The Impact of Tumor Fragmentation in Patients with Stage I Uterine Leiomyosarcoma on Patterns of Recurrence and Oncologic Outcome

Abstract

Objective:

To evaluate the impact of tumor fragmentation on oncologic outcomes in patients with stage I uterine leiomyosarcoma (uLMS).

Methods:

We identified all patients diagnosed with stage I uLMS presenting to our institution within three months of primary surgery, 1/2000-1/2019. Patients with recurrent disease were excluded. The non-morcellated group had total hysterectomy without documented specimen fragmentation; the morcellated group, total hysterectomy with documented specimen fragmentation. We defined fragmentation as manual fragmentation or morcellation (via power morcellator or otherwise) of the specimen in peritoneal cavity or vagina. Appropriate statistical analyses were performed.

Results:

152 patients met inclusion criteria. 107 (70%) underwent total hysterectomy (non-morcellated); 45 (30%) underwent morcellation. Median age at diagnosis for the entire cohort was 55 years (range 30-91). Median follow-up was 42.1 months (range 1.1-197.8). 40 (26.3%) patients had primary surgery at our institution, 112 (73.7%) at an outside hospital. In total 110 (72.3%) recurred: 72/107 (67.2%) non-morcellated; 38/44 (86.3%) morcellated. Median progression-free survival (PFS) for non-morcellated versus morcellated was 13.8 (95%CI 9.220.2) versus 7.3 months (95%CI 3-13.1), HR 1.5 (95%CI 1.02-2.24); P=0.04. Median overall survival (OS) for non-morcellated versus morcellated was 82.1 (95%CI 52.4-122) versus 47.8 months (95%CI 28.5-129.6), HR 1.1 (95%CI 0.67-1.82); P=0.7. Among patients with recurrence, 69.4% of non-morcellated recurred at hematogenous sites only, 18.1% recurred in peritoneum only; 28.9% of morcellated recurred at hematogenous sites, 63.2% in peritoneum. Race, lymphovascular invasion, postoperative chemotherapy, were independently associated with PFS. Mitotic index was independently associated with OS.

Conclusions:

Tumor fragmentation/morcellation was associated with significantly higher risk of recurrence and a nearly 4-fold increase in peritoneal recurrence. Prognostic biomarkers remain important in predicting oncologic outcomes, independent of fragmentation or treatment.

INTRODUCTION

Uterine leiomyosarcomas (uLMS) are rare tumors of the uterus, accounting for 1-2% of all uterine malignancies [1]. uLMS is associated with high rate of recurrence and poor prognosis [1]. Known prognostic biomarkers for oncologic outcome in high-grade uLMS include mitotic index, patient age, lymphovascular invasion (LVSI) and tumor size [2]. To date, no adjuvant treatment options have shown a survival benefit in patients with stage I disease, and predictive biomarkers of response to therapy remain to be determined [3].

While uLMS is the most common uterine sarcoma, these tumors are difficult to diagnose preoperatively due to their myometrial location and the limited accuracy of preoperative imaging in distinguishing them from benign leiomyomatous disease. Recent advances in MRI and texture analysis suggest that preoperative imaging will play an increasingly important role in distinguishing between benign and malignant degeneration of presumed myomas [4]. The gold standard treatment for uLMS is surgical excision with negative margins [5]. A total hysterectomy is preferred for known malignant disease. However, uLMS is often diagnosed after surgical procedures for presumed benign leiomyomatous disease. Often these procedures are performed using a minimally invasive (MIS) platform with tissue fragmentation for extraction. With the incorporation of power morcellation into minimally invasive gynecologic surgery in the early 2000s, MIS procedures became more common in the setting of large, bulky leiomyomatous disease. In 2008, our group was one of the first in the U.S. to publish the outcomes of uterine tumor morcellation and noted that approximately 15% of patients will be upstaged by reexploration, particularly those with LMS who underwent morcellation [6]. In 2013, a highly publicized medical case brought national attention to the potential risk of malignant disease dissemination associated with power morcellation during surgery for presumed benign leiomyomas [7]. Subsequently, the FDA conducted a literature review to determine the risk of benign leiomyomas harboring occult leiomyosarcoma. This analysis calculated a prevalence of 1 in 498 for unsuspected LMS in women undergoing hysterectomy or myomectomy for benign disease [8]. After this finding, the FDA issued a statement discouraging the use of power morcellation in both benign and malignant disease, as well as a “black box” warning for power morcellators [9]. After this, many institutions nationwide abandoned the use of power morcellators during gynecologic surgery.

In the wake of the “black box” warning, there have been multiple shifts in the approach to management of surgical leiomyoma, ranging from reverting to open procedures for myomectomy, to the incorporation of laparoscopic bags to facilitate extraction via tissue fragmentation while avoiding peritoneal spillage. However, there has not been a consensus on the true impact of tumor morcellation on oncologic outcomes in patients diagnosed with early-stage LMS. The primary aim of this study was to evaluate the impact of tumor morcellation on oncologic outcomes in patients with clinical stage I uLMS. The secondary aim was to identify and describe recurrence patterns in patients with morcellated versus non-morcellated specimens.

