IGF-II promoter methylation and ovarian cancer prognosis

Purpose The insulin-like growth factor-II (IGF-II) gene has four promoters that produce distinct transcripts which vary by tissue type and developmental stage. Dysregulation of normal promoter usage has been shown to occur in cancer; DNA methylation regulates promoter use. Thus, we sought to examine if DNA methylation varies among IGF-II promoters in ovarian cancer and if methylation patterns are related to clinical features of the disease. Study design Tumor tissue, clinical data, and follow-up information were collected from 215 patients diagnosed with primary epithelial ovarian cancer. DNA extracted from tumor tissues was analyzed for IGF-II promoter methylation with seven methylation specific PCR (MSP) assays: three for promoter 2 (P2) and two assays each for promoters 3 and 4 (P3 and P4).

Methylation was found to vary among the seven assays: 19.3% in P2A, 45.6% in P2B, 50.9% in P2C, 48.4% in P3A, 13.1% in P3B, 5.1% in P4A, and 6.1% in P4B. Methylation in any of the three P2 assays was associated with high tumor grade (P = 0.043), suboptimal debulking (P = 0.036), and disease progression [hazards ratio (HR) = 1.73, 95% confidence interval (CI) 1.09–2.74]. When comparing promoter methylation patterns, differential methylation of P2 and P3 was found to be associated with disease prognosis; patients with P3 but not P2 methylation were less likely to have disease progression (HR = 0.39, 95% CI 0.17–0.91) compared to patients with P2 but not P3 methylation.

This study shows that methylation varies among three IGF-II promoters in ovarian cancer and that this variation seems to have biologic implications as it relates to clinical features and prognosis of the disease.

IGF-II, Promoter methylation, Ovarian cancer prognosis

Insulin-like growth factor-II (IGF-II) is a peptide hormone that regulates cellular proliferation and differentiation via autocrine and paracrine mechanisms, and has been shown to have potent mitogenic and anti-apoptotic effects (Toretsky and Helman 1996; Yu and Rohan 2000). While it is crucial for normal development (Daughaday and Rotwein 1989), IGF-II is also implicated in the development and progression of a variety of cancers (Toretsky and Helman 1996; Yu and Rohan 2000; Khandwala et al. 2000; Pollak et al. 2004; Lu et al. 2006a, b). High levels of IGF-II have been found in tumors of the breast, prostate, bladder, stomach, colon, prostate, and ovary (Lu et al. 2006a, b; Li et al. 1998a; Shiraishi et al. 1998; Lambert et al. 1990; Tennant et al. 1996; Yun et al. 1996). Analyzing expression profiles of 516 genes in ovarian cancer revealed that IGF-II had the highest degree of up-regulation (Sawiris et al. 2002). High levels of IGF-II expression were found to be associated with advanced stage diseases, high grade tumors, and suboptimal debulking results (Sayer et al. 2005). Recent experiments in our laboratory also showed that high IGF-II expression was associated with poor prognosis of the disease (Lu et al. 2006b). The IGF-II gene has four promoters (P1–P4), and each can initiate transcription producing a distinct IGF-II transcript with different 5′-untranslated regions (UTR) (Gray et al. 1987; van Dijk et al. 1991). These transcripts also differ in their translatability and stability (Nielsen and Christiansen 1992, 1995; Holthuizen et al. 1993). Additionally, the four promoter regions may have differences in their capacity to recruit and bind necessary transcription factors. Thus, while the resulting IGF-II proteins will be the same, the quantity of protein produced by the four promoters may not be. IGF-II promoter usage has been shown to be both temporal and tissue specific (Gray et al. 1987; Holthuizen et al. 1993; Irminger et al. 1987; Sussenbach et al. 1993; Mineo et al. 2000). P2, P3, and P4 are active in fetal tissues, but silent after birth; P1 is the only promoter active in the healthy adult liver (Holthuizen et al. 1993; Sussenbach et al. 1993). Dysregulation of IGF-II promoter usage has been found in several cancers, including those of the breast, prostate, liver, and bladder (Mineo et al. 2000; Li et al. 1997; Hartmann et al. 2001). In hepatocellular carcinomas, expression from P1 was lost, while fetal transcripts were shown to be expressed (Li et al. 1997; Hartmann et al. 2001). In ovarian cancer, transcripts from P3 and P4 were detected more frequently, and at significantly higher levels, than transcripts from P1 and P2 (Lu et al. 2006a). DNA methylation is the predominant epigenetic alteration of the mammalian genome, and is known as a mechanism to silence gene expression (Issa 2003). Tumor suppressor genes, DNA repair genes, and genes involved in cancer invasion and metastasis are frequently hypermethylated in cancers (Esteller et al. 2001). Additionally, it has been postulated that hypomethylation of oncogenes may be a mechanism by which expression is increased. Given that altered IGF-II expression and dysregulation of promoter usage has been implicated in ovarian cancer, and that methylation is a mechanism by which promoter usage and expression may be regulated, this study was undertaken to evaluate IGF-II promoter methylation among ovarian cancer patients. We characterized IGF-II promoter methylation among 215 patients diagnosed with epithelial ovarian cancer, evaluated the associations between IGF-II promoter methylation and clinical and pathological features of the disease, and assessed IGF-II promoter methylation in relation to ovarian cancer progression free and overall survival.

