cancer. To reach a positive predictive value of 10%, (1 true cancer detected for every 10 tested), a test must be 75% specific and 99.6% sensitive [3]. CA125 (cancer antigen 125, MUC16), a glycoprotein, has been widely studied for detection of ovarian carcinoma [4], but it is neither sensitive nor specific enough to screen the general population. Approximately 20% of early stage ovarian carcinomas do not express CA125, thus the most sensitive it can be is 80% [3,4]. Additionally, CA125 can be elevated in many other conditions, such as endometriosis, menstruation, ovarian cysts, and pregnancy, reducing the specificity of the test [3,5]. Many imaging technologies have been studied to determine their effectiveness at detecting early stage ovarian cancer. MRIs, CAT scans, as well as transvaginal ultrasounds have all been used [3,5]. At present, none are able to detect early stage tumor [3].

In recent years, much focus has been placed on biomarkers for early stage ovarian cancer detection. Many groups have looked in patient specimens (sera, plasma, and urine) to find overexpressed proteins that would specifically be from the tumor and could be used in a screening test [3,5-7]. Often, these are compared to and used in conjunction with CA125 levels to get improved sensitivity and specificity. However, when combining more than one test, while the sensitivity usually goes up, the specificity is decreased, as there will be different false positives for each marker.

CRABP-1 is a cytoplasmic carrier protein for retinoic acid, regulating its ability to enter the nucleus and bind to nuclear receptors. Previous studies have reported a decrease in expression of CRABP-1 in other cancers, including ovarian, thyroid, and esophageal cancers, primarily through hypermethylation of the gene [8-10]. Recently, it has been shown to be increased in ovarian cancer tissues by IHC analysis [11].

Based on the aforementioned literature evidence, availability of monoclonal antibodies, and previous reference to being deregulated in ovarian cancer patients [8], we evaluated the expression of CRABP-1 and spondin 1 in 500 ovarian cancer cases.

Materials and methods

Tissue sections

FFPE tissue blocks were obtained from the Changhai Hospital of Second Military Medical University’s Tissue Procurement Facility after IRB approval and patient consent. Diagnoses were determined by the surgical pathologist and confirmed by two independent pathologists. Five micron sections of the FFPE tissues were adhered onto slides, then dried at 30°C overnight and for 30 min at 60°C immediately prior to staining to remove any residual water. Sections of normal ovary were only included in analysis if surface epithelium cells were present.

Tissue microarrays

Tissue microarray slides of 500 cases of ovarian cancer (containing 0.6 mm duplicate core samples for each patient) were provided by the Changhai Hospital of Second Military Medical University. Patients included in the TMA were chosen based on the fact that they were optimally debulked at initial surgery and had no macroscopic residual disease, thereby increasing the proportion of early stage cases on the TMA relative to the general population. None of the patients received neoadjuvant therapy, but all received platinum based chemotherapy following surgery. Because the 500 cases included on the ovarian TMA were originally diagnosed up to 18 years ago and the classification of ovarian cancer histologies has shifted over the years, care was taken to ensure that the current diagnostic criteria for subclassification of ovarian carcinoma based on cell type were uniformly applied [12,13]. Hematoxylin and eosin stained slides for all cases were reviewed by a gynecologic pathologist to confirm diagnosis, stage, tumor cell type, and grade prior to TMA inclusion; samples displaying multiple cell types (mixed tumors) were excluded from the study. Complete details about the cohort used for these TMAs are provided in Table 1. Patients were followed for a median of 4.6 (0.1-18) years after the initial surgery. Relapse was defined by visible disease progression by a variety of diagnostic modalities including radiology and physical exam.

Table 1.

