Follow-up MRI recommendations for adnexal lesions on pelvic ultrasound: frequency, associations and outcomes

To assess the frequency and outcomes of follow-up MRI recommendations for adnexal lesions on pelvic US interpreted by US experts versus non-US expert abdominal radiologists.

In this IRB-approved, HIPAA-compliant retrospective cohort study, all pelvic US reports in adult women between 2018-2023 at a tertiary academic center were screened using a validated NLP algorithm for follow-up MRI recommendations for incidental adnexal lesions. MRI recommendation rates, compliance with the recommendation, and impact of MRI on management (malignancy risk upgrade/no change/downgrade and final outcomes) were compared between US expert and non-US expert abdominal radiologists.

Of 19,335 pelvic US in 14,328 patients (15,609, 80.7% read by US experts, 3726, 19.3% by non-US expert abdominal radiologists), 458 (2.3%) had follow-up MRI recommendations. US experts made fewer recommendations (248, 1.5% vs 210, 5.6%; p < 0.001) and had higher compliance rate (157, 63.3% vs 108, 51.4%; p = 0.01). MRI caused risk downgrade, no change, and upgrade in 191 (72.1%), 62 (23.4%), and 12 (4.5%) respectively; without significant difference between US experts and non-US experts (113, 72%, 39, 24.8%, 5, 3.2% vs 78, 72.2%, 23, 21.3%, 7, 6.5%; p = 0.96). Of 191 lesions with risk-downgrade, 124 underwent surgery (n = 60) or follow-up (n = 64) without any cancers; 67 had no further follow-up. Of 12 lesions with risk-upgrade on MRI, 6 (50%) were malignant, 4 (33.3%) benign, 2 (16.7%) borderline. Of 193 cases where MRI was not performed, 60 were resected (53, 88.3% benign, 3, 5.0% borderline, 4, 6.7% malignant).

US experts made fewer follow-up MRI recommendations, with no difference in the impact of MRI on risk-assessment between US experts and non-US expert radiologists. MRI resulted in malignancy risk-downgrade in most lesions and can potentially avoid unnecessary surgeries.