European Journal of Obstetrics & Gynecology and Reproductive Biology 231 (2018) 35–42
Contents lists available at ScienceDirect
European Journal of Obstetrics & Gynecology and Reproductive Biology journal homepage: www.elsevier.com/locate/ejogrb
Review article
Diagnostic value of HE4 in ovarian cancer: A meta-analysis Jinbing Huang, Junying Chen* , Qiaoqiao Huang Department of Gynecology, The First Affiliated Hospital of Guangxi Medical University, Nanning, Guangxi Province, 530022, China
A R T I C L E I N F O
A B S T R A C T
Article history: Received 30 June 2018 Received in revised form 26 September 2018 Accepted 2 October 2018 Available online xxx
Objective: To analyze and evaluate the overall diagnostic value of Serum Human Epididymis Protein 4 (HE4) in ovarian cancer. Methods: We searched PubMed, EMBASE, Cochrane Library and Web of Science databases to collect articles in English that evaluated the diagnostic value of HE4 in patients with gynecological or pelvic masses. Two reviewers independently assessed the methodological quality of each study using the QUADAS-2 tool. A chart of literature quality was made using Revman 5.3 software. Finally, we built Summary Receiver Operating Characteristic (SROC) curves, Hierarchical Summary Receiver Operating Characteristic (HSROC) models, a Deek’s funnel figure as well as a meta-analysis of included studies using STATA12.0 and Meta-Disc1.4 software. Results: Eighteen studies from the published literature met all inclusion criteria for this analysis. Remarkably, no publication bias was found among the included studies. HE4 had a pooled sensitivity of 81% (95% confidence interval (CI): 77–85) and a pooled specificity of 91% (95% CI: 86–93,). Overall, the positive likelihood ratio (PLR) was 8.2, (95% CI: 5.60–12.00,) the negative likelihood ratio (NLR) was 0.21 (95% CI: 0.17–0.26), the diagnostic odds ratio (DOR) was 39 (95% CI: 25–62), the AUC of SROC was 0.91, and Cochrane-Q value was 86.02. Conclusions: HE4 is a valuable marker in the clinical diagnosis of ovarian cancer with both high AUC and Cochrane-Q. More studies are needed to determine if HE4 in the range of 100 mmol/L cutoff150 mmol/L than 60 mmol/L
Keywords: HE4 Ovarian cancer Diagnostic Meta-analysis
Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Inclusion and exclusion criteria . . . . . . . . . . . . . . . . . . . . . . . . . . Retrieved database . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Quality assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Statistical analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Characteristics and quality evaluation of the included literature Pooled diagnostic values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Subgroup analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sensitivity analysis and publication bias . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Disclosure of conflicts of interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
35 36 36 36 36 36 37 37 37 38 38 38 42 42
Introduction * Corresponding author at: Department of Gynecology, The First Affiliated Hospital of Guangxi Medical University, China. E-mail address:
[email protected] (J. Chen). https://doi.org/10.1016/j.ejogrb.2018.10.008 0301-2115/© 2018 Elsevier B.V. All rights reserved.
Ovarian cancer accounts for 2.5% of all malignancies among females but 5% of female cancer deaths due to low survival rates, largely driven by advance stage diagnoses [1]. Although early stage
36
J. Huang et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 231 (2018) 35–42
ovarian cancer has a 5-year relative survival rate of 93% [2], only 15% of ovarian cancer patients are diagnosed early because the early stages of ovarian cancer have no obvious symptoms. When ovarian cancer is diagnosed at III, IV stage, the 5-year survival rate drops sharply [3,4] to about 20–25% [5]. According to the American Cancer Society, in 2018 there will be approximately 22,240 new cases of ovarian cancer diagnosed and 14,070 ovarian cancer deaths in the United States. Therefore, improving prevention and early detection strategies is key for improving clinical outcomes with ovarian cancer. Tumor biomarkers for early detection of tumor and disease monitoring are therefore of great significance and so searching for high sensitivity and specificity of the serum tumor markers as an auxiliary diagnosis method has been an often-discussed research topic in academic circles. The value of CA125 as a serum biomarker in the preoperative diagnosis and clinical monitoring of ovarian cancer is clear. Unfortunately, it also has low diagnosis value in early stage ovarian cancer due to its low specificity, a high rate of false positives and low sensitivity [6]. For these reasons a new serum biomarker for ovarian cancer is being explored as a critical issue to ovarian cancer researchers. After