Ovarian carcinoma (OC) is the fifth most frequent female cancer type and the fourth most frequent cause of death from cancer among women in Denmark. At the time they are diagnosed with OC, approximately 70% of patients have advanced disease. It is believed that loss of tumor suppressor gene activity plays an important role in the origin and progression of OC and other malignancies. Loss of heterozygosity (LOH) may be detected in individuals heterozygous for an allele and is associated with loss of function of tumor suppressor genes.
The polymorphic marker regions (TP53, CACNLB1, D18S58, DXS538, and DXS454) were amplified by polymerase chain reaction followed by separation using gel electrophoresis before LOH was identified. In total, 160 women with primary epithelial OC were included in the study.
Univariate analyses showed significant differences in survival between patients who had advanced OC with LOH or with retention using the microsatellite markers DXS454 (P = 0.04) and DXS538 (P = 0.01). Multivariate Cox regression analysis that included all patients showed that DXS454 (relative hazard [RH] = 3.5; P = 0.002; 95% confidence interval [95% CI], 1.6–7.8), radicality of primary surgery (RH = 5.5; P < 0.0001; 95% CI, 2.7–11.1), and serum tetranectin level (RH = 0.8; P = 0.009; 95% CI, 0.7–0.9) were independent prognostic factors for survival. Multivariate Cox regression analysis restricted to patients with International Federation of Obstetrics and Gynecology Stage III–IV disease showed that DXS454 (RH = 3.4; P = 0.007; 95% CI, 1.4–8.1), radicality of primary surgery (RH = 5.4; P < 0.0001; 95% CI, 2.2–12.9), and serum tetranectin level (RH = 0.8; P = 0.042; 95% CI, 0.7–1.0) were independent prognostic factors.
Ovarian carcinoma (OC) is the fifth most frequent female cancer type and the fourth most frequent cause of death from cancer among women in Denmark.1, 2 At the time of diagnosis, approximately 70% of patients have advanced disease (International Federation of Gynecology and Obstetrics [FIGO] Stage III or IV). Published 5-year survival rates for patients with OC range from > 80% for patients with Stage I OC to < 20% for patients with Stage III and IV OC.3–5 The high frequency and poor prognosis of OC emphasizes the need for both additional and better prognostic factors.
The molecular genetic events underlying ovarian neoplasms are complex and are understood poorly. It is believed that loss of tumor suppressor gene activity plays an important role in the origin and progression of malignant tumors. A commonly described mechanism leading to the loss of tumor suppressor gene function, originally reported by Knudson,6 requires the occurrence of two events: First, one of two alleles must contain a mutation (somatic or germline), rendering this allele incapable of producing a functional tumor suppressor protein. Second, the remaining allele must be lost, resulting in an overall absence of tumor suppressor gene function.7 This loss may be detected in individuals heterozygous for a marker closely linked to the tumor suppressor gene (loss of heterozygosity [LOH]) and is characteristic of tumor suppressor genes. LOH frequencies up to 80% in OC have been described on multiple chromosomes, most commonly involving chromosomes 3, 6, 9, 11, 17, 18, and 22, suggesting the involvement of many different tumor suppressor genes.7–13 However, limited information on LOH at chromosome X is available, and LOH has been reported in 40–60% of the informative OC cases tested.9, 14, 15 Some studies suggest that LOH is more common in ovarian tumors with more aggressive biologic behavior and in more advanced ovarian tumors.7, 16 Most LOH studies on patients with OC have involved relatively small number of tumors, a wide range of different chromosomal loci, and variable patient characteristics. Furthermore, genetic differences within the study population are possible, which may explain the discordant findings in the published studies. It has been hypothesized that tetranectin (TN) plays a role in proteolytic processes is an important factor in the ability of malignant cells to proliferate, infiltrate normal tissue, and metastasize.17–21 In patients with cancer, the decreases in serum or plasma TN (s/p-TN) levels may reflect the proteolytic activity of the tumor, and low s/p-TN levels may be due to absorption of TN from the blood to the tumor site for the purpose of proteolysis. This hypothesis is supported by the immunohistochemical findings of high extracellular TN concentrations in malignant tumors in combination with low p-TN values and a poor survival for OC patients with this combination.22 Previously, it was reported that TN should be included as a prognostic factor in OC studies evaluating new prognostic markers.23
Based on previous reports that LOH occurs frequently on chromosome 17p and 17q,7, 9, 24 chromosome 18q,25 and chromosomes Xp and Xq,9, 14, 15 we decided to perform LOH analyses with 5 microsatellite markers in 160 Danish women with primary OC: TP53 (17p13.1), CACNLB1 (17q21–q22), D18S58 (18q22–q23), DXS538 (Xp11.21–p21.1), and DXS454 (Xq21-q23). We were able to perform this study using a set of samples from the Malignant Ovarian Cancer Study (“MALOVA”) collection in Denmark.
