Relapse of endometrial hyperplasia after conservative treatment: a cohort study with long-term follow-up.

STUDY QUESTION What is the risk of relapse for women with endometrial hyperplasia treated with levonorgestrel-releasing intrauterine system (LNG-IUS) or oral progestogens? SUMMARY ANSWER Relapse of complex endometrial hyperplasia after initial regression occurs often and it occurs less often in women treated with LNG-IUS than with oral progestogens. WHAT IS KNOWN ALREADY The LNG-IUS and oral progestogens are used to treat women with endometrial hyperplasia and achieve regression. There is uncertainty over whether further surveillance for these women is necessary as the risk for relapse is unknown. STUDY DESIGN, SIZE, DURATION A cohort study of 219 women with complex non-atypical or atypical endometrial hyperplasia who were treated and achieved initial regression with LNG-IUS (n = 153) or oral progestogens (n = 66) from August 1998 until December 2007 and followed up for >5 years. The mean length of follow-up was 74.7 ± SD 31.8 months for the LNG-IUS versus 87.6 ± SD 42.2 months for the oral progestogen group. PARTICIPANTS/MATERIALS, SETTING, METHODS We evaluated the proportion of women who relapsed or had hysterectomy after initial regression with LNG-IUS compared with oral progestogens by logistic regression and adjusting for confounding. The time from regression to relapse was explored through a survival analysis. MAIN RESULTS AND THE ROLE OF CHANCE Relapse of hyperplasia occurred in 13.7% (21/153) of women treated with LNG-IUS compared with 30.3% (20/66) of women treated with oral progestogens [adjusted odds ratio (OR) = 0.34, 95% confidence interval (CI): 0.17-0.7, P = 0.005]. Relapse rates over long-term follow-up were lower for complex non-atypical hyperplasia compared with atypical hyperplasia for both LNG-IUS (12.7%, 18/142 versus 27.3%, 3/11, respectively; P ≤ 0.001) and oral progestogens (28.3%, 17/60 versus 50%, 3/6, respectively; P ≤ 0.001). The survival analysis indicates that relapse occurred less often with LNG-IUS at 12, 24, 36, 48, 60 and >60 months of follow-up (hazard ratio 0.37, 95% CI: 0.2-0.7, P = 0.0013). There were no events of relapse after 48 months from regression with oral progestogens, but 5 women treated with LNG-IUS relapsed after 60 months when treatment was discontinued. Hysterectomy rates were lower in the LNG-IUS than oral progestogen group during follow-up (19.6%, 30/153 versus 31.8%, 21/66, respectively, OR = 0.52, 95% CI: 0.27-1, P = 0.05). Endometrial cancer was diagnosed in 2 (11.8%) women who had hysterectomy (n = 17) because of relapse. LIMITATIONS, REASONS FOR CAUTION We are unable to accurately estimate the cancer risk in women who relapse during follow-up as only 17 out of 41 who relapsed underwent hysterectomy. WIDER IMPLICATIONS OF THE FINDINGS Relapse of endometrial hyperplasia after initial regression occurs often and long-term follow-up is advised. STUDY FUNDING/COMPETING INTEREST(S) Ioannis D. Gallos and this study were funded through a grant from Wellbeing of Women (ELS022). No competing interests.