Clinica Chimica Acta 440 (2015) 57–63
Contents lists available at ScienceDirect
Clinica Chimica Acta journal homepage: www.elsevier.com/locate/clinchim
Development of a multimarker assay for differential diagnosis of benign and malignant pelvic masses Xinliang Chen a,1, Hui Zhou b,1, Rui Chen c, Jian He a, Ying Wang a, Lisi Huang a, Longqiaozi Sun a, Chaohui Duan a, Xiaohong Luo a, Haiyan Yan a,⁎ a b c
Department of Clinical Laboratory, Sun Yat-Sen Memorial Hospital, Guangzhou, Guangdong 510080, PR China Department of Gynaecology, Sun Yat-Sen Memorial Hospital, Guangzhou, Guangdong 510080, PR China Department of Respiratory Medicine, Sun Yat-Sen Memorial Hospital, Guangzhou, Guangdong 510080, PR China
a r t i c l e
i n f o
Article history: Received 30 June 2014 Received in revised form 27 October 2014 Accepted 13 November 2014 Available online 15 November 2014 Keywords: Pelvic mass Ovarian cancer Uterine cancer Tumor markers HE4 ROMA
a b s t r a c t Background: HE4, a novel tumor marker for detecting ovarian cancer, has been recently applied to clinical practice. However, the comprehensive evaluation of HE4 combined with other markers is still missing. We evaluated an optimal mode of HE4 employment for differential diagnosis of benign and malignant pelvic masses. Methods: Serum HE4, CA125, CA153, CA199, CA211 and CA724 were measured from 232 patients with pelvic messes (100 malignant masses, 132 benign diseases), and the risk of ovarian malignancy algorithm (ROMA) was also calculated. Receiver operating characteristic curves (ROC), the area under the curve (AUC), sensitivity and specificity were estimated. Results: The combination of HE4 and CA125 (AUC of 0.963, sensitivity of 96.6%, specificity of 65.7%) provided the best differential power in diagnosing ovarian cancer. ROMA performed better in the diagnosis of pelvic masses (AUC of 0.917, sensitivity of 82.0%, specificity of 78.8%) and uterine cancer (AUC of 0.838, sensitivity of 82.0%, specificity of 60.0%) compared with applying HE4 and CA125 individually. Conclusion: The optimal cut-off values (CA125: 93.2 U/ml, HE4: 87.6 pmol/l, ROMA: 18.1% for pre- and 31.5% for postmenopausal women), simultaneous use of CA125 and HE4 complemented by ROMA showed better performance than the traditional detection modes for differential diagnosis of ovarian cancer. We also observed that ROMA added more accuracy for differentiating the benign and malignant pelvic masses and auxiliary diagnosis of uterine cancer. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Pelvic masses are often detected in pelvic examinations among females with ovarian cancer and uterine cancer, both of which are the most lethal types of cancer. However, due to non-specific early symptoms and lack of reliable screening tools, ovarian cancer often reaches an advanced stage (FIGO III–IV) when it is detected and its 5-year survival rate is only 20%–30% [1]. Thus, a diagnosis at early stage is critical, which requires a high sensitivity and specificity [2]. The screening and diagnosis of ovarian cancer are based on physical examination, semeiology, tumor markers, imaging and proteomics, etc. In the last few decades, several tumor markers have been evaluated in patients with ovarian cancer, among which, only CA125 correlates closely with ovarian cancer [3]. Since CA125 is also upregulated in many other tumors and therefore results in the reduction of sensitivity
and specificity, currently attention has been moved onto a novel marker with higher specificity: human epididymis-specific protein 4(HE4). HE4 is a glycoprotein overexpressed in epithelial ovarian cancer, especially serous and endometrioid adenocarcinoma [4]. HE4 is associated with cancer cell adhesion, migration and tumor growth [5] and serves as a valuable prognostic factor for the overall survival in patients with epithelial ovarian cancer and also an early indicator of the recurrence of ovarian cancer [3,6–9]. Previous reports showed that HE4 used in conjunction with CA 125 improves specificity for ovarian cancer [10]. The risk of ovarian malignancy algorithm (ROMA), a simple index based on CA125 and HE4, is helpful to improve the accuracy in differentiating benignancy from malignancy [4,11,12]. Besides, HE4 combined with CA125 has been reported as a new tool for preoperative evaluation and postoperative surveillance of endometrial cancer patients [13,14]. 2. Materials and methods
⁎ Corresponding author at: Department of Clinical Laboratory, Sun Yat-Sen Memorial Hospital, 107 Yaniiangxi Road, Guangzhou, Guangdong 510080, PR China. E-mail address: [email protected] (H. Yan). 1 These authors contributed to the work equally and should be regarded as co-first authors.
http://dx.doi.org/10.1016/j.cca.2014.11.013 0009-8981/© 2014 Elsevier B.V. All rights reserved.
2.1. Patients, and clinical pathology information A total of 232 female patients diagnosed with pelvic masses were scheduled to have surgeries between March 2012 and March 2014 in
58
X. Chen et al. / Clinica Chimica Acta 440 (2015) 57–63
our hospital. All pathologic tissue specimens were confirmed by histopathological evaluation postoperatively. Patients with preoperativelyknown relapse of a previous cancer or with an active cancer other than ovarian cancer were excluded. All of the patients provided informed consents. The cases were divided into the following four groups according to the pathological types. From this, 232 patients, 60 patients were pathologically confirmed postoperatively with ovarian cancer, including the following histological subtypes: serous (n = 27), mucinous (n = 6), endometrioid (n = 11), clear cell (n = 7), undifferentiated (n = 9) (aged 23–81 y, median 53 y). Seventy patients were diagnosed with benign ovarian diseases including ovarian chocolate cysts (n = 11), simple cyst (n = 5), ovarian endometriosis cysts (n = 23), mature teratoma (n = 20), mucinous cystadenoma (n = 5), serous cystadenoma (n = 6) (age 17–77 y, median 33 y). Forty patients were diagnosed with uterine cancer in the following histological subtypes: leiomyosarcoma (n = 3), endometrial cancer (n = 37), (age 33–81 y, median 51 y). Sixty-two patients had benign uterine diseases including hysteromyoma (n = 43), endometrial polyp (n = 10), endometrial hyperplasia (n = 5), and endometrioma (n = 4) (age 24–61 y, median 41 y) (Table 1). 2.2. Methods Serum samples were collected preoperatively from our patients. Three milliliters of fasting blood samples were collected in the morning. All samples were centrifuged at 2000 ×g for 10 min. The serum without hemolysis and lipemia was separated to be stored at − 20 °C until measurement. Following the standard operating procedure (SOP) provided by the manufacturer, serum HE4, CA125, CA153, CA199, CA211 and CA724 were quantitatively measured by electrochemiluminescence immunoassay analyzer (ECLIA, Roche E170). The reference values of each index are HE4 b 140 pmol/l, CA125 b35 U/ml, CA153 b 25 U/ml, CA199 b34 U/ml, CA211 b3.3 ng/ml and CA724 b 7 U/ml. Cases with marker levels above threshold levels were considered to have a positive result. Three clinical routine CDMs (combined detection modes) were used to Table 1 Patient age, menopause status and tumor characteristics. Tumor characteristics Benign ovarian diseases Median age (range) Simple Endometriosis Mature teratoma Serous Mucinous Chocolate cysts Total Ovarian cancer Median age (range) Serous Mucinous Endometrioid Clear cell Undifferentiated Total Benign uterine diseases Median age (range) Hysteromyoma Endometrial polyp Endometrial hyperplasia Endometrioma Total Uterine cancer Median age (range) Leiomyosarcoma Endometrial Total Total
Premenopausal
Postmenopausal
All (%)
31 (24–47) 3 21 15 2 2 5 48
57 (49–77) 2 2 5 4 3 6 22
34 (24–77) 5 (7.1) 23 (32.9) 20 (28.6) 6 (8.6) 5 (7.1) 11 (15.7) 70 (30.2)
45 (23–51) 5 3 5 3 4 20
57 (49–81) 22 3 6 4 5 40
53 (23–81) 27 (45.0) 6 (10) 11 (18.3) 7 (11.7) 9 (15.0) 60 (25.9)
39 (24–49) 39 9 4 4 56
54 (50–61) 4 1 1 0 6
