Hysterectomy and myomectomy are common gynecologic procedures, with more than 600,000 and 43,000 surgeries performed yearly in the United States, respectively.1–3 Nearly 90% of hysterectomies are performed for benign indications, whereas myomectomies are performed only for benign indications.3,4
Occasionally, however, there may be occult (ie, preoperatively undetected) malignancy in these procedures. Power morcellation used for specimen removal during minimally invasive hysterectomies and myomectomies may lead to seeding and dissemination of unrecognized cancer cells into the abdominopelvic cavity. This may upstage cancer, worsen prognosis, increase recurrence, and decrease survival.5–8 Concerns about this risk have prompted the U.S. Food and Drug Administration to issue a position statement discouraging use of intracorporeal electromechanical morcellation for specimen removal,9 resulting in significant shifts in clinical practice.10
Nevertheless, current evidence on the risk of occult uterine malignancy remains highly variable (with estimated prevalence ranging from 0% to 3.17% across studies)11–13 and is limited in scope (eg, surgical approach and patient samples included).13–15 Moreover, the focus of attention on uterine sarcoma and leiomyosarcoma in previous research6,7,11,16,17 also overlooked more common types of gynecologic cancers such as endometrial, cervical, and ovarian carcinomas, which, if preoperatively undiagnosed, may disseminate to the abdominopelvic cavity as well when morcellated.
To address these gaps in prior research, we aimed to estimate the prevalence of preoperatively undiagnosed corpus uteri, cervix uteri, and ovarian malignancy in women undergoing hysterectomy or myomectomy for benign indications. In addition, we assessed characteristics of patients with such incidental diagnosis of cancers to inform preoperative risk stratification.
MATERIALS AND METHODS
We conducted a secondary analysis of data from the 2014–2015 American College of Surgeons (ACS) National Surgical Quality Improvement Program.18 Hospitals across the country (both academic and community-based hospitals) can voluntarily participate in the ACS National Surgical Quality Improvement Program, and the ACS National Surgical Quality Improvement Program data contain deidentified patient-level information on major surgical procedures performed in both inpatient and outpatient settings for patients 18 years of age or older submitted by participating hospitals.18–21 A systematic sampling process was used to select cases submitted to the National Surgical Quality Improvement Program database at each hospital (eg, submission of a certain number of cases per 8-day cycle for larger institutions and submission of all eligible cases for lower volume hospitals).19,20 Because the data were deidentified and pre-existing, this study was determined as exempt from institutional review board review by the Yale University Human Investigation Committee.
The ACS National Surgical Quality Improvement Program main participant use data file contained measures on patients' preoperative characteristics, intraoperative findings, and postoperative outcomes up to 30 days after surgery. A subset of the hospitals also submitted data to the ACS National Surgical Quality Improvement Program Procedure Targeted—Hysterectomy program, which provided additional measures specific to hysterectomy and myomectomy procedures (including pathology report on diagnosis of gynecologic malignancies).18,19 Ninety-one hospitals in 2014 and 109 hospitals in 2015 participated in the Procedure Targeted—Hysterectomy program.18 A trained and certified surgical clinical reviewer at each participating hospital collected data through medical chart abstraction and automatic extraction of data elements from hospital information systems.
We limited our analysis to patients included in the Procedure Targeted—Hysterectomy program (with linkable information from the main data file) and identified hysterectomies and myomectomies performed from January 1, 2014, to December 31, 2015, for adult women based on Current Procedural Terminology codes. Given that information on surgical indication was not directly available in the data and our aim was to focus on women with benign and nonobstetric indications, we excluded hysterectomies and myomectomies in which 1) the procedure was related to obstetric conditions, 2) patients underwent radical hysterectomy or other cancer diagnostic or treatment procedures, 3) there was grossly visible malignancy at beginning of surgery, or 4) the surgeon's specialty was gynecologic oncology. The latter three criteria were intended to remove patients who might be known or suspected to have cancer at the beginning of surgery, although we recognize that not all cases performed by gynecologic oncologists might be for cancerous indications. Specific procedure and diagnosis codes used in defining our sample are reported in Appendix 1, available online at http://links.lww.com/AOG/B74.
