ANTICANCER RESEARCH 32: 3467-3472 (2012)

Cell-free DNA Level as a Prognostic Biomarker for Epithelial Ovarian Cancer

JAE HONG NO1, KIDONG KIM1, KI HO PARK2 and YONG-BEOM KIM1

1Department of Obstetrics and Gynecology, Seoul National University Bundang Hospital, Gyunggi, Republic of Korea; 2Biomedical Research Institute, Seoul National University Hospital, Seoul, Republic of Korea

Abstract. Background: The use of cell-free DNA (cfDNA) is controversial because in only 50% of stage I ovarian as a non-invasive biomarker has been evaluated in many cancer cases is CA125 elevated (2). In addition, an elevated types of cancer. This study investigated the prognostic CA125 level is frequently associated with benign significance of cfDNA level for ovarian cancer. Materials gynecological disease and non-gynecological malignancies, and Methods: Preoperative sera of 36 patients with ovarian causing a significant false-positive rate. Accordingly, there is cancer and of 16 with benign tumors were analyzed using a clear need to develop new biomarkers for ovarian cancer. commercially available copy number assay kits to measure Cell-free DNA (cfDNA) is extracellular DNA found in the cfDNA level of including beta-2-microglobulin serum or plasma that can be detected by sophisticated (B2M), member RAS oncogene family (RAB25), claudin 4 techniques such as polymerase chain reaction (PCR) (3). The (CLDN4) and ATP-binding cassette subfamily F member 2 elevation of cfDNA has been reported in various (ABCF2). Cox regression analysis was used to calculate inflammatory diseases, including acute pancreatitis, hazard ratios (HR) and 95% confidence intervals (CI). cardiomyopathy and rheumatoid arthritis (4-7). It is known Results: cfDNA level of these genes had no association with that elevation of cfDNA can also be found in patients with other prognostic factors of ovarian cancer. In particular, in various types of cancer, and these cfDNA are believed to be patients with advanced stage disease, a low RAB25 level was derived from the tumor itself through active metabolism or an independent prognostic factor for disease-free survival tumor cell death (8). The use of cfDNA as a screening (HR=18.2, 95% CI=2.0-170.0) and overall survival method has been evaluated in many types of cancer including (HR=33.6, 95% CI=1.8-634.8). Conclusion: Our findings renal cell carcinoma, lung cancer, testicular germ cell tumor, suggest that the preoperative serum cfDNA level of RAB25 colorectal cancer, and breast cancer, suggesting that it could could be a useful biomarker predicting survival outcomes in be used as a non-invasive biomarker for a wide variety of patients with advanced ovarian cancer. cancer types (9-11). It was also reported that the detection of a high level of cfDNA has a role in the prediction of Ovarian cancer is the most lethal gynecological malignancy prognosis in patients with bladder cancer (12). among all types of gynecological cancer (1). The majority of In ovarian cancer, a few studies have evaluated the role of patients are diagnosed with disease at advanced stages due cfDNA as a biomarker for ovarian cancer. In studies with to the absence of specific symptoms and screening patients, total cfDNA levels were higher in patients than in biomarkers. The prognosis of patients with advanced disease normal controls (13, 14). An animal study with female nude is still poor despite the use of extensive surgery and adjuvant mice demonstrated a correlation of tumor burden with the chemotherapy. Although the serum cancer antigen 125 level of cfDNA (15). It remains unclear whether cfDNA (CA125) level has been used as a biomarker to detect levels have an association with the prognosis of ovarian recurrence during follow-up, the role of CA125 in screening cancer. More than 20% of ovarian carcinomas are associated with a copy number abnormality at various chromosomal locations (16). Recent studies using comparative genomic Correspondence to: Yong-Beom Kim, MD, Ph.D., Department of hybridization have revealed specific copy number changes Obstetrics and Gynecology, Seoul National University Bundang and expression level of different genes including RAS Hospital, Gumi-ro 300, Seongnam-si, Gyunggi-do, Republic of Korea. Tel: +82 317877253, Fax: +82 317874054, e-mail: [email protected] oncogene family (RAB25), claudin 4 (CLDN4) and ATP- binding cassette subfamily F member 2 (ABCF2) (17, 18). Key Words: Cell-free DNA, ovarian cancer, biomarker. Although the cfDNA is assumed to originate from tumor, it

0250-7005/2012 $2.00+.40 3467 ANTICANCER RESEARCH 32: 3467-3472 (2012)

Table Ⅰ. Median expression of selected genes in serum cell-free DNA level of patients with epithelial ovarian cancer and those with benign cyst.