41 (24–61) 43 (69.4) 10 (16.1) 5 (8.1) 4 (6.4) 62 (26.7)
46 (33–50) 3 16 19 153
56 (50–81) 0 21 21 89
51 (33–81) 3 (7.5) 37 (92.5) 40 (17.2) 232
compare with HE4. These 3 modes include CDM1 (CA125 + CA153), CDM2 (CA125 + CA153 + CA724) and CDM3 (CA125 + CA153 + CA724 + CA211). The test was defined as positive if one of the markers was positive in combined detection, and negative if all of the markers were negative. Serum HE4, CA125 levels and menopausal status were used to calculate ROMA. Menopause was considered when there was suspension of menstrual bleeding for at least 12 months [15]. A predictive index for ovarian cancer was calculated using the following formulae established by Moore et al. [16]: premenopausal PI = − 12 + 2.38 × ln(HE4) + 0.0626 × ln(CA125), postmenopausal PI = − 8.09 + 1.04 × ln(HE4) + 0.732 × ln(CA125). The ROMA value (predictive value) was calculated using the following formula: ROMA (%) = ePI / (1 + ePI) × 100. According to the indications of the HE4 manufacturer, indexes of at least 11.4% and 29.9% indicate a high risk of ovarian cancer in both pre- and postmenopausal women. 2.3. Statistics All calculations were performed with SPSS17.0 statistical software. The Wilcoxon–Mann–Whitney test and the Kruskal–Wallis test were used to compare biomarker distributions across two and more than two subgroups of patients, respectively; The χ2 test was used to evaluate the qualitative data (the frequency of the positive or negative specimens); sensitivity and specificity for CA125, HE4, ROMA, and combined detection modes were calculated. The predicted probabilities for each mode were used to construct receiver operating characteristic (ROC) curves. The area under the curve (AUC) values and the optimal cut-off values were calculated. A p b 0.05 was considered statistically significant. 3. Results 3.1. The evaluation of serum tumor markers in patients with either ovarian cancer or uterine cancer Median levels of different tumor markers in pre- and postmenopausal women are shown in Table 2. The median values of HE4, CA125, ROMA and positive rates are significantly higher among the patients with ovarian cancer compared with the benign cases both in pre- and postmenopausal women (p b 0.05). The median values of CA153, CA724, CA211 and positive rates for patients with ovarian cancer are significantly higher than patients with benign ovarian diseases (p b 0.05) except CA199 (p N 0.05). The median values of CA125, HE4 and ROMA for postmenopausal women were not significantly different from premenopausal women in the ovarian cancer group (p N 0.05). The HE4 and ROMA are statistically higher in patients with uterine cancer than in patients with benign uterine diseases (p b 0.05). 35.0% of the patients with uterine cancer had elevated values of CA125, and only 20.0% had elevated values of HE4. 3.2. The characterization of CA125 and HE4 in different ovarian cancer types We evaluated the levels of CA125 and HE4 in various ovarian cancer types. We found that the median HE4 for 46 patients in late stages (stages III + IV) are significantly higher than 14 patients in early stages (stages I + II) (p b 0.05) (Figs. 1, 2). The classification of different stages is based on the criteria of the International Federation of Gynecology and Obstetrics (FIGO) [17]. The levels of CA125 and HE4 in ovarian cancer of various pathological types (Table 3) are significantly different (p b 0.05). In particular, the median values of CA125 and HE4 are higher in the serous of ovarian cancer compared with the clear cell and mucinous types (p b 0.05); the positive rates of HE4 for the latter 2 types are only 57.1% and 16.7%, respectively. Besides, the median value of CA125 is also higher in endometriosis and chocolate cyst, compared
X. Chen et al. / Clinica Chimica Acta 440 (2015) 57–63
59
Table 2 Baseline data. Benign ovarian diseases Numbers (pre/postmenopausal) HE4 (pmol/l) (median, range) All Premenopausal Postmenopausal All positive (%) Premenopausal positive (%) Postmenopausal positive (%) CA125 (U/ml) (median, range) All Premenopausal Postmenopausal All positive (%) Premenopausal positive (%) Postmenopausal positive (%) ROMA (%) All Premenopausal Postmenopausal All positive (%) Premenopausal positive (%) Postmenopausal positive (%) CA153 (U/ml) (median, range) CA153 positive (%) CA199 (U/ml) (median, range) CA199 positive (%) CA724 (U/ml) (median, range) CA724 positive (%) CA211 (ng/ml) (median, range) CA211 positive (%)
70 (48/22)
Ovarian cancer
p
60 (20/40)
Benign uterine diseases 63 (57/6)
Uterine cancer
p
40 (19/21)
52.8 (31.8–155.8) 52.6 (31.8–79.4) 54.0 (37.6–155.8) 1 (1.4%) 0 (0%) 1 (4.5%)
242.6 (45.8–3109.0) 201.1a (61.8–2574.0) 270.9a (45.8–3109.0) 44 (73.3%) 13 (65%) 31 (77.5%)
0.000 0.000 0.000 0.000 0.000 0.000
52.2 (30.9–86.8) 50.9 (30.9–86.8) 61.5 (49.2–86.8) 0 (100%) 0 (100%) 0 (100%)
82.3 (37.6–519.2) 68.2 (38.2–186.3) 90.0 (37.6–519.2) 8 (20%) 2 (10.5%) 6 (28.6%)
0.000 0.001 NS 0.000 0.015 NS
21.6 (4.2–228.0) 23.2 (8.7–228.0) 21.4 (4.2–175.8) 23 (32.8%) 18 (37.5%) 5 (22.7%)
390.0 (13.1–13,280) 285.6a (18.6–11,423) 444.8a (13.1–13,280) 56 (93.3%) 19 (95%) 37 (92.5%)
0.000 0.000 0.000 0.000 0.000 0.000
21.0 (5.4–827.7) 21.1 (5.4–827.7) 20.1 (5.7–153.1) 11 (17.7%) 10 (17.9%) 1 (16.7%)
25.3 (8.6–410.8) 24.1 (9.4–105.4) 28.4 (8.6–410.8 14 (35%) 5 (26.3%) 9 (42.9%)
NS NS NS 0.049 NS NS
8.9 (2.7–36.4) 8.7 (2.7–20.6) 14.0 (3.9–36.4) 12 (17.1%) 7 (14.6%) 5 (22.7%) 9.7 (4.0–50.4) 3/(4.3%) 18.3 (1.0–506.8) 20 (28.6%) 1.7 (0.8–22.0) 7 (10%) 1.5 (0.7–6.9) 2 (2.9%)
86.5 (6.2–99.9) 70.4a (13.5–99.9) 87.4a (6.2–99.9) 59 (98.3%) 20 (100%) 39 (97.5%) 31.9 (3.7–1823.0) 36 (60%) 15.9 (0.6–1000.0) 17 (28.3%) 14.9 (0.6–3000.0) 32 (53.3%) 2.2 (0.7–16.9) 15 (25%)
0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 NS NS 0.000 0.000 0.000 0.000
8.3 (2.5–48.0) 7.9 (2.5–23.6) 17.2 (5.9–48.0) 14 (22.6%) 12 (21.4%) 2 (33.3%)
16.0 (4.1–90.0) 14.6 (4.1–65.6) 27.2 (11.5–90.0) 24 (60%) 12 (63.2%) 12 (57.1%)
0.000 0.001 NS 0.000 0.001 NS
a The CA125, HE4 and ROMA values for postmenopausal women were not significantly different from premenopausal women in the ovarian cancer group (p = 0.572, p = 0.388, p = 0.230).
with the other types of benign ovarian diseases. However, there is no statistically significant difference of HE4 observed between different types of benign ovarian diseases (p N 0.05).
3.3. The application of normal cut-off values and optimal cut-off values for diagnosis of pelvic masses We used the normal cut-off values and the optimal cut-off values determined by ROC to calculate the diagnostic values. The sensitivity and specificity of CA125, HE4, ROMA and CDMs of 3 clinical routines are shown in Table 4. Three modes include CDM1 (CA125 + CA153), CDM2 (CA125 + CA153 + CA724) and CDM3 (CA125 + CA153 + CA724 + CA211). The ROC and AUC of the 6 markers are presented in Fig. 3. HE4 exhibits the highest AUC (0.963) in single detection, while the lowest AUC is associated with CA199 (AUC 0.488) (Fig. 3). Thus,
Fig. 1. CA125 of different stages.
due to the weak efficiency of diagnosis, CDM is not recommended for joint application. When used to analyze benign ovarian masses versus ovarian cancer, HE4 had the highest specificity (98.6%), but at the cost of an unsatisfactory low sensitivity (73.3%). CA125 had a higher sensitivity (93.3%), but a lower specificity (67.1%) and AUC (0.926) compared with HE4 (p b 0.05). When the optimal cut-off values of CA125 and HE4 were applied, the specificity of CA125 and the sensitivity of HE4 increased to 88.6% and 88.3%. ROMA had the higher specificity (80.0%) and AUC (0.972) when compared with the combined detection of CA125 and HE4 (65.7%, 0.963). The use of the optimal cut-off values increased the specificity of the combined detection of CA125 and HE4 (87.1%). The sensitivities of 3 CDMs had no significant difference from CA125 alone (p N 0.05), and the specificities were lower (p b 0.05). The AUCs for the 3 CDMs were also lower than that for ROMA (Fig. 4). Besides,
Fig. 2. HE4 of different stages.
