Abstract
Background/Aim: The aim of the study was to identify new non-invasive ovarian cancer (OC) tumor markers. Materials and Methods: In postmenopausal ovarian cancer patients and in a control group (benign ovarian lesions and healthy subjects), preoperative plasma levels of cytokines, metalloproteinases and their tissue inhibitors were determined using ELISA while those of CA125 and HE4 by chemiluminescent microparticle immunoassay methods. Results: The diagnostic sensitivity (SE) value was the highest for HE4 and MMP-7 (78.0%). The diagnostic specificity (SP) for M-CSF, VEGF and MMP-9 was 95.2%, 95.2% and 95.7%, respectively. The highest positive predictive value (PPV) for M-CSF and MMP-9 was ~84.6% and negative predictive value (NPV) for MMP-7 and HE4 was ~87.6%. The biggest areas under the ROC curve were obtained for the combination of VEGF, MMP-7 or MMP-9 with HE4+CA125 (0.9130-0.9234), but not for CA125+HE4 (0.8260). Conclusion: Our research confirms the validity of combining classic markers with new markers to improve the diagnostic power of CA125 and HE4.
Ovarian cancer (OC) occurs at all ages and has a high mortality rate attributable to its occult development (1, 2). Recently published research has suggested that the majority of ovarian carcinomas originate from high-grade intraepithelial serous carcinomas in the fallopian tube which then spread to the ovary (3). The best known and most widely used tumor markers in routine ovarian cancer diagnosis are CA125 (carbohydrate antigen 125) (4-7) and HE4 (human epididymis protein 4) (5, 8). Many researchers are working on detecting new markers useful for early diagnosis of epithelial ovarian cancer (9, 10). The overexpression and increased concentrations of metalloproteinases (MMPs), tissue inhibitors of metalloproteinases (TIMPs) or hematopoietic growth factors (HGFs) and cytokines have been observed in the course of various types of cancers (6, 7, 11, 12).
The aim of this study was to determine the diagnostic power, according to the analysis of area under the ROC curve, of selected cytokines (M-CSF- macrophage-colony stimulating factor; VEGF - vascular endothelial growth factor), MMPs (MMP-2 -metalloproteinase-2; MMP-7 - metalloproteinase-7; MMP-9 - metalloproteinase-9), and TIMPs (TIMP-1 - tissue inhibitor of metalloproteinase-1; TIMP-2 - tissue inhibitor of metalloproteinase-2) separately and in combination with established tumor markers for the best cancer detection. To better reflect the female population, the control group included healthy women and women with benign ovarian lesions.
Materials and Methods
Patients. The groups studied are presented in Table I. A group of 140 postmenopausal women with epithelial ovarian cancer patients was analyzed. Clinical stages and histological classification were established based on the criteria of the International Federation of Gynecology and Obstetrics (FIGO) and the World Health Organization (WHO). Physical and blood examinations, ultrasound scanning and chest X-rays were used in pretreatment staging procedures. Due to very high HE4 concentration levels, patients with renal failure were excluded. The patients had not received any therapy before blood sample collection.
In the control group, 140 postmenopausal women [70 benign ovarian tumor (BOT) and 70 healthy volunteers] were included (Table I). The histopathology of the BOT group was established by tissue biopsy. Before blood collection the healthy women group was examined also by a gynecologist and an ultrasound examination was performed in every case. We excluded subjects with prior endometriosis or with renal failure.
Women with ovarian cancer or with benign lesions were patients of the Department of Gynecology, University Hospital in Białystok, Poland, in the years 2009-2014. All research participants had given their permission to be part of the study. The local Ethics Committee of the Medical University in Białystok, approved the study: R-I-002/314/2009; R-I-002/262/2010 and R-I-002/239/2014.
Biochemical analyses. Plasma samples were obtained following centrifugation (1000 rpm/15 min.) of venous blood collected into heparin sodium tubes and stored at −85°C. Duplicate samples were measured for each patient with enzyme-linked immunosorbent assay (ELISA) (Quantikine Human Immunoassay, R&D systems) for cytokines, metalloproteinases (MMPs) and tissue inhibitors of metalloproteinase (TIMPs). The assay showed no significant cross-reactivity with other human cytokines, metalloproteinases and tissue inhibitors of metalloproteinases.
The concentrations of comparable markers were assayed by chemiluminescent microparticle immunoassay (CMIA) (Abbott, Chicago, IL, USA).
The intra-assay coefficient of variation (CV%) of M-CSF is reported to be 3.4% at a mean concentration of 227 pg/ml, SD=7.7; of VEGF, 4.5% at a mean concentration of 235 pg/ml, SD=10.6; of MMP-2, 3.8% at a mean concentration of 11.2 ng/ml, SD=0.420; of MMP-7, 3.7% at a mean concentration of 4.58 ng/ml, SD=0.168; of MMP-9, 2.9% at a mean concentration of 11.0 ng/ml, SD=0.316; of TIMP-1, 5.0% at a mean concentration of 6.95 ng/ml, SD=0.35; and of TIMP-2, 3.4% at a mean concentration of 3.45 ng/ml, SD=0.116. The intra-assay CV for CA125 is reported to be 2.4% at a mean concentration of 43.5 U/ml, SD=1.1 and that of HE4 3.7% at a mean concentration of 39.0 pmol/l, SD=1.4.
