IJC International Journal of Cancer

Transcription

1 IJC International Journal of Cancer Reference ranges for HE4 and CA125 in a large Asian population by automated assays and diagnostic performances for ovarian cancer Yongjung Park, Yoonjung Kim, Eun Young Lee, Jong-Han Lee and Hyon-Suk Kim Department of Laboratory Medicine, Yonsei University College of Medicine, Seoul, Korea Human epididymis protein 4 (HE4) is a new biomarker for the detection of ovarian cancer. We evaluated the analytical performance of a novel automated HE4 assay and established reference ranges of HE4 and CA125. We also compared the diagnostic performance of both biomarkers for ovarian cancer. Precision performances and linearity of the HE4 assay were assessed. Serum samples from 2,182 healthy and 72 pregnant women were also assayed for HE4 and CA125, and the 95%, 97.5% and 99% reference limits for both markers were calculated. Additionally, sera from 66 ovarian cancer and 257 benign gynecologic disease patients were tested to validate reference ranges and diagnostic performances. The total precision of the HE4 assay was <5% coefficient of variation for most of the levels evaluated. The linearity range of this assay was from 15.0 to pmol/l. The 97.5% upper reference limits for HE4 and CA125 were 33.2 pmol/l (95% confidence interval [CI], ) and 38.3 U/mL (95% CI, ), respectively. Using these values as cutoff points, the sensitivity and specificity of HE4 for differentiating ovarian cancer from benign gynecologic diseases and healthy individuals were 90.9% and 94.1%, and those of CA125 were 72.7% and 94.4%. The receiver operating characteristic-area under the curve values of HE4 and CA125 for discriminating ovarian cancer from age-matched control were 0.94 and 0.86, respectively, and they were statistically different (p ). The new automated HE4 assay showed good analytical and diagnostic performances. The reference limits established in our study could be used as cutoff levels to facilitate more accurate diagnosis of ovarian cancer in Asian population. Ovarian cancer is a common malignant disease and is ranked fifth in causes of cancer-related mortalities in women. 1 The American Cancer Society estimates that 21,880 new patients will suffer from ovarian cancer and 13,850 patients will die from it in the United States in Pelvic masses can be Key words: human epididymis protein 4 (HE4), CA125, ovarian cancer, biomarker Abbreviations: ANOVA: analysis of variance; AP-CT: abdomenpelvis computerized tomography; AUC: area under the curve; CI: confidence interval; CLIA: chemiluminescent immunoassay; CLSI: the Clinical and Laboratory Standards Institute; CMIA: chemiluminescent microparticle immunoassay; CV: coefficient of variation; ECLIA: electrochemiluminescence immunoassay; ELISA: enzyme-linked immunosorbent assay; FDA: Food and Drug Administration; HE4: human epididymis protein 4; QC: quality control; RLU: relative light unit; ROC: receiver operating characteristic; ROMA: risk of ovarian malignancy algorithm; U/S: ultrasonography DOI: /ijc History: Received 8 Dec 2010; Accepted 29 Mar 2011; Online 11 Apr 2011 Correspondence to: Hyon-Suk Kim, Department of Laboratory Medicine, Yonsei University College of Medicine, 250 Sungsan-no, Seodaemun-gu, Seoul, Korea , Tel: þ , Fax: þ , kimhs54@yuhs.ac found in 20% of women in their lifetime, 2 and 15% to 20% of these cases will be diagnosed with ovarian cancer. 3 However, more than three-fourths of patients with ovarian cancer are diagnosed in the advanced stages, which is associated with a poor survival rate of about 10 20%, 4 although ovarian cancer has a good prognosis if detected in its early stages and when patient care by specialized clinicians and surgeons is provided. 5 7 Therefore, early detection and accurate differential diagnosis of ovarian cancer from benign pelvic masses is essential for improving patients survival. CA125 is the most commonly used tumor marker for detecting and monitoring ovarian cancer in current clinical practice. However, this glycoprotein is not expressed in up to 20% of ovarian cancer patients 8 and can be elevated in various benign or malignant conditions other than ovarian cancer. 9 Moreover, diagnostic performances of CA125 were not good enough for screening to detect early stage ovarian cancer in some studies. 10,11 Therefore, many studies have been performed to improve the diagnostic performance of biomarkers or their combinations for ovarian cancer diagnosis. Some markers, including mesothelin, inhibin, kallikreins and osteopontin, have been investigated as possible improvements in the sensitivity and specificity for early and precise detection of ovarian cancer Human epididymis protein 4 (HE4), also known as WAP four disulphide core 2 (WFDC2), is a novel protein and is one of the more promising biomarkers for improving diagnostic

