Occult Malignancy Incidence and Preoperative Assessment in Hysterectomies with Morcellation

Introduction

Hysterectomy is one of the most commonly performed surgical procedures in the United States, with more than 7 million women undergoing hysterectomy between 1998 and 2010. ¹ Hysterectomies can be performed abdominally, vaginally, or laparoscopically (with or without robotic assistance). Minimally invasive approaches, when feasible, are often preferred for their decreased recovery time, shorter duration of hospital stay, and lower intraoperative blood loss. ² However, removing large uteri or leiomyomas through small incisions can be challenging. Morcellation can assist in the removal of large specimens through tissue fragmentation. ³

The use of morcellation in gynecologic surgery has come under recent scrutiny secondary to concerns for dissemination of occult malignancy. ⁴ The incidence of undiagnosed gynecologic malignancy in women undergoing hysterectomy for benign indications has been quoted as high as 2.7%. ⁵ Specifically, relevant literature suggests a range of sarcoma incidence in women with fibroids from 1:352 to 1:7450.^6,7

Sarcomas are difficult to diagnose preoperatively, compared to cervical and endometrial cancers,^8,9 and there are no preoperative diagnostic tests that can be used to detect sarcomas reliably. However, preoperative evaluation before hysterectomy with morcellation should at least include current cervical cytology testing and may include pelvic imaging and endometrial assessment when appropriate.^10–12

The objectives of the current study were to evaluate the incidence of occult malignancy in women undergoing hysterectomy for benign indications in which morcellation was and was not performed, and to compare associated preoperative evaluations.

Materials and Methods

This retrospective cohort study was a medical-record review performed after approval by the University of Massachusetts Medical School (UMMS) institutional review board. Data were obtained from the UMass Memorial Health Care (UMMHC) electronic medical-record systems. Inclusion and exclusion criteria were determined prior to data collection (Fig. 1). All women who had hysterectomies performed at UMMHC between October 1, 2007, and June 30, 2014, were considered for inclusion. These women were identified by searching the term hysterectomy in the institutional billing database. Given the focus on hysterectomies performed for benign indications, charts were excluded if the hysterectomy was performed by a gynecologic oncologist or a nongynecologic specialist (surgical oncology or general surgery) to avoid biasing the data with regard to the preoperative evaluation completed and to reflect practice patterns among general gynecologists. Additional exclusion criteria were: (1) surgery performed outside the UMMHC system and (2) the medical record unavailable for review. Abstracted data were managed using REDCap (Research Electronic Data Capture) ¹³ which is a web-based program hosted at UMMS. ¹³ Any discrepancies among data sources were resolved by returning to the original medical record.

OPEN IN VIEWER

FIG. 1.

Study flow diagram. UMMS/UMMHC, University of Massachusetts School/UMass Memorial Health Care.

Demographic characteristics collected included patient age, race/ethnicity, height, weight, body mass index (BMI), smoking status, postmenopausal status defined as amenorrhea for 12 months, ¹⁴ history of hormone replacement therapy use, history of pelvic radiation, history of premenopausal hormone use (i.e., for contraception or for abnormal uterine bleeding [AUB] management), history of tamoxifen use, history of female infertility, history of personal malignancy, and a family history of gynecologic malignancy.

The characteristics of the surgical approach to hysterectomy were also collected. The surgical data collected included mode of hysterectomy, if concurrent bilateral salpingo-oophorectomy was performed, if gynecologic oncology was consulted intraoperatively, if morcellation was used, and if procedure conversion occurred (e.g., laparoscopic to abdominal, robotic to abdominal, vaginal to laparoscopic, vaginal to abdominal, or other).

Pathology reports were reviewed for final diagnoses. Specimens were classified as benign, precancerous, or malignant. If multiple diagnoses were included on a patient's pathology report, the diagnosis meriting the greatest level of concern or follow-up, resulted in use of that category (i.e., malignant > precancerous > benign). If the patient had malignant final pathology, details regarding cancer stage/grade, treatment, and current disease status were also collected.