Expression (copies/ml)

Gene Benign (n=16) Epithelial ovarian cancer (n=36) p-Value*

Median (IQR) Median (IQR)

B2M 450.5 (280.1-560.0) 292.7 (121.8-551.2) 0.127 RAB25 69.2 (46.2-80.3) 61.3 (20.5-80.2) 0.276 CLDN4 77.2 (57.3-106.1) 55.2 (16.2-92.7) 0.085 ABCF2 1.2 (1.0-2.4) 2.0 (1.1-2.6) 0.968

*Mann-Whitney U-test. IQR: interquartile range; B2M: beta-2-microglobulin; RAB25: member RAS oncogene family; CLDN4: claudin 4; ABCF2: ATP-binding cassette subfamily F member 2.

remains unclear whether it could be useful in detecting the (assay ID: Hs07520413_cn) genes used in this study were purchased change in copy number for a particular gene. from Applied Biosystems (Carlsbad, CA, USA) and confirmed using The purpose of this study was to investigate the prognostic the Genbank database. A master mix for PCR was prepared with 2.5 μl of RNase DNase free water, 1.25 μl primers (9 μM) and probe significance of serum levels of total cfDNA and cancer- (2.5 μM), 12.5 μl TaqMan Master PCR 2× master mixture (Applied specific genes in patients with ovarian cancer. Biosystems). DNA in 5 μl was added as the PCR template. Quantitative real-time PCR was performed on 96-well optical plates Materials and Methods using the above reagents; the results were analyzed on an ABI Prism 7000 Sequence Detection System (Applied Biosystems, Carlsbad, Patients. This study included 36 patients with primary epithelial CA, USA). The following PCR conditions were used. After initial ovarian cancer and 16 patients with benign ovarian tumors, activation of uracyl-N-glycosylase at 50˚C for 2 minutes, AmpliTaq including five cases of serous cystadenoma, three of mucinous Gold was activated at 95˚C for 10 min. The subsequent PCR cystadenoma, six of endometriotic cyst, and four of mature cystic condition consisted of 45 cycles of denaturation at 95˚C for 15 teratoma. All patients underwent surgery between 2005 and 2009 at seconds and annealing extension at 60˚C for 1 minute per cycle. Seoul National University Bundang Hospital. The mean age of the During the PCR amplification, the amplified products were measured ovarian cancer group was 51.8 years; for the benign group, this continuously by determination of the fluorescence emission. A value was 36.5 years. Patients with ovarian cancer underwent standard curve consisting of known DNA concentration was created platinum-based adjuvant chemotherapy if indicated. The median by serial dilution to quantify the copy number of specific gene. duration of follow-up was 30.7 months. Data regarding surgical stage, the presence of residual tumor after cytoreductive surgery, and Statistical analysis. SPSS for Windows version 12.0 (SPSS Inc., histologic type were obtained from medical records retrospectively. Chicago, IL, USA) was used for data analysis. The mean values of Stages were determined according to the International Federation of serum DNA level were compared using Mann-Whitney U-tests. The Gynecology and Obstetrics (FIGO) criteria (17), and the four major relationships between categorical data were evaluated using the chi- groups were used for analysis. The FIGO stage distribution was as square test or Fisher’s exact test. Pearson’s correlation coefficient follows: stage I, 10 patients (19.2%); stage II, 2 patients (3.8%); was employed to investigate the correlation. The differences in stage III, 20 patients (38.5%); stage IV, 4 patients (7.7%). disease-free survival (DFS) and overall survival (OS) between positive and negative groups were analyzed by means of log-rank DNA extraction and quantitative real-time PCR. Venous blood test and Cox regression analysis. Cut-off points for each gene were samples were obtained preoperatively from the patients. After determined by median level of patients with ovarian cancer. A p- extracting serum from whole blood, the serum was stored in a deep values less than 0.05 was considered as being statistically significant. freezer until it was used for analysis. DNA was isolated from 1 ml of serum using the QIAamp DNA mini Kit (Qiagen Inc, Valencia, Results CA, USA) according to the manufacturer’s instruction. Extracted DNA was resuspended in 30 μl of deionized distilled water. Target genes were amplified, and the amplicons were analyzed with the cfDNA level in benign and malignant ovarian tumors. To Topo-TA Cloning kit (Invitrogen, Eugene, OR, USA). These investigate the role of cfDNA as a screening marker for plasmids were used for target gene quantification. Serial dilutions ovarian cancer, we compared the cfDNA levels in patients of plasmid DNA were used to determine the dynamic range of with benign tumors and those with ovarian cancer (Table Ⅰ). quantification. Commercially available copy number assay kits were used for The serum B2M gene level in cfDNA in ovarian cancer was DNA quantification. TaqMan primers and probes for beta-2- similar to that of those with benign ovarian tumors. There microglobulin (B2M: assay ID: Hs00119644_cn), ABCF2 (assay ID: was no significant difference in cfDNA level according to Hs03001077_cn), RAB25 (assay ID: Hs02686900_cn) and CLDN4 the histologic types of benign tumors.