2 Park et al performance in ovarian cancer detection. The WFDC2 gene was first found to be expressed in some ovarian cancers through a comparative hybridization study on 21,500 unknown ovarian cdnas; 17 thereafter, it was reported that HE4 could be used as a biomarker for ovarian cancer with higher specificity than CA This protein was especially highly expressed in histologic subtypes of serous or endometrioid ovarian carcinoma. 19 In recent studies, HE4 combined with or without CA125 yielded the highest sensitivity for detecting ovarian cancer especially in the early stages. 20 The Risk of Ovarian Malignancy Algorithm (ROMA) score, which can be calculated from HE4 and CA125 concentrations, was able to distinguish ovarian cancer from benign pelvic masses with 94.3% sensitivity at a set specificity of 75% and with a receiver operating characteristic (ROC)-area under the curve (AUC) value of HE4 was also detected in urine specimens of ovarian cancer patients with sensitivities of 86.6% in Stages I/II to 89.0% in Stages III/IV with a specificity of 94.4%. 22 Nevertheless, there are no reference ranges or cutoff levels of HE4 established with a large healthy population, and almost all of the previous studies on HE4 utilized enzymelinked immunosorbent assays (ELISA), which could be laborintensive and time-consuming. Also, ELISA generally has lower analytical sensitivity and precision performance than automated assays utilizing the chemiluminescent immunoassay (CLIA) method, and the results from ELISA can be affected by the proficiency of a test performer. Recently, an automated HE4 assay was developed to improve the drawbacks of ELISA in October 2009, and the United States Food and Drug Administration (FDA) approved this assay for detecting and monitoring of ovarian cancer in June In this study, we aimed to evaluate analytical performances of this novel automated HE4 assay and to establish reference limits for HE4 and CA125 in a large healthy Asian population. We also validated the established reference limits using samples from ovarian cancer and benign gynecologic disease patients and compared diagnostic performances of HE4 to CA125. Material and Methods Healthy individuals and patients Serum samples from 2,400 apparently healthy female employees of Severance Hospital were collected from May to June All subjects had taken a health inspection when they were first hired and had undergone annual checkups at least for 2 years. All subjects over age 44 years also had been examined by abdomen and pelvis ultrasonography (U/S) every 2 years. Written consents including self-descriptions regarding their medical status were obtained from all subjects. The sera were separated immediately after arrival and tested for biomarkers as prescribed in the checkup. Thereafter, all residual samples were stored at 70 C until assayed for HE4 and CA125. Medical records including results of radiographic and laboratory tests of all the enrolled individuals were retrospectively reviewed; 16 individuals with nongynecologic diseases and 130 subjects who were diagnosed with benign gynecologic diseases by pelvic U/S and/or abdomen-pelvis computerized tomography (AP-CT) were excluded from our study. The remaining 2,254 subjects were assayed for HE4 and CA125. These individuals were classified into two groups: healthy individuals (n ¼ 2,182) and pregnant women (n ¼ 72). Additionally, independent serum sample set from 393 nonduplicated patients was collected over the same period to validate the diagnostic performances of both markers with the reference ranges established in our study. All these samples were requested by the gynecologic department of Severance Hospital to assay biomarkers for suspicious or known gynecological malignancies. Immediately after the requested tests were finished, the remaining specimens were stored at 70 C until assayed for HE4 and CA125. The patients medical records were also retrospectively reviewed to collect patients information, including diagnosis and disease stages, and 65 subjects with gynecologic cancers other than ovarian cancer or without any gynecologic disorders were excluded. Patients with pathologically unconfirmed ovarian cancers were also omitted. Finally, 257 samples from patients with benign gynecologic diseases and 66 from ovarian cancer patients were assayed for HE4 and CA125. All the diagnoses had been made from the results of imaging studies including pelvic U/S and/or AP-CT and from the results of a surgical biopsy if available (all 66 ovarian cancer cases included in our study were pathologically confirmed). HE4 and CA125 assays HE4 concentrations of all our study subjects were measured using the Abbott Architect i2000sr analyzer with Architect HE4 kits (Abbott Laboratories, Abbott Park, IL). This assay utilizes a chemiluminescent microparticle immunoassay (CMIA) principle for quantitative detection of HE4 antigen in human serum. It is a