MATERIALS AND METHODS
The material was from the first 160 consecutive patients with primary epithelial OC who were included in the MALOVA study and for whom blood samples and corresponding tissue samples were available, including 37 patients (23%) with Stage I OC, 12 patients (8%) with Stage II OC, 93 patients (58%) with Stage III OC, and 18 patients (11%) with Stage IV OC, representing the similar frequencies between stages as found in the MALOVA study (Stage I, 28% of patients; Stage II, 10% of patients; Stage III, 53% of patients; and Stage IV, 9% of patients). The median patient age was 60.5 years (range, 35–79 years). The MALOVA study is a multidisciplinary Danish study on primary OC, covering epidemiology (lifestyle factors), biochemistry, and molecular biology with the objective of identifying risk factors and prognostic factors for OC. The MALOVA study is described in detail elsewhere.26–29 Briefly, preoperative blood samples as well as tumor tissue samples were obtained from the patients. Histopathologic classifications of the ovarian tumors were based on the typing criteria of the World Health Organization. Pathology reports and tissue specimens were collected from the different participating hospitals. One pathologist who specialized in ovarian tumors reviewed the tissue specimen without knowledge of the original diagnosis. Subsequently, the reviewed diagnosis was compared with the original diagnosis. In terms of invasiveness, agreement between the original hospital diagnosis and the review diagnosis was present in 98% of the specimens. Through the reviewing procedure, it was possible to define the histologic type of tumor more precisely in 14% of cases. FIGO stages were obtained from clinical records and were reviewed by two gynecologists, both of whom specialized in OC. Furthermore, in the clinical records, patients were classified according to whether they underwent radical surgery with no macroscopic residual tumor or nonradical surgery with macroscopic (visible) residual tumor after surgery. In this article, we used the reviewed information. The study was approved by the Scientific Ethical Committee in the study area (KF01-384/95).
In Denmark, all inhabitants have a unique personal identification number, which is used universally in the Danish society. These identification numbers, which comprise information concerning date of birth and gender, are registered in the computerized Danish National Central Population Register. The register contains information regarding, for example, dates of death and emigration. All patients in the current study were traced in this register and were followed until the date of death, emigration, or March 2001, whichever came first. Women who died during follow-up were linked to a Danish hospital reference system, and information regarding their last hospital admission was obtained. The relevant hospital files were collected and scrutinized, and the causes of death were established. At the end of follow-up, a total of 80 patients with OC had died from OC (median follow-up, 19 months; range, 1–61 months), and 80 patients with OC remained alive (median follow-up, 53 months; range, 28–74 months). Five patients with Stage I OC died of OC (median follow-up, 57 months; range, 9–57 months) and 32 patients with Stage I OC remained alive (median follow-up, 60 months; range, 28–74 months). Two patients with Stage II OC died (median follow-up, 47 months; range, 44–51 months) and 10 patients with Stage II OC remained alive (median follow-up, 57 months; range, 32–71 months). Of the patients with Stage III OC, 55 patients died of OC (median follow-up, 22 months; range, 1–61 months) and 38 patients remained alive (median follow-up, 50 months; range, 28–71 months), whereas all 18 patients with Stage IV OC patients died (median follow-up, 45 months; range, 1–46 months).