Our primary outcome measure was occult malignancy of the corpus uteri, cervix uteri, and ovary. Our secondary outcome measure was the stage of identified occult malignancies. Cancer type and stage were recorded based on the International Federation of Gynecology and Obstetrics staging guidelines per pathology report. No information about subtypes of cancer was available. We defined a patient as having cancer of the corpus uteri, cervix uteri, or ovary, respectively, if the corresponding stage variable indicated a valid cancer stage, that is, stage I-IVB or not otherwise specified (NOS) for cancer of the corpus uteri, stage I-IVB or NOS for cancer of the cervix uteri, and stage I-IV or NOS for cancer of the ovary. When pathologic data reported a specific stage for one or two of the cancers but noted stage NOS for the other cancer(s), we assumed the other cancer(s) were not present. Because diagnosis codes could not definitively distinguish pre-existing, preoperatively suspected, or postoperatively confirmed malignancy, it was not primarily used in defining cancer cases. Only in rare situations when there was inconsistency or uncertainty in pathologic data elements (eg, when all three cancers were reported as stage NOS), we deferred to the diagnosis code to clarify which gynecologic cancer was present (Appendix 2, available online at http://links.lww.com/AOG/B74).
Patient demographic characteristics included age (younger than 40, 40–54, or 55 years or older), race (white, black, other), and ethnicity (Hispanic vs non-Hispanic). Clinical characteristics included comorbid conditions (hypertension requiring medication, diabetes mellitus requiring oral agents or insulin), body mass index (BMI [calculated as weight (kg)/[height (m)]2]; less than 25, 25–29.9, or 30 or greater), smoking within the past year (yes or no), parity (zero, one, or two or greater), endometriosis (yes or no), pelvic inflammatory disease (yes or no), prior abdominal or pelvic operation (yes or no), and care setting (inpatient vs outpatient). For hysterectomies, we also measured uterine weight (250 g or less vs greater than 250 g) and surgical route, including total abdominal hysterectomy, total laparoscopic or laparoscopic-assisted vaginal hysterectomy, vaginal hysterectomy, abdominal supracervical hysterectomy, and laparoscopic supracervical hysterectomy. Myomectomies were categorized as abdominal, laparoscopic, or vaginal. Surgical route was determined based on Current Procedural Terminology codes (Appendix 1, http://links.lww.com/AOG/B74).
Hysterectomies and myomectomies were analyzed separately. We characterized the prevalence of occult corpus uteri, cervical, and ovarian cancer using frequencies and percentages for the overall sample of hysterectomies and myomectomies, respectively, as well as stratified by hysterectomy route. To assess uncertainty in these estimates, we also estimated their 95% CIs based on binomial distribution (ie, exact CI) when the number of relevant occult cancer cases was fewer than five or based on normal approximation otherwise. Distribution of these incidental cancer cases by stage was delineated using frequencies and percentages.
To help identify risk factors for having undiagnosed gynecologic cancer, we estimated logistic regression models to examine the association between available patient characteristics and presence of undiagnosed corpus uteri, cervical, and ovarian cancer, respectively, in women undergoing hysterectomy. We first conducted unadjusted, bivariate analysis of patient characteristics and selected those that were statistically significant at the P<.10 level and then included only these patient characteristics in the adjusted logistical regression to limit the number of variables in the final model. The number of events per predictor variable in each of our multivariable regression models met the conventional criterion (ie, 10 or greater events per predictor variable) for ensuring accuracy in estimation of coefficients.22 Because of the small number of occult cancers, we also used the Firth penalized likelihood method in estimation to reduce potential small sample bias. We did not perform regression analysis for myomectomies as a result of the small number of patients with incidental gynecologic cancer in myomectomies. All analyses were performed using SAS 9.4.
A total of 24,076 hysterectomies and 2,368 myomectomies met our sample eligibility criteria (Appendix 3, available online at http://links.lww.com/AOG/B74). Of the patients undergoing hysterectomy, median age was 46 years (interquartile range 41–52) and median BMI was 29.3 (interquartile range 25.2–34.4). Close to 12% of the women had endometriosis, 1.4% had pelvic inflammatory disease, and 64.9% had a prior abdominal or pelvic surgery (Table 1). Total laparoscopic or laparoscopic-assisted vaginal hysterectomy (48.0%) and total abdominal hysterectomy (21.2%) were the most common surgical routes; 7.3% were laparoscopic supracervical hysterectomies. Of the 2,368 myomectomies, most patients (70.7%) were younger than 40 years of age with a median BMI of 28.2 (interquartile range 24.4–33.1). Most myomectomies were performed abdominally (60.0%) and in the inpatient setting (62.5%).