60
X. Chen et al. / Clinica Chimica Acta 440 (2015) 57–63
Table 3 CA125 and HE4 of different pathological types of pelvic masses. CA125 (U/ml)
Ovarian cancer Serous Endometrioid Clear cell Mucinous p (Kruskal–Wallis) Benign ovarian diseases Chocolate cyst Endometriosis Teratoma Simple Serous Mucinous p (Kruskal–Wallis)
HE4 (pmol/l)
Median (range)
Positive (%)
Median (range)
Positive (%)
931.3⁎ (25.8–13,280) 301.5 (20.5–11,423) 80.1 (41.2–3458) 155.3 (13.1–198.9 0.004
26 (96.3%) 10 (90.9%) 7 (100.0%) 4 (66.7%)
390.5⁎ (97.9–1985) 216.8 (45.8–2574) 194.8 (65.8–1166) 79.1 (54.8–227.2) 0.010
22 (81.5%) 8 (72.7%) 4 (57.1%) 1 (16.7%)
45.0⁎⁎ (22.5–162.0) 49.7⁎⁎ (10.9–228.0) 13.8 (8.0–26.2) 13.6 (13.0–65.4) 14.9 (4.2–117.6) 22.6 (9.8–175.8) 0.000
7 (63.6%) 13 (56.5%) 0 (0%) 1 (20.0%) 1 (16.7%) 1 (20.0%)
53.9 (39.4–79.4) 53.0 (38.1–70.3) 49.1 (31.8–75.7) 43.4 (38.5–74.8) 52.3 (37.9–87.4) 76.4 (41.1–155.8) NS
0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 1 (20.0%)
⁎ p b 0.05 (compared with clear cell and mucinous). ⁎⁎ p b 0.05 (compared with other types of benign ovarian diseases).
ROMA had a high sensitivity (100.0%) and specificity (85.4%) in premenopausal women (p b 0.05), and the AUC for ROMA in premenopausal women (0.993) was higher than that in postmenopausal women (0.963) (Figs. 5, 6). When the cut-off values of ROMA were set as 18.1% for premenopausal women and 31.5% for postmenopausal women, ROMA presents both high sensitivity and specificity (91.7%, 97.1%). When analyzing uterine malignant tumors, the specificities for CA125, HE4 and ROMA were 82.3%, 100%, 77.4%, respectively, but the sensitivities were 35.0%, 20.0%, 60.0%, respectively. ROMA had the highest AUC (0.838) followed by HE4 (0.808) and CA125 (0.580) (Fig. 7, Table 4). When differentiating benign and malignant pelvic masses, HE4 had the highest specificity (99.2%), and the lowest sensitivity (52.0%);
ROMA had a low specificity (78.8%) and the highest sensitivity (82.0%); The highest AUC was calculated for ROMA (0.917), followed by HE4 (0.806) and CA125 (0.711) (Fig. 8, Table 4). When the cut-off values of ROMA were set as 24.6%, ROMA had the highest specificity (97.7%) for differential diagnosis of malignant pelvic masses. 4. Discussion The serum tumor marker CA125 is commonly used to predict pelvic masses. But due to the limited specificity and sensitivity, most clinicians use a combination of multiple tumor markers to differentiate benign and malignant pelvic masses. CA125 is often combined with CA153, CA199, CA724 and CA211 to screen patients for ovarian cancer. As a novel tumor marker in the diagnosis of ovarian cancer, HE4 has been
Table 4 Sensitivity, specificity for CA125, HE4, ROMA, CA153, CA724, CA211 and three clinical routine combine detection modes (CDM) for women with pelvic masses. Tumor marker
Normal cut-off values⁎ Sensitivity (%)
OC vs benign ovarian diseases CA125 All Premenopausal Postmenopausal HE4 All Premenopausal Postmenopausal ROMA All Premenopausal Postmenopausal CA153 CA724 CA211 CA125 + HE4 CDM1 (CA125 + CA153) CDM2 (CA125 + CA153 + CA724) CDM3 (CA125 + CA153 + CA724 + CA211) UC vs benign uterine diseases CA125 HE4 ROMA Malignant pelvic masses vs benign pelvic masses CA125 HE4 ROMA
Optimal cut-off values Specificity (%)
93.2 U/ml 93.2 U/ml 110.9 U/ml
Optimal cut-off values Sensitivity (%)
Specificity (%)
85.0 85.0 82.5
88.6 87.5 95.5
88.3 90.0 87.5
97.1 95.8 95.5
91.7 95.0 92.5 78.3 53.3 73.3 93.3 96.7 96.7 96.7
97.1 97.9 95.5 91.4 98.6 70.0 87.1 80.0 78.6 57.1
93.3 95.0 92.5
67.1 62.5 77.3
73.3 65.0 77.5
98.6 100.0 95.5
96.7 100.0 95.0 60.0 53.3 25.0 96.6 93.3 93.3 95.0
80.0 85.4 68.2 95.7 90.0 97.1 65.7 64.3 57.0 55.7
35.0 20.0 60.0
82.3 100 77.4
33.6 U/ml 66.4 pmol/l 14.2%
40.0 70.0 75.0
82.3 90.3 85.5
70.0 52.0 82.0
73.5 99.2 78.8
78.3 U/ml 81.9 pmol/l 24.6%
58.0 75.0 75.0
88.6 96.2 97.7
87.6 pmol/l 72.3 pmol/l 97.5 pmol/l
18.1% 31.5% 19.3 U/ml 11.4 U/ml 1.83 ng/ml
⁎ Normal cut-off value: CA125 b35 U/ml; HE4 b140 pmol/l; premenopausal women: ROMA b11.4%, postmenopausal women: ROMA b29.9%; CA153 b25 U/ml; CA724 b7 U/ml; and CA211 b3.3 ng/ml.
X. Chen et al. / Clinica Chimica Acta 440 (2015) 57–63
Fig. 3. The ROC of single tumor marker detection of ovarian cancer vs benign ovarian diseases.
used in the clinical practice only for a short period of time. In this study, we aimed to investigate the diagnosis efficiency and the optimal detection mode of HE4. As shown in Table 2, the levels of HE4, CA125 and ROMA were significantly higher in the ovarian cancer group (p b 0.05), thus these 3 are good candidates for predicting ovarian cancer. But since the AUC for CA199 alone was only 0.488, it is not recommended for the combined detection. The difference of HE4 between pre- and postmenopausal women in the ovarian cancer group did not achieve statistical significance
Fig. 4. The ROC of multiple tumor marker combined detection of ovarian cancer vs benign ovarian diseases.
61
Fig. 5. The ROC of ROMA for premenopausal women ovarian cancer vs benign ovarian diseases.
(p N 0.05) (Table 2), which is contradictory to what Yang et al. and Moore et al. [15,18] observed. The sample size chosen for the study may have caused the difference. We also showed that the median value of HE4 of 46 patients in late stages was significantly higher than in 14 patients in the early stages (p b 0.05) (Figs. 1, 2), which is consistent with the study of Yang et al. [2]. It indicates the role of HE4 in the progress of developing ovarian cancer. Moreover, the median CA125 and HE4 values were higher in the serous ovarian cancer compared with the clear cell and mucinous
Fig. 6. The ROC of ROMA for postmenopausal women ovarian cancer vs benign ovarian diseases.
62
X. Chen et al. / Clinica Chimica Acta 440 (2015) 57–63
Fig. 7. The ROC of CA125, HE4 and ROMA for uterine cancer, uterine cancer vs benign uterine diseases.