CAl25 was described as an auxiliary diagnosis for ovarian cancer and widely used in clinical, human epididymis protein 4 (HE4) became a prominent research focus for the auxiliary diagnosis of ovarian cancer. In 2003, Hellstrom et al. [7] found HE4 to be a secreted glycoprotein overexpressed in epithelial ovarian cancer (EOC). Multiple studies have confirmed that HE4 has higher sensitivity and specificity than CA125 in the early stages of EOC [8–10]. The HE4 protein is expressed in malignant ovarian tissues at a significantly higher rate than that of benign tumors and normal ovarian tissue [11]. HE4 is regarded as a serum tumor marker which has high sensitivity and specificity in screening tests. HE4 can be used in the detection and staging of treatment for ovarian cancer [12]. HE4 has been used as an indicator of ovarian cancer in screening and diagnostic for the past 20 years. However, only from 2010 to 2013 were there many meta-analyses published about the diagnosis value of HE4 in ovarian cancer. All confirmed that HE4 is highly valuable in the diagnosis of ovarian cancer, with the lowest AUC being 0.89 [13]. However, the early detection rate and diagnosed early state ovarian cancer are still very low. Our study surveyed evidence-based medicine to verify and further explore the diagnostic value of serum HE4 in ovarian cancer. Materials and methods Inclusion and exclusion criteria Studies were included in this analysis if they fulfilled the following inclusion criteria [1]: clinical diagnostic test described the expression level of the serum HE4 for the diagnosis of ovarian cancer, including all study types (not excluding retrospective studies) [2]; criterion for diagnosis of ovarian cancer and benign ovarian tumor was established by a histopathological review and diagnosis [3]; study population consisted of patients with ovarian cancer, benign pelvic diseases and healthy women [4]; blood samples were collected from patients before surgical or cytotoxic therapies [5]; study data was reliable and could be used to extract four key clinical parameters (number of true-positive, falsepositive, true-negative, and false-negative testing cases) [6]; diagnostic cutoff was described [7]; source of reagent and test methods was explicit [8]; articles were published in English [9]; 10 or more patients made up the study population. Exclusion criteria were as follows [1]: data was insufficient to construct a 2*2 table of the test results (serum HE4 concentration) or data was obviously inconsistent [2]; abstracts, reviews, talks and
review class documentations [3]; all or part of the study population did not have a reported histopathological diagnosis [4]; Study population had recurrent ovarian cancer or had been treated with cytotoxic therapies [5]; literature did not meet inclusion criteria. Retrieved database Two investigators (H.JB and H.QQ) independently searched for articles published about ovarian cancer and HE4 in English up to May 2018. The following in English databases were used: PubMed, EMBASE, Cochrane, Web of Science. Index Words and search strategy The following keywords were used: ovarian neoplasms, ovary cancer, ovarian cancers, cancer of ovary, cancer of the ovary, ovary neoplasm, ovarian carcinoma, ovarian tumor, human epididymis secretory protein 4, human epididymis protein 4, HE4 and WFDC2. Our search strategy was as follows: ("ovarian neoplasms"[MeSH Terms] OR " ovarian neoplasms "[All Fields] OR" ovary cancer "[All Fields]) OR" ovarian cancers "[All Fields]) OR" cancer of ovary "[All Fields]) OR" cancer of the ovary "[All Fields]) OR "ovary neoplasm "[All Fields]) OR "ovarian tumor" AND (human epididymis secretory protein 4 [All Fields] OR "human epididymis protein 4 "[All Fields]) OR " HE4"[All Fields]) OR " WFDC2 "[All Fields]). Literature and data extraction Two researchers independently filtered all literature from each database according to the inclusion criteria. Firstly, duplicates were removed using Endnote, then publication titles, abstracts, and fulltext articles were screened to identify eligible studies. Any disagreements between the reviews were resolved through consensus with a third reviewer (C.JY). The data retrieved from the studies included study design, blinding, reagent source, assay methods, cutoff design, cutoff point, menstruation status, and outcome data (true positive, false positive, true negative, false negative). Quality assessment We used the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) and Revman 5.3 software to assess the quality of included studies. We then used quantitative method to assess the studies. The QUADAS-2 included 14 items [14]. Each key domain included two sections: risk of bias and applicability. If answers to all signaling questions for a domain were ‘yes’, then we could judge the risk of bias was low. If any