Of women who underwent hysterectomy, 1.44% (95% CI 1.29–1.59%) were found to have malignancy of the corpus uteri, 0.60% (95% CI 0.50–0.70%) had malignancy of the cervix uteri, and 0.19% (95% CI 0.14–0.25%) had malignancy of the ovary (Table 2). However, prevalence varied substantially among women undergoing different surgical routes. For example, among women who underwent a total laparoscopic or laparoscopic-assisted vaginal hysterectomy, 1.89% (95% CI 1.65–2.14%) were found to have occult malignancy of the corpus uteri. In contrast, the prevalence of occult corpus uteri cancer was considerably lower in women undergoing laparoscopic supracervical hysterectomy: 0.23% (95% CI 0.06–0.58%).
The occult corpus uteri and ovarian cancers identified tended to be in early stages (Table 3). Eighty percent of these corpus uteri cancers and 60.9% of the ovarian malignancies were stage I-IC neoplasms. Among patients with occult cancer of the cervix uteri, 44.1% were diagnosed at stage I-IB2; 40.7% had stage NOS.
The risk of occult gynecologic cancer was considerably higher in women aged 55 years or older in hysterectomies in general and across most surgical routes examined (Table 4). For example, in total abdominal hysterectomy, 9.72% (95% CI 7.30–12.14%) of the women aged 55 years or older had occult corpus uteri cancer as compared with 1.06% (95% CI 0.72–1.39%) in those aged 40–54 years. Multivariable logistic regression analysis adjusting for other patient characteristics also showed that, compared with women aged 40–54 years, those aged 55 years or older were consistently more likely to have occult malignancy of the corpus uteri (odds ratio [OR] 6.46, 95% CI 4.96–8.41), cervix uteri (OR 1.81, 95% CI 1.21–2.69), and ovary (OR 3.54, 95% CI 1.91–6.54) (Table 5). In addition, several other patient characteristics such as race, smoking status, BMI, parity, hypertension, diabetes, prior abdominal or pelvic surgery, endometriosis, and uterine weight were identified as significantly associated with a woman's risk of having some of the occult gynecologic malignancies (Table 5).
Of women undergoing myomectomies, five (0.21%, 95% CI 0.03–0.40%) were found to have cancer of the corpus uteri. One was stage I, and the other four had stage NOS. None of these patients underwent laparoscopic myomectomy (three had abdominal myomectomy and two had vaginal myomectomy). No patients were found to have occult cervical or ovarian malignancy. We were unable to assess risk factors for occult malignancy in the myomectomy sample as a result of the small number of cancer cases.
Our analysis of occult gynecologic malignancy revealed several important findings. In women undergoing hysterectomies for presumed benign indications, 1.44%, 0.60%, and 0.19% had malignancy of the corpus uteri, cervix uteri, and ovary, respectively. The prevalence varied considerably by patient age and surgical route. Additionally, occult corpus uteri cancer was found in 0.21% of women undergoing myomectomies.
Our study improves on prior work in multiple ways. First, we included a large and diverse sample of hospitals with pathology data abstracted from clinical charts. Conversely, previous research often relied on small samples from selected institutions5,12,23–25 and geographic regions4,26,27 or on administrative data.15 Second, we assessed three major gynecologic cancers rather than corpus uteri cancer alone as did most prior studies. This expands our understanding of potential adverse consequences of intracorporeal electromechanical morcellation.28 Third, we included women with a variety of surgical routes, which helped inform surgical planning for a broader patient population.