Study subjects From a consecutive series of 264 patients who underwent surgery for suspected ovarian tumors in the Gynecologic Oncology Unit of the Department of Gynecology at the University of Turin in Italy between October 1991 and February 2000, we identified 215 women with histologically confirmed primary epithelial ovarian cancer. Patients excluded were those with metastatic tumors of non-ovarian origins (n = 23), borderline tumors of low malignant potential (n = 19), primary ovarian tumors of non-epithelial origin (n = 6), or endometriosis (n = 1). This study was approved by the Institutional Review Board of the University of Turin. The average age of patients at surgery was 57.6 years (SD = 11.5; range 26–82). Follow-up information was collected after surgery through June 2001 for 205 patients (95.3%). Overall survival (OS), defined as the duration of time from the date of surgery to the date of death or last contact if the patient was still alive, ranged from 0.6 to 114.1 months, with a median of 31.1 months. Disease progression-free survival (PFS) was defined as the duration from the date of surgery to the date of first evidence of local recurrence or distant metastasis, or last contact. PFS times ranged from 0.6 to 108.8 months, with a median of 20.6 months. Of the 205 patients with complete follow-up information, 42.4% (n = 87) remained alive without disease progression, 12.7% (n = 26) remained alive after disease progression, 33.2% (n = 68) died after relapse, and 11.7% (n = 24) died of the disease without remission. Disease stage was determined according to the staging criteria of the International Federation of Gynecologists and Obstetricians (FIGO) (FIGO 1987). Of the 214 patients who had stage information, 53 (24.8%) had stage I disease, 12 (5.6%) had stage II, 134 (62.6%) had stage III, and 15 (7.0%) had stage IV. Histological grade and type were determined using the World Health Organization (WHO) criteria (Serov et al. 1973). Well-differentiated grade 1 tumors were identified in 15.9% of the patients (n = 34); grade 2 and poorly differentiated grade 3 tumors were found in 18.7% (n = 40) and 65.4% (n = 140) of the patients, respectively. Histological types included serous (40.0%, n = 86), and non-serous (7.4% clear cell, n = 16, 19.5% endometrioid, n = 42, 8.4% mucinous, n = 18, and 24.7% other epithelial tumors, n = 53). Among the patients who had cytoreductive procedures (debulking), 52.2% (n = 109) achieved optimal results indicating that either no detectable, or less than 1 cm of residual disease remained after surgery, while 47.9% (n = 100) had a suboptimal debulking outcomes indicating that one centimeter or more of detectable residual lesions remained after surgery. The prevailing post-operative treatment for these patients was chemotherapy, which was administered to the majority of patients (n = 182, 84.7%). Chemotherapeutic agents used included adriamycin, carboplatin, cisplatin, cyclophosphamide, epirubicin, and taxol, which were administered in a variety of combinations as previously described (Beeghly et al. 2006). Response to treatment was clinically defined, also as previously described (Beeghly et al. 2006). One hundred and forty-seven patients (68.4%) had a complete response, 38 patients (17.7%) had a partial response, 4 patients (1.9%) had no response (stable disease), 14 patients (6.5%) had disease progression during treatment, and 12 patients (5.6%) had no information available on their treatment response. When analyzed, patients with partial response, no response, and disease progression were classified as one group, and compared to patients who had a complete response. Laboratory analysis Fresh tumor tissue was collected from each patient during surgery. The tissue samples were snap-frozen in liquid nitrogen immediately after surgical resection, and then stored at −80°C until analysis. Two pathologists examined the frozen sections of each tissue specimen to confirm tumor content; the majority of tissue samples contained 80–90% tumor cells. Genomic DNA was extracted from approximately 100 mg of pulverized ovarian tumor tissue and was treated with sodium bisulfite to convert unmethylated cytosines to thymines while leaving methylated cytosine bases unchanged. Sodium bisulfite treatment has been described in detail elsewhere (Herman et al. 1996); briefly, it included DNA denaturing with 2 N NaOH, followed by treatment with 10 mM hydroquinone and 3 M sodium bisulfite. Modified DNA was then purified using the QIAEX II Gel Extraction Kit (QIAGEN, Valencia, CA, USA) and concentrated with Pellet Paint NF Co-Precipitant (Novagen, San Diego, CA, USA). Final suspension volume was in 100 μl of TE buffer. Methylation specific PCR (MSP) was used to assess CpG site methylation within three promoter regions of the IGF-II gene, P2, P3, and P4. A scarcity of CpG sites within P1 limited its evaluation by MSP. Three MSP assays, P2A, P2B and P2C, were developed to evaluate methylation in P2; two assays were developed for both the P3 and P4 regions, P3A, P3B, P4A and P4B, respectively. Relative locations of the MSP assays are shown in Fig. 1. Each MSP assay included two pairs of PCR primers; one specific for methylated and one specific for unmethylated bisulfite converted sequences. Primer sequences are listed in Table 1, along with annealing temperatures and product sizes. Bases in bold indicate a C to T transition due to bisulfite treatment of the DNA, and underlined bases indicate a cytosine base within a CpG site. Table 1. | Primer | Primer sequence 5′→3′ | Ta (°C) | Product (bp) | |---|---|---|---| | P2A, M, F | CGGATTTTTTACGCGGGGATCG | 59 | 127 | | P2A, M, R | CCGAACTTAACG AACGTCCG | || | P2A, U, F | GTTTGTGGATTTTTTATGTGGGGAT TG | 137 | | | P2A, U, R | ACAMCCCAAACTTAACAAACATCCA | || | P2B, M, F | TTAGT CGGT CGTGCGCGTTA | 62 | 138 | | P2B, M, R | CG AACAACG AAAAAAAAACGCTACCG | || | P2B, U, F | AGAATTTTAGT TGGT TGTGTGTGTTA | 148 | | | P2B, U, R | TTCCCA AACAACA AAAAAAAAACACTACCA | || | P2C, M, F | GACGTTAATTTT CGGGGACGTT | 55 | 363 | | P2C, M, R | AACGCGAAATAAAACGAACGTATACGA | || | P2C, U, F | TTTTGGATGTTAATTTT TGGGGATGTT | 373 | | | P2C, U, R | AAAAAAACACAAAATAAAACAAACATATACAA | || | P3A, M, F | CGTAGTCGGTTTTTCGCGCG | 64 | 149 | | P3A, M, R | CG AACG AAACGCGCAACCG | || | P3A, U, F | TGGTGTAGTTGGTTTTTTGTGTGTT | 159 | | | P3A, U, R | CCTAAACACA AACA AAACACACAACCA | || | P3B, M, F | TTTACGAGTATAGGTCGTTCGGGTATT | 60 | 367 | | P3B, M, R | CGCCG AAAAACAAACGAACTACG | || | P3B, U, F | TTTTAGTTTATGAGTATAGGTTGTTTGGGTATT | 380 | | | P3B, U, R | CAACTCCACCA AAAAACAAACAAACTACA | || | P4A, M, F | CGGTAGCGTCGTCGGTTTTTAG | 66 | 208 | | P4A, M, R | CGCGCAAAAAACG AAACCCG | || | P4A, U, F | GTGGGTGGTAGTGTTGT TGGTTTTTAG | 219 | | | P4A, U, R | AAACTCCACACAAAAAACAAAACCCACA | || | P4B, M, F | CGAGATCGCGAGAGGAAGACG | 62 | 217 | | P4B, M, R | CGCG ACCTCTCTCCG AATCG | || | P4B, U, F | GTGTTAGTGAGAT TGTGAGAGGAAGATGGG | 226 | | | P4B, U, R | ACCACAACCTCTCTCCAAATCAAAAACT | Primer name: P promoter region, M methylated, U unmethylated, F forward, R reverse Primer sequence: bases in bold indicate a C to T transition due to bisulfite treatment of the DNA, and underlined bases indicate a cytosine base within a CpG site. If unmethylated, these bases undergo a C to T bisulfite-induced change, but if methylated, are protected from conversion All MSP reactions began with initial denaturing at 94°C for 5 min, and concluded with a final extension at 72°C for 7 min. Annealing temperatures and times varied from assay to assay, ranging from 55 to 66°C and from 45 s to 1 min, respectively. The number of amplification cycles varied, from 35 to 40. Four control samples were included in each batch of MSP analysis: two positive controls (known methylated and unmethylated samples), and two negative controls (un-treated DNA and water). MSP products were visualized on 2.0–2.5% agarose gels and stained with ethidium bromide. Representative reactions for each MSP assay are shown in Fig. 2. Determination of the methylation status of each sample was made without knowledge of patients’ clinical and tumor characteristics, survival outcomes, and results of other MSP assays. A subset of methylated and unmethylated MSP products from each assay was sequenced to confirm methylation status. Statistical analysis MSP results within each region were combined by dichotomizing each promoter as either positive or negative for methylation. Associations between patient or tumor characteristics and IGF-II promoter methylation were analyzed using the χ 2 or Fisher’s exact test where appropriate. Promoter methylation patterns were constructed as follows: Group 1: P2 and P3 both negative for methylation; Group 2: P2 positive for methylation and P3 negative for methylation; Group 3: P2 negative for methylation and P3 positive for methylation; and Group 4: P2 and P3 both positive for methylation. Effects of methylation on PFS and OS were evaluated using Kaplan–Meier survival curves; survival differences were examined using the log-rank test. Cox proportional hazards regression was also used to assess the effect of methylation on ovarian cancer survival; hazards ratios (HRs) and their 95% confidence intervals (CIs) were calculated. Regression analysis was done both with and without adjustment; Covariates included patient’s age at surgery, disease stage, and tumor grade. All statistical tests were two sided, and an alpha value of 0.05 was used to determine significance. Statistical analysis was conducted with SAS vs 9.1 (SAS Institute, Cary, NC, USA).

Figure 1 shows the relative position of the seven MSP assays in three IGF-II promoters, the number of CpG sites covered by each pairs of primers, and the percentage of ovarian cancer samples found to be methylated by each assay. The highest percentage of methylation determined by MSP was found in the most 3′-region of P2 (P2C); 50.9% of tumor samples were positive for methylation. Other regions with frequent methylation included P2B (45.6%) and P3A (48.4%). Methylation in the P2A and P3B regions was less frequent, 19.3 and 13.1% of ovarian cancer samples, respectively. Promoter 4 had the lowest frequency of methylation among the three promoters, only 5.1% in P4A and 6.1% in P4B. As expected, methylation determined by different MSP assays within the same promoter regions was positively associated with each other (data not shown). To best approximate total methylation across a region, each promoter was dichotomized as either positive or negative for methylation for each sample. As shown in Table 2, of the patient samples successfully tested, 64.2% (n = 136) were found to have methylation within promoter 2, 52.1% (n = 111) were found methylation within promoter 3, and 9.8% (n = 21) were found methylation within promoter 4. Further, when methylation patterns of P2 and P3 were evaluated, 23.1% (n = 49) of the samples were negative for methylation in both promoters (Group 1), 24.5% (n = 52) were positive for P2 methylation but not P3 (Group 2), 12.7% (n = 27) were positive for P3 but not P2 (Group 3), and 39.6% (n = 84) positive for methylation in both P2 and P3 (Group 4). Table 2. | IGF-II promoter methylation | Totala | n | % | |---|---|---|---| | Promoter 2—any methylation | 212 | 136 | 64.2 | | Promoter 3—any methylation | 213 | 111 | 52.1 | | Promoter 4—any methylation | 215 | 21 | 9.8 | | P2 and P3 methylation patterns | ||| | Group 1: P2 U P3 U | 212 | 49 | 23.1 | | Group 2: P2 M P3 U | 212 | 52 | 24.5 | | Group 3: P2 U P3 M | 212 | 27 | 12.7 | | Group 4: P2 M P3 M | 212 | 84 | 39.6 | aNumber of samples with informative MSP results Table 3 shows IGF-II promoter region methylation in relation to clinical and pathological features of ovarian cancer, including patient age, disease stage, histological grade, histological type, and results of surgical debulking. Methylation in P2 was associated with poorly differentiated tumors, (P = 0.043) and suboptimal debulking results (P = 0.036). Although not significant, methylation in P3 tended to occur more frequently among patients with better prognostic factors, including earlier disease stage, well-differentiated tumors, and optimal debulking results. Methylation in P4 was not found to vary with any patient or tumor characteristics. Table 3. | Variable | Promoter 2 | Promoter 3 | Promoter 4 | ||| |---|---|---|---|---|---|---| | Any methylation | Any methylation | Any methylation | |||| | n | %b | n | %b | n | %b | | | Age at diagnosis | |||||| | <50 | (38/57) | 66.7 | (27/60) | 45.0 | (4/57) | 7.0 | | 50–60 | (39/64) | 60.9 | (29/65) | 44.6 | (6/66) | 9.1 | | ≥70 | (57/87) | 65.5 | (53/88) | 60.2 | (10/88) | 11.4 | | P-valuea | 0.775 | 0.085 | 0.678 | ||| | Stage of disease | |||||| | I and II | (40/65) | 61.5 | (38/64) | 59.4 | (6/65) | 9.2 | | III and IV | (95/146) | 65.1 | (72/147) | 49.0 | (14/149) | 9.4 | | P-valuea | 0.622 | 0.165 | 0.970 | ||| | Histological grade | |||||| | Differentiated (1–2) | (40/73) | 54.8 | (42/74) | 56.8 | (5/74) | 6.8 | | Undifferentiated (3) | (95/138) | 68.8 | (68/137) | 49.6 | (15/140) | 1.1 | | P-valuea | 0.043 | 0.323 | 0.344 | ||| | Histological