Subtype and CRABP-1 scores of tissue microarrays

Subtype	Median Age (range)	Median PreOP*	CRAB-1 Positive Overall
Serous (n=212)	59.6 (33.5-86.0)	108 (0-23,000)	33 (15.6%)
Endometroid (n=125)	54.1 (29.4-88.1)	130 (8-13,000)	10 (8.0%)
Clear Cell (n=132)	55.0 (28.1-89.0)	64 (4-7,750)	9 (6.8%)
Mucinous (n=31)	56.4 (25.4-76.7)	45 (7-650)	1 (3.2%)
Total (n=500)	56.6 (25.4-89.0)	98 (0-23,000)	53 (10.6%)

^* PreOP, preoperative.

Immunohistochemical staining of tissues

Tissue sections were deparaffinized and rehydrated through a series of xylene and ethanol washes. Antigen retrieval was performed in a 1x solution of Reveal citrate buffer (Biocare, Concord, CA) for 20 minutes for and spondin 1, but not for CRABP-1. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 10 minutes. Slides were incubated with mouse anti-human CRABP-1 monoclonal antibody (clone C-1; AbCam, 1/750), normal mouse IgG1 (clone 3-5D1-C9; AbCam, 1/2000) overnight at 4°C or chicken-anti-human spondin 1 polyclonal antibody (clone ab14271; AbCam, at 5 ug/ml), or normal chicken IgY (ab50579, AbCam) for 1 hour at room temperature. All primary antibodies were diluted in 1:5 Sniper (Biocare):PBS. Slides were washed 2 times with PBS for 5 minutes, then incubated with biotinylated horse-anti-mouse secondary antibody (Vector Laboratories, Burlingame, CA) per the manufacturer’s protocol or biotinylated rabbit anti-chicken secondary antibody (1/500, ab6752, AbCam), followed with an avidin:biotin complex (Vector Laboratories) for 30 minutes. Staining was visualized with Vulcan Fast Red (Biocare) for CRABP-1 and spondin 1 following the manufacturer’s directions. Slides were examined by a pathologist and assigned a score of 0 (no staining); 1 (<10% of neoplastic cells staining); 2 (10-50% of neoplastic cells staining); or 3 (>50% of neoplastic cells staining). Positive control tissues and staining patterns were: retina (CRABP-1, cytoplasmic), ovary (spondin 1, using two ovarian cancer tissues positive and one tissue negative by Western immunoblotting, not shown, cytoplasmic, extracellular).

Results and discussion

CRABP-1 expression in the control tissue, eye, was localized to the retina (not shown). In IHC staining of whole sections of tissues, CRABP-1 was 60% sensitive for serous tumors (Figure 1, 15 of 25 tissues positive), 25% sensitive for clear cell tumors (1 of 4 tissues positive), and 92% specific (23 of 25 normal tissues negative). However, only a small number of TMA tissues stained positively for CRABP-1 overall (10.6%) (Table 1). Expression was highest in the serous subtype (15.6%), followed by endometrioid (8.0%), clear cell (6.8%) and mucinous (3.2%) subtypes (Figure 2A). Expression was significantly higher in serous tumors compared to the other subtypes (p=0.04). No significant difference in CRABP-1 staining was present by stage (Figure 2B, p=0.23), Silverberg grade (Figure 2C, p=0.14), relapse-free survival (p=0.09), or overall survival (p=0.14).

Figure 1.

CRABP-1 IHC staining in (A) normal ovarian tissue (ovarian surface epithelial cells), and (B) ovarian cancer (tumor). Staining in ovarian cancer cells was localized to the cytoplasm. All pictures taken at 20x magnification.

Figure 2.

CRABP-1 staining of ovarian cancer tissue microarrays. A: Percent of each subtype of ovarian cancer staining positively for CRABP-1; B: Percent positive for CRABP-1 by stage; C: Percent positive for CRABP-1 by Silverberg grade.