Outpatient medical records were reviewed for the preoperative diagnoses, which were categorized as AUB, fibroids/leiomyoma, pelvic pain, pelvic organ prolapse (POP), pelvic mass, precancerous changes (including cervical intraepithelial neoplasia [CIN] or endometrial hyperplasia), or other. If subjects had multiple preoperative diagnoses, all were included. Subject preoperative evaluation was also abstracted from the charts. Abstraction focused on the following evaluation components: cervical cytology; endometrial assessment within 1 year prior to surgery; and pelvic imaging within 1 year prior to surgery. Cervical cytology was evaluated to determine if a patient had an up-to-date Papanicolau test per current American Cancer Society guidelines, specifically cytology alone within the past 3 years, cytology with human papilloma virus cotesting within the past 5 years, or last Papanicolau test at the age of 65 with no prior abnormal screening results. ¹⁵ Endometrial assessment was categorized as endometrial biopsy or dilation and curettage (D&C). Imaging modality was collected and included ultrasound (US; transvaginal or transabdominal), magnetic resonance imaging, computed tomography (CT) scan, sonohysterogram, hysterosalpingogram, and positron-emission tomography scan.

Outcomes of interest included use of morcellation during hysterectomy as noted in the operative report. This was defined as either manual fragmentation or use of any electromechanical device to fragment the tissue into smaller pieces to aid in the removal of the specimen. ³ The study population was then divided into patients who had hysterectomy with morcellation, compared to those who had hysterectomy without morcellation.

Categorical variables were compared using χ ² and Fisher's exact test as appropriate, and continuous variables were compared using a Student's t-test or a Wilcoxon rank-sum test. Statistical significance set at p < 0.05 for all comparisons. All analyses were performed using Stata/MP 13.1 (StataCorp. 2013. Stata Statistical Software: Release13. College Station, TX: StataCorp LP).

Results

A total of 5461 hysterectomies were performed between October 1, 2007, and June 30, 2014, at UMMHC as identified from the billing database. To restrict the sample to hysterectomies performed for benign indications and to reflect practice patterns for preoperative evaluations completed by general gynecologists, subjects were excluded if the surgeons were gynecologic oncologists or nongynecologic specialists (surgical oncology or general surgery; n = 3128), for a preoperative diagnosis of malignancy (n = 23), and 1 due to operative report not indicating a hysterectomy was performed. Therefore, a total of 2309 women had hysterectomy, with 396 (17.1%) having the procedure with morcellation and 1916 (82.9%) having the operation without morcellation.

Baseline characteristics are shown in Table 1. The 2 groups were similar with no significant differences in race, BMI, smoking status, personal malignancy history, family history of gynecologic malignancy, history of hormone replacement use, tamoxifen use, premenopausal hormone use, and pelvic radiation history. Women undergoing hysterectomy with morcellation versus those that did not have it were, on average, younger (33.4 versus 39.7 years; p < 0.001) but were more often postmenopausal (40.4% versus 32.2%; p = 0.002). When morcellation was performed, the vast majority of the time it was uncontained (99.2%) and accomplished with the use of power morcellation (93.4%). Manual morcellation in the abdomen or vagina accounted for 6.6% of morcellation cases. Of 396 morcellation cases, 4.80% (n = 19) had intraoperative complications that were not necessarily related to the act of morcellation itself. There were 5 cases with ureteral/genitourinary injury, 1 case with a vascular injury, 3 cases requiring return to the operating room, 11 cases requiring blood transfusion, and 3 cases with an estimated blood loss (EBL) >2 L.

Surgical characteristics of the analyzed cohort are shown in Table 2. Patients having hysterectomy with morcellation had significantly different surgical characteristics, including more robotic supracervical (48.2% versus 0.3%, p < 0.001) and laparoscopic supracervical (38.4% versus 0.1%; p < 0.001) hysterectomies, less EBL (212.1 mL versus 280.3 mL; p < 0.001), fewer intraoperative gynecologic oncology consultations (0.0% versus 1.2%; p = 0.023), fewer frozen-section pathology evaluations (2.5% versus 5.5%; p = 0.013), and fewer bilateral salpingo-oophorectomies (14.9% versus 21.9%; p = 0.002). The most common preoperative diagnosis for patients with morcellation was pelvic organ prolapse (POP; 42.7% versus 24.4%; p < 0.001), with 98.2% of these cases having power morcellation used in the abdominal cavity. The most common indication for patients without morcellation was AUB (45.4% versus38.1%; p = 0.008). Those with morcellation had significantly different final pathologies than those without morcellation (p < 0.001), including higher rates of benign pathology (98.5% versus 89.9%), and lower rates of postoperative precancerous pathology (1.3% versus 8.1%) and malignant pathology (0.3% versus 2.0%).