type | |||||| | Serous | (57/85) | 67.1 | (44/85) | 51.8 | (9/86) | 10.5 | | Non-serous | (79/127) | 62.2 | (67/128) | 52.3 | (12/129) | 9.3 | | P-valuea | 0.470 | 0.934 | 0.778 | ||| | Debulking outcome | |||||| | Optimal | (62/108) | 57.4 | (61/108) | 56.5 | (10/109) | 9.2 | | Suboptimal | (70/98) | 71.4 | (47/98) | 48.0 | (9/100) | 9.0 | | P-valuea | 0.036 | 0.221 | 0.965 | Significant values are bold aP-values from χ2 test bPercent of patients with any IGF-II promoter methylation Associations between P2 and P3 methylation patterns and patient and tumor characteristics are shown in Table 4. Significant differences were observed in the distribution of patients with optimal versus suboptimal debulking (P = 0.029), and this difference existed mainly for those who were positive for methylation in P3 but not P2 (Group 3). Of this group of patients, 77.8% had optimal debulking, while among the other groups only 43.1–53.2% of patients had optimal debulking. When comparing the debulking results of Group 3 with each of the other groups, we found significant differences between Group 3 and Group 1 (P = 0.036), Group 2 (P = 0.003), or Group 4 (P = 0.010), but no differences between Groups 1, 2, and 4 (data not shown). Differences between Group 2 and Group 3 patients were also suggested by tumor grade. A higher percentage of patients positive for P2 methylation and negative for P3 methylation (Group 2) had grade 3 tumors than patients positive for methylation in P3 but not P2 (Group 3), 78.9% versus 51.9% (P = 0.066) respectively. Together, these analyses suggest that a pattern of P2 methylation but not P3 differs from a pattern of P3 methylation but not P2 with respect to ovarian cancer prognosis. Table 4. | Variable | Group 1 P2 U P3 U n (%)a | Group 2 P2 M P3 U n (%)a | Group 3 P2 U P3 M n (%)a | Group 4 P2 M P3 M n (%)a | P-valueb | |---|---|---|---|---|---| | Age at diagnosis | 0.526 | |||| | <50 | 14 (29.8) | 16 (30.8) | 5 (18.5) | 22 (26.8) | | | 50–60 | 17 (36.2) | 18 (34.6) | 8 (29.6) | 21 (25.6) | | | ≥70 | 16 (34.0) | 18 (34.6) | 14 (51.9) | 39 (47.6) | | | Stage of disease | 0.516 | |||| | I and II | 15 (30.6) | 12 (23.1) | 10 (37.0) | 28 (33.7) | | | III and IV | 34 (69.4) | 40 (76.9) | 17 (63.0) | 55 (66.3) | | | Histological grade | 0.066 | |||| | Differentiated (1–2) | 20 (40.8) | 11 (21.2) | 13 (48.2) | 29 (34.9) | | | Undifferentiated (3) | 29 (59.2) | 41 (78.9) | 14 (51.9) | 54 (65.1) | | | Histological type | 0.871 | |||| | Serous | 18 (36.7) | 23 (44.2) | 10 (37.0) | 34 (40.5) | | | Non-serous | 31 (63.3) | 29 (55.8) | 17 (63.0) | 50 (59.5) | | | Debulking outcome | 0.029 | |||| | Optimal | 25 (53.2) | 22 (43.1) | 21 (77.8) | 40 (49.4) | | | Suboptimal | 22 (46.8) | 29 (56.9) | 6 (22.2) | 41 (50.6) | Significant values are bold aColumn totals may not sum to 100% due to rounding error bP-values from χ2 test To assess IGF-II promoter methylation in relation to prognosis, proportional hazards regression analysis was conducted for each promoter region, as well as for the P2 and P3 methylation patterns (Table 5). Methylation in the P2 promoter region was associated with an increased risk of disease progression (HR = 1.77, 95% CI 1.12–2.79), which remained significant after adjusting for patient’s age at surgery, disease stage, and histological grade (HR = 1.73, 95% CI: 1.09–2.74). Neither P3 nor P4 methylation was significantly associated with disease progression, and none of the three promoters independently indicated that IGF-II methylation was associated with a significantly increased risk of death. However, survival differences were observed between patients with different P2 and P3 methylation patterns; patients negative for P2 methylation and positive for P3 methylation (P2U/P3M) had a more than 50% reduction in risk for disease progression compared to those positive for P2 methylation and negative for P3 methylation (P2M/P3U). When P2U/P3M patients were compared to P2M/P3U patients, the difference was significant in both univariate (HR = 0.36, 95% CI 0.16–0.83) and multivariate analysis (HR = 0.39, 95% CI 0.17–0.91). A similar association was also suggested for OS although the difference was not statistically significant. Kaplan–Meier survival curves for Groups 2 and 3 are shown in Fig. 3; the curves were significantly different for disease free survival (P = 0.008) but not for OS (P = 0.116). The relationship between IGF-II promoter methylation and ovarian cancer prognosis did not differ among all patients, and those treated with either platinum, or taxol containing chemotherapeutic regimens (data not shown). Table 5. | IGF-II promoter methylation | Progression free survivalb | Overall survivalb | |---|---|---| | P2 methylation | || | Unadjusted | 1.77 (1.12–2.79) | 1.33 (0.86–2.07) | | Adjusteda | 1.73 (1.09–2.74) | 1.28 (0.82–2.00) | | P3 methylation | || | Unadjusted | 0.90 (0.60–1.35) | 0.92 (0.61–1.39) | | Adjusteda | 0.96 (0.63–1.46) | 0.96 (0.63–1.47) | | P4 methylation | || | Unadjusted | 1.24 (0.62–2.46) | 0.96 (0.46–1.98) | | Adjusteda | 1.43 (0.71–2.88) | 0.98 (0.41–2.03) | | IGF-II promoter methylation patterns | Progression free survivalb | Overall survivalb | |---|---|---| | Group 1: P2 U P3 U | || | Unadjusted | 0.66 (0.37–1.17) | 0.89 (0.50–1.60) | | Adjusteda | 0.71 (0.40–1.27) | 0.92 (0.51–1.65) | | Group 2: P2 M P3 U | || | Reference group | 1.00 | 1.00 | | Group 3: P2 U P3 M | || | Unadjusted | 0.36 (0.16–0.83) | 0.53 (0.24–1.17) | | Adjusteda | 0.39 (0.17–0.91) | 0.57 (0.25–1.31) | | Group 4: P2 M P3 M | || | Unadjusted | 0.90 (0.56–1.46) | 0.99 (0.60–1.65) | | Adjusteda | 0.98 (0.60–1.59) | 1.00 (0.60–1.70) | aAdjusted for age, stage, and histological grade bBold values indicate significant associations The associations between patient response to treatment and IGF-II promoter methylation were also examined (data not shown). Of the 203 patients with information on their response to treatment, 72.4% had a complete response, and 27.6% had either a partial or no response. Methylation patterns of P2 and P3 were found to be related to the clinical measurement of patient response to treatment. Patients negative for methylation in P2 and positive for methylation in P3 (Group 3) had an 88.9% response rate, whereas those positive for P2 methylation and negative for methylation in P3 (Group 2) had a 61.2% response rate (P = 0.011). This relationship between response to treatment and IGF-II promoter methylation is consistent with the associations found for survival: Group 3 patients fare significantly better than Group 2 patients.