Tissue samples that were obtained at primary cytoreductive surgery were examined closely by the MALOVA pathologist. All tissues used for DNA extraction were scored to contain > 80% tumors cells with no calcification. Furthermore, two different tissue samples from each patient were analyzed, making sure of a reproducible LOH result in each case. Chromosomal DNA was extracted from peripheral leukocytes and from 112 snap-frozen OC tissues using a 5% chelex solution, as described previously30; subsequently digested with proteinase K; extracted with phenol/chloroform; and finally precipitated with 70% ethanol. From the remaining 48 OC tissues, chromosomal DNA was extracted from archival tissue, as described previously.31 The quality of the chromosomal DNA was tested on an agarose gel before further use. Of this, 1 μL (containing 50–100 ng) was used in polymerase chain reaction (PCR).
Five polymorphic markers, all known to cover regions with a high degree of heterozygosity, were tested by PCR: TP53 (17p13.1), CACNLB1 (17q11.2–q22), D18S58 (18q22–q23), DXS538 (Xp11.21–p21.1), and DXS454 (Xq21–q23). Amplification was performed in a Hybaid Thermocycler (U.K.). PCR was performed in a 20 μL volume containing approximately 50–100 ng DNA; PCR buffer (10mmol/L Tris-HCl, [pH 8.3]; 50 mmol/L KCl; 1.5 mmol/L MgCl2; and 0.01% weight/volume gelatin); 0.1–0.5 μmol/L primers (forward and reverse); 200 μmol/L each of deoxycytidine triphosphate (dCTP), deoxyguanine triphosphate, deoxythymidine triphosphate, and deoxyadinosine triphosphate; 2 μCi [32P] dCTP (6000 Ci/mmol/L); and 0.1 U Ampli Taq Gold DNA polymerase (Perkin Elmer, Foster City, CA). The target DNA initially was denatured at 95 °C for 5 minutes. Thirty-five cycles of amplification were performed for each primer pair under following conditions: TP53 at 95 °C for 30 seconds, 58 °C for 30 seconds, and 68 °C for 30 seconds; CACNLB1 at 95 °C for 30 seconds, 53 °C for 30 seconds, and 68 °C for 30 seconds; D18S58 at 95 °C for 30 seconds, 52 °C for 30 seconds, and 68 °C for 30 seconds; DXS538 at 95 °C for 30 seconds, 50 °C for 30 seconds, and 68 °C for 30 seconds; and DXS454 at 95 °C for 30 seconds, 50 °C for 30 seconds, and 68 °C for 30 seconds. For all reactions, a final extension step was performed for 10 minutes at 72 °C.
Detection of PCR products
Ten microliters of amplified DNA solution were added to 10 μL of loading dye (95% formamide, 12 mmol/L ethylenediamine tetraacetic acid [pH 8.0], 0.025% bromphenol blue, and 0.025% xylene cyanol). The samples were heated for 5 minutes at 99 °C then immediately cooled on ice. Five microliters of the denatured samples were electrophoresed on a 7% polyacrylamide sequencing gel (Sequagel, National Diagnostic, U.K.), and alleles were detected by autoradiography. LOH was considered present when diminution (visually) of > 75% was estimated in 1 allele of the tumor DNA compared with the paired constitutional DNA. Allelic imbalance was scored when 1 allele appeared 50–75% weaker than the other allele compared with the products of the normal DNA. The appearance of different sized alleles in the tumor DNA compared with the normal DNA was scored as microsatellite instability. All results were scored independently by two experienced physicians who had no knowledge of clinical data or patient outcomes. If the scoring differed, then each case was discussed with a conclusive score for each patient. Examples of tumors that showed LOH, retention, or noninformative results are shown in Figure 1.
Blood samples were left on the clot at room temperature, and serum was separated by centrifugation at 3000 × g for 20 minutes at room temperature. Serum was stored in aliquots at −80 °C.
