Uterine sarcomas, malignancies of the myometrium or endometrial connective tissue, occur rarely and are associated with poor prognosis.1–3 They may be indistinguishable from uterine leiomyomas, which is the most common indication for hysterectomy in the United States.4 The use of laparoscopic hysterectomy and power morcellation has substantially reduced surgical morbidity, but concern over peritoneal dissemination of occult uterine sarcoma in which there was no preoperative or operative suspicion for malignancy has prompted calls for reevaluation of the incidence of occult uterine sarcoma and risk associated with morcellation.5–7 In November 2014 the U.S. Food and Drug Administration issued a warning against use of power morcellators in the majority of women undergoing hysterectomy for uterine leiomyomas.8
The current data on the incidence of uterine sarcoma among women undergoing hysterectomy for leiomyomas are based primarily on studies from single-center or tertiary-based practices and include a broad range of publication dates (1990–2012), small numbers of patients (101–1,429), and sarcomas (0–7) per study.9,10 In addition, interpretation of existing studies of the risk associated with power morcellation of occult sarcomas is challenging because of small numbers of sarcomas and morcellation procedures per study and methodology flaws.11
To fill gaps in the literature, we conducted this study to: 1) estimate the age- and race-specific incidence of occult uterine sarcoma among women undergoing hysterectomy for suspected benign leiomyomas; and 2) assess the association of morcellation and length of survival among women with leiomyosarcoma.
MATERIALS AND METHODS
This population-based cohort study was performed using data abstracted from Kaiser Permanente's electronic health record and regional claims systems. Kaiser Permanente is an integrated health care delivery system in which health plan members obtain all care from Permanente physicians and allied health professionals with few exceptions. The three regions included in this study, northern California, southern California, and Colorado, served approximately 3.1 million adult women annually during the study period. Oversight and approval to conduct this study were provided by the institutional review board of the Kaiser Foundation Research Institute.
We identified all women 18 years and older who underwent hysterectomy from January 1, 2006, to December 31, 2013, at a Kaiser Permanente-owned, contracted, or referral hospital with a principal discharge diagnosis of leiomyomas based on International Classification of Diseases, 9th Revision and Current Procedural Terminology coding of surgery indication. Women with a new diagnosis of uterine sarcoma during the same time period were identified through the Kaiser Permanente regional cancer registries using the International Classification of Diseases for Oncology codes (Appendix 1, available online at http://links.lww.com/AOG/A728).
All sarcomas that were incident at hysterectomy or at the time of myomectomy that subsequently led to a hysterectomy during the study period were included in our analyses. Sarcomas that were incident at a biopsy only and hysterectomy was not performed were excluded.
The pathology reports were reviewed and malignancies were categorized into histopathology subgroups based on the 2014 World Health Organization classification of uterine mesenchymal malignancies as follows: 1) leiomyosarcomas (including epithelioid and myxoid types); 2) low-grade endometrial stromal sarcomas; and 3) high-grade endometrial stromal sarcomas, undifferentiated uterine sarcomas, including very rare miscellaneous mesenchymal malignancies (rhabdomyosarcomas and perivascular epithelioid cell tumors).12–14 We excluded mixed epithelial–mesenchymal tumors (carcinosarcoma and adenosarcoma).15,16
Clinical and surgical information on sarcomas was obtained by chart review. Hysterectomies were categorized as: 1) no morcellation; 2) power morcellation defined as use of an electronic device at laparoscopy; and 3) nonpower morcellation defined as intraoperative tissue disruption with a scalpel or cautery at laparotomy or vaginal hysterectomy (including facilitating myomectomy) or myomectomy done at a separate procedure that led to the hysterectomy. Uterine sarcomas were considered occult if the surgeon did not report a preoperative suspicion for malignancy in the preoperative clinic notes, admitting history, and physical or operative note. They were considered nonoccult if known or suspected malignancy was documented before the hysterectomy as a result of suspicious or positive endometrial or other biopsy, suspicious imaging (reported by the interpreting radiologist), or other reason, including cancer found at a separate myomectomy procedure. American Joint Committee on Cancer criteria for staging were used and were based on the findings at the original surgery unless postoperative imaging in occult sarcomas revealed distant metastases making them stage IV. Follow-up information including adjuvant treatments, time to and site of recurrence, death, and latest medical contact was also obtained by chart review. Information on vital status was obtained from the tumor registries, which utilize standard reporting sources including state death certificate and Social Security databases.