The majority of TMA samples were positive for spondin 1 (Table 2). Both endometrioid and mucinous tumor stained 100% positive for spondin 1, followed by serous (99.5%) and clear cell (99.2%) tumors. There was no significant difference in staining between subtypes (Figure 3A, p=0.56). There were no significant differences in spondin 1 staining by stage (Figure 3B, p=0.83), Silverberg grade (Figure 3C, p=0.31), or relapse-free survival (not shown, p=0.25). Overall survival was inversely associated with expression of spondin 1 (not shown, p=0.03); however, as only two of the 500 TMA samples were negative for spondin 1, this result should be interpreted with caution.

Table 2.

Subtype and spondin 1 scores of tissue microarrays

Subtype	Median Age (range)	Median PreOP*	CRAB-1 Positive Overall
Serous (n=210)	59.6 (33.5-86.0)	108 (0-23,000)	209 (99.5%)
Endometroid (n=123)	54.1 (29.4-88.1)	130 (8-13,000)	123 (100%)
Clear Cell (n=132)	55.0 (28.1-89.0)	64 (4-7,750)	131 (99.2%)
Mucinous (n=31)	56.4 (25.4-76.7)	45 (7-650)	31 (100%)
Total (n=496)	56.6 (25.4-89.0)	98 (0-23,000)	494 (99.6%)

^* PreOP, preoperative.

Figure 3.

Spondin 1 staining of ovarian cancer tissue microarrays. A: Percent of each subtype of ovarian cancer staining positively for spondin-1; B: Percent positive for spondin-1 by stage; C: Percent positive for spondin-1 by Silverberg grade.

The results above demonstrate that spondin 1 is a useful marker for ovarian cancer; additionally, the high percentages of tumors that are positive for spondin 1 make it an ideal target for therapy. CRABP-1 was not expressed at high levels in any subtype of ovarian cancer, making it a poor marker. It was earlier reported that mRNA expression of CRABP-1 was increased in ovarian cancer tissues compared to normal ovaries [7], however we were not able to confirm this at the protein expression level. It is possible that while the mRNA of CRABP-1 is increased in ovarian cancer tissues, the protein may not be translated or it may be degraded quickly.

When determining the optimal conditions for staining for CRABP-1 using FFPE tissues, we noted that expression of CRABP-1 in tumors was at times heterogeneous; some areas of the tumor stained and others did not. It is possible that the small 0.6 mm punch biopsies of the TMAs do not represent the entire tumor and more tumors may be positive if larger sections of tissue were stained.

Spondin 1 was expressed at high levels in nearly all of the tissues on the TMA. No tissue is listed as positive control on Protein Atlas (accessed on 3/24/13, http://www.proteinatlas.org/gene_info.php?ensembl_gene_id=ENSG00000152268). Thus, we choose to use ovarian cancer tissues we found positive (T050599) and negative (T050586) by Western immunoblotting (not shown). Although optimal conditions for IHC staining were determined using these tissues, it appears that the TMA staining had a high level of background staining present. This could be due to different fixation protocols for the tissues. Fixation of tissues can alter the antigenic site recognized by antibodies, thus differing methods of fixations could change how well an antibody would be able to detect its target. Therefore, these results should be interpreted with caution.

In summary, we sought to validate the overexpression of two genes in ovarian cancer. We verified the overexpression of spondin 1. Expression of CRABP-1 in whole ovarian cancer tissue sections was higher than in the TMA tissue cores. There are a few possible explanations for this discrepancy. First, the staining on the whole tissue sections was heterogeneous; some tumor cells stained positively while others did not stain within the same tissue section. Because of this, it is possible that the tissue cores on the TMA sampled areas that were negative although other areas may have been positive, decreasing the overall sensitivity of the staining. A second possibility is that like CRABP-2, CRABP-1 can readily diffuse out of cells, leading to a decrease in staining. Although conditions for spondin 1 staining were worked out, staining in the TMAs was highly non-specific, with only two of the samples on the TMA were negative for spondin 1. It is highly unlikely that only two samples are negative, however, and thus the results should be interpreted with caution.

Disclosure of conflict of interest

None.