two-step immunoassay with anti-he4 coated microparticles in the first step and anti-he4 acridinium-labeled conjugate in the second step (two antibodies specific to the HE4 antigen are used). The resulting chemiluminescent reaction is measured as relative light units (RLUs), which are directly related to the amount of HE4 antigen in the samples. The displayed unit of HE4 concentration is pmol/l. CA125 levels of the same subjects were quantified by using the Roche Cobas e411 analyzer with Elecsys CA125 II kits (Roche Diagnostics Corp., Indianapolis, IN). This assay utilizes electrochemiluminescence immunoassay (ECLIA) method, and the unit used for the results is U/mL. Total precision claimed by the manufacturer is between 2.5% and 4.2% coefficient of variation (CV) when specimens with CA125 levels from 7.8 to U/mL were assayed. Precision performance evaluation Precision performance of the HE4 assay was assessed based on guidelines from the Clinical and Laboratory Standards

3 1138 Reference Ranges for HE4 and CA125 Table 1. Precision data of the HE4 assay over the 20-day evaluation period Precision (% CV) Material Mean (pmol/l) Repeatability Between-run Between-day Total QC Low QC Medium QC High Pooled serum Low Pooled serum High 1, Institute (CLSI) document EP4-A2. Testing was performed using a single lot of reagents, calibrators and controls with one analyzer. Three levels of quality control (QC) materials and two levels of pooled serum panels were assayed in replicates of two at two separate times per day for 20 days. The analyzer used a single calibration curve throughout the study. Validation of the linearity range Tests for validating the linearity range of the HE4 assay were performed based on the CLSI document EP6-A. Samples with five levels were prepared by mixing high and low levels of pooled patients sera and were tested in duplicate. The linearity range of the HE4 assay demonstrated by the manufacturer is from 20.0 to 1,500.0 pmol/l. Reference range establishment The 95th, 97.5th and 99th percentile upper reference limits of HE4 and CA125 were analyzed according to the CLSI document C28-A2. This document provides guidelines for collecting and selecting clinical samples, minimum requirements of sample number and applicable statistical methods to establish reference limits. As the markers levels of the healthy population in our study were not in a normal distribution, we calculated the upper reference limits with a nonparametric method, in which the upper reference limit is regarded as certain percentile value on the test results of a population. Among the 2,254 subjects without any diseases, 72 pregnant women were excluded in the reference range establishment, and the HE4 and CA125 levels of all 2,182 samples were analyzed. This population was divided into four subgroups of ages 20 to 29 (n ¼ 858), 30 to 39 (n ¼ 835), 40 to 49 (n ¼ 368) and 50 to 65 (n ¼ 121), and the upper reference limits of each group were also calculated. The levels of HE4 and CA125 between these subgroups were compared to identify the differences by age in the population. Diagnostic performances of HE4 and CA125 We validated the upper reference limits established in our study using samples from patients with ovarian cancer or benign gynecologic diseases. The sensitivities of HE4 and CA125 for detecting ovarian cancer (n ¼ 66) were analyzed, and their specificities for discriminating ovarian cancer from benign gynecologic diseases (n ¼ 257) or from healthy individuals (n ¼ 2,182) or from both groups (n ¼ 2,439) were also estimated. ROC curve analysis was performed to evaluate the diagnostic performance of both markers for differentiating ovarian cancer (n ¼ 66) from benign gynecologic diseases (n ¼ 67) and healthy subjects (n ¼ 165), which were selected according to older age to have equal mean age to ovarian cancer group (mean age ¼ 52.1 years), and the AUC values of the markers were compared. Statistical Analysis All statistical analyses were performed using the Analyse-it Method Evaluation Edition version 2.22 software (Analyse-it Software, City West Business Park, Leeds, UK). Kolmogorov- Smirnov test was used to define the distributions of markers levels in the healthy population and those of the groups with diseases. Multiple comparisons between the study groups were performed using the Kruskal-Wallis test or one-way analysis of variance (ANOVA) with pairwise comparison and the Bonferroni correction to compensate for alpha statistical errors. p-values of less than 0.05 were considered to be statistically significant in all analyses. Results Precision performance of the HE4 assay The mean concentrations of materials tested for the precision evaluation were ranging from 15.8 pmol/l to pmol/l. The HE4 assay showed a total precision performance of less than 5% CV for all assessed panels except the one with the highest HE4 level for which total precision was 6.5% CV. Repeatability, between-run and between-day precision are also summarized in Table 1. Linearity range of the HE4 assay The HE4 assay was demonstrated to be linear in the range from 15.0 to pmol/l. When measured values were compared to expected values, the linearity was found to be the following equation: y ¼ x , r 2 ¼ We could not evaluate the upper level of linearity claimed by the manufacturer, because there was not enough serum with the highest level. Characteristics of study groups Age, HE4 and CA125 concentrations of each study group are shown in Table 2. The median HE4 and CA125 levels of healthy individuals were 21.2 pmol/l and 15.2 U/mL and those