Analyses were performed using SAS 9.3. P values were considered statistically significant at <.05. We examined characteristics of uterine sarcomas by histopathology and morcellation subgroups using χ2 tests or Fisher exact tests for categorical variables and Wilcoxon-Mann-Whitney test for continuous variables. We calculated age- and race–ethnicity-specific incidence rates for occult uterine sarcoma. Proportions and 95% confidence intervals (calculated using Wald or Wilson Score method) were reported per 1,000 hysterectomies by histopathology, age, and race–ethnicity subgroups. Poisson regression with robust variance was used to estimate the risk ratios of occult uterine sarcoma across age groups (adjusted for race–ethnicity) and across race–ethnicity groups (adjusted for age).17 We used the Kaplan-Meier method to estimate the unadjusted cumulative 3-year disease-free and overall survival probabilities by morcellation group for sarcomas with disease confined to the uterus (American Joint Committee on Cancer stage I). For the disease-free survival analysis, women were censored at last clinical encounter or disenrollment from the health plan, the time of death resulting from all causes, or the administrative end of study (May 31, 2015). For overall survival, women were censored at the administrative end of study. We evaluated differences in disease-free and overall survival among morcellation groups using the log-rank test.
Propensity score methods were used to evaluate the effect of morcellation on recurrence and death while adjusting for baseline confounding factors using Poisson regression fitted by inverse probability weighting estimation.18 We estimated the propensity scores using logistic regression models to predict morcellation group with the covariates age (younger than 50, 50–59, or 60 years or older), race–ethnicity (white, Hispanic, African American, or Asian or other), American Joint Committee on Cancer substage (IA, tumor size 5 cm or less or IB greater than 5 cm), occult sarcoma (yes or no), body mass index ([calculated as weight (kg)/(height [m])2] 24.9 or less, 25–34.9, or 35 or greater), uterine weight (500 or less, 500–1,000, or greater than 1,000 g), and Charlson Index of comorbid conditions score greater than 2 (yes or no).19 The set of baseline covariates was reduced using the Akaike information criterion.20 For disease-free and overall survival, at 1, 2, and 3 years, to contrast risks of different types of morcellation, we generated two separate Poisson regression models fitted by inverse probability weighting estimation comparing: 1) nonpower morcellation with no morcellation; and 2) power and nonpower morcellation groups combined with no morcellation. Adjusted risk ratios for power morcellation compared with no morcellation could not be estimated as a result of small numbers (n=8). The adjusted risk ratios for recurrence and death at 1, 2, and 3 years were compared with their crude analog obtained based on unweighted Poisson regression.
From 2006 to 2013, 93,153 women underwent hysterectomy; 34,603 (37.2%) had a principal discharge diagnosis of leiomyomas. Among hysterectomies for leiomyomas, 45.8% were abdominal, 44.7% laparoscopic, and 9.5% vaginal and 20.1% vaginal. We identified 321 new uterine sarcomas, of which 293 were in patients who underwent hysterectomy (Fig. 1). Twenty-eight uterine sarcomas were excluded because they were identified at the time of biopsy but did not undergo a hysterectomy as a result of advanced disease or other comorbidities. Of the 293 hysterectomies, seven were preceded by separate myomectomy procedures (four abdominal, two hysteroscopic, and one vaginal myomectomy of a prolapsing leiomyoma) in which uterine sarcomas were identified postoperatively. There were no uterine sarcomas identified at myomectomy that did not go on to hysterectomy.
Less than half (42.7%) of uterine sarcomas were occult before hysterectomy (Fig. 1). Among the 187 stage I sarcomas, 94 (50.3%) were occult. Power and nonpower morcellation occurred in only a small number of stage I leiomyosarcomas (8 [7.2%] and 27 [24.3%]) and low-grade endometrial stromal sarcomas (8 [17.0%] and 6 [12.8%]); no high-grade endometrial stromal sarcomas were morcellated.
Characteristics of uterine sarcomas that underwent hysterectomy are presented in Table 1. The majority (65.9%) of women were age 50 years and older and half (50.9%) were postmenopausal. Patients with low-grade endometrial stromal sarcomas were significantly more likely to be younger (P=.005), premenopausal (P=.006), and American Joint Committee on Cancer stage I (P=.003). Leiomyosarcomas (47.1%) and low-grade endometrial stromal sarcomas (53.2%) were significantly more likely to be occult than high-grade endometrial stromal sarcomas (18.6%) (P<.001).
