Letter to the Editor: Re: Raspagliesi F, Maltese G, Bogani G, et al. Morcellation worsens survival outcomes in patients with undiagnosed uterine leiomyosarcomas: A retrospective MITO group study. Gynecol Oncol. 2017; 144: 90-5

Abstract

Dear Dr. Karlan, We read with interest the recent article by Raspagliesi et al., entitled “Morcellation worsens survival outcomes in patients with undiagnosed uterine leiomyosarcomas,” which reviews outcomes after morcellation at the MITO centers (Raspagliesi et al., 2017). After reading this article we have concerns about the data that are presented in this manuscript and would ask the authors to clarify their findings. The authors describe 4000 consecutive women who underwent hysterectomy or myomectomy for uterine fibroids and found 125 sarcomas including 91 cases of uterine leiomyosarcoma (uLMS). Thus, the calculated incidence of sarcoma in this population is 3.1% and the rate of uLMS is 2.3%. This represents an incidence of uterine sarcoma of 1 in 32 surgeries and a risk of uLMS of 1 in 43. This is more than tenfold the rate of uterine sarcoma noted in the FDA report on this matter where the incidence of uLMS in surgery for all cases of leiomyoma was 0.20% (FDA, 2014). In another report from Kaiser Permanente on a cohort of 34,603 hysterectomies performed for presumed leiomyomata, the incidence of uLMS was 0.32% and the rate of occult uLMS was 0.18% (Raine-Bennett et al., 2016). Raspagliesi gives no details of the evaluation or selection of these patients with regard to imaging or biopsy. This is important as preoperative biopsies are abnormal in 86% of cases of uterine sarcoma (Bansal et al., 2008). Clinical evaluation can also help prevent unanticipated morcellation. Despite the rate of uLMS of 0.32% in the Kaiser cohort study, the actual rate of power morcellation of uLMS was only 0.02% or 1 case in 4325, demonstrating that morcellation can be prevented in over 90% of uLMS based on proper selection. Therefore, it is difficult to understand why the rate of power morcellation of sarcoma is almost 40 times higher in the Raspagliesi study (31 per 4000). This is especially concerning since 3/4 of cases involved a uterine sparing procedure and the average age of these women was 43.6 years. What were the criteria for selecting cases for morcellation? How did the selection criteria alter the rate of unintended morcellation? What percentage of women who had a myomectomy could reasonably expect to maintain fertility? If fertility were not an issue, could morcellation have been avoided if a hysterectomy was performed instead? Given the data presented, we suspect the practice patterns in this report may not be representative of those in other parts for the world and may not be generalizable to other populations. The endpoints that the authors indicate throughout this report are 2-year DFS and 2-year OS despite the fact that this study spans 10 years, and longer follow-up is available for the majority of this population. The methods section indicates that the survival outcomes were limited to 2 years to “reduce temporal bias”. Was this pre-specified in the statistical plan? The presentation of the results indicates they did evaluate overall DFS and OS (within a 2 year follow up period) rather than evaluating survival outcomes as a 2-year landmark analysis. Referring to these endpoints as “2 year DFS” and “2 year OS” suggests that 2-year landmark analyses were used, but the presentation of the results suggest otherwise. Clarification needs to be provided on how the survival endpoints were derived. Were all subjects who were alive (OS) and disease free (DFS) censored at 2 years regardless of what year surgery occurred? Other endpoints or a full Kaplan-Meyer analysis should be reported to determine if a temporal bias indeed does exist as the authors hypothesize. A puzzling aspect of this report is that morcellation did not significantly alter the local recurrence rate, yet overall survival was significantly diminished. The proposed mechanism by which power morcellation causes harm is dissemination of tumor in the abdominal cavity, which would increase local recurrence. How do the authors explain this inconsistency with previous studies that show an increase in local recurrence and decreased PFS but little or no effect on overall survival? (Raine-Bennett et al., 2016, Park et al., 2011) The most intuitive answer to this question would be that the higher risk of death in the morcellation group would be due to tumor biology, or manipulation of the tumor during myomectomy and not morcellation per se. The authors do not provide the factors that were evaluated in the multivariate analysis. Did this include known pathologic risk factors for recurrence such as mitotic count? If any known risk factor for recurrence were excluded from the analysis, the authors should disclose this and why those factors were omitted, as well as the impact of the omissions on their conclusions. While this article is provocative, many questions remain, and we would ask the authors to speculate as to why their results are so different from previous reports. We look forward to further clarification of these data. Conflict of interest R. Wendel Naumann M.D. - Honorarium from Ethicon. Jubilee Brown, M.D. - Honorarium from Ethicon, Honorarium from Olympus. James Symanowski Ph.D. - No conflicts