Including an alternately spliced exon (Mineo et al. 2000; Ikejiri et al. 1991), the IGF-II gene has a total of ten exons (E1–E10); the four promoters precede E1, E4, E6, and E7, respectively (Engstrom et al. 1998). Few CpG sites were found to exist within the first promoter (P1), while the latter three promoters (P2–P4) are contained within a large CpG rich region (Issa et al. 1996). Several MSP assays were designed in P1, but none were successfully optimized on samples; assessment of P1 methylation was therefore not conducted in this study. Seven MSP assays were developed to assess methylation across P2, P3, and P4. The results showed high percentages (>45%) of methylation among two of the P2 and one of the P3 assays. Methylation in the remaining P2 and P3 assays was between 10 and 20%. The two P4 assays detected relatively little methylation. Among the three promoters, CpG site methylation increased from P2A to P3A, and then declined from P3B to P4B as the regions went from 5′ to 3′ along the IGF-II gene. Only the last three exons of the IGF-II gene encode the mature protein (E8-E10); differences in promoter usage and transcription initiation allow for tissue-specific and time-dependent variations in IGF-II expression (Gray et al. 1987; van Dijk et al. 1991; Irminger et al. 1987; Sussenbach et al. 1993). For example, during embryonic development and fetal growth, IGF-II can be transcribed from all four promoters although promoter activity may vary from organ to organ. After birth, the role of IGF-II diminishes, as IGF-I becomes the dominant regulator of postnatal growth and development. In most adult tissues, IGF-II is silent, except for transcription from P1 in the liver (Holthuizen et al. 1993; Sussenbach et al. 1993). This developmental-specific regulation has been shown to be altered in tumor tissues. Normal transcription of IGF-II from P1 in the liver was found to be disrupted in the majority (70–90%) of hepatocellular carcinoma cases (Li et al. 1997; Ng et al. 1998). In addition to cessation of P1 activity, researchers also found that expression of IGF-II was switched to fetal promoters P2, P3, and P4 in hepatocellular carcinomas (Ng et al. 1998; Uchida et al. 1997). Additionally, human hepatocellular carcinoma and colon carcinoma derived cell lines were shown to express IGF-II transcripts from P3 and P4, respectively (Schneid et al. 1993). When comparing primary colon cancers to corresponding healthy tissue, Winkler et al. (1999) found that IGF-II was overexpressed, and that the transcripts were derived from P3 and P4. When ovarian tumors were analyzed, transcripts from P3 and P4 were found to be more frequent and abundant than transcripts from P1 and P2; further high levels of expression from P3 and P4 were associated with significantly increased risks of death (Lu et al. 2006a). While promoter usage of the IGF-II gene has been shown to vary, the activities of these promoters may also differ. Mineo et al. (2000) used semiquantitative RT-PCR to examine IGF-II expression, and found that promoter usage was highly variable in breast, prostate, and bladder cancers; when expression was quantified, the P3 and P4 promoters were found to be much more active than the P2 promoter (Mineo et al. 2000). This may be due to a combination of many factors. First, the DNA sequences of the promoters contain different regulatory elements and transcription factor binding sites (van Dijk et al. 1991; Holthuizen et al. 1993; Sussenbach et al. 1993), and so may differ in their ability to initiate transcription. Additionally, differences in the length or constitution of the resulting transcripts may also confer variation in the rates at which transcripts are produced (West et al. 1996; Neilsen et al. 1990). Further, the IGF-II transcripts may also differ in their translatability and stability (Nielsen and Christiansen 1992, 1995; West et al. 1996). Nielsen et al. studied both rat and human IGF-II mRNA transcripts, and found that those with different 5′ UTRs were subject to different processes; some sedimented with ribosomes and were thought to be engaged in protein synthesis, while others were subject to degradation by endonucleolysis (Nielsen and Christiansen 1992, 1995). Together, all these processes affect the rate at which IGF-II is produced, such that promoter usage may determine the relative level of IGF-II that is expressed. These variations underscore how one gene product may have different actions that are both tissue-specific and time-dependent. Studies have shown that IGF-II is normally expressed in the ovary, and that expression is substantially elevated in ovarian cancer (Yun et al. 1996; Sawiris et al. 2002). Promoter usage is one mechanism by which IGF-II expression may be regulated, and methylation is a mechanism by which promoter usage may be directed. Effects of promoter methylation on IGF-II expression was first suggested by Raizis et al. (1993) after early structural analysis showed that P3 was hypermethylated in DNA from cultured cells. Li et al. (1998b) investigated the methylation status of the IGF-II P1 and P3 promoters during normal human liver development, and found that promoter methylation was correlated with expression; prenatal liver tissues with hypomethylation of P3 were reported to express high levels of P3 transcripts, whereas in postnatal liver tissues where the P3 promoter was hypermethylated, P3 transcripts were undetectable. The relationship between P3 methylation and expression was further confirmed in human hepatoblastomas where hypermethylation in the proximal region of P3 was associated with low levels of IGF-II expression, while hypomethylation of P3 corresponded to higher expression levels of P3 transcripts (Eriksson et al. 2001). The variation of DNA methylation of the IGF-II promoters