Serum levels of tumor-associated antigen were determined using a CA 125 immunoassay (CA125-EIA; Abbott Laboratories, Chicago, IL), according to the manufacturer's instructions. The within-assay coefficient of variation (CV), using a control sample of 30 U/mL, was 6.6% (n = 60 samples), and the between-assay CV was 6.2% (n = 10 samples).
An enzyme-linked immunoadsorbent assay based on the polyclonal rabbit antihuman TN antibody A371 (DAKO A/S, Glostrup, Denmark) was used for quantitation of TN.32, 33 The within-assay CV was measured at 5.1% (n = 123 samples), whereas the between-assay CV was measured at 8.8% (n = 25 samples) at 10.9 mg/L.
Statistical comparisons between groups were performed using the rank sum test or the Mann–Whitney U test. Survival differences were estimated by using the method of Kaplan and Meier and were tested with a log-rank test.
The relative hazard (RH), as the exponential function of the respective regression coefficient with the 95% confidence interval (95% CI), was derived from multivariate Cox analyses that were used to test jointly the relative importance of variables as predictors of survival.34 The factors were FIGO stage (Cox 1 analysis: 19 patients with Stage I–II OC vs. 48 patients with Stage III–IV OC; Cox 2 analysis: 48 patients with Stage III–IV OC), the two markers DXS454 and DXS538, radicality of primary surgery, histologic tumor type, patient age, serum TN level, and serum CA 125 level. These analyses were performed using the SPSS 11.5 statistical software package (SPSS, Chicago, IL).
LOH Status and Prognosis
Univariate survival analyses were limited to patients with advanced stage OC (FIGO Stage III–IV), because the number of events (i.e., deaths) were too small in the Stage I–II subgroups. Significant differences in survival were found when comparing loss to retention of the DXS454 marker in the 65 informative Stage III–IV OC cases (P = 0.04) (Fig. 2).
Similarly, a shorter survival was observed for patients with a loss in DXS538 (80 patients with Stage III/IV OC; P = 0.01) (Fig. 3). In contrast, there was no significant difference noted between patients who had tumors with a loss or retention of the marker TP53 (89 patients with Stage III–IV OC; P = 0.73), the marker CACNLB1 (85 patients with Stage III–IV OC; P = 0.58), and the marker D18S58 (74 patients with Stage III–IV OC; P = 0.49). Furthermore, we performed a univariate survival analysis on patients with FIGO Stage III–IV OC of the serous adenocarcinoma type. A significantly reduced survival was observed in patients with loss at the marker DXS538 (55 patients with Stage III–IV OC, serous adenocarcinoma; P = 0.0095) and the marker DXS454 (43 patients with Stage III–IV OC, serous adenocarcinoma; P = 0.03) compared with patients who had retention of those markers. No significant difference in survival were observed between patients who had tumors with loss or retention of the markers D18S58 (59 patients with Stage III/IV OC, serous adenocarcinoma; P = 0.69), CACNLB1 (62 patients with Stage III–IV OC, serous adenocarcinoma; P = 0.12), and TP53 (64 patients with Stage III–IV OC, serous adenocarcinoma; P = 0.64).
In the Cox 1 analysis, we included 67 patients with OC (19 patients with Stage I–II OC vs. 48 patients with Stage III–IV OC), who had informative results at both marker DXS454 and marker DXS538, which showed prognostic value in the univariate analyses. This multivariate Cox regression revealed that the independent prognostic factors were DXS454 (LOH vs. retention; RH = 3.5; P = 0.002 [95% CI, 1.6–7.8]), radicality of primary surgery (RH = 5.5; P < 0.0001 [95% CI, 2.7–11.1]), and s-TN (RH = 0.8; P = 0.009 [95% CI, 0.7–0.9]). FIGO stage (P = 0.85), histologic tumor type (P = 0.90), patient age (P = 0.96), DXS538 (P = 0.52), and serum-CA 125 (P = 0.94) were found to have no independent prognostic value.