4 Park et al Table 2. The results of HE4 and CA125 concentrations by study group Group (n) Age (years) 1 p-value 2 HE4 (pmol/l) 1 p-value 2 CA125 (U/mL) 1 p-value 2 Healthy individuals (2,182) 32.0 ( ) 21.2 ( ) 15.2 ( ) Pregnant women (72) 31.0 ( ) ( ) ( ) < Benign gynecologic diseases (257) 39.0 ( ) < ( ) < ( ) < Ovarian cysts (121) 34.0 ( ) ( ) < ( ) < Myoma of uterus (99) 42.0 ( ) < ( ) < ( ) Benign ovarian tumors (19) 34.0 ( ) ( ) < ( ) Adenomyosis (12) 44.0 ( ) < ( ) < ( ) < Endometrial pathologies (6) ( ) ( ) ( ) Ovarian cancer (66) ( ) < ( ) < ( ) < Data are shown as median (first to third quartiles). 2 p-values were calculated by comparison of each study group with healthy individuals. 3 Includes endometrial polyp or hyperplasia, and endometrioma. 4 Includes 54 cases of surface epithelial-stromal tumors (21 serous cystadenocarcinomas, 21 papillary serous cystadenocarcinomas, 5 mucinous cystadenocarcinomas, 4 clear cell carcinomas, 2 endometrioid adenocarcinomas, and 1 adenocarcinoma, not otherwise specified), 1 case of carcinoma, not otherwise specified, 3 cases of sex cord-stromal tumors, 3 cases of germ cell tumors and 5 cases of metastatic adenocarcinomas. of patients with ovarian cancer were 64.4 pmol/l and U/mL, respectively. Distribution of the assayed values of the study groups are illustrated in Figure 1. Levels of both markers in the groups of women who were pregnant and who had diseases were higher than those of healthy individuals. Additionally, both markers were elevated in different patterns according to histologic subtypes; for example, the median HE4 level of patients with the subtype serous cystadenocarcinoma was pmol/l and that of mucinous cystadenocarcinoma was 26.7 pmol/l (data not shown). Reference limits of HE4 and CA125 With 2,182 healthy subjects, the 95th, 97.5th and 99th percentile upper reference limits for both markers were calculated according to age group and are displayed in Table 3. The reference limits for HE4 in the group with 30- to 65- year-old subjects were analyzed separately, because HE4 levels between subjects aged and were statistically different. In the same manner, the reference limit for CA125 in 20- to 49-year-old subjects was analyzed separately. For all healthy subjects, regardless of age, the 97.5th percentile upper reference limits of HE4 and CA125 were 33.2 pmol/l (95% confidence interval [CI], ) and 38.3 U/mL (95% CI, ), respectively. The 97.5th percentile reference limits of HE4 ranged from 31.8 to 34.0 pmol/l by age group, and those of CA125 were between 27.8 and 41.1 U/mL. Neither marker showed parametric distribution; for example, the Kolmogorov-Smirnov D value for distribution for HE4 levels in healthy subjects was 0.08, and its p-value was < Diagnostic performances of HE4 and CA125 Sensitivities and specificities of HE4 and CA125 for detecting ovarian cancer are summarized in Table 4. When using the 97.5% upper reference limits as cut-off points, the sensitivity and specificity of HE4 for differentiating ovarian cancer (n ¼ 66) from benign gynecologic diseases (n ¼ 257) and healthy individuals (n ¼ 2,182) were 90.9% and 94.1%, and those of CA125 were 72.7% and 94.4%. The ROC-AUC values for discriminating ovarian cancer (n ¼ 66) from benign gynecologic diseases (n ¼ 67) and healthy subjects (n ¼ 165) were (95% CI, ; p < ) for HE4 and (95% CI, ; p < ) for CA125, which showed statistically significant difference (p ¼ ) (Fig. 2). Mean ages were not different between those three groups for ROC analysis by one-way ANOVA test (for all three groups, mean age ¼ 52.1 years, p ¼ ). Discussion In our study, we evaluated the analytical and diagnostic performance of a new automated HE4 assay. This assay showed good precision performance with less than 5% CV in all assessed materials except the one with the highest mean concentration, 1,228.3 pmol/l. However, this concentration may be not clinically important because the cutoff level of HE4 for detecting ovarian cancer is far less than that. Most previous research on HE4 utilized the ELISA method to quantify levels of HE4, and there is only one available study regarding the precision performances of ELISA for quantifying HE4 in the literature, of which the within-run and between-run CVs were 11.7% and 23.8%, respectively. 