observed in our study appears to be consistent with current understanding of their transcriptional activities. We found that patients with a pattern of P2 methylation without P3 methylation had worse survival outcomes compared to those with a pattern of P3 methylation without P2 methylation. These methylation patterns may indicate the importance of these regions in determining the activity of IGF-II. High P3 activity might result from either P3 hypomethylation or P2 hypermethylation, and may be indicative of strong IGF-II expression, which would in turn promote tumor growth, and lead to an unfavorable prognosis. We are currently further characterizing the relationship between methylation patterns and IGF-II expression in ovarian cancer. It should also be noted that IGF-II transcription is normally maternally imprinted, such that expression only occurs from the paternal allele (Giannoukakis et al. 1993). However, in many types of cancer, including Wilms’ tumor, renal cell carcinoma, cervical cancer, gastric cancer, colon cancer, lung cancer, and head and neck cancer, the imprinting regulation is disrupted, leading to biallelic expression of IGF-II (Ogawa et al. 1993; Douc-Rasy et al. 1996; Wu et al. 1997; Nonomura et al. 1997; el-Naggar et al. 1999; Kohda et al. 2001; Cui et al. 2002). This dysregulation of transcription and increased expression of IGF-II is believed to be involved in cancer development and tumor growth (Reeve 1996; Morison et al. 1996). However, loss of imprinting (LOI) has not been shown to be a primary mechanism for increased IGF-II expression in epithelial ovarian cancer. Xiong et al. (2002) reported that none of 20 benign and six of 22 (27.3%) malignant informative ovarian tumors had IGF-II LOI, although whether the ovarian tumors were epithelial in origin was not stated. Kim et al. (1998) found that six of 11 ovarian tumors (54.5%) had IGF-II LOI, although only two samples were from epithelial ovarian cancers, and of these, only one was informative for LOI analysis. Chen et al. (2000) reported that five of 20 informative epithelial ovarian tumors (25.0%) had IGF-II LOI, and Murphy et al. (2006) recently found that five of 23 serous epithelial ovarian cancers (21.7%) had lost IGF-II imprinting. Finally, Yun et al. (1996) found only monoallelic IGF-II expression in 11 informative ovarian tumors. Therefore, IGF-II LOI may have a lesser impact on expression and disease outcome in ovarian cancer than IGF-II promoter methylation and usage. To verify our assessment of methylation, we sequenced a subset of products from each MSP assay using forward primers and an automated sequencer (ABI-3730XL). All sequencing results were in agreement with the bisulfite treated IGF-II sequences, confirming that the targeted regions of IGF-II were indeed amplified during MSP. Additionally, sequenced MSP products were evaluated for percentage of CpG sites found to be positive for methylation, by C to T bisulfite conversion. The median percentage of CpG methylation among the PCR products generated from methylation specific primers was 97.4%, whereas the median percentage of methylation among the products generated with primers specific for unmethylated sequences was only 4.5%. This difference was substantial (P < 0.001), and supports the idea that samples scored as positive for methylation indeed have many CpG sites modified by methylation across the entire assayed region. Thus, even though the real methylation status of a region is actually a percentage of CpG sites, the MSP assay can be considered to be a valid proxy for determining methylation of a region as a dichotomous variable. Further, methylation of each promoter region was assessed by more than one MSP assay, and any misclassification of methylation would be independent of outcome assessment and therefore only serve to attenuate the actual association toward the null. While non-quantitative, the MSP assay is a rapid, cost-effective, and appropriate method useful for population-based studies. In summary, this study showed that methylation varied across three promoter regions of the IGF-II gene in epithelial ovarian cancer. Methylation in P4 was found relatively infrequently, while methylation in P2 and P3 was detected more frequently. Hypermethylation of P2 was associated with unfavorable disease characteristics. Patterns of methylation in promoters 2 and 3 were found to be associated with disease prognosis. Patients negative for P2 methylation and positive for P3 methylation fared better than patients positive for P2 methylation and negative for P3 methylation. These results suggest that methylation pattern of the IGF-II promoters are important for the regulation of IGF-II expression, and may have potential clinical implications in ovarian cancer prognosis. Acknowledgments This research was supported in part by the NIH grant R01AG21392, the Anna Fuller Fund from the Yale University School of Medicine, Department of Pathology (A. Beeghly), and the Associazione Italiana per la Ricerca sul Cancro (D. Katsaros).