Finally, in the Cox 2 analysis, which was restricted to patients with advanced OC (48 patients with Stage III–IV OC) who had results that were informative for the markers DXS454 and DXS538, the independent prognostic factors were DXS454 (RH = 3.4; P = 0.007 [95% CI, 1.4–8.1]), radicality of primary surgery (RH = 5.4; P < 0.0001 [95% CI, 2.2–12.9]), and s-TN (RH = 0.8; P = 0.042 [95% CI, 0.7–1.0]). DXS538 (P = 0.94), histologic tumor type (P = 0.90), patient age (P = 0.91), and serum-CA 125 (P = 0.65) were not found to have any prognostic value.
The Frequency of LOH in Relation to FIGO Stages and the Histopathologic Tumor Types
The frequency of LOH ranged from 20% at the DXS538 marker on chromosome Xp to 39% at the TP53 marker on chromosome 17p (Table 1). Table 2 presents the frequencies of LOH according to disease stage. Most tumors that displayed LOH were confined to patients with Stage III OC and also comprised the largest number of patients. For two of the five investigated markers, significant differences in frequencies of LOH occurred with increasing OC stage (TP53, P = 0.002; CACNLB1: P = 0.003). For the remaining 3 markers (D18S58, DXS454, and DXS538), the frequency of LOH with increasing OC stage did not reach statistical significance (D18S58: P = 0.41; DXS454: P = 0.07; and DXS538: P = 0.34) (Table 2).
Table 1. Summary of the Results Obtained from Loss of Heterozygosity Analyses Using Microsatellite Polymerase Chain Reaction
LOH: loss of heterozygosity; NOS: not otherwise specified.
Loss of heterozygosity analysis (the number of samples with Loss of heterozygosity/the number informative cases tested).
Papillary adenocarcinoma NOS
Clear cell neoplasms
Frequency of LOH in Relation to Serum TN Levels, Serum CA 125 Levels, and Age in Patients with Stage III OC
We were able to compare the pattern of LOH with levels of two serum biomarkers, CA 125 and TN, which have a known relation to the prognosis of patients with OC. Only 2 of the studied chromosome regions demonstrated a significant association between LOH frequency and the median levels of s-TN in patients with Stage III OC (TP53: P = 0.026; CACNLB1: P = 0.005) (Table 4). None of the comparisons between the median s-CA 125 levels in patients with Stage III OC that displayed LOH or retention yielded significant differences (Table 4).
Table 4. Serum Tetranectin, Serum CA 125, and Age in Patients with Stage III Ovarian Carcinoma that Displayed Loss of Heterozygosity or Retention
P values were calculated using the Mann–Whitney U test.
The age at the time of diagnosis in patients with Stage III OC who displayed LOH at the CACNLB1 marker on chromosome 17q (median age, 66 years; range, 45–77 years) differed significantly from the age of patients with Stage III OC who had retention (median age, 56 years; range, 35–77 years; CACNLB1: P = 0.018). No significant differences in the age of patients were demonstrated for the other 4 microsatellite markers (TP53: P = 0.69; D18S58: P = 0.94; DXS538: P = 0.94; and DXS454: P = 0.99) (Table 4).
In the current study, LOH analysis was performed to evaluate the importance of five microsatellite markers from three different chromosomes as prognostic markers for OC. The markers analyzed were TP53 (17p13.1), CACNLB1 (17q21–q22), DXS538 (Xp11.21–p21.1), and DXS454 (Xq21–q23). Results from one of the microsatellite markers, D18S58 (chromosome region 18q22–23) did not correlate with any of the tested clinical variables and, thus, is not discussed further.
The overall frequencies found in this study were lower compared with the frequencies reported in the literature. One explanation may be the use of a stricter definition to confirm LOH. This criteria for LOH was chosen to ensure a low person-to-person variability when scoring LOH. When comparing the observed frequencies in this study with reported frequencies, it is of relevance to consider the different scales used.