3468 No et al: Cell-free DNA in Ovarian Cancer

Table Ⅱ. Association of cell-free DNA level according to prognostic factors of ovarian cancer.

Expression (Copies/ml)

Characteristic n (%) B2M RAB25 CLDN4 ABCF2

Median p-Value median p-Value median p-Value median p-Value

Age, years 0.876 0.616 0.851 0.778 <50 17 (47.2) 339.4 58.4 62.3 1.8 ≥50 19 (52.8) 327.0 58.6 58.6 2.0 Stage 0.655 0.934 0.655 0.476 I-II 12 (33.3) 375.0 59.5 64.3 1.8 III-IV 24 (66.7) 311.8 58.1 58.3 2.0 Histology 0.046 0.011 0.018 0.899 Serous 21 (58.3) 263.3 44.8 45.2 1.9 Other* 15 (41.7) 430.2 77.7 81.6 1.9 Residual mass 0.899 0.776 0.899 0.374 <1 cm 21 (58.3) 331.0 60.7 61.5 2.0 ≥1 cm 15 (41.7) 335.4 55.5 58.8 1.7 Recurrence 0.424 0.938 0.767 0.389 No 18 (50.0) 370.6 60.5 62.6 1.8 Yes 18 (50.0) 295.1 57.0 58.1 2.0

*Including mucinous (n=8), endometrioid (n=2), clear cell carcinoma (n=2) and undifferentiated carcinoma (n=3).

Table Ⅲ. Survival analysis according to serum cell-free DNA level in patients with advanced-stage ovarian cancer.

Disease-free survival Overall survival

HR 95% CI p-Value HR 95% CI p-Value

Age (>50 years) 1.2 0.3-4.4 0.12 4.0 0.7-23.8 0.12 Serous type 2.3 0.3-18.2 0.41 28.7 2.0-404.7 0.01 Residual mass (>1 cm) 3.6 0.8-17.2 0.11 5.9 1.0-36.1 0.05 Gene level High B2M 2.1 0.5-8.2 0.29 0.9 0.2-5.0 0.90 Low RAB25 18.2 2.0-170.0 0.01 33.6 1.8-634.8 0.02 High CLDN4 2.7 0.6-13.1 0.22 7.2 0.6-87.3 0.12 High ABCF2 2.2 0.4-10.9 0.34 4.2 0.6-30.3 0.16

HR: Hazard ratio, CI: confidence interval.

To test whether cfDNA has a role for predicting the Interestingly, a low level of RAB25 was also an independent prognosis of ovarian cancer, we investigated the association of prognostic factor for poor DFS (hazard ratio, HR=8.6, 95% prognostic factors with the evel of different genes in cfDNA confidence interval, CI=1.3-57.1). According to a subsequent (Table Ⅱ). Compared to other histological types, patients with subgroup analysis on patients with advanced stages of serous ovarian cancer had lower levels of B2M, RAB25 and disease, RAB25 was an independent prognostic factor for CLDN4 in cfDNA. Other prognostic factors including age>50 DFS (HR=18.2, 95% CI=2.0-170.0) and OS (HR=33.6, 95% years, advanced stage and presence of residual tumor had no CI=1.8-634.8) (Table Ⅲ). association with the cfDNA level. No correlation between cfDNA and CA125 or CA19-9 was found. Discussion

Survival analysis according to cfDNA level. In multivariate It has recently been reported that circulating cfDNA could survival analysis using Cox regression analysis, advanced be used as a non-invasive biomarker for a wide variety of stage was an independent prognostic factor for poor DFS. cancer types. In this study, we investigated the role of cfDNA

