Risk of Ovarian Malignancy Algorithm (ROMA) through time and space: a meta-analysis

Objectives:

Each year in the U.S. over two million women will be diagnosed with an ovarian cyst or pelvic mass of which approximately 200,000 will require surgery. Of these women, 22,000 (~11%) will ultimately be diagnosed with epithelial ovarian cancer (EOC). Multiple studies have demonstrated that women with EOC managed by a gynecologic oncologist and at high volume institutions for EOC have improved survival rates. Despite these facts, less than 50% of women diagnosed with EOC are cared for by gynecologic oncologists or at high volume institutions. The Risk of Ovarian Malignancy Algorithm (ROMA) was developed to assist in the triage of women at high risk for EOC to gynecologic oncologists. ROMA has been cleared by the FDA for risk assessment and triage in women presenting with an ovarian cyst or pelvic mass. The objective of this study was to demonstrate the performance of ROMA in a large population of women presenting with a pelvic mass. This is the largest collection of patients evaluated with ROMA to date.

Methods:

A meta-analysis was conducted by pooling data from 5 pelvic mass trials. This study was institutional review board (IRB)-exempt due to the secondary use of preexisting data. Data was combined into one cohort for analysis of ROMA. As per ROMA design, the sensitivity, PPV and NPV were calculated at a set specificity of ~75%. All samples were run on the Abbott ARCHITECT platform for both CA125 and HE4.

Results:

A total of 2,004 patients were included in the analysis (837 premenopausal and 1,167 postmenopausal). There were 448 EOCs (89 stage I, 45 stage II, 284 stage III, 24 stage IV and 6 unstaged), 83 Borderline/LMP tumors, 1,224 benign masses, 23 non-epithelial ovarian cancers, 165 other gynecological cancers, 59 non-gynecologic metastatic cancers, and 2 mesotheliomas. For the detection of benign vs EOC only in pre and postmenopausal women, at a specificity of 75.0%, ROMA achieved a sensitivity of 91.3%, a PPV of 57.2% and a NPV 95.9%. In postmenopausal women, there were 554 benign cases and 379 EOCs. At a specificity of 75.3%, ROMA achieved a sensitivity of 92.3%, a PPV of 71.9% and a NPV of 93.5%. In premenopausal patients, there were 670 benign cases and 69 EOCs. At a specificity of 74.8%, ROMA achieved a sensitivity of 85.5%, a PPV of 25.9% and a NPV of 98.0%. An analysis of the ability of ROMA to differentiate early stage EOC from benign in pre and post-menopausal women revealed a sensitivity of 76.1%, a specificity of 75.0%, a PPV of 25.0% a NPV of 96.6%. ROMA detected over 98% of all late stage III and IV EOCs.

Conclusions:

ROMA has been shown in multiple independent prospective trials to be an accurate tool for pelvic mass risk assessment. This is the largest collection to date to validate the performance characteristics of ROMA. ROMA should be used as one of the tools for pelvic mass risk assessment to assist in the triage of patients at high risk for EOC to gynecologic oncologists.