types (p b 0.05) (Table 3). In line with our report, Drapkin et al. [19] found overexpressed HE4 in ovarian cancer, mainly in serous carcinoma and endometrial carcinoma, but not in clear cell carcinoma and mucinous carcinoma. Thus, the combination of CA125 and HE4 is specific to serous ovarian cancer. It has been reported that those patients with a renal failure have the highest HE4 positive rate in the non-ovarian benign diseases [20], which may explain why the single detection mode using HE4 with the highest AUC (0.963) still causes misdiagnosis (e.g. 8 patients with endometrial cancer had positive HE4 results in the uterine cancer group). To
optimize the specificity of HE4, we applied a new cut-off value for HE4 with a boundary value of 87.6 pmol/l. As we expected, it resulted in an increased sensitivity of 88.3%. Also, CA125 increased in many patients with ovarian benign diseases, including chocolate cyst (45.0 U/ml, 63.6%) and ovarian endometriosis (49.7 U/ml, 56.5%) and consequently had a relatively low specificity of 67.1%. But with the optimal cut-off value of 93.2 U/ml, the specificity of CA125 increased to 88.6% while retaining high sensitivity (85.0%). Next, we applied the combination of HE4 and CA125. Upon the combination, the sensitivity reached as high as 96.6%, but the specificity was only 65.7% (Table 4). The use of optimal cut-off values increased the specificity to 87.1%. Also, clinical doctors can calculate ROMA through HE4 and CA125 to classify the risk of epithelial ovarian cancer in female patients with pelvic masses. The sensitivity for ROMA in the identification of benign and malignant ovarian tumors is 96.7%. Compared with the combined detection of CA125 and HE4, ROMA had higher specificity (80.0%) and AUC (0.972). In particular, ROMA had the highest sensitivity (100.0%), specificity (85.4%) and AUC (0.993) for premenopausal women in the diagnosis of ovarian cancer (Table 4, Fig. 5). The optimal cut-off values of ROMA were 18.1% for premenopausal women and 31.5% for postmenopausal women separately. Obviously, ROMA is very useful for clinical application. Traditionally, clinicians tended to use the combined detection of multiple tumor markers for the diagnosis of ovarian cancer. However, in our study, we showed that the sensitivities of 3 kinds of routine combined detection modes (CA125 + CA153, CA125 + CA153 + CA724, CA125 + CA153 + CA724 + CA211) are similar as CA125 alone, but the specificities are lower. Besides, the AUC of these 3 modes is not significantly different from that of ROMA and the combined detection of HE4 and CA125 (Fig. 4). Thus, the traditional combined detection of multiple tumor markers is no longer an ideal diagnosis of ovarian cancer. The value of CA125 and HE4 for the diagnosis of uterine malignant tumors is not clear. Although the specificities for CA125 and HE4 were 82.3% and 100% respectively, the sensitivities were only 35.0% and 20.0% respectively (Table 4, Fig. 7). But ROMA well differentiated uterine cancer and benign uterine diseases with higher sensitivity (60.0%), specificity (77.4%) and AUC (0.838). Thus, ROMA is suitable for the auxiliary diagnosis of uterine cancer. HE4, with higher specificities (99.2%) and higher AUC (0.899), performed better than CA125 in the identification of benign and malignant pelvic masses. The sensitivity, specificity and AUC for ROMA were 82.0%, 78.8% and 0.917 respectively (Table 4, Fig. 8), which makes ROMA a good candidate for differential diagnosis of pelvic masses. Early stages of ovarian cancer show minor symptoms and can be easily misdiagnosed. But the detection of HE4 and CA125, and the application of ROMA have shown promising sensitivity and specificity for screening patients with pelvic masses. Moreover, this detection mode also differentiates benign pelvic masses from malignant ones. The current cut-off values of tumor markers which are used in China were provided by the reagent producers. The statistic data originates in a foreigner. So we suggested new cut-off values of CA125, HE4 and ROMA should be used in clinics for the Chinese. When it comes to the limitations of this study, because the cases only came from the Department of Gynecology of our hospital, a small number of cases may influence the representativeness of the experimental group. In addition, the study belongs to the retrospective study. We didn't have a follow-up of the patient's prognosis and survival rates. So whether HE4 could be used as prognostic factors of ovarian cancer is to be discussed. We suggest that a prospective study could be designed to explore the value of HE4 in clinics. 5. Conclusion
Fig. 8. The ROC of CA125, HE4 and ROMA for pelvic masses, malignant pelvic masses vs benign pelvic masses.
Compared with the traditional detection modes that combined multiple tumor markers, the use of CA125, HE4 and ROMA shows better
X. Chen et al. / Clinica Chimica Acta 440 (2015) 57–63
performance in terms of specificity and sensitivity in our study, and therefore becomes a new promising application for the differential diagnosis of ovarian cancer, especially among premenopausal women. Besides, it also differentiates the benign pelvic masses from the malignant ones, which is another advantages of our newly applied way of detection in clinical practice. HE4, with high specificity for ovarian cancer, will reduce the misdiagnosis rate of ovarian cancer and help estimate the progression of disease. ROMA is a good choice for auxiliary diagnosis of uterine cancer, principally for endometrial cancer. The optimal cut-off values (CA125: 93.2 U/ml, HE4: 87.6 pmol/l, ROMA: 18.1% for pre- and 31.5% for postmenopausal women) could largely improve the diagnosis of pelvic masses. List of abbreviations (ROMA) the risk of ovarian malignancy algorithm (HE4) human epididymis-specific protein 4 (FIGO) International Federation of Gynecology and Obstetrics (CDM) combined detection mode
References [1] Yang Z, Luo Z, Zhao B, et al. Diagnosis and preoperative predictive value of serum HE4 concentrations for optimal debulking in epithelial ovarian cancer. Oncol Lett 2013;6:28–34. [2] Kristjansdottir B, Levan K, Partheen K, Sundfeldt K. Diagnostic performance of the biomarkers HE4 and CA125 in type I and type II epithelial ovarian cancer. Gynecol Oncol 2013;131:52–8. [3] Anastasi E, Marchei GG, Viggiani V, Gennarini G, Frati L, Reale MG. HE4: a new potential early biomarker for the recurrence of ovarian cancer. Tumour Biol 2010; 31:113–9. [4] Karlsen MA, Sandhu N, Hogdall C, et al. Evaluation of HE4, CA125, risk of ovarian malignancy algorithm (ROMA) and risk of malignancy index (RMI) as diagnostic tools of epithelial ovarian cancer in patients with a pelvic mass. Gynecol Oncol 2012;127:379–83.
63
[5] Lu R, Sun X, Xiao R, Zhou L, Gao X, Guo L. Human epididymis protein 4 (HE4) plays a key role in ovarian cancer cell adhesion and motility. Biochem Biophys Res Commun 2012;419:274–80. [6] Bandiera E, Romani C, Specchia C, et al. Serum human epididymis protein 4 and risk for ovarian malignancy algorithm as new diagnostic and prognostic tools for epithelial ovarian cancer management. Cancer Epidemiol Biomarkers Prev 2011;20: 2496–506. [7] Paek J, Lee SH, Yim GW, et al. Prognostic significance of human epididymis protein 4 in epithelial ovarian cancer. Eur J Obstet Gynecol Reprod Biol 2011;158:338–42. [8] Trudel D, Tetu B, Gregoire J, et al. Human epididymis protein 4 (HE4) and ovarian cancer prognosis. Gynecol Oncol 2012;127:511–5. [9] Zheng H, Gao YN, Gao WJ, Gao M, Yan X. Expression of serum human epididymis protein 4 in epithelial ovarian cancer and its correlation with prognosis. Zhonghua Zhong Liu Za Zhi 2013;35:445–9. [10] Chang X, Ye X, Dong L, et al. Human epididymis protein 4 (HE4) as a serum tumor biomarker in patients with ovarian carcinoma. Int J Gynecol Cancer 2011;21:852–8. [11] Presl J, Kucera R, Topolcan O, et al. HE4 a biomarker of ovarian cancer. Ceska Gynekol 2012;77:445–9. [12] Wang MJ, Qi J, Wang H, et al. Diagnostic value of combining detection of human epididymis protein 4 and CA125 in patients with malignant ovarian carcinoma. Zhonghua Zhong Liu Za Zhi 2011;33:540–3. [13] Angioli R, Plotti F, Capriglione S, et al. The role of novel biomarker HE4 in endometrial cancer: a case control prospective study. Tumour Biol 2013;34:571–6. [14] Bignotti E, Ragnoli M, Zanotti L, et al. Diagnostic and prognostic impact of serum HE4 detection in endometrial carcinoma patients. Br J Cancer 2011;104:1418–25. [15] Yang J, Sa M, Huang M, et al. The reference intervals for HE4, CA125 and ROMA in healthy female with electrochemiluminescence immunoassay. Clin Biochem 2013; 46:1705–8. [16] Moore RG, McMeekin DS, Brown AK, et al. A novel multiple marker bioassay utilizing HE4 and CA125 for prediction of ovarian cancer in patients with a pelvic mass. Gynecol Oncol 2009;112:40–6. [17] Shepherd JH. Revised FIGO staging for gynaecological cancer. Br J Obstet Gynaecol 1989;96:889–92. [18] Moore RG, Miller MC, Eklund EE, Lu KH, Bast RJ, Lambert-Messerlian G. Serum levels of the ovarian cancer biomarker HE4 are decreased in pregnancy and increase with age. Am J Obstet Gynecol 2012;206:341–9. [19] Drapkin R, von Horsten HH, Lin Y, et al. Human epididymis protein 4 (HE4) is a secreted glycoprotein that is overexpressed by serous and endometrioid ovarian carcinomas. Cancer Res 2005;65:2162–9. [20] Escudero JM, Auge JM, Filella X, Torne A, Pahisa J, Molina R. Comparison of serum human epididymis protein 4 with cancer antigen 125 as a tumor marker in patients with malignant and nonmalignant diseases. Clin Chem 2011;57:1534–44.