3469 ANTICANCER RESEARCH 32: 3467-3472 (2012) as a screening and prognostic biomarker for ovarian cancer. associated with poor DFS and OS in patients with ovarian Our findings showed that there was no statistical difference cancer. It has been suggested that RAB25 could play key in the levels of B2M, RAB25, CLDN4 and ABCF2 genes in roles in cell proliferation and in the prevention of apoptosis serum cfDNA between patients with ovarian cancer and by reducing the expression of proapoptotic molecules (22). patients with benign ovarian tumors. Our finding is not The association of poor survival with low RAB25 in serum consistent with a previous study showing that cfDNA levels cfDNA might be explained by reduced apoptosis of tumor in patients with ovarian cancer were higher than that of 12 cells and lower release of RAB25 gene in serum in our study age-matched normal controls (13). However, another study population (18, 22). Recently, several studies have shown the showed that plasma cfDNA in patients with ovarian cancer tumor suppressor function of RAB25 in breast and colon was lower compared to patients with benign tumors (14). cancer (23, 24). These results also could explain the impact These findings are consistent with previous studies showing of RAB25 on the survival in our study. that the discriminatory power of cfDNA between benign and In conclusion, our study suggests that the detection of malignant disease has been found to be modest for various RAB25 gene copy number variation could be useful as a types of cancer (3). One explanation for these results is the prognostic biomarker in patients with advanced ovarian elevation of cfDNA by other causes, including variation in cancer, although measuring total cfDNA levels has limited the source. Circulating cfDNA in serum is reported to be value as a screening marker for ovarian cancer. higher than that in plasma because the release of DNA from leukocytes during blood clotting leads to nonspecific Acknowledgements elevation of total cfDNA (20). The use of serum in this study might have limited the detection of differences in cfDNA This work was supported by grant number 11-2009-032 from Seoul levels between benign and malignant tumors. Further National University Bundang Hospital Research Fund. research into better understanding the role of cfDNA in screening of ovarian cancer is warranted. References Next we investigated whether cfDNA could predict the prognosis of patient with ovarian cancer. Our finding showed 1 Jemal A, Siegel R, Xu J and Ward E: Cancer statistics, 2010. CA that B2M gene measured in cfDNA had no association with Cancer J Clin 60: 277-300, 2010. prognosis of ovarian cancer. This result is similar to that of a 2 Jacobs I and Bast RC Jr.: The CA125 tumour-associated antigen: previous study showing that the serum cfDNA level had no a review of the literature. Hum Reprod 4: 1-12, 1989. 3 Jung K, Fleischhacker M and Rabien A: Cell-free DNA in the association with prognostic factors in patients with ovarian blood as a solid tumor biomarker-a critical appraisal of the cancer (14). literature. Clin Chim Acta 411: 1611-1624, 2011. Recently, researchers have sought to find specific 4 Lazar L, Rigo J Jr., Nagy B, Balogh K, Mako V, Cervenak L, alterations in cfDNA, such as cfDNA strand integrity, Mezes M, Prohaszka Z and Molvarec A: Relationship of mutation of tumor suppressor genes, methylation alterations, circulating cell-free DNA levels to cell-free fetal DNA levels, or microsatellite alterations. In ovarian cancer, one study clinical characteristics and laboratory parameters in pre- showed that tumor-specific polymorphic microsatellite loci eclampsia. BMC Med Genet 10: 120-125, 2009. 5 Gornik I, Wagner J, Gasparovic V, Lauc G and Gornik O: Free were detected in the cfDNA of 85% of patients with ovarian serum DNA is an early predictor of severity in acute pancreatitis. cancer. Another study using a microarray-based methylation Clin Biochem 42: 38-43, 2009. assay showed that differential methylation of promoters in 6 Bulicheva N, Fidelina O, Mkrtumova N, Neverova M, Bogush Ras association domain family member 1A (RASSF1A), A, Bogush M, Roginko O and Veiko N: Effect of cell-free DNA calcitonin-related polypeptide alpha (CALCA) and E1A of patients with cardiomyopathy and rDNA on the frequency of binding p300 (EP300) in cfDNA may be a useful contraction of electrically paced neonatal rat ventricular biomarker to differentiate between certain benign and myocytes in culture. Ann NY Acad Sci 1137: 273-277, 2008. 7 Zhong XY, von Muhlenen I, Li Y, Kang A, Gupta AK, Tyndall malignant ovarian tumors (21). Thus, we investigated A, Holzgreve W, Hahn S and Hasler P: Increased concentrations whether copy number variance of a specific gene in cfDNA of antibody-bound circulatory cell-free DNA in rheumatoid could predict the prognosis of patients with ovarian cancer. arthritis. Clin Chem 53: 1609-1614, 2007. Interestingly, we found that a low serum cfDNA level of 8 Anker P and Stroun M: Circulating DNA in plasma or serum. RAB25 was associated with increased risk of death from Medicina (B Aires) 60: 699-702, 2000. ovarian cancer. A previous study using array comparative 9 Gang F, Guorong L, An Z, Anne GP, Christian G and Jacques T: genomic hybridization showed an increase in DNA copy Prediction of Clear Cell Renal Cell Carcinoma by Integrity of Cell-free DNA in Serum. Urology 75: 262-265, 2010. number of the RAB25 gene on 1q22 in 54% in 10 Deligezer U, Eralp Y, Akisik EE, Akisik EZ, Saip P, Topuz E and ovarian cancer (18). In that study, RAB25 copy number was Dalay N: Size distribution of circulating cell-free DNA in sera of directly correlated with the expression level, and the breast cancer patients in the course of adjuvant chemotherapy. increased expression of RAB25 in tumor tissue was Clin Chem Lab Med 46: 311-317, 2008.

