The continuous variables ROMA and CPH-I in this study have several non-standard distributions. Therefore, the statistics of ROMA and CPH-I will be presented as median values (Q25% – Q75%).
The median values of CPH-I and ROMA in the OC group, as shown in the Table 3, are higher than those of the benign tumor group, and borderline tumor group. Specifically, for the study population (n = 475), the median value of CPH-I in the malignant group was 24.81% (3.49 – 81.21%), whereas in the benign tumor group it was 0.42% (0.29 – 24.39%), and those in the borderline tumor was 0.42% (0.29 – 24.39%). The median values of ROMA in the cancer group, borderline group, and the benign group were 49.93% (12.78 – 81.22%), 5.48% (3.28 – 16.38%), and 5.03% (3.46 – 8.71%), respectively (p < 0.05). Compared to previous studies, the median values of CPH-I and ROMA from our research are lower than those some other studies in the world. According to Adriana Yoshida (2016), the median values of CPH-I for benign tumors, ovarian carcinomas were 1.4%, and 83.4%, respectively (12). Meanwhile, in Lubos Minar’s study, the median values of CPH-I in the benign and malignant groups were 2.2% and 75.4%, respectively (14) (Table 5). More detailed analysis in the pre-menopausal and post-menopausal groups, which examine the differences between CPH-I and ROMA, showed that the median values of CPH-I and ROMA were higher in the post-menopausal group, compared with the pre-menopausal group (Table 3). Median values of CPH-I and ROMA in the post-menopausal group were higher than what in the pre-menopausal group. The Sensitivity/Specificity (Se/Sp) of CPH-I in the absence of marginal ovarian tumors, non-epithelial OC, and OC metastasis, was 89.7%/ 85.3%, but if the above objects were included, the corresponding Se/ Sp became lower at 73.1%/ 84.4% (12).
The Se/Sp of ROMA and CPH-I in the diagnosis of OC were 76.1%/ 87.0% and 83.6%/ 78.7% respectively. The optimal cut-off point of the CPH-I was 1.89%, the AUCs of ROMA and CPH-I were respectively 0.860 (95%CI: 0.825 – 0.890) and 0.868 (95% CI: 0.833 – 0.896). Research by T. Nikola (2017) differential diagnosis between ovarian endometriosis and ovarian carcinoma showed that the accuracy of the Copenhagen Index was higher than ROMA, 93.75% and 85.42%, respectively (15).
Zhiheng Wang et al. (2019) argued that the HE4 level, ROMA and CPH-I values of epithelial ovarian cancer (EOC) stages I and II (I + II) were all higher than that of borderline ovarian tumor (BOT) I + II and benign groups whether in all, pre-, or postmenopausal groups (p < 0.01). When distinguishing BOT I+II from EOC I+II, the AUC-ROC of CPH-I and HE4 were larger than CA-125 (p < 0.001). CPH-I is more valuable than CA-125 when distinguishing marginal ovarian tumors with stage I – II ovarian carcinoma, while HE4 may be better than CA-125 in post-menopausal group; HE4 and CPH-I have been more advantageous than CA-125 when differentiating a borderline ovarian tumor with an early-stage ovarian carcinoma (I + II) in the absence of histology or type of serum fluid. The AUC of CPH-I and ROMA in the pre-menopausal group are respectively 0.779 and 0.760, and in the post-menopausal group are 0.802 and 0.774. In the pre-menopausal group, the Se/ Sp of ROMA and CPH-I are respectively 78.69%/ 64.75% and 70.49%/ 78.69%. In the postmenopausal group, the Se/ Sp of ROMA and CPH–I are respectively 82.98%/ 68.18% and 85.11%/ 68.18% (16).
According to Estrid Høgdall, ROMA and CPH–I can be used to the differential diagnosis between benign and malignant ovarian tumors [13]. Since family doctors might be unable to perform an ultrasound test, therefore both ROMA and CPH-I could provide the initial reliable information which helps the patient to get early diagnosis and proper treatment from specialized centers. In general, CPH-I and ROMA have similar sensitivity and accuracy. CPH-I is not identical to ROMA and RMI because it is not independent from ultrasound test and menopausal status. Menopausal status can be determined based on age, hormone concentration or amenorrhea per year, so that, the diagnosis of menopausal status has not been standardized. Therefore, CHP-I could be a simpler method to optimize the management when assessing women with suspected OC, including age instead of menopausal status (11,17).
Table 5
Diagnostic validity of CPH-I and ROMA from literature.
Authors
|
Copenhagen Index
|
ROMA
|
AUC
|
Se/Sp (%)
|
AUC
|
Se/Sp (%)
|
A. Yoshida (2016) [12]
|
0.84
|
73.1/ 84.4
|
0.82
|
71.2/ 83.5
|
L. Minar (2017) [14]
|
0.81
|
69.0/ 85.0
|
0.83
|
71.0/ 88.0
|
T. Nikolova (2017) [15]
|
0.91
|
81.8/ 97.3
|
0.90
|
90.9/ 83.8
|
Z. Wang (2019) [16]
|
0.810
|
78.7/ 74.3
|
0.807
|
62.9/ 88.2
|
Estrid Høgdall (2016) [17]
|
0.960
|
–
|
0.954
|
–
|
Nguyen Vu Quoc Huy (2018) (18)
|
–
|
–
|
0.912
|
86.7/ 88.7
|
This study
|
0.862
|
80.4/ 80.3
|
0.848
|
69.5/ 92.2
|
The Copenhagen Index is a new indicator, which has been introduced in several studies around the world. The ROMA algorithm is an index that the US Food and Drug Administration has introduced in clinical practice to distinguish benign and malignant ovarian tumors, based on the three variables: CA-125, serum HE4, and menopausal status (19). These two indexes have quite similar values since both are partially based on CA-125, HE4. Since the serum CA-125, HE4 concentrations were affected by many factors including age, smoking, uterine fibroids, pregnancy, endometriosis, pelvic inflammatory disease, gallbladder stone, this will affect the values of the Copenhagen index and ROMA (20,21). In the future, more research about these two indicators on different target groups to clarify these differences, aiming to overcome the limitations of these indicators and avail in clinical practice.
To the best of our knowledge, this is the first cohort study from Vietnam with large number of ovarian tumor subjects included, examining the validity of CPH-I and comparing with those from ROMA in risk stratification for ovarian tumor malignancy. Although being rigorously designed and implemented, the OC cases were still limited, and the laboratory equipements were different at the two facilities where the works were done, this could partially affect the homogeneity of the data analysis.