The incidence of occult uterine sarcoma and leiomyosarcoma among women with hysterectomy for leiomyomas was 1 of 278 or 3.60 (95% confidence interval [CI] 2.97–4.23) and 1 of 429 or 2.33 (95% CI 1.83–2.84) per 1,000 hysterectomies, respectively (Table 2). The risk of occult uterine sarcoma and leiomyosarcoma adjusted for race–ethnicity in women aged 50 years and older was significantly greater compared with younger women. The risk of occult low-grade endometrial stromal sarcomas and high-grade endometrial stromal sarcomas in women aged 60 years and older adjusted for race–ethnicity was significantly greater compared with younger women.
Characteristics and follow-up information on stage I uterine sarcomas by histopathology subgroup and morcellation status are presented in Table 3. Aside from the significant differences in surgical route and other procedures performed that would be expected in hysterectomies with morcellation (power and nonpower groups combined) compared with no morcellation, there were no significant differences in surgical characteristics. Median follow-up months for leiomyosarcomas, low-grade endometrial stromal sarcomas, and high-grade endometrial stromal sarcomas were 35.1 (interquartile range 21.1–62.9), 51.3 (interquartile range 34.6–77.7), and 29.9 (interquartile range 9.8–59.1), respectively. There were no significant differences by morcellation group for leiomyosarcoma (P=.15) and low-grade endometrial stromal sarcomas (P=.76) and there was minimal loss to follow-up. Although the proportion of recurrences was similar for leiomyosarcomas with and without morcellation, leiomyosarcomas without morcellation were significantly less likely to have abdominal recurrences compared with leiomyosarcomas with morcellation (41.2% compared with 72.2%, P=.03) (Table 3).
Kaplan–Meier survival plots were generated for stage I leiomyosarcoma by morcellation groups but not for low-grade endometrial stromal sarcomas and high-grade endometrial stromal sarcomas because of the small number of events and absence of morcellation procedures, respectively. The unadjusted probability of 3-year disease-free survival for stage I leiomyosarcomas with no morcellation, power morcellation, and nonpower morcellation was 0.54 (95% CI 0.42–0.64), 0.19 (95% CI 0.01–0.54), and 0.51 (95% CI 0.30–0.69), respectively. Although the plots suggest decreased probability of survival for power morcellation, there was no significant difference among all groups (P=.15) or between power morcellation and no morcellation (P=.29) or between nonpower morcellation and no morcellation (P=.88) (Fig. 2A).
The unadjusted probability of 3-year overall survival for stage I leiomyosarcoma for no morcellation, power morcellation, and nonpower morcellation groups was 0.64 (95% CI 0.52–0.74), 0.75 (95% CI 0.31–0.93), and 0.68 (95% CI 0.46–0.83), respectively; there was no significant difference among all groups (P=.97) or between power morcellation and no morcellation (P=1.0) or between nonpower morcellation and no morcellation (P=1.0) (Fig. 3A).
The crude and adjusted risk ratios of recurrence at 1, 2, and 3 years comparing power, nonpower, and power and nonpower morcellation groups combined to no morcellation are presented in Figure 2B–D. The crude risk ratios comparing power morcellation with no morcellation at 1 and 3 years were 2.38 (95% CI 1.05–5.38, P=.04) and 1.88 (95% CI 1.50–2.37, P<.001), respectively. All other risk ratios for recurrence had 95% CIs that included 1 and were not significant.
The crude and adjusted risk ratios of death at 1, 2, and 3 years comparing power, nonpower, and power and nonpower groups combined to no morcellation are presented in Figure 3B–D. The crude and adjusted risk ratios comparing power and nonpower morcellation groups combined to no morcellation at 1 year were 3.45 (95% CI 1.04–11.44, P=.04) and 5.12 (95% CI 1.33–19.76, P=.02), respectively. All other risk ratios for death had 95% CIs that included 1 and were not significant. The crude risk ratio for death comparing power morcellation with no morcellation at 1 year approached significance at 4.75 (95% CI 1.03–22.0, P=.05).