3470 No et al: Cell-free DNA in Ovarian Cancer

11 Schwarzenbach H, Stoehlmacher J, Pantel K and Goekkurt E: 19 Santin AD, Zhan F, Bellone S, Palmieri M, Cane S, Bignotti E, Detection and monitoring of cell-free DNA in blood of patients Anfossi S, Gokden M, Dunn D, Roman JJ, O’Brien TJ, Tian E, with colorectal cancer. Ann N Y Acad Sci 1137: 190-196, 2008. Cannon MJ, Shaughnessy J, Jr. and Pecorelli S: 12 Ellinger J, Bastian PJ, Ellinger N, Kahl P, Perabo FG, Buttner R, profiles in primary ovarian serous papillary tumors and normal Muller SC and Ruecker A: Apoptotic DNA fragments in serum ovarian epithelium: identification of candidate molecular of patients with muscle-invasive bladder cancer: a prognostic markers for ovarian cancer diagnosis and therapy. Int J Cancer entity. Cancer Lett 264: 274-280, 2008. 112: 14-25, 2004. 13 Kamat AA, Sood AK, Dang D, Gershenson DM, Simpson JL 20 Zhong XY, Hahn S, Kiefer V and Holzgreve W: Is the quantity and Bischoff FZ: Quantification of total plasma cell-free DNA of circulatory cell-free DNA in human plasma and serum in ovarian cancer using real-time PCR. Ann NY Acad Sci 1075: samples associated with gender, age and frequency of blood 230-234, 2006. donations? Ann Hematol 86: 139-143, 2007. 14 Zachariah RR, Schmid S, Buerki N, Radpour R, Holzgreve W 21 Liggett TE, Melnikov A, Yi Q, Replogle C, Hu W, Rotmensch J, and Zhong X: Levels of circulating cell-free nuclear and Kamat A, Sood AK and Levenson V: Distinctive DNA mitochondrial DNA in benign and malignant ovarian tumors. methylation patterns of cell-free plasma DNA in women with Obstet Gynecol 112: 843-850, 2008. malignant ovarian tumors. Gynecol Oncol 120: 113-120, 2011. 15 Kamat AA, Bischoff FZ, Dang D, Baldwin MF, Han LY, Lin YG, 22 Fan Y, Xin XY, Chen BL and Ma X: Knockdown of RAB25 Merritt WM, Landen CN Jr., Lu C, Gershenson DM, Simpson expression by RNAi inhibits growth of human epithelial ovarian JL and Sood AK: Circulating cell-free DNA: a novel biomarker cancer cells in vitro and in vivo. Pathology 38: 561-567, 2006. for response to therapy in ovarian carcinoma. Cancer Biol Ther 23 Cheng JM, Volk L, Janaki DK, Vyakaranam S, Ran S and Rao 5: 1369-1374, 2006. KA: Tumor suppressor function of RAB25 in triple-negative 16 Sonoda G, Palazzo J, du Manoir S, Godwin AK, Feder M, breast cancer. Int J Cancer 126: 2799-2812, 2010. Yakushiji M and Testa JR: Comparative genomic hybridization 24 Goldenring JR and Nam KT: RAB25 as a tumour suppressor in detects frequent overrepresentation of chromosomal material colon carcinogenesis. Br J Cancer 104: 33-36, 2011. from 3q26, 8q24, and 20q13 in human ovarian carcinomas. Genes Cancer 20: 320-328, 1997. 17 Odicino F, Pecorelli S, Zigliani L and Creasman WT: History of the FIGO cancer staging system. Int J Gynaecol Obstet 101: 205-210, 2008. 18 Cheng KW, Lahad JP, Kuo WL, Lapuk A, Yamada K, Auersperg N, Liu J, Smith-McCune K, Lu KH, Fishman D, Gray JW and Received March 19, 2012 Mills GB: The RAB25 small GTPase determines aggressiveness Revised April 24, 2012 of ovarian and breast cancers. Nat Med 10: 1251-1256, 2004. Accepted April 25, 2012

3471