Several reasons may explain our observed variation in risk of occult malignancy across hysterectomy routes. First, it may reflect inherent differences in the underlying condition of patients undergoing different surgical routes. For instance, women with prolapse are often treated with supracervical or vaginal hysterectomy and have been shown to have a lower risk for unexpected gynecologic malignancy.4 Second, surgeons may systematically select patients to undergo different types of procedures based on their risk profile. Preoperative concerns for elevated cancer risk (that may not be observable in our data) could prompt surgeons to perform complete removal of the uterus, resulting in the higher prevalence of occult malignancy observed in total (vs supracervical) hysterectomies.
We noted a particularly higher risk of occult malignancy in women aged 55 years and older. This is consistent with prior literature showing a positive relationship between advancing age and occult cancer risk,2,4,28 highlighting the clinical importance of thorough preoperative evaluations for older women. Unfortunately, we had limited information on patients' preoperative conditions precluding us from delineating the potential confounding effect of surgical indications (eg, postmenopausal bleeding) in the relationship between age and occult cancer risk. Additional research with more specifics about patient preoperative characteristics will provide additional insights and facilitate development of risk prediction or risk stratification tools to guide clinical care.
Our results also suggest potential utility of power morcellation in selected patient populations. For instance, no occult corpus uteri cancer was identified in women undergoing laparoscopic myomectomy. With careful screening, benefits of minimally invasive procedures enabled by power morcellation (eg, smaller incisions, lower complication rates, and more rapid recovery) may outweigh the risk of occult malignancy in these patients. Rather than completely dismissing its use, there may be a role for selective use of power morcellation, especially with advancement in techniques such as contained power morcellation inside an isolation bag,29 which offers additional protection.
Findings from this study should be interpreted with some limitations in mind. First, our data lacked information on surgical indication. Thus, some patients at higher risk for cancer (eg, those with endometrial hyperplasia) might have been inadvertently included. To minimize its effect, we excluded procedures performed by gynecologic oncologists because the American College of Obstetricians and Gynecologists recommends involvement of a gynecologic oncologist in the care of these patients.2 However, this may limit generalizability of our sample because gynecologic oncologists often care for patients with complex benign conditions. Second, although we captured three major gynecologic cancer types, further delineation of identified malignancies (eg, differentiating endometrial adenocarcinoma from uterine sarcoma) was not feasible within our data and requires further research. Likewise, because cervical or ovarian tissue was not always included in pathology at the time of hysterectomy and myomectomy, our analysis might underestimate the prevalence of occult cervical and ovarian cancer. Finally, after the U.S. Food and Drug Administration safety warning about power morcellation in 2014, there might have been more rigorous preoperative screening and more restrictive selection of patients for supracervical hysterectomy. Hence, our analysis of 2014–2015 data may reflect conservative estimates of occult cancer risk in women undergoing supracervical hysterectomy.
Despite these limitations, findings from our study have important implications for clinical care. Currently, screening methods for uterine sarcoma or ovarian cancer are limited, whereas preoperative screening for cervical (cervical cytology) and endometrial cancer (endometrial sampling or imaging) is available but not sufficiently sensitive.24,30 Given the nontrivial risk of occult gynecologic cancer, it is imperative to further advance gynecologic cancer screening techniques. Moreover, higher risk of occult cancer in women at older age or undergoing total hysterectomy calls for more careful preoperative evaluation and counseling for these patients. The potential role for selective use of power morcellation also warrants further evaluation.
1. Tsui C, Klein R, Garabrant M. Minimally invasive surgery: national trends in adoption and future directions for hospital strategy. Surg Endosc 2013;27:2253–7.
2. Endometrial cancer. Practice Bulletin No. 149. American College of Obstetricians and Gynecologists. Obstet Gynecol 2015;125:1006–26.
3. Cohen SL, Vitonis AF, Einarsson JI. Updated hysterectomy surveillance and factors associated with minimally invasive hysterectomy. JSLS 2014;18. pii: e2014.00096.
4. Mahnert N, Morgan D, Campbell D, Johnston C, As-Sanie S. Unexpected gynecologic malignancy diagnosed after hysterectomy performed for benign indications. Obstet Gynecol 2015;125:397–405.
5. Graebe K, Garcia-Soto A, Aziz M, Valarezo V, Heller PB, Tchabo N, et al. Incidental power morcellation of malignancy: a retrospective cohort study. Gynecol Oncol 2015;136:274–7.