This study within a diverse, community-based U.S. population confirms uterine sarcomas are rare and occur more commonly in older women. It also shows less than half were occult. Our estimate of the incidence of occult leiomyosarcoma of 1 in 429 among women having hysterectomies for leiomyomas probably overestimates the incidence because we did not capture in the denominator women with hysterectomies with a principal discharge diagnosis of abnormal bleeding when leiomyomas was listed as a secondary diagnosis. Our estimates of the incidence of occult leiomyosarcoma are not directly comparable with the previous estimates by the U.S. Food and Drug Administration (1/498 women undergoing hysterectomy or myomectomy for presumed benign leiomyomas) and Pritts et al (1/1,700 women undergoing surgeries in retrospective studies and 1/8,300 in prospective studies) because these studies were meta-analyses in which inclusion criteria for leiomyosarcomas and the denominators varied widely across studies included.9,21
Our survival analyses do not demonstrate significant differences in the unadjusted 3-year disease-free and overall probability of survival among stage I leiomyosarcomas with power or nonpower morcellation and those with no morcellation; however, clinically significant effects cannot be ruled out as a result of small numbers. The crude and adjusted risk ratios for death at 1 year, however, were significantly increased for the power and nonpower morcellation groups combined compared with no morcellation indicating that morcellation is associated with decreased early survival. We may not have demonstrated a corresponding significantly increased risk of recurrence at 1 year as a result of better ascertainment of mortality status through cancer registry databases or alternatively there may have been differential survival associated with recurrences for the power and nonpower morcellation groups combined compared with no morcellation.
Although the significantly increased crude risk ratios of recurrence at 1 and 3 years for power morcellation compared with no morcellation are highly suggestive of an adverse effect of power morcellation, we cannot rule out the possibility of confounding. Although the total number of uterine sarcomas in this case series is the largest that we are aware of from a community-based U.S. population, the number of leiomyosarcomas with power morcellation was too small to calculate adjusted risk ratios or make inferences about the risk of power morcellation. However, it is plausible that power morcellation would be equally as harmful as nonpower morcellation.
The lack of demonstration of a significantly increased adjusted risk ratios for recurrence or death at 3 years for the power and nonpower morcellation groups combined may be the result of poor outcomes as time elapses in leiomyosarcomas without morcellation or, alternatively, lack of power to detect differences at 3 years as a result of the smaller sample size with 3 years of follow-up. The upper bounds of the CIs for the adjusted risk ratios of recurrence and death at 3 years for the power and nonpower morcellation groups combined compared with no morcellation (1.77 and 2.32, respectively) provide a guide of the range of potential adverse risk of morcellation.
Survival data from our series provide more representative information on the risk associated with morcellation than recent reports by Perri et al, Park et al, and George et al.3,7,22–24 These studies have important methodologic flaws that lead to bias, which most likely overestimate the adverse effect of morcellation including: 1) selection bias associated with referral populations with unknown background incidence of leiomyosarcoma; 2) inclusion of hysterectomies with disease beyond the uterus; 3) use of comparison groups that lacked laparoscopic or vaginal hysterectomies that did not undergo morcellation; and 4) lack of specification of types of morcellation. Our study has several strengths. The integrated health care system and comprehensive electronic health records enabled capture of hysterectomies performed, comprehensive identification of patients with uterine sarcoma, specific histopathology, morcellation, and clinical outcomes with minimal loss to follow-up.
Our study also has limitations. We also may not have had sufficient power to detect differences in the incidence of occult uterine sarcomas by race. There was no central pathology review. We could not control for mitotic index, because although elevated mitotic activity sufficient for diagnosis of leiomyosarcoma was specified in the pathology reports, the amount was not always specified. Finally, we were not able to estimate uterine sarcoma incidence at myomectomy given only seven sarcomas that were incident at myomectomy over 8 years.
We hope our data on the estimated incidence of occult uterine sarcoma in combination with the demonstrated association of morcellation with decreased early survival for women with leiomyosarcomas may be useful to patients and health care providers who wish to weigh the risks against the known benefits of minimally invasive surgery.5,6
1. Nordal RR, Thoresen SO. Uterine sarcomas in Norway 1956–1992: incidence, survival and mortality. Eur J Cancer 1997;33:907–11.