The inter-assay coefficient of variation (CV%) of M-CSF is reported to be 3.1% at a mean concentration of 232 pg/ml, SD=7.3; of VEGF, 7.0% at a mean concentration of 250 pg/ml, SD=17.4; of MMP-2, 6.6% at a mean concentration of 11.1 ng/ml, SD=0.738; of MMP-7, 4.1% at a mean concentration of 4.82 ng/ml, SD=0.198; of MMP-9, 6.9% at a mean concentration of 12.2 ng/ml, SD=0.845; of TIMP-1, 4.9% at a mean concentration of 6.90 ng/ml, SD=0.34; and of TIMP-2, 5.7% at a mean concentration of 3.45 ng/ml, SD=0.197. The inter-assay CV for CA125 is reported to be 3.9% at a mean concentration of 43.5 U/ml, SD=1.7, and that of HE4 2.8% at a mean concentration of 39.0 pmol/l, SD=1.1.
Statistical analysis. We performed statistical analysis using the STATISTICA 12.0 PL program. The diagnostic power of all studied markers was compared by assessing the significance of differences between the areas under their ROC curves (p<0.05), (the GraphRoc Program for Windows).
The cut-off values were calculated by Youden's index and were as follows: M-CSF 1004.9 pg/ml; VEGF 402.6 pg/ml; MMP-2 194.8 ng/ml; MMP-7 3.9 ng/ml; MMP-9 519.8 ng/ml; TIMP-1 170.0 ng/ml; TIMP-2 49.4 ng/ml; HE4 72.3 pmol/l and CA125 81.3 U/ml.
Results
Tables II and III present the sensitivity (SE), specificity (SP), positive (PPV) and negative (NPV) predictive values of the tested parameters. The SE value in the ovarian cancer group was the highest for HE4 (78.0%) and MMP-7 (78.0%). The diagnostic SP was the highest for M-CSF, VEGF and MMP-9 (95.2%; 95.2%, 95.7% respectively) and was higher than that for CA125 (91.0%) and HE4 (82.2%). We indicated also the highest PPV value for M-CSF (84.5%) and MMP-9 (84.6%) and the highest NPV value for MMP-7 and HE4 (87.6% and 87.4%) (Table II). The combined analysis of the investigated parameters resulted in a high increase in the SE value and the maximum ranges were obtained for the combination of MMP-7, MMP-9 or M-CSF with both conventional tumor markers (95.0%; 94.0%; 93.0%, respectively). The SP and PPV values dropped slightly during the combined analysis (Table III). In the OC cancer group the NPV values were the highest for the combination of: VEGF+CA125 (97.8%), MMP-7+CA125+HE4 (96.2%) and M-CSF+CA125+HE4 (95.2%). Interestingly, the diagnostic criteria demonstrated for the combined CA125 and HE4 analysis reached lower values: SE-89.0%, SP-87.0%, PPV-68.5% and NPV-92.9% (Table III).
The ROC (receiver-operating characteristics) is a commonly used method for comparing the diagnostic power of laboratory tests. The AUCs (area under the ROC curve) of every biomarker compared with the remaining group were significantly higher compared to AUC=0.5 (with exception of MMP-2 and TIMP-2) (Table IV). HE4 (0.8647), CA125 (0.8301) and MMP-7 (0.8260) areas under the ROC curve were the largest in the OC group (Table IV; Figure 1). The combination of biomarkers studied resulted in a further increase in the area under the ROC curve (Table IV, Figures 2, 3 and 4). Especially for the combination of VEGF, MMP-7 or MMP-9 with HE4+CA125, increased to the value: 0.9130-0.9234 (Table IV). It should be emphasized that the areas under the ROC for the tested biomarkers in combination with HE4 or CA125 were larger than those for the CA125 and HE4 combination (0.8260) (Table III; Figure 5).