23 ELISA is generally known to have lower precision and analytical sensitivity than CMIA or CLIA methods. Therefore, an automated HE4 assay could improve the analytical performances and facilitate an accurate diagnosis of ovarian cancer. We analyzed HE4 and CA125 concentrations of sera collected from 2,182 healthy individuals, 72 pregnant women and 323 patients with benign gynecologic diseases or ovarian cancer to establish reference ranges of both markers and to validate the utility of these reference limits as cutoff levels for detecting ovarian cancer. In the laboratory statistics, minimum of 120 healthy subjects are generally required to establish 97.5 percentile upper reference limits with only 90% CIs,

5 1140 Reference Ranges for HE4 and CA125 Figure 1. Distributions of HE4 and CA125 levels by study group. (a and b) Levels of both markers differed significantly between study groups. The upper and lower ends of boxes and box inner lines correspond to the upper and lower quartiles and median values, respectively. Whiskers denote minimum and maximum values, and circles indicate individual values. and greater statistical power can be obtained with results from larger population. Thus, we tried to include as many healthy individuals as possible to enhance reliability of our results, and the reference limits calculated from our data could have narrow ranges of 95% CIs. Additionally, our results showed that HE4 and CA125 levels in pregnant women and in patients with gynecologic diseases were higher than those of healthy individuals. Especially in ovarian cancer, the median level of HE4 was about three times higher than that of healthy individuals, and the median CA125 concentration was elevated approximately seven times higher than that of healthy subjects. Levels of HE4 in pregnant women or patients with benign gynecologic diseases were also higher than in the healthy population, but differences in median levels were no greater than 7 pmol/l (i.e., increments were less than 33% in proportion to the median HE4 level of healthy subjects). On the other hand, median CA125 concentrations in pregnant women or patients with benign gynecologic diseases were, respectively, 55.3% and 43.4% higher than those in the healthy population. This

6 Park et al Table 3. Reference ranges of HE4 and CA125 according to age group Upper reference limits (95% CI) Marker Age (n) 95th percentile 97.5th percentile 99th percentile HE4 (pmol/l) (858) 29.7 ( ) 31.8 ( ) 36.0 ( ) (835) 30.3 ( ) 34.0 ( ) 37.9 ( ) (368) 31.6 ( ) 33.4 ( ) 38.1 ( ) (121) 31.2 ( ) 32.6 ( ) (1,324) ( ) 33.6 ( ) 37.7 ( ) All (2,182) 30.3 ( ) 33.2 ( ) 36.6 ( ) CA125 (U/mL) (858) 33.7 ( ) 39.4 ( ) 59.6 ( ) (835) 32.7 ( ) 41.1 ( ) 64.4 ( ) (368) 31.6 ( ) 35.3 ( ) 45.1 ( ) (121) 22.8 ( ) 27.8 ( ) (2,061) ( ) 39.0 ( ) 54.6 ( ) All (2,182) 32.8 ( ) 38.3 ( ) 53.5 ( ) 1 Only 90% CI can be calculated because of the number of subjects. 2 99% reference limit cannot be statistically calculated to 65 year-old group was analyzed separately, because HE4 levels of and groups were different to 49 year-old group was analyzed separately, as CA125 concentrations in the and year-old groups were different. Table 4. Sensitivities and specificities of HE4 and CA125 for detecting ovarian cancer HE4 CA125 Cutoff level 1 (percentile of upper reference limit) Parameter 30.3 (95) 33.2 (97.5) 36.6 (99) 32.8 (95) 38.3 (97.5) 53.5 (99) Sensitivity (%) (95% CI) ( ) ( ) ( ) ( ) ( ) ( ) Specificity (%) (95% CI) ( ) ( ) ( ) ( ) ( ) ( ) Specificity (%) (95% CI) ( ) ( ) ( ) ( ) ( ) ( ) Specificity (%) (95% CI) ( ) ( ) ( ) ( ) ( ) ( ) 1 Units are pmol/l for HE4 and U/mL for CA Specificities in discriminating ovarian cancer (n ¼ 66) from benign gynecologic diseases (n ¼ 257). 3 Specificities in differentiating ovarian cancer (n ¼ 66) from healthy individuals (n ¼ 2,182). 4 Specificities in distinguishing ovarian cancer (n ¼ 66) from healthy individuals (n ¼ 2,182) and benign gynecologic diseases (n ¼ 257). means that HE4 is less falsely elevated in pregnant women and patients with benign gynecologic diseases and is more specific to ovarian cancers than CA125. In the literature, CA125 levels were found to be high in one-fourth of pregnant women and one-third of patients with acute pelvic inflammatory disease. 