2. Brooks SE, Zhan M, Cote T, Baquet CR. Surveillance, epidemiology, and end results analysis of 2677 cases of uterine sarcoma 1989–1999. Gynecol Oncol 2004;93:204–8.
3. Tropé CG, Abeler VM, Kristensen GB. Diagnosis and treatment of sarcoma of the uterus. A review. Acta Oncol 2012;51:694–705.
4. Wright JD, Herzog TJ, Tsui J, Ananth CV, Lewin SN, Lu YS, et al.. Nationwide trends in the performance of inpatient hysterectomy in the United States. Obstet Gynecol 2013;122:233–41.
5. Nieboer TE, Johnson N, Lethaby A, Tavender E, Curr E, Garry R, et al.. Surgical approach to hysterectomy for benign gynaecological disease. The Cochrane Database of Systematic Reviews 2009, Issue 3. Art No.: CD003677. DOI: 10.1002/14651858.CD003677.pub4.
6. Walsh CA, Walsh SR, Tang TY, Slack M. Total abdominal hysterectomy versus total laparoscopic hysterectomy for benign disease: a meta-analysis. Eur J Obstet Gynecol Reprod Biol 2009;144:3–7.
7. George S, Barysauskas C, Serrano C, Oduyebo T, Rauh-Hain JA, Del Carmen MG, et al.. Retrospective cohort study evaluating the impact of intraperitoneal morcellation on outcomes of localized uterine leiomyosarcoma. Cancer 2014;120:3154–8.
10. Bojahr B, De Wilde RL, Tchartchian G. Malignancy rate of 10,731 uteri morcellated during laparoscopic supracervical hysterectomy (LASH). Arch Gynecol Obstet 2015;292:665–72.
11. Bogani G, Cliby WA, Aletti GD. Impact of morcellation on survival outcomes of patients with unexpected uterine leiomyosarcoma: a systematic review and meta-analysis. Gynecol Oncol 2015;137:167–72.
12. Robert J, Kurman EA, editors. WHO classification of tumours of female reproductive organs. 4th ed. Lyon (France): International Agency for Research on Cancer; 2014.
13. Tavassoli FA, Devilee P, editors. Pathology and genetics of tumours of the breast and female genital organs. Lyon (France): International Agency for Research on Cancer; 2003.
14. Lee CH, Nucci MR. Endometrial stromal sarcoma—the new genetic paradigm. Histopathology 2014;67:1–19.
15. Sreenan JJ, Hart WR. Carcinosarcomas of the female genital tract. A pathologic study of 29 metastatic tumors: further evidence for the dominant role of the epithelial component and the conversion theory of histogenesis. Am J Surg Pathol 1995;19:666–74.
16. Friedlander ML, Covens A, Glasspool RM, Hilpert F, Kristensen G, Kwon S, et al.. Gynecologic Cancer InterGroup (GCIG) consensus review for mullerian adenosarcoma of the female genital tract. Int J Gynecol Cancer 2014;24(suppl 3):S78–82.
17. Spiegelman D, Hertzmark E. Easy SAS calculations for risk or prevalence ratios and differences. Am J Epidemiol 2005;162:199–200.
18. Hernán MA, Robins JM. Estimating causal effects from epidemiological data. J Epidemiol Community Health 2006;60:578–86.
19. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373–83.
20. Cole SR, Hernán MA. Constructing inverse probability weights for marginal structural models. Am J Epidemiol 2008;168:656–64.
21. Pritts EA, Vanness DJ, Berek JS, Parker W, Feinberg R, Feinberg J, et al.. The prevalence of occult leiomyosarcoma at surgery for presumed uterine fibroids: a meta-analysis. Gynecol Surg 2015;12:165–77.
22. D'Angelo E, Prat J. Uterine sarcomas: a review. Gynecol Oncol 2010;116:131–9.
23. Perri T, Korach J, Sadetzki S, Oberman B, Fridman E, Ben-Baruch G. Uterine leiomyosarcoma: does the primary surgical procedure matter? Int J Gynecol Cancer 2009;19:257–60.
24. Park JY, Kim DY, Kim JH, Kim YM, Kim YT, Nam JH. The impact of tumor morcellation during surgery on the outcomes of patients with apparently early low-grade endometrial stromal sarcoma of the uterus. Ann Surg Oncol 2011;18:3453–61.
Supplemental Digital Content
© 2016 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.