question were answered with ‘no’, potential bias existed. Concerns about applicability were judged as ‘low’, ‘high’, or ‘unclear’. Statistical analysis We used the STATA12 software to perform all statistical analyses. The heterogeneity within studies was evaluated using the I2 test and Q Test, and I 2 > 50% presented the existence of heterogeneity [15]. A bivariate regression model was used to calculate the pooled sensitivity, specificity, positive and negative likelihood ratios (PLRs and NLRs, respectively), diagnostic odds ratio (DOR) and their respective 95% confidence intervals (CIs) [16]. We also calculated the area under the summary receiver operator characteristic curve (SROC, AUC). AUC ranged from 0 to 1, and an AUC of 1 represented perfect discrimination ability, while an AUC < 0.5 showed poor diagnostic ability [17]. We also conducted subgroup analysis based on menstrual status, pathological type, cut-off value and so on. We then used Deek’s funnel plot to assess publication bias, and Fagan plots to show the relationship between prior probability, likelihood ration, and posterior test probability [18]. P < 0.05 was considered to be statistically significant. We also
J. Huang et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 231 (2018) 35–42
used meta-Disc 1.4 software to calculate AUC and the standard error of AUC. If necessary, compared different AUC tests to estimate diagnostic performance with z-test. Results Characteristics and quality evaluation of the included literature The flowchart of our selection strategy is presented in Fig. 1. Our initial search obtained 1067 records. Then 320 duplicate records were removed, additionally 36 records about review, meta, report were removed, and 588 records were excluded after reviewing titles, abstract and topic. Next, 43 full-text articles were assessed for eligibility, after further review we removed 23 records unrelated to HE4 diagnostic value. Similarly, two articles were excluded for obvious contradictory data. Finally, 18 articles were included in the final qualitative and quantitative analyses. In Table 1, we summarize the characteristics of the 18 included articles on the meta-analysis of the diagnostic value of HE4. The publication year of these studies ranged from 2008 to 2016. In total the included studies assessed 3815 individuals (1237 ovarian cancer patients, 245 healthy subjects and 2333 benign pelvic tumor patients). The detailed characteristics of the included studies are presented in Table 1.The results of quality assessment are present in Figs. 2 and 3. All of the included studies received
37
moderately high scores on the QUADAS-2 quality assessment therefore the quality of this meta-analysis was good. During patient selection, risk of bias was relatively higher for some retrospective studies. During the part of index test, risk of bias was relatively higher for some cutoff set in advance. Those have little influence on the diagnosis evaluation of this article. Before reviewing the literature, the specific interpretation of this study was designed according to the scoring criteria of QUADAS-2, as shown in Fig. 2. Exploring heterogeneity Using STATA12.0 software, I2 values were measured to be 98% (p = 0.000), with a Spearman correlation coefficient r = -0.33 (p = 0.11), revealed that there was no threshold effect in our study. In the forest plot of DOR, every study and mergers are not distributing along the same line, Cochrane -Q = 5.8e10, p = 0.00, the beta of HSROC model p = 0.055 > 0.05, and SROC was asymmetric. All of the above indicate that heterogeneity was caused by nonthreshold effect therefore we combined statistics from all studies. Pooled diagnostic values The details of our overall pooled diagnostic values including DOR, sensitivity, specificity, PLR, NLR are shown in Table 2 and sensitivity and specificity forest plots are shown in Fig. 4. SROC curve are shown in Fig. 5, and AUC was 0.91 (95%CI: 0.88-0.93).
Fig. 1. Flowchart of selection of eligible studies.
38
J. Huang et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 231 (2018) 35–42
Table 1 The general characteristics of the18 included studies. Num.
author
year
Area
design
TP
FP
FN
TN
cutoff
blind
method
histological
1 2 3 4 5 6 7
Moore, R.G. [19] Montagnana, M [20]. Abdel-Azeez, H.A [21]. Holcomb, K. [22] Chang, X. H. [23] Jacob, F. [24] Van Gorp, T. [25] premenopausal postmenopausal Chan, K. K. [26] premenopausal postmenopausal Ortiz-Munoz, B. [27] premenopausal postmenopausal Winarto, H. [28] Ghasemi, N. [29] Hye Yon Cho [30] Chen, X. [31] premenopausal postmenopausal Romagnolo, C. [32] premenopausal postmenopausal Wei, S. [33] premenopausal postmenopausal Xu, Y [34] premenopausal postmenopausal Zheng, L. E. [35] Yanaranop, M [36]
2008 2009 2010 2011 2011 2011 2011
USA Italy Egypt USA china Switzerland. Belgium