Contents lists available at ScienceDirect
Clinica Chimica Acta journal homepage: www.elsevier.com/locate/clinchim
Development of a multimarker assay for differential diagnosis of benign and malignant pelvic masses Xinliang Chen a,1, Hui Zhou b,1, Rui Chen c, Jian He a, Ying Wang a, Lisi Huang a, Longqiaozi Sun a, Chaohui Duan a, Xiaohong Luo a, Haiyan Yan a,⁎ a b c
Department of Clinical Laboratory, Sun Yat-Sen Memorial Hospital, Guangzhou, Guangdong 510080, PR China Department of Gynaecology, Sun Yat-Sen Memorial Hospital, Guangzhou, Guangdong 510080, PR China Department of Respiratory Medicine, Sun Yat-Sen Memorial Hospital, Guangzhou, Guangdong 510080, PR China
a r t i c l e
i n f o
Article history: Received 30 June 2014 Received in revised form 27 October 2014 Accepted 13 November 2014 Available online 15 November 2014 Keywords: Pelvic mass Ovarian cancer Uterine cancer Tumor markers HE4 ROMA
a b s t r a c t Background: HE4, a novel tumor marker for detecting ovarian cancer, has been recently applied to clinical practice. However, the comprehensive evaluation of HE4 combined with other markers is still missing. We evaluated an optimal mode of HE4 employment for differential diagnosis of benign and malignant pelvic masses. Methods: Serum HE4, CA125, CA153, CA199, CA211 and CA724 were measured from 232 patients with pelvic messes (100 malignant masses, 132 benign diseases), and the risk of ovarian malignancy algorithm (ROMA) was also calculated. Receiver operating characteristic curves (ROC), the area under the curve (AUC), sensitivity and specificity were estimated. Results: The combination of HE4 and CA125 (AUC of 0.963, sensitivity of 96.6%, specificity of 65.7%) provided the best differential power in diagnosing ovarian cancer. ROMA performed better in the diagnosis of pelvic masses (AUC of 0.917, sensitivity of 82.0%, specificity of 78.8%) and uterine cancer (AUC of 0.838, sensitivity of 82.0%, specificity of 60.0%) compared with applying HE4 and CA125 individually. Conclusion: The optimal cut-off values (CA125: 93.2 U/ml, HE4: 87.6 pmol/l, ROMA: 18.1% for pre- and 31.5% for postmenopausal women), simultaneous use of CA125 and HE4 complemented by ROMA showed better performance than the traditional detection modes for differential diagnosis of ovarian cancer. We also observed that ROMA added more accuracy for differentiating the benign and malignant pelvic masses and auxiliary diagnosis of uterine cancer. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Pelvic masses are often detected in pelvic examinations among females with ovarian cancer and uterine cancer, both of which are the most lethal types of cancer. However, due to non-specific early symptoms and lack of reliable screening tools, ovarian cancer often reaches an advanced stage (FIGO III–IV) when it is detected and its 5-year survival rate is only 20%–30% [1]. Thus, a diagnosis at early stage is critical, which requires a high sensitivity and specificity [2]. The screening and diagnosis of ovarian cancer are based on physical examination, semeiology, tumor markers, imaging and proteomics, etc. In the last few decades, several tumor markers have been evaluated in patients with ovarian cancer, among which, only CA125 correlates closely with ovarian cancer [3]. Since CA125 is also upregulated in many other tumors and therefore results in the reduction of sensitivity
and specificity, currently attention has been moved onto a novel marker with higher specificity: human epididymis-specific protein 4(HE4). HE4 is a glycoprotein overexpressed in epithelial ovarian cancer, especially serous and endometrioid adenocarcinoma [4]. HE4 is associated with cancer cell adhesion, migration and tumor growth [5] and serves as a valuable prognostic factor for the overall survival in patients with epithelial ovarian cancer and also an early indicator of the recurrence of ovarian cancer [3,6–9]. Previous reports showed that HE4 used in conjunction with CA 125 improves specificity for ovarian cancer [10]. The risk of ovarian malignancy algorithm (ROMA), a simple index based on CA125 and HE4, is helpful to improve the accuracy in differentiating benignancy from malignancy [4,11,12]. Besides, HE4 combined with CA125 has been reported as a new tool for preoperative evaluation and postoperative surveillance of endometrial cancer patients [13,14]. 2. Materials and methods
⁎ Corresponding author at: Department of Clinical Laboratory, Sun Yat-Sen Memorial Hospital, 107 Yaniiangxi Road, Guangzhou, Guangdong 510080, PR China. E-mail address: [email protected] (H. Yan). 1 These authors contributed to the work equally and should be regarded as co-first authors.
http://dx.doi.org/10.1016/j.cca.2014.11.013 0009-8981/© 2014 Elsevier B.V. All rights reserved.
2.1. Patients, and clinical pathology information A total of 232 female patients diagnosed with pelvic masses were scheduled to have surgeries between March 2012 and March 2014 in
58
X. Chen et al. / Clinica Chimica Acta 440 (2015) 57–63
our hospital. All pathologic tissue specimens were confirmed by histopathological evaluation postoperatively. Patients with preoperativelyknown relapse of a previous cancer or with an active cancer other than ovarian cancer were excluded. All of the patients provided informed consents. The cases were divided into the following four groups according to the pathological types. From this, 232 patients, 60 patients were pathologically confirmed postoperatively with ovarian cancer, including the following histological subtypes: serous (n = 27), mucinous (n = 6), endometrioid (n = 11), clear cell (n = 7), undifferentiated (n = 9) (aged 23–81 y, median 53 y). Seventy patients were diagnosed with benign ovarian diseases including ovarian chocolate cysts (n = 11), simple cyst (n = 5), ovarian endometriosis cysts (n = 23), mature teratoma (n = 20), mucinous cystadenoma (n = 5), serous cystadenoma (n = 6) (age 17–77 y, median 33 y). Forty patients were diagnosed with uterine cancer in the following histological subtypes: leiomyosarcoma (n = 3), endometrial cancer (n = 37), (age 33–81 y, median 51 y). Sixty-two patients had benign uterine diseases including hysteromyoma (n = 43), endometrial polyp (n = 10), endometrial hyperplasia (n = 5), and endometrioma (n = 4) (age 24–61 y, median 41 y) (Table 1). 2.2. Methods Serum samples were collected preoperatively from our patients. Three milliliters of fasting blood samples were collected in the morning. All samples were centrifuged at 2000 ×g for 10 min. The serum without hemolysis and lipemia was separated to be stored at − 20 °C until measurement. Following the standard operating procedure (SOP) provided by the manufacturer, serum HE4, CA125, CA153, CA199, CA211 and CA724 were quantitatively measured by electrochemiluminescence immunoassay analyzer (ECLIA, Roche E170). The reference values of each index are HE4 b 140 pmol/l, CA125 b35 U/ml, CA153 b 25 U/ml, CA199 b34 U/ml, CA211 b3.3 ng/ml and CA724 b 7 U/ml. Cases with marker levels above threshold levels were considered to have a positive result. Three clinical routine CDMs (combined detection modes) were used to Table 1 Patient age, menopause status and tumor characteristics. Tumor characteristics Benign ovarian diseases Median age (range) Simple Endometriosis Mature teratoma Serous Mucinous Chocolate cysts Total Ovarian cancer Median age (range) Serous Mucinous Endometrioid Clear cell Undifferentiated Total Benign uterine diseases Median age (range) Hysteromyoma Endometrial polyp Endometrial hyperplasia Endometrioma Total Uterine cancer Median age (range) Leiomyosarcoma Endometrial Total Total
Premenopausal
Postmenopausal
All (%)
31 (24–47) 3 21 15 2 2 5 48
57 (49–77) 2 2 5 4 3 6 22
34 (24–77) 5 (7.1) 23 (32.9) 20 (28.6) 6 (8.6) 5 (7.1) 11 (15.7) 70 (30.2)
45 (23–51) 5 3 5 3 4 20
57 (49–81) 22 3 6 4 5 40
53 (23–81) 27 (45.0) 6 (10) 11 (18.3) 7 (11.7) 9 (15.0) 60 (25.9)
39 (24–49) 39 9 4 4 56
54 (50–61) 4 1 1 0 6
41 (24–61) 43 (69.4) 10 (16.1) 5 (8.1) 4 (6.4) 62 (26.7)