To our knowledge, this is the first detailed study large enough to clarify the prognostic value of LOH using microsatellite markers at chromosomes 17, 18, and X. We found that LOH at Xq (DXS454) was associated with reduced survival in patients with OC, as illustrated in both univariate and multivariate analyses, and that LOH at DXS538 (Xp11.21–p21.1) was associated with poor survival in patients with OC, as illustrated in the univariate analysis but not in the multivariate analysis. In this study, we used our selected combination of variables and compared the relative strength of each variable with respect to prognosis.
Disease stage in patients with OC was not found to be an independent prognostic factor in the multivariate Cox analysis. One explanation may be that very few early-stage tumors were informative for both markers. However, 19 early-stage tumors were informative for both markers and were included in the Cox analysis. Therefore, one-third of the included tumors were early-stage OC. However, this is not why disease stage was of no prognostic value in the Cox analysis performed in this study. Because of this finding, it may be important to evaluate the DXS454 marker as a prognostic factor in other, larger studies.
Because OC is a female malignancy, it has been speculated often that there may be a specific role for the X chromosome in ovarian carcinogenesis.9 If a tumor suppressor gene is located on the X chromosome, and if females with a germline mutation in one copy of that tumor suppressor gene experienced nonrandom X chromosome inactivation, then some or all of the tissues of such women may lack the wild-type suppressor gene function. Buller et al. found that nonrandom X chromosome inactivation may be a predisposing factor for developing an invasive ovarian tumor.35, 36 Our study, which was performed on tissues and corresponding blood samples from women with OC, may support this hypothesis; furthermore, our findings suggest that loss of the X chromosome may be a primary or early event in ovarian tumorigenesis, because LOH at DXS454 (Xq21–q23) occurred in both early and late stages of OC. These findings raise the possibility of a tumor suppressor gene locus on the X chromosome (21q–23q) that may have prognostic value.
We have not identified any other studies that investigated the markers DXS454 and DXS538 in patients with OC with respect to survival. Furthermore, due to limitations of sample size, few previous studies included prognostic evaluation of LOH markers at all.
Manderson et al. described LOH on both arms of the X chromosome in 75 patients with OC, among whom 62 patients had Stage III or IV disease.37 Those authors reported a 7% frequency of LOH of Xp compared with 36% at Xq. Buekers et al. investigated LOH on the X chromosome using 6 different microsatellite markers in 81 OC patients with no information of stage and found that LOH was more frequent on Xp, with the highest frequency of LOH demonstrated at the region Xp 22.2-3 (37.7%).15 Shelling et al. described that LOH of Xp was 38%, whereas loss at Xq was 29%.9 In the current study, the frequency of LOH at Xp was 20%, and the frequency of LOH at Xq was 29%, consistent with the most recent of the studies by Manderson et al.37
Different frequencies of LOH can be demonstrated between the various histologic subgroups of OC with borderline tumors that display a lower overall frequency of LOH compared with invasive tumors.13 We observed greater frequencies of LOH in serous adenocarcinoma compared with mucinous adenocarcinoma, findings that are in agreement with earlier published studies13, 16, 25 and that add further to the growing evidence that the pathways of molecular carcinogenesis and progression may differ among the histologic subtypes of OC.
Manderson et al. investigated whether there was significant difference in the frequency of LOH with respect to age. They found one locus at 3p14 for which the frequency of LOH increased significantly with age.37 Pieretti and Turker38 reported that the frequency of LOH of an entire chromosome 17 homologue increased with older age at the time of diagnosis of OC, and Pieretti et al.13 found that patients with chromosome 17 loss were older than patients without this alteration. We found a significant difference between the median age of patients with LOH and patients with retention in the CACNLB1 locus. One possible explanation for this phenomenon may be that LOH of certain chromosomal loci on chromosome 17 are nonrandom events that reflect the contribution of multiple tumor suppressor genes to a more aggressive phenotype, resulting in poorer survival often being noted in women of older age.39
In the current study, LOH at DXS454 appeared to correlate with shorter survival in Danish patients with epithelial OC.
The authors thank all nurses and physicians in the Gynecology and Pathology Departments for their tremendous work. The authors are grateful to Vibeke Reese for technical assistance.