24 Huhtinen et al. 25 also reported that HE4 was less increased in benign conditions and more specific to ovarian cancer than CA125. In our study, the reference limits for HE4 and CA125 were established by age group for a homogeneous population consisting of healthy Asian female hospital workers. Reference limits of HE4 tended to be slightly elevated with age, whereas those of CA125 were considerably decreased in the oldest age group. For example, the 97.5th percentile upper reference limits of HE4 according to the age groups ranged from 31.8 to 34.0 pmol/l, while those of CA125 were between 27.8 and 41.1 U/mL; thus, the age-related variation was greater for CA125. Lowe et al. 26 previously researched the effects of personal characteristics on serum biomarker levels in postmenopausal women or women older than 50 and reported that, while age did not influence CA125 levels, HE4 concentrations were slightly affected. In the same study, levels of both markers were not associated with other personal factors including height, weight, ethnic group and hormone replacement therapy. Therefore, the slight elevation of upper reference limits for HE4 in older age groups could be mainly affected by the subjects age. However, number of subjects in older age group is relatively smaller than younger

7 1142 Reference Ranges for HE4 and CA125 Figure 2. ROC curves of HE4 and CA125 for differentiating ovarian cancer (n ¼ 66) from benign gynecologic diseases (n ¼ 67) and healthy subjects (n ¼ 165). The ROC-AUC values were 0.94 (95% CI, ; p < ) for HE4 and 0.86 (95% CI, ; p < ) for CA125, and these were statistically different (p ¼ ). Mean ages were not different between those three groups by one-way ANOVA test (for all three groups, mean age ¼ 52.1 years, p ¼ ). age groups in our study, and this difference of sample size between the age groups might also affect our result. Additionally, lower CA125 concentrations in the oldest age group of our study might be caused by their menopausal status. Unfortunately, we could not collect data regarding the menopausal status of the study subjects. Nevertheless, CA125 levels in postmenopausal women have been found to be significantly lower than those of premenopausal individuals in a few studies, 27,28 and accordingly, the sensitivity and specificity of CA125 for detecting ovarian cancer were higher in postmenopausal women than in premenopausal subjects. 28 When using the 97.5th percentile reference limits calculated from our data as a cutoff level (33.2 pmol/l for HE4 and 38.3 U/mL for CA125), the sensitivity and specificity for discriminating patients with ovarian cancer from healthy individuals and benign gynecologic diseases were 90.9% and 94.1% for HE4, and those for CA125 were 72.7% and 94.4%. Hence, HE4 showed greater sensitivity for detecting ovarian cancer than CA125, with similar specificities. However, other cutoff points subdivided for older age groups could be necessary, as the median age of ovarian cancer patients in our study was 55 years old. From our data, the 97.5th percentile reference limits specific to the age group of 50 to 65 year-old subjects (n ¼ 121) were 32.6 pmol/l (90% CI, ) for HE4 and 27.8 U/mL (90% CI, ) for CA125. When using these values as cutoff points for differentiating ovarian cancer from healthy controls and patients with benign diseases, the sensitivity and specificity of HE4 were 90.9% (95% CI, 81.3% 96.6%) and 93.5% (95% CI, 92.5% 94.5%), which show equal sensitivity and slightly decreased specificity compared to those at the cutoff level calculated from the group of all healthy subjects in our study. The sensitivity and specificity of CA125 by the cutoff from the oldest age group were 81.8% (95% CI, 70.4% 90.2%) and 87.2% (95% CI, 85.8% 88.5%). HE4 consequently showed better diagnostic performance than CA125 when using cutoffs specific to the oldest age group. Overall, HE4 seems to be a more useful biomarker for detecting ovarian cancer than CA125 because it showed lesser age-related variation and greater sensitivity and/or specificity. Additionally, we compared the diagnostic performances of HE4 and CA125 using ROC analysis, and the resultant ROC AUC value of HE4 for distinguishing ovarian cancer from benign gynecologic diseases and healthy individuals with the same mean age as ovarian cancer patients of our study was 0.94 and that of CA125 was Similar to our data, the ROC-AUC values of HE4 for differentiating ovarian cancer from benign diseases ranged from 0.86 to 0.91 and those of CA125 were between 0.81 and 0.84 in several previous studies. 18,20,29,30 In some of the same studies and a few others, the ROC-AUC values for discriminating ovarian cancers from healthy controls were 0.93 to 0.99 for HE4 and 0.86 to 0.94 for CA ,29 31 Differences of the results between studies might have been caused by differing numbers