46 (33–50) 3 16 19 153
56 (50–81) 0 21 21 89
51 (33–81) 3 (7.5) 37 (92.5) 40 (17.2) 232
compare with HE4. These 3 modes include CDM1 (CA125 + CA153), CDM2 (CA125 + CA153 + CA724) and CDM3 (CA125 + CA153 + CA724 + CA211). The test was defined as positive if one of the markers was positive in combined detection, and negative if all of the markers were negative. Serum HE4, CA125 levels and menopausal status were used to calculate ROMA. Menopause was considered when there was suspension of menstrual bleeding for at least 12 months [15]. A predictive index for ovarian cancer was calculated using the following formulae established by Moore et al. [16]: premenopausal PI = − 12 + 2.38 × ln(HE4) + 0.0626 × ln(CA125), postmenopausal PI = − 8.09 + 1.04 × ln(HE4) + 0.732 × ln(CA125). The ROMA value (predictive value) was calculated using the following formula: ROMA (%) = ePI / (1 + ePI) × 100. According to the indications of the HE4 manufacturer, indexes of at least 11.4% and 29.9% indicate a high risk of ovarian cancer in both pre- and postmenopausal women. 2.3. Statistics All calculations were performed with SPSS17.0 statistical software. The Wilcoxon–Mann–Whitney test and the Kruskal–Wallis test were used to compare biomarker distributions across two and more than two subgroups of patients, respectively; The χ2 test was used to evaluate the qualitative data (the frequency of the positive or negative specimens); sensitivity and specificity for CA125, HE4, ROMA, and combined detection modes were calculated. The predicted probabilities for each mode were used to construct receiver operating characteristic (ROC) curves. The area under the curve (AUC) values and the optimal cut-off values were calculated. A p b 0.05 was considered statistically significant. 3. Results 3.1. The evaluation of serum tumor markers in patients with either ovarian cancer or uterine cancer Median levels of different tumor markers in pre- and postmenopausal women are shown in Table 2. The median values of HE4, CA125, ROMA and positive rates are significantly higher among the patients with ovarian cancer compared with the benign cases both in pre- and postmenopausal women (p b 0.05). The median values of CA153, CA724, CA211 and positive rates for patients with ovarian cancer are significantly higher than patients with benign ovarian diseases (p b 0.05) except CA199 (p N 0.05). The median values of CA125, HE4 and ROMA for postmenopausal women were not significantly different from premenopausal women in the ovarian cancer group (p N 0.05). The HE4 and ROMA are statistically higher in patients with uterine cancer than in patients with benign uterine diseases (p b 0.05). 35.0% of the patients with uterine cancer had elevated values of CA125, and only 20.0% had elevated values of HE4. 3.2. The characterization of CA125 and HE4 in different ovarian cancer types We evaluated the levels of CA125 and HE4 in various ovarian cancer types. We found that the median HE4 for 46 patients in late stages (stages III + IV) are significantly higher than 14 patients in early stages (stages I + II) (p b 0.05) (Figs. 1, 2). The classification of different stages is based on the criteria of the International Federation of Gynecology and Obstetrics (FIGO) [17]. The levels of CA125 and HE4 in ovarian cancer of various pathological types (Table 3) are significantly different (p b 0.05). In particular, the median values of CA125 and HE4 are higher in the serous of ovarian cancer compared with the clear cell and mucinous types (p b 0.05); the positive rates of HE4 for the latter 2 types are only 57.1% and 16.7%, respectively. Besides, the median value of CA125 is also higher in endometriosis and chocolate cyst, compared
X. Chen et al. / Clinica Chimica Acta 440 (2015) 57–63
59
Table 2 Baseline data. Benign ovarian diseases Numbers (pre/postmenopausal) HE4 (pmol/l) (median, range) All Premenopausal Postmenopausal All positive (%) Premenopausal positive (%) Postmenopausal positive (%) CA125 (U/ml) (median, range) All Premenopausal Postmenopausal All positive (%) Premenopausal positive (%) Postmenopausal positive (%) ROMA (%) All Premenopausal Postmenopausal All positive (%) Premenopausal positive (%) Postmenopausal positive (%) CA153 (U/ml) (median, range) CA153 positive (%) CA199 (U/ml) (median, range) CA199 positive (%) CA724 (U/ml) (median, range) CA724 positive (%) CA211 (ng/ml) (median, range) CA211 positive (%)
70 (48/22)
Ovarian cancer
p
60 (20/40)
Benign uterine diseases 63 (57/6)
Uterine cancer
p
40 (19/21)
52.8 (31.8–155.8) 52.6 (31.8–79.4) 54.0 (37.6–155.8) 1 (1.4%) 0 (0%) 1 (4.5%)
242.6 (45.8–3109.0) 201.1a (61.8–2574.0) 270.9a (45.8–3109.0) 44 (73.3%) 13 (65%) 31 (77.5%)
0.000 0.000 0.000 0.000 0.000 0.000
52.2 (30.9–86.8) 50.9 (30.9–86.8) 61.5 (49.2–86.8) 0 (100%) 0 (100%) 0 (100%)
82.3 (37.6–519.2) 68.2 (38.2–186.3) 90.0 (37.6–519.2) 8 (20%) 2 (10.5%) 6 (28.6%)
0.000 0.001 NS 0.000 0.015 NS
21.6 (4.2–228.0) 23.2 (8.7–228.0) 21.4 (4.2–175.8) 23 (32.8%) 18 (37.5%) 5 (22.7%)
390.0 (13.1–13,280) 285.6a (18.6–11,423) 444.8a (13.1–13,280) 56 (93.3%) 19 (95%) 37 (92.5%)
0.000 0.000 0.000 0.000 0.000 0.000
21.0 (5.4–827.7) 21.1 (5.4–827.7) 20.1 (5.7–153.1) 11 (17.7%) 10 (17.9%) 1 (16.7%)
25.3 (8.6–410.8) 24.1 (9.4–105.4) 28.4 (8.6–410.8 14 (35%) 5 (26.3%) 9 (42.9%)
NS NS NS 0.049 NS NS
8.9 (2.7–36.4) 8.7 (2.7–20.6) 14.0 (3.9–36.4) 12 (17.1%) 7 (14.6%) 5 (22.7%) 9.7 (4.0–50.4) 3/(4.3%) 18.3 (1.0–506.8) 20 (28.6%) 1.7 (0.8–22.0) 7 (10%) 1.5 (0.7–6.9) 2 (2.9%)
86.5 (6.2–99.9) 70.4a (13.5–99.9) 87.4a (6.2–99.9) 59 (98.3%) 20 (100%) 39 (97.5%) 31.9 (3.7–1823.0) 36 (60%) 15.9 (0.6–1000.0) 17 (28.3%) 14.9 (0.6–3000.0) 32 (53.3%) 2.2 (0.7–16.9) 15 (25%)
0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 NS NS 0.000 0.000 0.000 0.000
8.3 (2.5–48.0) 7.9 (2.5–23.6) 17.2 (5.9–48.0) 14 (22.6%) 12 (21.4%) 2 (33.3%)
16.0 (4.1–90.0) 14.6 (4.1–65.6) 27.2 (11.5–90.0) 24 (60%) 12 (63.2%) 12 (57.1%)
0.000 0.001 NS 0.000 0.001 NS
a The CA125, HE4 and ROMA values for postmenopausal women were not significantly different from premenopausal women in the ovarian cancer group (p = 0.572, p = 0.388, p = 0.230).
with the other types of benign ovarian diseases. However, there is no statistically significant difference of HE4 observed between different types of benign ovarian diseases (p N 0.05).
3.3. The application of normal cut-off values and optimal cut-off values for diagnosis of pelvic masses We used the normal cut-off values and the optimal cut-off values determined by ROC to calculate the diagnostic values. The sensitivity and specificity of CA125, HE4, ROMA and CDMs of 3 clinical routines are shown in Table 4. Three modes include CDM1 (CA125 + CA153), CDM2 (CA125 + CA153 + CA724) and CDM3 (CA125 + CA153 + CA724 + CA211). The ROC and AUC of the 6 markers are presented in Fig. 3. HE4 exhibits the highest AUC (0.963) in single detection, while the lowest AUC is associated with CA199 (AUC 0.488) (Fig. 3). Thus,
Fig. 1. CA125 of different stages.
due to the weak efficiency of diagnosis, CDM is not recommended for joint application. When used to analyze benign ovarian masses versus ovarian cancer, HE4 had the highest specificity (98.6%), but at the cost of an unsatisfactory low sensitivity (73.3%). CA125 had a higher sensitivity (93.3%), but a lower specificity (67.1%) and AUC (0.926) compared with HE4 (p b 0.05). When the optimal cut-off values of CA125 and HE4 were applied, the specificity of CA125 and the sensitivity of HE4 increased to 88.6% and 88.3%. ROMA had the higher specificity (80.0%) and AUC (0.972) when compared with the combined detection of CA125 and HE4 (65.7%, 0.963). The use of the optimal cut-off values increased the specificity of the combined detection of CA125 and HE4 (87.1%). The sensitivities of 3 CDMs had no significant difference from CA125 alone (p N 0.05), and the specificities were lower (p b 0.05). The AUCs for the 3 CDMs were also lower than that for ROMA (Fig. 4). Besides,
Fig. 2. HE4 of different stages.