Prospective retrospective Prospective Prospective Prospective Prospective Prospective
CMIA
EOC
Spain
Retrospective
not clear
ECLIA
OC
2014 2014 2015 2015
Indonesia Iran. Korea China
Retrospective Retrospective Prospective Prospective
not not not not
CMIA EIA CMIA ECLIA
OC OC OC EOC
2016
Italy
Prospective
not clear
EIA
EOC
2016
China
retrospective
not clear
CLIA
EOC
2016
China
retrospective
not clear
ECLIA
EOC
2016 2016
China Thailand
retrospective Prospective
70 30 72 70 102.6 70 72.2 66 74.2 70 70 140 140 77 140 103.4 150 72.3 87.6 72.3 97.5 63.1 102.3 150 140 140 140 70 70 84.8 31.08 72
not clear
2014
146 12 21 179 168 27 194 129 67 312 260 52 104 32 81 52 29 63 68 46 21 257 175 82 92 60 32 292 254 44 30 246
EOC OC EOC EOC OC EOC OC
prospective
18 1 7 2 8 4 42 14 30 28 8 20 4 3 2 7 6 5 7 2 5 13 4 9 16 8 8 55 21 29 10 26
EIA ELISA ECLIA CMIA EIA ELISA EIA
Asia
20 0 3 16 9 6 34 13 19 10 9 1 15 2 4 9 5 4 2 2 1 33 27 6 2 1 1 19 10 3 29 70
yes not clear yes yes yes yes not clear
2013
49 45 34 16 44 25 119 28 89 37 14 23 25 7 17 43 26 18 53 18 35 60 19 41 48 19 29 155 86 74 48 133
not clear not clear
ECLA ECLIA
OC OC
8
9
10 11 12 13
14
15
16
17 18
clear clear clear clear
note:Holcomb, K. only has the premenopausal patient. FN, false-negative; FP, false-positive; TP, true-positive; TN, true-negative, EOC: epithelial ovarian cancer; OC: ovarian cancer.
Fagan plot showed that the prior probability was 20%, and the posttest probability was 67% of PLR, and 5% of NLR. Additionally, the diagnostic accuracy of HE4 for detecting ovarian cancer between patients with adnexal mass was very high. Meta-regression analysis We divided 18 studies for subgroup analysis by:control (benign ovarian tumor groups and healthy, benign ovarian tumor groups), area (within and without Asia), HE4 testing method (enzyme linked immunosorbent assay (ELISA) and chemiluminescence assay (CMIA),), histopathology (EOC and OC), cutoff design (according to ROC curve, optimal value and set advance) and cutoff (cutoff<50 mmol/L and cutoff>50 mmol/L). These subgroups were used for meta-regression analysis. Through meta-regression analysis, we found except p-value of specificity of cutoff subgroup was 0.38 > 0.05, P values of sensitivity and specificity of all subgroups were less than 0.05. In the Joint Model the p-value of cutoff was < 0.05 (p = 0.04), which suggested the cutoff value was probably the main source of heterogeneity for HE4 in OC. Subgroup analysis According to the clinical application and result of metaregression, 18 studies were divided into the additional subgroups for analysis: premenopausal and premenopausal, EOC and OC (calculation out proportion of non-epithelial ovarian cancer was 15.57%); benign ovarian tumors and benign mixed the healthy, cutoff design (optimal cutoff and set advance cutoff), cutoff < 50 and cutoff > 50. Analysis diagnostic value of HE4 of each subgroup, because the group cutoff < 50 only has two studies, so no further analysis discussion. Each subgroup with indicators are presented in Table 3. Considering different cutoff values for HE4 may have different diagnostic values in ovarian cancer they may also have a significant
impact on a clinician’s assessment on the nature of adnexal tumors. We subdivided group of cutoff>50 mmol/L into two group as follow:60 mmol/L < cutoff < 100 mmol/L, 100 mol/L cutoff 150 mmol/L. The results of statistical analysis are shown in Table 4. Sensitivity analysis and publication bias Meta-analysis was carried out again after removing large samples or studies with large differences within their results. Our results showed no significant changes, indicating that the stability of this study was satisfactory and the results were reliable. Additionally, the constructed Deek’s plot shows there was no publication bias (t = 1.53, P = 0.15, Fig. 6). Conclusions In 2010–2013 saw heavy research about the diagnostic value of serum HE4 for ovarian cancer. These reports had the following results: an AUC range between 0.89–0.94, and subgroup analysis of the included study populations were divided into premenopausal and postmenopausal, early and advance OC, benign tumor patients and healthy people as control group, area of patient, kit recommended cutoff and 95% confidence interval, immunosorbent assay and chemiluminescence immunoassay method. All total studies and subgroups suggested that HE4 has high diagnostic value in the diagnosis of ovarian cancer. Our study used meta-analytic methods on evidence-based medicine to further explore HE4 value in the diagnosis of ovarian cancer. Compared with previous related meta-analysis, considering the HE4 expression mainly in the epithelial ovarian cancer, we added these subgroups: EOC and OC (proportion of non-epithelial
J. Huang et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 231 (2018) 35–42
39
Fig. 2. The tabular Presentation of QUADAS-2 Results.