8 Park et al References 1. Jemal A, Siegel R, Ward E, Hao Y, Xu J, Thun MJ. Cancer statistics, CA Cancer J Clin 2009;59: Curtin JP. Management of the adnexal mass. Gynecol Oncol 1994;55:S National Institutes of Health Consensus Development Conference Statement. Ovarian cancer: screening, treatment, and follow-up. Gynecol Oncol 1994;55:S Schink JC. Current initial therapy of stage III and IV ovarian cancer: challenges for managed care. Semin Oncol 1999;26: Earle CC, Schrag D, Neville BA, Yabroff KR, Topor M, Fahey A, Trimble EL, Bodurka DC, Bristow RE, Carney M, Warren JL. Effect of surgeon specialty on processes of care and outcomes for ovarian cancer patients. J Natl Cancer Inst 2006;98: Paulsen T, Kjaerheim K, Kaern J, Tretli S, Trope C. Improved short-term survival for advanced ovarian, tubal, and peritoneal cancer patients operated at teaching hospitals. Int J Gynecol Cancer 2006; 16(Suppl 1): Carney ME, Lancaster JM, Ford C, Tsodikov A, Wiggins CL. A populationbased study of patterns of care for ovarian cancer: who is seen by a gynecologic oncologist and who is not? Gynecol Oncol 2002;84: Rosen DG, Wang L, Atkinson JN, Yu Y, Lu KH, Diamandis EP, Hellstrom I, Mok SC, Liu J, Bast RC, Jr. Potential markers that complement expression of CA125 in epithelial ovarian cancer. Gynecol Oncol 2005;99: Maggino T, Gadducci A, D Addario V, Pecorelli S, Lissoni A, Stella M, Romagnolo C, Federghini M, Zucca S, Trio D, et al. Prospective multicenter study on CA 125 in postmenopausal pelvic masses. Gynecol Oncol 1994;54: Kobayashi H, Yamada Y, Sado T, Sakata M, Yoshida S, Kawaguchi R, Kanayama S, Shigetomi H, Haruta S, Tsuji Y, Ueda S, Kitanaka T. A randomized study of screening for ovarian cancer: a multicenter study in Japan. Int J Gynecol Cancer 2008; 18: and characteristics of patients and controls, including types of benign diseases, cancer stages and histologic subtypes. In our data, elevation of HE4 and CA125 was obvious in some histologic subtypes, including serous cystadenoma (n ¼ 21, median HE4 ¼ pmol/l, median CA125 ¼ U/ ml) and papillary serous cystadenocarcinoma (n ¼ 21, median HE4 ¼ 61.2 pmol/l, median CA125 ¼ 65.4 U/mL), but elevation of either or both of the markers were not evident in other histologies; for example, mucinous cystadenocarcinoma (n ¼ 5, median HE4 ¼ 26.7 pmol/l, median CA125 ¼ 39.5 U/mL), clear cell carcinoma (n ¼ 4, median HE4 ¼ 46.4 pmol/l, median CA125 ¼ U/mL), or endometrioid adenocarcinoma (n ¼ 2, median HE4 ¼ 54.4 pmol/l, median CA125 ¼ 33.8 U/mL). However, levels of markers between the histologic subtypes in our study may not be compared due to small numbers of cases and uneven distributions of FIGO stages in respective subtypes. Previously Kobel et al. 32 had investigated the expression patterns of 21 different biomarkers in accordance with histologic subtypes and reported that most of the markers differed significantly between subtypes. Our study was conducted to establish reference limits for HE4 and CA125, which can be useful for interpreting the results of these biomarkers. Analytical and diagnostic performances were also evaluated to validate the utilities of biomarkers and their reference limits. However, we were not able to obtain information on phases of the menstrual cycle or menopausal status, and thus we were not able to establish separate reference limits for individuals with or without menstruation, nor for individuals with phases of the menstrual cycle. In a recent study, HE4 was expressed more in ovulatory phases of the hormonal cycle (mean 45.3 pmol/l) than in follicular phases (mean 39.1 pmol/l). 33 Therefore, conditions affecting ovulation or hormonal regulation (i.e., menopausal status, medication of oral contraceptives) could also influence on the serum levels of HE4. Nevertheless, our study showed that levels of CA125 were obviously lower in the 50 to 65 year-old group, and those of HE4 were not considerably different between age groups, which might be explained by the complementary effects of ovulation and age on HE4 levels. Therefore, we suggest that using reference limits of HE4 for all adult ages and applying the age-specific reference limits of CA125 to result interpretation can be helpful to detecting ovarian cancer in various age sets. Additionally, subjects of this study was limited to Asian, thus our results may not be applicable to other ethnic groups. Further studies on reference limits according to menopausal status, phases of menstrual cycle and ethnic differences could also be helpful for defining the characteristics of both markers in various populations. In conclusion, the new automated HE4 assay evaluated in our study showed good analytical performance, and HE4 had superior diagnostic performances to CA125 for discriminating ovarian cancer from benign gynecologic diseases. The reference limits established with our data could be practically used as cutoff levels to facilitate more accurate diagnosis of ovarian cancer and to rule out benign gynecologic diseases in Asian population. Acknowledgements The authors thank Abbott Laboratories (Abbott Park, IL, USA) for providing HE4 detection kits. 11. Bast RC, Jr. Status of tumor markers in ovarian cancer screening. J Clin Oncol 2003;21:200s-5s. 12. Yousef GM, Polymeris ME, Yacoub GM, Scorilas A, Soosaipillai A, Popalis C, Fracchioli S, Katsaros D, Diamandis EP. Parallel overexpression of seven kallikrein genes in ovarian cancer. Cancer Res 2003; 63: Kim JH, Skates SJ, Uede T, Wong KK, Schorge JO, Feltmate CM, Berkowitz RS, Cramer DW, Mok SC. Osteopontin as a potential diagnostic biomarker for ovarian cancer. JAMA 2002;287: Healy DL, Burger HG, Mamers P, Jobling T, Bangah M, Quinn M, Grant P, Day AJ, Rome R, Campbell JJ. Elevated serum inhibin concentrations in postmenopausal women with ovarian tumors. N Engl J Med 1993;329: Gadducci A, Ferdeghini M, Prontera C, Moretti L, Mariani G, Bianchi R, Fioretti P. The concomitant determination of different tumor markers in patients with epithelial ovarian cancer and benign ovarian masses:

9 1144 Reference Ranges for HE4 and CA125 relevance for differential diagnosis. Gynecol Oncol 1992;44: Hellstrom I, Hellstrom KE. SMRP and HE4 as biomarkers for ovarian carcinoma when used alone and in combination with CA125 and/or each other. Adv Exp Med Biol 2008;622: Schummer M, Ng WV, Bumgarner RE, Nelson PS, Schummer B, Bednarski DW, Hassell L, Baldwin RL, Karlan BY, Hood L. Comparative hybridization of an array of 21,500 ovarian cdnas for the discovery of genes overexpressed in ovarian carcinomas. Gene 1999;238: Hellstrom I, Raycraft J, Hayden-Ledbetter M, Ledbetter JA, Schummer M, McIntosh M, Drescher C, Urban N, Hellstrom KE. The HE4 (WFDC2) protein is a biomarker for ovarian carcinoma. Cancer Res 2003;63: Drapkin R, von Horsten HH, Lin Y, Mok SC, Crum CP, Welch WR, Hecht JL. Human epididymis protein 4 (HE4) is a secreted glycoprotein that is overexpressed by serous and endometrioid ovarian carcinomas. Cancer Res 2005;65: Moore RG, Brown AK, Miller MC, Skates S, Allard WJ, Verch T, Steinhoff M, Messerlian G, DiSilvestro P, Granai CO, Bast RC, Jr. The use of multiple novel tumor biomarkers for the detection of ovarian carcinoma in patients with a pelvic mass. Gynecol Oncol 2008;108: Moore RG, Jabre-Raughley M, Brown AK, Robison KM, Miller MC, Allard WJ, Kurman RJ, Bast RC, Skates SJ. Comparison of a novel multiple marker assay vs the Risk of Malignancy Index for the prediction of epithelial ovarian cancer in patients with a pelvic mass. Am J Obstet Gynecol 2010;203:228 e Hellstrom I, Heagerty PJ, Swisher EM, Liu P, Jaffar J, Agnew K, Hellstrom KE. Detection of the HE4 protein in urine as a biomarker for ovarian neoplasms. Cancer Lett 2010;296: Scholler N, Lowe KA, Bergan LA, Kampani AV, Ng V, Forrest RM, Thorpe JD, Gross JA, Garvik BM, Drapkin R, Anderson GL, Urban N. Use of yeast-secreted in vivo biotinylated recombinant antibodies (Biobodies) in bead-based ELISA. Clin Cancer Res 2008;14: Halila H, Stenman UH, Seppala M. Ovarian cancer antigen CA 125 levels in pelvic inflammatory disease and pregnancy. Cancer 1986;57: Huhtinen K, Suvitie P, Hiissa J, Junnila J, Huvila J, Kujari H, Setala M, Harkki P, Jalkanen J, Fraser J, Makinen J, Auranen A, et al. Serum HE4 concentration differentiates malignant ovarian tumours from ovarian endometriotic cysts. Br J Cancer 2009;100: Lowe KA, Shah C, Wallace E, Anderson G, Paley P, McIntosh M, Andersen MR, Scholler N, Bergan L, Thorpe J, Urban N, Drescher CW. Effects of personal characteristics on serum CA125, mesothelin, and HE4 levels in healthy postmenopausal women at high-risk for ovarian cancer. Cancer Epidemiol Biomarkers Prev 2008;17: Bon GG, Kenemans P, Verstraeten R, van Kamp GJ, Hilgers J. Serum tumor marker immunoassays in gynecologic oncology: establishment of reference values. Am J Obstet Gynecol 1996;174: Malkasian GD, Jr., Knapp RC, Lavin PT, Zurawski VR, Jr, Podratz KC, Stanhope CR, Mortel R, Berek JS, Bast RC, Jr, Ritts RE. Preoperative evaluation of serum CA 125 levels in premenopausal and postmenopausal patients with pelvic masses: discrimination of benign from malignant disease. Am J Obstet Gynecol 1988;159: Scholler N, Crawford M, Sato A, Drescher CW, O Briant KC, Kiviat N, Anderson GL, Urban N. Bead-based ELISA for validation of ovarian cancer early detection markers. Clin Cancer Res 2006; 12: Shah CA, Lowe KA, Paley P, Wallace E, Anderson GL, McIntosh MW, Andersen MR, Scholler N, Bergan LA, Thorpe JD, Urban N, Drescher CW. Influence of ovarian cancer risk status on the diagnostic performance of the serum biomarkers mesothelin, HE4, and CA125. Cancer Epidemiol Biomarkers Prev 2009;18: Montagnana M, Lippi G, Ruzzenente O, Bresciani V, Danese E, Scevarolli S, Salvagno GL, Giudici S, Franchi M, Guidi GC. The utility of serum human epididymis protein 4 (HE4) in patients with a pelvic mass. J Clin Lab Anal 2009; 23: Kobel M, Kalloger SE, Boyd N, McKinney S, Mehl E, Palmer C, Leung S, Bowen NJ, Ionescu DN, Rajput A, Prentice LM, Miller D, et al. Ovarian carcinoma subtypes are different diseases: implications for biomarker studies. PLoS Med 2008;5:e Anastasi E, Granato T, Marchei GG, Viggiani V, Colaprisca B, Comploj S, Reale MG, Frati L, Midulla C. Ovarian tumor marker HE4 is differently expressed during the phases of the menstrual cycle in healthy young women. Tumour Biol 2010; 31:411 5.