Of all 2309 reviewed charts, 39 women undergoing hysterectomy for preoperative benign indications were diagnosed with occult malignancy. Characteristics of these patients are summarized in Table 3. The overall incidence of malignancy on final pathology was 1.7%. More specifically, the overall incidence of sarcoma was 0.09%. The overall rate of malignancy differed between nonmorcellated versus morcellated specimens (2.0% versus 0.3%, p < 0.001). One patient having a hysterectomy with morcellation was diagnosed with leiomyosarcoma, making the incidence of sarcoma in the morcellated group 0.3%. Of the 39 cases with occult malignancy, 51.3% (n = 20) had preoperative diagnoses of precancerous changes. The majority of these (n = 15; 75%) were complex endometrial hyperplasias with or without atypia, 20% (n = 4) with CIN, and 5.0% (n = 1) with BRCA-mutation–carrier status. In terms of preoperative evaluations, 74.4% (n = 29) had up-to-date cervical cytology, 66.7% (n = 26) had some form of endometrial assessment, and 41.0% (n = 16) had preoperative pelvic imaging. Of patients with presenting with AUB, 92.3% (n = 12) had sampling of the uterine lining, with 1 not having sampling due to inability to obtain an office biopsy.

There were no statistically significant differences between the total morcellated and nonmorcellated groups with any preoperative evaluation inclusive of cervical cytology, imaging, or endometrial assessment (p = 0.875; Fig. 2). There were no differences in rates of current cervical cytology (68.9% versus 71.0%; p = 0.403) or imaging (39.6% versus 35.2%; p = 0.092) between both groups. Patients with morcellation were significantly less likely to have endometrial assessments, compared to patients without morcellation (21.7% versus 34.1%; p < 0.001).

OPEN IN VIEWER

FIG. 2.

Comparison of preoperative evaluation between women undergoing hysterectomy with and without morcellation. *, p-Value <0.001.

Discussion

The overall incidence of occult malignancy in women having hysterectomies for benign indications was 1.7%, which is similar to other reports in the literature. ^6,7,16 The current study confirmed that the malignancy incidence is low, with the incidence of sarcoma being rarer. In the current patient population, the incidence of sarcoma among women having hysterectomy was 0.09%. Specifically, among women having hysterectomy with morcellation, the incidence of sarcoma was 0.3%. This rate is less than the reported sarcoma frequency used by the U.S. Food and Drug Administration (FDA) in its statement regarding morcellation use in hysterectomy. ⁶

The case of a morcellated malignancy in the current study was a leiomyosarcoma occurring in a postmenopausal patient with a known history of uterine fibroids, who had presented with pelvic pain. Transvaginal US 9 months prior to surgery had shown a large anterior fibroid, and repeat US the month of surgery showed interval growth of that fibroid by 3 cm in multiple dimensions. She underwent total laparoscopic hysterectomy with power morcellation, and final pathology showed high-grade leiomyosarcoma. An immediate postoperative chest CT showed pulmonary nodules consistent with metastatic disease. She was started on adjuvant chemotherapy. Ultimately, 24 months following her initial surgery, she had diffused disease progression and died 27 months after initial diagnosis. Although this patient did not have endometrial sampling, sampling has lower predictive value for detecting sarcomas than uterine epithelial malignancies. ⁹

There is also no reliable imaging modality for differentiating sarcomas from leiomyomas. ¹⁷ This patient did have growth of her presumed fibroids raising the suspicion of malignancy, ⁹ although this also is not reliable for detecting sarcomas.^18–20 It does not appear that morcellation of the specimen influenced the stage of disease, as diffuse pulmonary nodules were present on the immediate postoperative chest CT. Uterine leiomyosarcoma is rare but is an aggressive disease with an overall poor prognosis. The current authors believe that this case illustrates a patient for whom preoperative consultation with gynecologic oncology or intraoperative frozen section was warranted, such that appropriate surgical staging by an oncologist as well as counseling could have been performed.