ovarian cancer was 15.57%) and cutoff design (optimal cutoff and set advance cutoff), cutoff < 50 and cutoff > 50. We subdivided the cutoff group cutoff > 50 mmol/L into two group to better explore cutoff value range. This study incorporated 18 studies to create a meta-analysis. Threshold effect examination showed that there was no obvious threshold effect and the Deek’s plot showed there was no publication bias. These findings also indicated that the study was of reliable quality and good credibility. The results of pooled analysis were: AUC = 0.91, Q = 86.02, sensitivity 81%, specificity
91%, positive likelihood ratio 8.2, negative likelihood ratio 0.21, DOR 39. According to the above results, serum HE4 is a valuable marker for the clinical diagnosis of ovarian cancer, but some inadequacies still exist, which we discuss below. Heterogeneity statistics of this study (I2 = 98, p = 0.000), the beta of HSROC model p = 0.055>0.05, indicating a non-threshold effect caused by heterogeneity is larger. According meta-regression analysis we found that subgroup cutoff was the main source of overall heterogeneity for this study, however test method, area, control, histopathologic, cutoff design were not mains source of
40
J. Huang et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 231 (2018) 35–42
Fig. 3. The graphical of QUADAS-2 results.
Table 2 Combined statistics of included 18 studies. Index
Merger value
95%CI
I2(%)
Cochran-Q
p
DOR Sen Spe PLR NLR
39.00 0.81 0.91 8.20 0.21
25.00–62.00 0.77–0.85 0.86–0.93 5.60–12.00 0.17–0.26
100.00 70.49 91.69 88.84 71.67
5.8e+09 57.62 204.70 203.69 60.00
0.00 0.00 0.00 0.00 0.00
Sen: sensitivity; spe: specificity; DOR: diagnostic odds ratio;PLR: positive likelihood ratio; NLR: negative likelihood ratio; CI: confidence interval.
literature cannot provide enough data on early and advanced ovarian cancer to do further meta-analysis. This study was carried out on the menstrual state for premenopausal and postmenopausal subgroup analysis. Our results show that the diagnostic value HE4 in premenopausal and postmenopausal subgroup both has high diagnostic performance but has little different to each other. The diagnostic value of HE4 in the subgroup of benign ovarian tumors or benign mixed the healthy as control group have little difference. In other words, HE4 can use for early detection of ovarian tumor and ovarian disease
Fig. 4. Forest Plots of paired sensitivity and specificity for HE4.
heterogeneity. Previous studies revealed test method, histopathologic, cutoff design and so on are the main source of overall heterogeneity, but all of the above indicators affected overall heterogeneity in meta-regression of this study with p values > 0.05. We consider the reason of that most because of the smaller number studies limits. According to the clinical application of analysis, premenopausal and postmenopausal, early and advance OC may be the main source of heterogeneity, but the current
monitoring. The diagnostic value of HE4 in the subgroup of optimal cutoff or set advance cutoff also have little difference. This study compared the diagnostic value of HE4 in EOC with OC subgroup, the AUC of the epithelial ovarian cancer subgroup was 0.93, the AUC of the ovarian cancer subgroup was 0.90. The AUC of epithelial ovarian cancer group is higher than the ovarian cancer group (proportion of non-epithelial ovarian cancer was 15.57%). Using the Z test, P values > 0.05, revealed that there was no statistically significant
J. Huang et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 231 (2018) 35–42
Fig. 5. The SROC curve of the 18 included studies.