60
X. Chen et al. / Clinica Chimica Acta 440 (2015) 57–63
Table 3 CA125 and HE4 of different pathological types of pelvic masses. CA125 (U/ml)
Ovarian cancer Serous Endometrioid Clear cell Mucinous p (Kruskal–Wallis) Benign ovarian diseases Chocolate cyst Endometriosis Teratoma Simple Serous Mucinous p (Kruskal–Wallis)
HE4 (pmol/l)
Median (range)
Positive (%)
Median (range)
Positive (%)
931.3⁎ (25.8–13,280) 301.5 (20.5–11,423) 80.1 (41.2–3458) 155.3 (13.1–198.9 0.004
26 (96.3%) 10 (90.9%) 7 (100.0%) 4 (66.7%)
390.5⁎ (97.9–1985) 216.8 (45.8–2574) 194.8 (65.8–1166) 79.1 (54.8–227.2) 0.010
22 (81.5%) 8 (72.7%) 4 (57.1%) 1 (16.7%)
45.0⁎⁎ (22.5–162.0) 49.7⁎⁎ (10.9–228.0) 13.8 (8.0–26.2) 13.6 (13.0–65.4) 14.9 (4.2–117.6) 22.6 (9.8–175.8) 0.000
7 (63.6%) 13 (56.5%) 0 (0%) 1 (20.0%) 1 (16.7%) 1 (20.0%)
53.9 (39.4–79.4) 53.0 (38.1–70.3) 49.1 (31.8–75.7) 43.4 (38.5–74.8) 52.3 (37.9–87.4) 76.4 (41.1–155.8) NS
0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 1 (20.0%)
⁎ p b 0.05 (compared with clear cell and mucinous). ⁎⁎ p b 0.05 (compared with other types of benign ovarian diseases).
ROMA had a high sensitivity (100.0%) and specificity (85.4%) in premenopausal women (p b 0.05), and the AUC for ROMA in premenopausal women (0.993) was higher than that in postmenopausal women (0.963) (Figs. 5, 6). When the cut-off values of ROMA were set as 18.1% for premenopausal women and 31.5% for postmenopausal women, ROMA presents both high sensitivity and specificity (91.7%, 97.1%). When analyzing uterine malignant tumors, the specificities for CA125, HE4 and ROMA were 82.3%, 100%, 77.4%, respectively, but the sensitivities were 35.0%, 20.0%, 60.0%, respectively. ROMA had the highest AUC (0.838) followed by HE4 (0.808) and CA125 (0.580) (Fig. 7, Table 4). When differentiating benign and malignant pelvic masses, HE4 had the highest specificity (99.2%), and the lowest sensitivity (52.0%);
ROMA had a low specificity (78.8%) and the highest sensitivity (82.0%); The highest AUC was calculated for ROMA (0.917), followed by HE4 (0.806) and CA125 (0.711) (Fig. 8, Table 4). When the cut-off values of ROMA were set as 24.6%, ROMA had the highest specificity (97.7%) for differential diagnosis of malignant pelvic masses. 4. Discussion The serum tumor marker CA125 is commonly used to predict pelvic masses. But due to the limited specificity and sensitivity, most clinicians use a combination of multiple tumor markers to differentiate benign and malignant pelvic masses. CA125 is often combined with CA153, CA199, CA724 and CA211 to screen patients for ovarian cancer. As a novel tumor marker in the diagnosis of ovarian cancer, HE4 has been
Table 4 Sensitivity, specificity for CA125, HE4, ROMA, CA153, CA724, CA211 and three clinical routine combine detection modes (CDM) for women with pelvic masses. Tumor marker
Normal cut-off values⁎ Sensitivity (%)
OC vs benign ovarian diseases CA125 All Premenopausal Postmenopausal HE4 All Premenopausal Postmenopausal ROMA All Premenopausal Postmenopausal CA153 CA724 CA211 CA125 + HE4 CDM1 (CA125 + CA153) CDM2 (CA125 + CA153 + CA724) CDM3 (CA125 + CA153 + CA724 + CA211) UC vs benign uterine diseases CA125 HE4 ROMA Malignant pelvic masses vs benign pelvic masses CA125 HE4 ROMA
Optimal cut-off values Specificity (%)
93.2 U/ml 93.2 U/ml 110.9 U/ml
Optimal cut-off values Sensitivity (%)
Specificity (%)
85.0 85.0 82.5
88.6 87.5 95.5
88.3 90.0 87.5
97.1 95.8 95.5
91.7 95.0 92.5 78.3 53.3 73.3 93.3 96.7 96.7 96.7
97.1 97.9 95.5 91.4 98.6 70.0 87.1 80.0 78.6 57.1
93.3 95.0 92.5
67.1 62.5 77.3
73.3 65.0 77.5
98.6 100.0 95.5
96.7 100.0 95.0 60.0 53.3 25.0 96.6 93.3 93.3 95.0
80.0 85.4 68.2 95.7 90.0 97.1 65.7 64.3 57.0 55.7
35.0 20.0 60.0
82.3 100 77.4
33.6 U/ml 66.4 pmol/l 14.2%
40.0 70.0 75.0
82.3 90.3 85.5
70.0 52.0 82.0
73.5 99.2 78.8
78.3 U/ml 81.9 pmol/l 24.6%
58.0 75.0 75.0
88.6 96.2 97.7
87.6 pmol/l 72.3 pmol/l 97.5 pmol/l
18.1% 31.5% 19.3 U/ml 11.4 U/ml 1.83 ng/ml
⁎ Normal cut-off value: CA125 b35 U/ml; HE4 b140 pmol/l; premenopausal women: ROMA b11.4%, postmenopausal women: ROMA b29.9%; CA153 b25 U/ml; CA724 b7 U/ml; and CA211 b3.3 ng/ml.
X. Chen et al. / Clinica Chimica Acta 440 (2015) 57–63
Fig. 3. The ROC of single tumor marker detection of ovarian cancer vs benign ovarian diseases.
used in the clinical practice only for a short period of time. In this study, we aimed to investigate the diagnosis efficiency and the optimal detection mode of HE4. As shown in Table 2, the levels of HE4, CA125 and ROMA were significantly higher in the ovarian cancer group (p b 0.05), thus these 3 are good candidates for predicting ovarian cancer. But since the AUC for CA199 alone was only 0.488, it is not recommended for the combined detection. The difference of HE4 between pre- and postmenopausal women in the ovarian cancer group did not achieve statistical significance
Fig. 4. The ROC of multiple tumor marker combined detection of ovarian cancer vs benign ovarian diseases.
61
Fig. 5. The ROC of ROMA for premenopausal women ovarian cancer vs benign ovarian diseases.
(p N 0.05) (Table 2), which is contradictory to what Yang et al. and Moore et al. [15,18] observed. The sample size chosen for the study may have caused the difference. We also showed that the median value of HE4 of 46 patients in late stages was significantly higher than in 14 patients in the early stages (p b 0.05) (Figs. 1, 2), which is consistent with the study of Yang et al. [2]. It indicates the role of HE4 in the progress of developing ovarian cancer. Moreover, the median CA125 and HE4 values were higher in the serous ovarian cancer compared with the clear cell and mucinous
Fig. 6. The ROC of ROMA for postmenopausal women ovarian cancer vs benign ovarian diseases.
62
X. Chen et al. / Clinica Chimica Acta 440 (2015) 57–63
Fig. 7. The ROC of CA125, HE4 and ROMA for uterine cancer, uterine cancer vs benign uterine diseases.