- Beeghly A, Katsaros D, Chen H, Fracchioli S, Zhang Y, Massobrio M, Risch H, Jones B, Yu H (2006) Glutathione S-transferase polymorphisms and ovarian cancer treatment and survival. Gynecol Oncol 100:330–337 [DOI] [PubMed] [Google Scholar] - Chen CL, Ip SM, Cheng D, Wong LC, Ngan HYS (2000) Loss of imprinting of the IGF-II and H19 genes in epithelial ovarian cancer. Clin Cancer Res 6:474–479 [PubMed] [Google Scholar] - Cui H, Onyango P, Brandenburg S, Wu Y, Hsieh CL, Feinberg AP (2002) Loss of imprinting in colorectal cancer linked to hypomethylation of H19 and IGF2. Cancer Res 62:6442–6446 [PubMed] [Google Scholar] - Daughaday WH, Rotwein P (1989) Insulin-like growth factors I and II. Peptide, messenger ribonucleic acid and gene structures, serum, and tissue concentrations. Endocrine Reviews 10:68–91 [DOI] [PubMed] [Google Scholar] - Douc-Rasy S, Barrois M, Fogel S, Ahomadegbe JC, Stehelin D, Coll J, Riou G (1996) High incidence of loss of heterozygosity and abnormal imprinting of H19 and IGF2 genes in invasive cervical carcinomas. Uncoupling of H19 and IGF2 expression and biallelic hypomethylation of H19. Oncogene 12:423–430 [PubMed] [Google Scholar] - el-Naggar AK, Lai S, Tucker SA, Clayman GL, Goepfert H, Hong WK, Huff V (1999) Frequent loss of imprinting at the IGF2 and H19 genes in head and neck squamous carcinoma. Oncogene 18:7063–7069 [DOI] [PubMed] [Google Scholar] - Engstrom W, Shokrai A, Otte K, Granerus M, Gessbo A, Bierke P, Madej A, Sjolund M, Ward A (1998) Transcriptional regulation and biological significance of the insulin like growth factor II gene. Cell Prolif 31:173–189 [DOI] [PMC free article] [PubMed] [Google Scholar] - Eriksson T, Frisk T, Gray SG, von Schweinitz D, Pietsch T, Larsson C, Sandstedt B, Ekstrom TJ (2001) Methylation changes in the human IGF2 p3 promoter parallel IGF2 expression in the primary tumor, established cell line, and xenograft of a human hepatoblastoma. Exp Cell Res 270:88–95 [DOI] [PubMed] [Google Scholar] - Esteller M, Corn PG, Baylin SB, Herman JG (2001) A gene hypermethylation profile of human cancer. Cancer Res 61:3225–3229 [PubMed] [Google Scholar] - FIGO (1987) Changes in definition of clinical staging for carcinoma of the cervix and ovary. Am J Obstet Gynecol 56:263–264 [PubMed] [Google Scholar] - Giannoukakis N, Deal C, Paquette J, Goodyer CG, Polychronakos C (1993) Parental genomic imprinting of the human IGF2 gene. Nat Genet 4:98–101 [DOI] [PubMed] [Google Scholar] - Gray A, Tam AW, Dull TJ, Hayflick J, Pintar J, Cavenee WK, Koufos A, Ullrich A (1987) Tissue-specific and developmentally regulated transcription of the insulin-like growth factor 2 gene. DNA 6:283–295 [DOI] [PubMed] [Google Scholar] - Hartmann W, Waha A, Koch A, Albrecht S, Gray SG, Ekstrom TJ, von Schweinitz D, Pietsch T (2001) Promoter-specific transcription of the IGF2 gene: a novel rapid, non-radioactive and highly sensitive protocol for mRNA analysis. Virchows Arch 439:803–807 [DOI] [PubMed] [Google Scholar] - Herman JG, Graff JR, Myohanen S, Nelkin BD, Baylin SB (1996) Methylation-specific PCR: a novel PCR assay for methylation status of CpG islands. Proc Natl Acad Sci USA 93:9821–9826 [DOI] [PMC free article] [PubMed] [Google Scholar] - Holthuizen P, Van Dijk MA, Rodenburg RJ, Koonen-Reemst AM, Sussenbach JS (1993) Transcriptional regulation of the major promoters of the human IGF-II gene. Mol Reprod Dev 35:391–393 [DOI] [PubMed] [Google Scholar] - Ikejiri K, Wasada T, Haruki K, Hizuka N, Hirata Y, Yamamoto M (1991) Identification of a novel transcription unit in the human insulin-like growth factor-II gene. Biochem J 280:439–444 [DOI] [PMC free article] [PubMed] [Google Scholar] - Irminger JC, Rosen KM, Humbel RE, Villa-Komaroff L (1987) Tissue-specific expression of insulin-like growth factor II mRNAs with distinct 5′ untranslated regions. Proc Natl Acad Sci USA 84:6330–6334 [DOI] [PMC free article] [PubMed] [Google Scholar] - Issa JP (2003) Methylation and prognosis: of molecular clocks and hypermethylator phenotypes. Clin Cancer Res 9:2879–2881 [PubMed] [Google Scholar] - Issa JP, Vertino PM, Boehm CD, Newsham IF, Baylin SB (1996) Switch from monoallelic to biallelic human IGF2 promoter methylation during aging and carcinogenesis. Proc Natl Acad Sci USA 93:11757–11762 [DOI] [PMC free article] [PubMed] [Google Scholar] - Khandwala HM, McCutcheon IE, Flyvbjerg A, Friend KE (2000) The effects of insulin-like growth factors on tumorigenesis and neoplastic growth. Endocr Rev 21:215–244 [DOI] [PubMed] [Google Scholar] - Kim HT, Choi BH, Niikawa N, Lee TS, Chang SI (1998) Frequent loss of imprinting of the H19 and IGF-II genes in ovarian tumors. Am J Med Genet 80:391–395 [DOI] [PubMed] [Google Scholar] - Kohda M, Hoshiya H, Katoh M, Tanaka I, Masuda R, Takemura T, Fujiwara M, Oshimura M (2001) Frequent loss of imprinting of IGF2 and MEST in lung adenocarcinoma. Mol Carcinog 31:184–191 [DOI] [PubMed] [Google Scholar] - Lambert S, Vivario J, Boniver J, Gol-Winkler R (1990) Abnormal expression and structural modification of the insulin-like growth-factor-II gene in human colorectal tumors. Int J Cancer 46:405–410 [DOI] [PubMed] [Google Scholar] - Li X, Nong Z, Ekstrom C, Larsson E, Nordlinder H, Hofmann WJ, Trautwein C, Odenthal M, Dienes HP, Ekstrom TJ, Schirmacher P (1997) Disrupted IGF2 promoter control by silencing of promoter P1 in human hepatocellular carcinoma. Cancer Res 57:2048–2054 [PubMed] [Google Scholar] - Li SL, Goko H, Xu ZD, Kimura G, Sun Y, Kawachi MH, Wilson TG, Wilczynski S, Fujita-Yamaguchi Y (1998a) Expression of insulin-like growth factor (IGF)-II in human prostate, breast, bladder, and paraganglioma tumors. Cell Tissue Res 291:469–479 [DOI] [PubMed] [Google Scholar] - Li X, Gray SG, Flam F, Pietsch T, Ekstrom TJ (1998b) Developmental-dependent DNA methylation of the IGF2 and H19 promoters is correlated to the promoter activities in human liver development. Int J Dev Biol 42:687–693 [PubMed] [Google Scholar] - Lu L, Katsaros D, Wiley A, Rigault de la Longrais IA, Puopolo M, Schwartz P, Yu H (2006a) Promoter-specific transcription of insulin-like growth factor-II in epithelial ovarian cancer. Gynecol Oncol 103(3):990–995 [DOI] [PubMed] [Google Scholar] - Lu L, Katsaros D, Wiley A, Rigault de la Longrais IA, Risch HA, Puopolo M, Yu H (2006b) The relationship of insulin-like growth factor-II, insulin-like growth factor binding protein-3, and estrogen receptor-alpha expression to disease progression in epithelial ovarian cancer. Clin Cancer Res 12:1208–1214 [DOI] [PubMed] [Google Scholar] - Mineo R, Fichera E, Liang SJ, Fujita-Yamaguchi Y (2000) Promoter usage for insulin-like growth factor-II in cancerous and benign human breast, prostate, and bladder tissues, and confirmation of a 10th exon. Biochem Biophys Res Commun 268:886–892 [DOI] [PubMed] [Google Scholar] - Morison IM, Becroft DM, Taniguchi T, Woods CG, Reeve AE (1996) Somatic overgrowth associated with overexpression of insulin-like growth factor II. Nat Med 2:311–316 [DOI] [PubMed] [Google Scholar] - Murphy SK, Huang Z, Wen Y, Spillman MA, Whitaker RS, Simel LR, Nichols TD, Marks JR, Berchuck A (2006) Frequent IGF2/H19 domain epigenetic alterations and elevated IGF2 expression in epithelial ovarian cancer. Mol Cancer Res 4:283–292 [DOI] [PubMed] [Google Scholar] - Nielsen FC, Christiansen J (1992) Endonucleolysis in the turnover of insulin-like growth factor II mRNA. J Biol Chem 267:19404–19411 [PubMed] [Google Scholar] - Nielsen FC, Christiansen J (1995) Posttranscriptional regulation of insulin-like growth factor II mRNA. Scand J Clin Lab Invest Suppl 220:37–46 [PubMed] [Google Scholar] - Neilsen FC, Gammeltoft S, Christiansen J (1990) Translational discrimination of mRNAs coding for human insulin-like growth factor II. J Biol Chem 265:13431–13434 [PubMed] [Google Scholar] - Ng IO, Lee JM, Srivastava G, Ng M (1998) Expression of insulin-like growth factor II mRNA in hepatocellular carcinoma. J Gastroenterol Hepatol 13:152–157 [DOI] [PubMed] [Google Scholar] - Nonomura N, Nishimura K, Miki T, Kanno N, Kojima Y, Yokoyama M, Okuyama A (1997) Loss of imprinting of the insulin-like growth factor II gene in renal cell carcinoma. Cancer Res 57:2575–2577 [PubMed] [Google Scholar] - Ogawa O, Eccles MR, Szeto J, McNoe LA, Yun K, Maw MA, Smith PJ, Reeve AE (1993) Relaxation of insulin-like growth factor II gene imprinting implicated in Wilms’ tumour. Nature 362:749–751 [DOI] [PubMed] [Google Scholar] - Pollak MN, Schernhammer ES, Hankinson SE (2004) Insulin-like growth factors and neoplasia. Nat Rev Cancer 4:505–518 [DOI] [PubMed] [Google Scholar] - Raizis AM, Eccles MR, Reeve AE (1993) Structural analysis of the human insulin-like growth factor-II P3 promoter. Biochem J 289:133–139 [DOI] [PMC free article] [PubMed] [Google Scholar] - Reeve AE (1996) Role of genomic imprinting in Wilms’ tumour and overgrowth disorders. Med Pediatr Oncol 27:470–475 [DOI] [PubMed] [Google Scholar] - Sawiris GP, Sherman-Baust CA, Becker KG, Cheadle C, Teichberg D, Morin PJ (2002) Development of a highly specialized cDNA array for the study and diagnosis of epithelial ovarian cancer. Cancer Res 62:2923–2928 [PubMed] [Google Scholar] - Sayer RA, Lancaster JM, Pittman J, Gray J, Whitaker R, Marks JR, Berchuck A (2005) High insulin-like growth factor-2 (IGF-2) gene expression is an independent predictor of poor survival for patients with advanced stage serous epithelial ovarian cancer. Gynecol Oncol 96:355–361 [DOI] [PubMed] [Google Scholar] - Schneid H, Holthuizen PE, Sussenbach JS (1993) Differential promoter activation in two human insulin-like growth factor-II-producing tumor cell lines. Endocrinology 132:1145–1150 [DOI] [PubMed] [Google Scholar] - Serov SF, Soully RF, Sobin LH (1973) Histological typing of ovarian tumors: international histological classification of tumors. World Health Organization, Geneva, pp 37–42 [Google Scholar] - Shiraishi T, Mori M, Yamagata M, Haraguchi M, Ueo H, Sugimachi K (1998) Expression of insulin-like growth factor 2 mRNA in human gastric cancer. Int J Oncol 13:519–523 [DOI] [PubMed] [Google Scholar] - Sussenbach JS, Rodenburg RJ, Scheper W, Holthuizen P (1993) Transcriptional and post-transcriptional regulation of the human IGF-II gene expression. Adv Exp Med Biol 343:63–71 [DOI] [PubMed] [Google Scholar] - Tennant MK, Thrasher JB, Twomey PA, Drivdahl RH, Birnbaum RS, Plymate SR (1996) Protein and messenger ribonucleic acid (mRNA) for the type 1 insulin-like growth factor (IGF) receptor is decreased and IGF-II mRNA is increased in human prostate carcinoma compared to benign prostate epithelium. J Clin Endocrinol Metab 81:3774–3782 [DOI] [PubMed] [Google Scholar] - Toretsky JA, Helman LJ (1996) Involvement of IGF-II in human cancer. J Endocrinol 149:367–372 [DOI] [PubMed] [Google Scholar] - Uchida K, Kondo M, Takeda S, Osada H, Takahashi T, Nakao A, Takahashi T (1997) Altered transcriptional regulation of the insulin-like growth factor 2 gene in human hepatocellular carcinoma. Mol Carcinog 18:193–198 [PubMed] [Google Scholar] - van Dijk MA, van Schaik FM, Bootsma HJ, Holthuizen P, Sussenbach JS (1991) Initial characterization of the four promoters of the human insulin-like growth factor II gene. Mol Cell Endocrinol 81:81–94 [DOI] [PubMed] [Google Scholar] - West CA, Arnett TR, Farrow SM (1996) Expression of insulin-like growth factor I (IGF-I) mRNA variants in rat bone. Bone 19:41–46 [DOI] [PubMed] [Google Scholar] - Winkler R, Delacroix L, Bensbaho K, Lambert S, Collette J, Hodzic D (1999) IGF-II in primary human colorectal tumors: peptide level, activated promoters, parental imprinting and gene rearrangement. Horm Metab Res 31:148–154 [DOI] [PubMed] [Google Scholar] - Wu MS, Wang HP, Lin CC, Sheu JC, Shun CT, Lee WJ, Lin JT (1997) Loss of imprinting and overexpression of IGF2 gene in gastric adenocarcinoma. Cancer Lett 120:9–14 [DOI] [PubMed] [Google Scholar] - Xiong Y, Sun Y, Li H (2002) Different imprinting status of IGF-2 in epithelial ovarian tumors. J Huazhong Univ Sci Technolog Med Sci 22:255–256 [DOI] [PubMed] [Google Scholar] - Yu H, Rohan T (2000) Role of the insulin-like growth factor family in cancer development and progression. J Natl Cancer Inst 92:1472–1489 [DOI] [PubMed] [Google Scholar] - Yun K, Fukumoto M, Jinno Y (1996) Monoallelic expression of the insulin-like growth factor-2 gene in ovarian cancer. Am J Pathol 148:1081–1087 [PMC free article] [PubMed] [Google Scholar]