to the two AUC differences, and the diagnostic value of HE4 in EOC, OC subgroup has no difference. HE4 was mainly secreted by epithelial carcinoma, in clinical thinking of HE4 in the diagnosis of ovarian epithelial carcinoma has higher diagnostic performance, the conclusion of this research could be further analysis via larger of intake studies to make sure we can get more accurate conclusions. In this study we subdivided the cutoff>50 mmol/L group into two subgroup as follow:60 mmol/L
41
our analysis was small, more studies are required to determine if such difference makes sense. Cutoff values have a great influence on a clinician’s judgment of the nature of ovarian tumors. We will continue to collect more studies to verify this result and explore a smaller range of higher accurate diagnostic cutoff value in the future. Some research indicated that SROC and DOR are not easily used in clinical practice, however, the likelihood ratio is considered more clinically meaningful [37]. A larger positive likelihood ratio is considered to indicate an accurate diagnosis and helps with the discovery of OC patients. Negative likelihood ratio is considered the smaller the better at in negative correct diagnosis and distinguish false patients. A PLR > 10 or NLR < 0.1, show a decisive change from a priori probability to a posteriori probability and basic can confirm or exclude the diagnosis. The subgroup 5 < PLR <10 or 0.1 < NLR <0.2 indicated that there is a moderate change from the pretest probability to the post-test probability. Furthermore 2 < PLR <5 or 0.2 < NLR <0.5 indicated a small degree of change from the pretest probability to the post-test probability. Finally, 1 < PLR < 2 or 0.5 < NLR < 1, indicated that a priori probability to a posteriori probability basic don't change, and have limited use for disease diagnosis [38,39]. In this study, PLR was 8.2 group, and NLR was 0.21. When HE4 assay results were negative, the probability that the patient had ovarian cancer was 21%, which is not low enough to rule out ovarian cancer. Therefore, the likelihood ratio analysis reveals that the diagnosis performance of HE4 is still not perfect. Taken together, although the diagnostic performance of HE4 is not perfect, HE4 remains a valuable marker in the clinical diagnosis of ovarian cancer. HE4 is still can be used for early detection of tumor and disease monitoring. Galgano et al. [40] have reported that HE4 could be expressed in the distal convoluted tubules of the kidney. Yuan et al. [41] have reported that elevated levels of serum HE4 are associated with decreased kidney function, and also with advanced stages of renal fibrosis. Therefore, when we used HE4 to evaluate the diagnosis of ovarian tumors, patients with renal insufficiency should be excluded first. The limitations of this study included the fact that researchers report that the diagnosis value of HE4 reported for early and advance ovarian cancer are obviously different, but our study only included three studies separately describe the early diagnosis of
Table 3 meta-analysis statistics of each subgroup. Diagnosis index
subgroup
AUC
Q
Sen
Spe
PLR
NLR
DOR
Menstrua
premenopausal postmenopausal EOC OC health benign Set advance Optimal <50 >50
0.90 0.91 0.93 0.90 0.92 0.91 0.91 0.90
5.86 9.92 21.40 18.75 12.62 52.99 19.98 22.69
0.24 0.25 0.23 0.18 0.14 0.24 0.25 0.18
54.00 50.00 45.00 36.00 58.00 34.00 37.00 46.00
57.74
0.94 0.94 0.93 0.87 0.89 0.90 0.92 0.89 0.70 0.91
13.20 12.60 10.60 6.60 8.20 8.10 9.40 8.00
0.91
0.77 0.76 0.78 0.84 0.87 0.79 0.77 0.84 0.90 0.79
8.70
0.23
38.00
Histological control cutoff design Cutoff
EOC: epithelial ovarian cancer; OC: ovarian cancer; Sen: sensitivity; spe: specificity; PLR: positive likelihood ratio; NLR: negative likelihood ratio; DOR: diagnostic odds ratio.
Table 4 Diagnosis value of different ranges of cutoff. Group
AUC (95%CI)
Sen (95%CI)
Spe (95%CI)
PLR (95%CI)
NLR (95%CI)
DOR (95%CI)
60
0.90 [0.87–0.92] 0.90 [0.87–0.93]
0.78 [0.73–0.83] 0.83 [0.76–0.88]
0.90 [0.86–0.93] 0.92 [0.86–0.96]
8.10 [5.8–11.5] 10.30 [5.70–18.30]
0.24 [0.20–0.30] 0.19 [0.14–0.26]
34.00 [23.00–50.00] 55.00 [29.00–103.00]
100 cutoff150
Sen: sensitivity; spe: specificity; PLR: positive likelihood ratio; NLR: negative likelihood ratio; DOR: diagnostic odds ratio; CI: confidence interval.
42
J. Huang et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 231 (2018) 35–42
Fig. 6. Deek’s Funnel figure.
early and advance ovarian cancer. In the future, more articles that meet the inclusion criteria should be included for further analysis. Additionally, the quantity of included literature in our systematic review is little, so a larger amount of studies would significantly help verify the diagnostic value of HE4 in differentiating between epithelial ovarian cancer and mixed ovarian cancer. Disclosure of conflicts of interest The authors have no conflict of interest to disclose. This study was supported by the National natural science foundation of China (grant nos. 81460398, 81860457) and the Natural Science Foundation of Guangxi Province, China (grant no. 2017GXNSFAA198106). References [1] Siegel R.L., Miller KD, Jemal A. Cancer statistics, 2018. CA Cancer J Clin 2018;68 (1):7–30. [2] N H. SEER Cancer statistics review, 1975-2014. Bethesda, MD: National Cancer Institute; 2017 seer.cancer.gov/csr/1975_2014/. Accessed March 1, 2018. 2018. [3] Goff BA, Mandel LS, Drescher CW, Urban N, Gough S, Schurman KM, et al. Development of an ovarian cancer symptom index: possibilities for earlier detection. Cancer 2007;109(2):221–7. [4] Gershenson DM, Sun CC, Lu KH, Coleman RL, Sood AK, Malpica A, et al. Clinical behavior of stage II-IV low-grade serous carcinoma of the ovary. Obstet Gynecol 2006;108(2):361–8. [5] Morgan RJ. Epithelial ovarian cancer. JNCCN–J Natl Compr Cancer Netw 2011;82–113. [6] Andersen MR, Goff BA, Lowe KA, Scholler N, Bergan L, Drescher CW, et al. Use of a symptom index, CA125, and HE4 to predict ovarian cancer. Gynecol Oncol 2010;116(3):378–83. [7] Hellstrom I, Raycraft J, Hayden-Ledbetter M, Ledbetter JA, Schummer M, McIntosh M, et al. The HE4 (WFDC2) protein is a biomarker for ovarian carcinoma. Cancer Res 2003;63(13):3695–700. [8] Wang JW, Gao J, Yao HW, Wu ZY, Wang MJ, Qi J. Diagnostic accuracy of serum HE4, CA125 and ROMA in patients with ovarian cancer: a meta-analysis. Tumour Biol 2014;35(6):6127–38. [9] Karlsen MA, Hogdall EV, Christensen IJ, Borgfeldt C, Kalapotharakos G, Zdrazilova-Dubska L, et al. A novel diagnostic index combining HE4, CA125 and age may improve triage of women with suspected ovarian cancer - an international multicenter study in women with an ovarian mass. Gynecol Oncol 2015;138(3):640–6. [10] Hamed EO, Ahmed H, Sedeek OB, Mohammed AM, Abd-Alla AA. Abdel Ghaffar HM. Significance of HE4 estimation in comparison with CA125 in diagnosis of ovarian cancer and assessment of treatment response. Diagn Pathol 2013;8:11. [11] Zhuang HY, Gao J, Hu ZH, Liu JJ, Liu DW, Lin B. Co-expression of Lewis y antigen with human epididymis protein 4 in ovarian epithelial carcinoma. PLoS One 2013;8(7)e68994. [12] Drapkin R, von Horsten HH, Lin Y, Mok SC, Crum CP, Welch WR, et al. Human epididymis protein 4 (HE4) is a secreted glycoprotein that is overexpressed by serous and endometrioid ovarian carcinomas. Cancer Res 2005;65(6):2162–9. [13] Lin J, Qin J, Sangvatanakul V. Human epididymis protein 4 for differential diagnosis between benign gynecologic disease and ovarian cancer: a systematic review and meta-analysis. Eur J Obstet Gynecol Reprod Biol 2013;167(1):81–5.
[14] Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 2011;155(8):529–36. [15] Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ (Clinical research ed) 2003;327(7414):557–60. [16] Reitsma JB, Glas AS, Rutjes AW, Scholten RJ, Bossuyt PM, Zwinderman AH. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. J Clin Epidemiol 2005;58(10):982–90. [17] Hamza TH, Arends LR, van Houwelingen HC, Stijnen T. Multivariate random effects meta-analysis of diagnostic tests with multiple thresholds. BMC Med Res Methodol 2009;9:73. [18] Song F, Khan KS, Dinnes J, Sutton AJ. Asymmetric funnel plots and publication bias in meta-analyses of diagnostic accuracy. Int J Epidemiol 2002;31(1):88–95. [19] Moore RG, Brown AK, Miller MC, Skates S, Allard WJ, Verch T, et al. The use of multiple novel tumor biomarkers for the detection of ovarian carcinoma in patients with a pelvic mass. Gynecol Oncol 2008;108(2):402–8. [20] Montagnana M, Lippi G, Ruzzenente O, Bresciani V, Danese E, Scevarolli S, et al. The utility of serum human epididymis protein 4 (HE4) in patients with a pelvic mass. J Clin Lab Anal 2009;23(5):331–5. [21] Abdel-Azeez