In the current study cohort, women undergoing hysterectomy with and without morcellation had similar rates of preoperative cervical cytology and pelvic imaging despite varied presentations and symptoms. Patients having hysterectomy without morcellation had higher rates of endometrial assessments but presented more commonly with AUB. Given the clinical and medico–legal concerns regarding morcellation and the accuracy of endometrial biopsy for identifying hyperplasia or malignancy, ²¹ it would not be unreasonable to perform an endometrial biopsy prior to an operation with planned morcellation.

Of patients with occult malignancy, 51.3% (n = 20) had preoperative diagnoses of precancerous changes, with 75.0% (n = 15) of these patients having endometrial hyperplasia (93.3% [n = 14]; hyperplasia with atypia). The preoperative diagnosis of endometrial hyperplasia should alert the surgeon to increased malignancy risk on final pathology given that 38.5% (n = 15) of the population with occult malignancy had endometrial hyperplasia preoperatively. Current literature suggests that 50% of women diagnosed with atypical hyperplasia have concurrent carcinomas and that the risk of malignancy with simple or complex hyperplasia without atypia is not well known. ²²

Interestingly, patients with preoperative indications of POP and postoperative occult malignancy were frequently diagnosed with endometrial adenocarcinoma (12.8%, n = 5), with only 1 patient having an endometrial biopsy. Women undergoing surgery for POP had lower rates of endometrial assessments because of their initial presentations; thus, it is not surprising that endometrial cancer was the frequently diagnosed occult malignancy. POP surgery is often performed via a minimally invasive approach with morcellation. Sacral colpopexy for POP after supracervical hysterectomy has superior outcomes to other vaginal procedures as well as a lower rate of mesh erosion, ²³ which has caused an increase in supracervical hysterectomies, compared to total hysterectomies at UMMHC. Given the concerns of occult malignancy dissemination, an argument could be made that endometrial biopsy should be included in the evaluation of patients undergoing hysterectomy for prolapse. This would improve the preoperative detection of endometrial cancer, ²⁴ and reduce the risk of dissemination and cancer upstaging. Endometrial biopsy is safe and well-tolerated with a low risk of complications, such as uterine perforation. ²⁵ The cost-efficacy and clinical efficacy of these approaches has not yet been determined, and warrants further investigation.

A strength of the current study was the detailed data collected regarding the preoperative evaluation of patients undergoing hysterectomy with and without morcellation. While many studies have quoted an incidence of occult malignancy, few have investigated each patient's preoperative evaluation. ¹⁶

A limitation of this study included the retrospective nature of its design. There were missing data and possible misclassification of the data in the electronic health records, limiting some of the analyses. Given that private-practice patients were included in this study, their Papanicolau test results were not always visible in the medical records. Therefore, the 29.3% of the population without documented current cervical cytology could reflect low rates of preoperative Papanicolau testing or a limitation of the medical records. To oppose this, comprehensive chart abstractions supplemented what was initially available through automatic electronic inquiry. Additionally, a large proportion of women undergoing hysterectomy were excluded, as their primary surgeons were gynecologic oncologists or surgical oncologists. A large majority of the hysterectomies performed at UMMHC are performed for oncologic reasons, and they were excluded to avoid skewing the preoperative evaluation data. Therefore, 7 years of data were reviewed and collected to develop a large sample size of women having hysterectomy for benign indications.

Conclusions

The risk of occult malignancy—and, in particular, sarcoma—is very low in women undergoing hysterectomy for benign indications. Given that the most common occult malignancy was endometrial adenocarcinoma, it may be reasonable for all patients, especially postmenopausal women, undergoing hysterectomy to have endometrial biopsies as part of their preoperative evaluations. Again, the cost-efficacy of this approach has not yet been determined, and warrants further investigation.

Morcellation also appears to be a safe surgical technique for approaching minimally invasive hysterectomy that should be discussed with patients. The optimal preoperative evaluation when performing a hysterectomy for presumed benign indications will be based on clinician judgment and patient chief complaint, but more liberal use of endometrial biopsy could reduce the risk of occult malignancy.