Introduction/Background*Human epididimis protein 4 (HE4) has been reported as a promizing biomarker in the assessment of the risk of malignancy in patients, diagnosed with pelvic mass. Howewer, reference limits of HE4 do not provide clinically relevant discrimination between malignant and benign ovarian diseases. The superiority of well-known Risk of Ovarian Malignancy Algorithm (ROMA), based on both HE4 and Carbohidrate Antigene 125 (CA125), over CA125 alone and its clinical significance are still debated.

The aim of this study was to elaborate a new algorithm, based on serum CA125, HE4 and age, to assess the risk of malignancy in premenopausal patients with pelvic mass.

Methodology The training dataset included 284 premenopausal patients operated because of presence of pelvic mass, out of which 35 and 249 had malignant and benign disease respectively. A new algorithm based on serum HE4, CA125 and age as variables has been developed by using the scenario of discrimination “benign disease versus epithelial ovarian cancer (all stages) together with borderline ovarian tumors FIGO stage 1c2-3c”. This new algorithm, named Risk of Ovarian Cancer Kazan Index (ROCK-I), was further compared with ROMA and CA125 alone. A recruitment of a validating dataset is presently being concluded.

Result(s)*The original ROCK-I has demonstrated greater diagnostic performance than ROMA (table 1, figure 1). When the above-mentioned scenario of discrimination was used the specificities (95% confidence interval (CI)) of ROCK-I and ROMA were 93,7% (90 – 96,1%) and 85,7% (80.8 – 89.5%) respectively, while the sensitivities (95%CI) were 89.6% (75.3 – 96.0) and 83.95 (68.5 – 92.6) respectively and accuracies (95%CI) were 93.2% (89.7 – 95.6) and 85.5% (80.9 – 89.1). The superiority of ROCK-I in specificity, accuracy and positive predictive value was statistically significant (p<0.05), while the tendency towards the superiority in sensitivity and negative predictive value does not reach the statistical significance. Areas under receiver-operating-characteristic curves (95%CI) of ROCK-I and ROMA were 0.97 (0.948 – 0.992) and 0.94 (0.899 – 0.981) respectively.

• Download figure
• Open in new tab
• Download powerpoint

Abstract 723 Figure 1

Pre-test and post-test probabilities of Epithelial ovarian cancer (all stage) together with borderline ovarian tumours stage lc2-lllc

Conclusion*The proposed ROCK-I has demonstrated greater diagnostic performance than ROMA in the analyzed dataset. If an external validation can show similar or even slightly lower difference between ROCK-I and ROMA it may provide a new basis of routine-use of HE4 in patients with pelvic mass.