Discussion
In this study, the analysis of the area under the ROC curve was utilized to determine the diagnostics usefulness of selected cytokines, metalloproteinases and tissue inhibitors of metalloproteinases separately and in combination with accepted markers of ovarian cancer. The diagnostic performance of researched markers in discriminating OC from the control group comprised of BOT and healthy subjects showed the best results for comparative markers. Moreover, HE4 (0.8647) was better in discriminating between the aforementioned groups than CA125 (0.8301). Our data are in accordance with (13-15) or different from (16-18) the results from other studies. The differences of the results between the studies might be due to differences in the disease stages and histologic types of ovarian cancer group and the composition of control groups enrolled in each study. The MMP-7 AUC value (0.8260) was slightly lower than that of CA125 and it was the best result among all tested cytokines, metalloproteinases and their tissue inhibitors. These results correspond to our previous study (19) though the ROC curve was plotted for the healthy women group vs. malignant cases group. The combination of CA125+HE4 with MMP-7, MMP-9 or VEGF resulted in the best diagnostic power with the highest AUC value, up to 0.9234. Comparable results were obtained regarding: VEGF, MMP-9 or MMP-7 in the ovarian (19-22) breast (23, 24) or gastric cancer (25). Interestingly, the AUC value was smaller during the combined analysis of both commonly used tumor markers (0.8260). The work of Jacob et al. (12) in a group of postmenopausal women confirmed our observations. M-CSF, MMP-2, TIMP-1 and TIMP-2 separately showed limited diagnostic power in discriminating between the groups mentioned above. In contrast, other investigators found that increased serum levels of TIMP-1 and the ratio of TIMP-1 to MMP-2 as well as the ratio of TIMP-1 to the complex: MMP-2–TIMP-2 are useful in discriminating between malignant ovarian tumors and ovarian tumors of low malignant potential (26). However, the tested group was composed mainly of serous and mucinous malignant ovarian tumors and was far smaller (61 patients) (26). Our data are also in opposition to the studies of other researchers, who compared urinary TIMP-1 and MMP-2 levels in patients with pancreatic malignancies (27). Our present observations about the diagnostic power of M-CSF are in disagreement with the results of our previous studies in ovarian (AUC=0.8562-0.8864) (6, 28), endometrial (AUC=0.794) (29) or breast cancer (AUC=0.769-0.801) (24, 30-32). These data were obtained after statistical analysis of the ROC curve between the group of women with cancer disease versus healthy individuals. We believe that the composition of the presented control group reflects better the current population of women and makes our analysis more reliable. According to calculated by Youden's index cut-off for selected biomarkers, HE4 and MMP-7 presented the highest and equal values of diagnostic SE (78%). Furthermore, a maximum increase in the diagnostic SE was obtained for the combination of M-CSF, MMP-7 or MMP-9 with both ovarian tumor markers to 93-95% as compared with the use of both CA125 and HE4 together. Our data are similar to the published results of other investigators who found that the combination of CA125, MMP-7, CCL11 (CC chemokine 11) and CCL18 (CC chemokine 18) improves the diagnostic SE value in the early stages of ovarian cancer (94%) (33). Zhang et al. (34, 35) postulated the usefulness of the combined detection of MMP-9, Hpa (heparanase) and CL (cathepsin L) for patients' clinical evaluation and determination of the extent of OC metastasis before surgery. The conclusion about M-CSF is also in line with our previous studies, in which diagnostic value of the presence of M-CSF in various diagnostic panels with established tumor markers was evaluated (6, 23, 28, 31, 36). Diagnostic specificity (SP) reached the highest values for both cytokines and MMP-9 (95.2%). Review of the existing literature, indicated that similar results have been obtained by other investigators regarding the course of ovarian cancer (81.4-100%) (21, 28, 35). It should be underlined, that among the examined factors MMP-7 showed comparable or higher values of PPV and NPV compared to those presented by HE4, while M-CSF and MMP-9 presented the highest values of PPV in the whole group studied (~84.5%). In the present study, the combination of CA125 with MMP-7 had undoubtedly the highest NPV value, ~98%. These findings as well as the results on TIMP-1 are partially in accordance with our previous publications (19, 28) probably as a result of differences in the composition of control groups. Unfortunately, we could not compare our data regarding the poor diagnostic utility of MMP-2 and TIMP-2 in ovarian cancer with the results of other researchers due to a lack of publications on the subject, although their diagnostic power was demonstrated in breast (32) or pancreatic malignancies (27).
In summary, to the best of our knowledge, the current study is the first to evaluate the diagnostic utility of preoperative plasma levels of 9 carefully selected markers: cytokines, MMPs and TIMPs independently and in combination with CA125 or HE4, based on the area under the ROC curve analysis. The strength of the current results lies in the fact that the study groups (OC patients and control BOT group) were homogeneous, only serous and endometrioid sub-types of epithelial ovarian tumors were enrolled, and all participants were postmenopausal. Statistical analysis showed that HE4 was superior to CA125 in discriminating between OC and control group. The results of this study also suggest that combining VEGF, MMP-7 or MMP-9 in the diagnostic panels with HE4 and/or CA125 measurements might minimize the rate of misdiagnosis and improve the diagnostic power of both commonly used tumor markers.
Acknowledgements
This research was financed by a Grant for Scientific Research (nr N N407 530738) in the years 2010-2014 from the Polish Ministry of Science and Higher Education.
Footnotes
Authors' Contributions
GEB conceived the study, performed the immunoassays, conducted the statistical analysis and drafted the manuscript; BP performed the immunoassays and assisted in drafting the manuscript; EG conducted data acquisition and participated in sequence alignment; MZ performed the immunoassays; JO participated in data interpretation; MS participated in data interpretation; MD participated in the design and coordination of the study; SŁ participated in the design and coordination of the study, data interpretation and assisted in drafting the manuscript. All Authors have read and approved the final manuscript.
Conflicts of Interest
The Authors declare that they have no competing interests.
- Received March 7, 2019.
- Revision received April 4, 2019.
- Accepted April 8, 2019.
- Copyright© 2019, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved