Efficacy of levonorgestrel intrauterine device (LIUD) and transvaginal ultrasound (TVUS) as treatment and surveillance of women with early endometrial cancer (EC) and complex atypical hyperplasia (CAH) who are poor surgical candidates

Objectives: The purpose was to investigate (1) the efficacy of levonorgestrel intrauterine device (LIUD) for treatment of early-stage endometrial cancer (EC) or complex atypical hyperplasia (CAH) in women with high body mass index (BMI) or multiple significant comorbidities and (2) thickness of endometrial lining (EL) on transvaginal ultrasound (TVUS) as a surrogate marker for pathologic treatment response. Methods: An institutional review board–approved, single-institution retrospective study identified patients who had high BMI or poor surgical candidates with biopsy-proven early-stage EC and CAH and subsequently treated with LIUD. Measurements of EL were obtained before treatment and serial TVUS (at the same institution) were performed to evaluate EL every 3 months. Endometrial biopsies (EMB) were performed concordantly with TVUS to evaluate for clinical response. Clinical benefit rates (CBR) (complete response [CR] + partial response [PR] + stable disease [SD]) and progression-free survival (PFS) were calculated using follow-up EMB results. Maximum likelihood ordinal logistic model was used to predict a clinical response based on EMB compared with changes in EL on serial TVUS. Results: A total of 32 patients were included: 9 (28.1%) with CAH and 23 (71.9%) with early EC. The mean BMI was 49.3 kg/m2 and the mean Revised Cardiac Risk Index was 2.5. CR was achieved in 13 (50.0%), PR in 4 (15.4%), SD in 5 (19.2%), and progressive disease in 4 (15.4%). CBR was 84.6%, with a mean PFS of 3.3 years. The ordinal logistic model showed a 20% reduction in EL from baseline predicted an 85.4% rate of no progression and 55.9% CR/PR. A 50% reduction of EL from baseline predicted a 94.8% nonprogression rate and 79.6% CR/PR. Importantly, of the 4 patients who progressed and failed, the EL was thickened or increasing and the model correctly predicted progression. Conclusions: In select patients with high BMI or significant comorbidities, LIUD is a reasonable first alternative to treat EC or CAH. EL, as measured on serial TVUS, is a good surrogate marker to predict response in this unique approach to a difficult patient population.