6. Park JY, Park SK, Kim DY, Kim JH, Kim YM, Kim YT, et al. The impact of tumor morcellation during surgery on the prognosis of patients with apparently early uterine leiomyosarcoma. Gynecol Oncol 2011;122:255–9.
7. Pritts EA, Parker WH, Brown J, Olive DL. Outcome of occult uterine leiomyosarcoma after surgery for presumed uterine fibroids: a systematic review. J Minim Invasive Gynecol 2015;22:26–33.
8. Einstein MH, Barakat RR, Chi DS, Sonoda Y, Alektiar KM, Hensley ML, et al. Management of uterine malignancy found incidentally after supracervical hysterectomy or uterine morcellation for presumed benign disease. Int J Gynecol Cancer 2008;18:1065–70.
10. Desai VB, Guo XM, Xu X. Alterations in surgical technique after FDA statement on power morcellation. Am J Obstet Gynecol 2015;212:685–7.
11. 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;12:165–177.
12. Hill AJ, Carroll AW, Matthews CA. Unanticipated uterine pathologic finding after morcellation during robotic-assisted supracervical hysterectomy and cervicosacropexy for uterine prolapse. Female Pelvic Med Reconstr Surg 2014;20:113–5.
13. Liu FW, Galvan-Turner VB, Pfaendler KS, Longoria TC, Bristow RE. A critical assessment of morcellation and its impact on gynecologic surgery and the limitations of the existing literature. Am J Obstet Gynecol 2015;212:717–24.
14. Wright JD, Chen L, Burke WM, Hou JY, Tergas AI, Ananth CV, et al. Trends in use and outcomes of women undergoing hysterectomy with electric power morcellation. JAMA 2016;316:877–8.
15. Wright JD, Tergas AI, Burke WM, Cui RR, Ananth CV, Chen L, et al. Uterine pathology in women undergoing minimally invasive hysterectomy using morcellation. JAMA 2014;312:1253–5.
17. Parker WH, Kaunitz AM, Pritts EA, Olive DL, Chalas E, Clarke-Pearson DL, et al. U.S. Food and Drug Administration's guidance regarding morcellation of leiomyomas: well-intentioned, but is it harmful for women? Obstet Gynecol 2016;127:18–22.
22. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 1996;49:1373–9.
23. Kho KA, Lin K, Hechanova M, Richardson DL. Risk of occult uterine sarcoma in women undergoing hysterectomy for benign indications. Obstet Gynecol 2016;127:468–73.
24. Frick AC, Walters MD, Larkin KS, Barber MD. Risk of unanticipated abnormal gynecologic pathology at the time of hysterectomy for uterovaginal prolapse. Am J Obstet Gynecol 2010;202:507.e1–4.
25. Ackenbom MF, Giugale LE, Wang Y, Shepherd JP. Incidence of occult uterine pathology in women undergoing hysterectomy with pelvic organ prolapse repair. Female Pelvic Med Reconstr Surg 2016;22:332–5.
26. Picerno TM, Wasson MN, Gonzalez Rios AR, Zuber MJ, Taylor NP, Hoffman MK, et al. Morcellation and the incidence of occult uterine malignancy: a dual-institution review. Int J Gynecol Cancer 2016;26:149–55.
27. Ramm O, Gleason JL, Segal S, Antosh DD, Kenton KS. Utility of preoperative endometrial assessment in asymptomatic women undergoing hysterectomy for pelvic floor dysfunction. Int Urogynecol J 2012;23:913–7.
28. Harris JA, Swenson CW, Uppal S, Kamdar N, Mahnert N, As-Sanie S, et al. Practice patterns and postoperative complications before and after US Food and Drug Administration safety communication on power morcellation. Am J Obstet Gynecol 2016;214:98.e1–13.
29. Cohen SL, Einarsson JI, Wang KC, Brown D, Boruta D, Scheib SA, et al. Contained power morcellation within an insufflated isolation bag. Obstet Gynecol 2014;124:491–7.
30. Chapman JA, Mannel RS, DiSaia PJ, Walker JL, Berman ML. Surgical treatment of unexpected invasive cervical cancer found at total hysterectomy. Obstet Gynecol 1992;80:931–4.