HA, Labib HA, Sharaf SM, Refaie AN. HE4 and mesothelin: novel biomarkers of ovarian carcinoma in patients with pelvic masses. Asian Pac J Cancer Prev 2010;11(1):111–6. [22] Holcomb K, Vucetic Z, Miller MC, Knapp RC. Human epididymis protein 4 offers superior specificity in the differentiation of benign and malignant adnexal masses in premenopausal women. Am J Obstet Gynecol 2011;205(4)358 e1-6. [23] Chang XH, Ye X, Dong L, Cheng HY, Cheng YX, Zhu LR, et al. Human epididymis protein 4 (HE4) as a serum tumor biomarker in patients with ovarian carcinoma. Int J Gynecol Cancer 2011;21(5):852–8. [24] Jacob F, Meier M, Caduff R, Goldstein D, Pochechueva T, Hacker N, et al. No benefit from combining HE4 and CA125 as ovarian tumor markers in a clinical setting. Gynecol Oncol 2011;121(3):487–91. [25] Van Gorp T, Cadron I, Despierre E, Daemen A, Leunen K, Amant F, et al. HE4 and CA125 as a diagnostic test in ovarian cancer: prospective validation of the Risk of Ovarian Malignancy Algorithm. Br J Cancer 2011;104(5):863–70. [26] Chan KK, Chen CA, Nam JH, Ochiai K, Wilailak S, Choon AT, et al. The use of HE4 in the prediction of ovarian cancer in Asian women with a pelvic mass. Gynecol Oncol 2013;128(2):239–44. [27] Ortiz-Munoz B, Aznar-Oroval E, Garcia AG, Peris AC, Ballestero PP, Yepes MS, et al. HE4, Ca125 and ROMA algorithm for differential diagnosis between benign gynaecological diseases and ovarian cancer. Tumor Biol 2014;35 (7):7249–58. [28] Winarto H, Laihad BJ, Nuranna L. Modification of cutoff values for HE4, CA125, the risk of malignancy index, and the risk of malignancy algorithm for ovarian cancer detection in Jakarta, Indonesia. Asian Pac J Cancer Prev : APJCP 2014;15 (5):1949–53. [29] Ghasemi N, Ghobadzadeh S, Zahraei M, Mohammadpour H, Bahrami S, Ganje MB, et al. HE4 combined with CA125: favorable screening tool for ovarian cancer. Med Oncol (Northwood, London, England) 2014;31(1):808. [30] Cho HY. Comparison of HE4, CA125, and risk of ovarian malignancy algorithm (ROMA) in the prediction of ovarian cancer in Korean women. Int J Gynecol Cancer 2015;25(9):991. [31] Chen XL, Zhou H, Chen R, He J, Wang Y, Huang LS, et al. Development of a multimarker assay for differential diagnosis of benign and malignant pelvic masses. Clin Chim Acta 2015;440:57–63. [32] Romagnolo C, Leon AE, Fabricio ASC, Taborelli M, Polesel J, Del Pup L, et al. HE4, CA125 and risk of ovarian malignancy algorithm (ROMA) as diagnostic tools for ovarian cancer in patients with a pelvic mass: an Italian multicenter study. Gynecol Oncol 2016;141(2):303–11. [33] Wei S, Li H, Zhang B. The diagnostic value of serum HE4 and CA-125 and ROMA index in ovarian cancer. Biomed Rep 2016;5(1):41–4. [34] Xu Y, Zhong RH, He J, Ding R, Lin HX, Deng YW, et al. Modification of cut-off values for HE4, CA125 and the ROMA algorithm for early-stage epithelial ovarian cancer detection: Results from 1021 cases in South China. Clin Biochem 2016;49(1-2):32–40. [35] Zheng LE, Qu JY, He F. The diagnosis and pathological value of combined detection of HE4 and CA125 for patients with ovarian cancer. Open Med (Warsaw, Poland) 2016;11(1):125–32. [36] Yanaranop M, Tiyayon J, Nakrangsee S, Thinkhamrop B. Diagnostic accuracy and optimal cutoff value of serum HE4 to predict ovarian cancer in Thai women with pelvic masses. J Med Assoc Thai 2016;99(12):1263–7. [37] Deeks JJ. Systematic reviews in health care: systematic reviews of evaluations of diagnostic and screening tests. BMJ (Clinical research ed) 2001;323 (7305):157–62. [38] Scales Jr. CD, Dahm P, Sultan S, Campbell-Scherer D, Devereaux PJ. How to use an article about a diagnostic test. J Urol 2008;180(2):469–76. [39] Jaeschke R, Guyatt GH, Sackett DL. Users’ guides to the medical literature. III. How to use an article about a diagnostic test. B. What are the results and will they help me in caring for my patients? The Evidence-Based Medicine Working Group. JAMA 1994;271(9):703–7. [40] Galgano MT, Hampton GM, Frierson Jr. HF. Comprehensive analysis of HE4 expression in normal and malignant human tissues. Mod Pathol 2006;19 (6):847–53. [41] Yuan TF, Li Y. Human epididymis protein 4 as a potential biomarker of chronic kidney disease in female patients with normal ovarian function. Lab Med 2017;48(3):238–43.