types (p b 0.05) (Table 3). In line with our report, Drapkin et al. [19] found overexpressed HE4 in ovarian cancer, mainly in serous carcinoma and endometrial carcinoma, but not in clear cell carcinoma and mucinous carcinoma. Thus, the combination of CA125 and HE4 is specific to serous ovarian cancer. It has been reported that those patients with a renal failure have the highest HE4 positive rate in the non-ovarian benign diseases [20], which may explain why the single detection mode using HE4 with the highest AUC (0.963) still causes misdiagnosis (e.g. 8 patients with endometrial cancer had positive HE4 results in the uterine cancer group). To
optimize the specificity of HE4, we applied a new cut-off value for HE4 with a boundary value of 87.6 pmol/l. As we expected, it resulted in an increased sensitivity of 88.3%. Also, CA125 increased in many patients with ovarian benign diseases, including chocolate cyst (45.0 U/ml, 63.6%) and ovarian endometriosis (49.7 U/ml, 56.5%) and consequently had a relatively low specificity of 67.1%. But with the optimal cut-off value of 93.2 U/ml, the specificity of CA125 increased to 88.6% while retaining high sensitivity (85.0%). Next, we applied the combination of HE4 and CA125. Upon the combination, the sensitivity reached as high as 96.6%, but the specificity was only 65.7% (Table 4). The use of optimal cut-off values increased the specificity to 87.1%. Also, clinical doctors can calculate ROMA through HE4 and CA125 to classify the risk of epithelial ovarian cancer in female patients with pelvic masses. The sensitivity for ROMA in the identification of benign and malignant ovarian tumors is 96.7%. Compared with the combined detection of CA125 and HE4, ROMA had higher specificity (80.0%) and AUC (0.972). In particular, ROMA had the highest sensitivity (100.0%), specificity (85.4%) and AUC (0.993) for premenopausal women in the diagnosis of ovarian cancer (Table 4, Fig. 5). The optimal cut-off values of ROMA were 18.1% for premenopausal women and 31.5% for postmenopausal women separately. Obviously, ROMA is very useful for clinical application. Traditionally, clinicians tended to use the combined detection of multiple tumor markers for the diagnosis of ovarian cancer. However, in our study, we showed that the sensitivities of 3 kinds of routine combined detection modes (CA125 + CA153, CA125 + CA153 + CA724, CA125 + CA153 + CA724 + CA211) are similar as CA125 alone, but the specificities are lower. Besides, the AUC of these 3 modes is not significantly different from that of ROMA and the combined detection of HE4 and CA125 (Fig. 4). Thus, the traditional combined detection of multiple tumor markers is no longer an ideal diagnosis of ovarian cancer. The value of CA125 and HE4 for the diagnosis of uterine malignant tumors is not clear. Although the specificities for CA125 and HE4 were 82.3% and 100% respectively, the sensitivities were only 35.0% and 20.0% respectively (Table 4, Fig. 7). But ROMA well differentiated uterine cancer and benign uterine diseases with higher sensitivity (60.0%), specificity (77.4%) and AUC (0.838). Thus, ROMA is suitable for the auxiliary diagnosis of uterine cancer. HE4, with higher specificities (99.2%) and higher AUC (0.899), performed better than CA125 in the identification of benign and malignant pelvic masses. The sensitivity, specificity and AUC for ROMA were 82.0%, 78.8% and 0.917 respectively (Table 4, Fig. 8), which makes ROMA a good candidate for differential diagnosis of pelvic masses. Early stages of ovarian cancer show minor symptoms and can be easily misdiagnosed. But the detection of HE4 and CA125, and the application of ROMA have shown promising sensitivity and specificity for screening patients with pelvic masses. Moreover, this detection mode also differentiates benign pelvic masses from malignant ones. The current cut-off values of tumor markers which are used in China were provided by the reagent producers. The statistic data originates in a foreigner. So we suggested new cut-off values of CA125, HE4 and ROMA should be used in clinics for the Chinese. When it comes to the limitations of this study, because the cases only came from the Department of Gynecology of our hospital, a small number of cases may influence the representativeness of the experimental group. In addition, the study belongs to the retrospective study. We didn't have a follow-up of the patient's prognosis and survival rates. So whether HE4 could be used as prognostic factors of ovarian cancer is to be discussed. We suggest that a prospective study could be designed to explore the value of HE4 in clinics. 5. Conclusion
Fig. 8. The ROC of CA125, HE4 and ROMA for pelvic masses, malignant pelvic masses vs benign pelvic masses.
Compared with the traditional detection modes that combined multiple tumor markers, the use of CA125, HE4 and ROMA shows better
X. Chen et al. / Clinica Chimica Acta 440 (2015) 57–63
performance in terms of specificity and sensitivity in our study, and therefore becomes a new promising application for the differential diagnosis of ovarian cancer, especially among premenopausal women. Besides, it also differentiates the benign pelvic masses from the malignant ones, which is another advantages of our newly applied way of detection in clinical practice. HE4, with high specificity for ovarian cancer, will reduce the misdiagnosis rate of ovarian cancer and help estimate the progression of disease. ROMA is a good choice for auxiliary diagnosis of uterine cancer, principally for endometrial cancer. The optimal cut-off values (CA125: 93.2 U/ml, HE4: 87.6 pmol/l, ROMA: 18.1% for pre- and 31.5% for postmenopausal women) could largely improve the diagnosis of pelvic masses. List of abbreviations (ROMA) the risk of ovarian malignancy algorithm (HE4) human epididymis-specific protein 4 (FIGO) International Federation of Gynecology and Obstetrics (CDM) combined detection mode
References [1] Yang Z, Luo Z, Zhao B, et al. Diagnosis and preoperative predictive value of serum HE4 concentrations for optimal debulking in epithelial ovarian cancer. Oncol Lett 2013;6:28–34. [2] Kristjansdottir B, Levan K, Partheen K, Sundfeldt K. Diagnostic performance of the biomarkers HE4 and CA125 in type I and type II epithelial ovarian cancer. Gynecol Oncol 2013;131:52–8. [3] Anastasi E, Marchei GG, Viggiani V, Gennarini G, Frati L, Reale MG. HE4: a new potential early biomarker for the recurrence of ovarian cancer. Tumour Biol 2010; 31:113–9. [4] Karlsen MA, Sandhu N, Hogdall C, et al. Evaluation of HE4, CA125, risk of ovarian malignancy algorithm (ROMA) and risk of malignancy index (RMI) as diagnostic tools of epithelial ovarian cancer in patients with a pelvic mass. Gynecol Oncol 2012;127:379–83.
63
[5] Lu R, Sun X, Xiao R, Zhou L, Gao X, Guo L. Human epididymis protein 4 (HE4) plays a key role in ovarian cancer cell adhesion and motility. Biochem Biophys Res Commun 2012;419:274–80. [6] Bandiera E, Romani C, Specchia C, et al. Serum human epididymis protein 4 and risk for ovarian malignancy algorithm as new diagnostic and prognostic tools for epithelial ovarian cancer management. Cancer Epidemiol Biomarkers Prev 2011;20: 2496–506. [7] Paek J, Lee SH, Yim GW, et al. Prognostic significance of human epididymis protein 4 in epithelial ovarian cancer. Eur J Obstet Gynecol Reprod Biol 2011;158:338–42. [8] Trudel D, Tetu B, Gregoire J, et al. Human epididymis protein 4 (HE4) and ovarian cancer prognosis. Gynecol Oncol 2012;127:511–5. [9] Zheng H, Gao YN, Gao WJ, Gao M, Yan X. Expression of serum human epididymis protein 4 in epithelial ovarian cancer and its correlation with prognosis. Zhonghua Zhong Liu Za Zhi 2013;35:445–9. [10] Chang X, Ye X, Dong L, et al. Human epididymis protein 4 (HE4) as a serum tumor biomarker in patients with ovarian carcinoma. Int J Gynecol Cancer 2011;21:852–8. [11] Presl J, Kucera R, Topolcan O, et al. HE4 a biomarker of ovarian cancer. Ceska Gynekol 2012;77:445–9. [12] Wang MJ, Qi J, Wang H, et al. Diagnostic value of combining detection of human epididymis protein 4 and CA125 in patients with malignant ovarian carcinoma. Zhonghua Zhong Liu Za Zhi 2011;33:540–3. [13] Angioli R, Plotti F, Capriglione S, et al. The role of novel biomarker HE4 in endometrial cancer: a case control prospective study. Tumour Biol 2013;34:571–6. [14] Bignotti E, Ragnoli M, Zanotti L, et al. Diagnostic and prognostic impact of serum HE4 detection in endometrial carcinoma patients. Br J Cancer 2011;104:1418–25. [15] Yang J, Sa M, Huang M, et al. The reference intervals for HE4, CA125 and ROMA in healthy female with electrochemiluminescence immunoassay. Clin Biochem 2013; 46:1705–8. [16] Moore RG, McMeekin DS, Brown AK, et al. A novel multiple marker bioassay utilizing HE4 and CA125 for prediction of ovarian cancer in patients with a pelvic mass. Gynecol Oncol 2009;112:40–6. [17] Shepherd JH. Revised FIGO staging for gynaecological cancer. Br J Obstet Gynaecol 1989;96:889–92. [18] Moore RG, Miller MC, Eklund EE, Lu KH, Bast RJ, Lambert-Messerlian G. Serum levels of the ovarian cancer biomarker HE4 are decreased in pregnancy and increase with age. Am J Obstet Gynecol 2012;206:341–9. [19] Drapkin R, von Horsten HH, Lin Y, et al. Human epididymis protein 4 (HE4) is a secreted glycoprotein that is overexpressed by serous and endometrioid ovarian carcinomas. Cancer Res 2005;65:2162–9. [20] Escudero JM, Auge JM, Filella X, Torne A, Pahisa J, Molina R. Comparison of serum human epididymis protein 4 with cancer antigen 125 as a tumor marker in patients with malignant and nonmalignant diseases. Clin Chem 2011;57:1534–44.