METHODS

We retrospectively identified all patients with clinical stage I high-grade uLMS who presented to our institution from January 2000 to January 2019. Patients who had their primary surgery performed at an outside institution were included in our analysis only if they presented to our institution within three months of the primary surgery date. We excluded all patients who presented with a recurrence or suspected recurrence. The cohort was divided into two groups: the non-morcellated group, defined as patients who underwent total hysterectomy, without any documented specimen fragmentation or power morcellation; and the morcellation group, defined as patients who underwent any type of specimen fragmentation or power morcellation in the peritoneal cavity or vagina. Patients who were deemed to have greater than stage I disease at time of diagnosis, patients who underwent incomplete hysterectomy without knowledge of margin status, or patients for whom complete pathology reports or surgical reports were unavailable, were excluded from this study. Recurrence was found either on secondary surgery (i.e. re-operation), imaging, or physical examination.

Pattern of recurrence was recorded during data collection and analysis. Three patterns of recurrence were defined: hematogenous (lung, liver, bone or brain); peritoneal; and retroperitoneal (lymph nodes).

Statistical Analyses

Differences in clinicodemographic factors between the non-morcellated and morcellation groups were tested using Fisher’s exact test for categorical variables and the Wilcox on Rank-Sum test for continuous variables.

Progression-free survival (PFS) was calculated from date of initial surgery to date of first recurrence or progression, date of death, or date of last follow-up, whichever occurred first. Both progression or death were considered as events in PFS. Overall survival (OS) was calculated from initial surgery to date of death or last follow-up date, whichever occurred first. The Kaplan-Meier method was used to estimate survival rate and median survival time. In univariate survival analyses, categorical variables were tested using the Log-rank test, and continuous variables were tested using the Wald test based on the Cox Proportional-Hazards (CoxPH) model. The Time-dependent method was applied to variables which occurred after surgery (i.e. postoperative chemotherapy, etc.). Multivariate CoxPH models for PFS and OS were built based on the results of univariate analyses.

Recurrence patterns for different groups were analyzed using Competing risks methodology. When estimating the cumulative incidence failure rate (CIF) of hematogenous recurrence (or peritoneal recurrence), the other type of recurrence such as peritoneal recurrence (or hematogenous recurrence) and death were considered as competing events. The Gray test was applied to test the difference in this setting [10].

RESULTS

We initially identified 301 patients with uLMS. Seventy-nine patients presented to our institution more than three months from the primary surgery; 70 patients had undergone subtotal hysterectomy (supracervical hysterectomy or myomectomy) with incomplete data regarding surgical procedures and type of fragmentation and were thus excluded. One hundred fifty-two patients met the inclusion criteria and were included in the final analysis. Forty patients (26.3%) had primary surgery at our institution; 112 (73.7%) had primary surgery at an outside hospital. Clinicodemographic data are shown and compared in Table 1: 107 patients (70%) had a total hysterectomy and 45 (30%) underwent tumor fragmentation. Race, BMI, tumor size, mitotic count and postoperative treatment, were comparable between the groups. Patients in the non-morcellated group were older (median age 56.9 years) than those in the morcellation group (median 50.4 years) (p≤0.001), more likely to undergo open surgery, less likely to undergo reoperation, and more likely to have LVSI on pathology review.

Table 1:

Description of Clinicodemographic Variables

	All	Non-Morcellated	Morcellated	p-value
N	152	107	45
Age (years)				<0.001
Median (range)	55 (30-91.7)	56.9 (34.6-91.7)	50.7 (30.69.8)
Race
White	125(82.2%)	89(83.2%)	36(80%)	-
Black	12(7.9%)	9(8.4%)	3(6.7%)
Asian	8(5.3%)	5(4.7%)	3(6.7%)
Hispanic	5(3.3%)	4(3.7%)	1(2.2%)
Other	2(1.3%)	0(0%)	2(4.4%)
BMI (kg/m2)
Median (range)	25.2 (17.3-44)	26 (17.3-44)	24.3 (17.7-43.9)	0.275
Tumor Size (cm)
Median (range)	9 (2-28)	9 (2-23)	9.5 (3-28)	0.142
LVSI				0.005
Negative	99(65.1%)	72(67.3%)	27(60%)
Positive	33(21.7%)	27(25.2%)	6(13.3%)
Not Done	20(13.2%)	8(7.5%)	12(26.7%)
Mitotic Count				0.106
Median (range)	20 (3-90)	20 (3-90)	18 (5-43)
Diagnosis Method				<0.001
EMB/D&C/Hysteroscopic Myomectomy	41(27%)	38(35.5%)	3(6.7%)
Abd/MIS Myomectomy	7(4.6%)	0(0%)	7(15.6%)
Supracervical Hysterectomy	11(7.2%)	0(0%)	11(24.4%)
Total Hysterectomy	93(61.2%)	69(64.5%)	24(53.3%)
Primary Surgery Location				<0.001
MSK	40(26.3%)	38(35.5%)	2(4.4%)
OSH	112(73.7%)	69(64.5%)	43(95.6%)
Primary Surgery Type				<0.001
TAH	107(70.4%)	94(87.9%)	13(28.9%)
SCH	5(3.3%)	0(0%)	5(11.1%)
TLH/TVH	13(8.6%)	6(5.6%)	7(15.6%)
RA-TLH	13(8.6%)	7(6.5%)	6(13.3%)
LA-SCH	6(3.9%)	0(0%)	6(13.3%)
Abd/MIS Myomectomy	8(5.3%)	0(0%)	8(17.8%)
Surgical Approach
Open	119 (78.3%)	94 (87.9%)	25 (55.6%)	<0.001
Minimally Invasive	33 (21.7%)	13 (12.1%)	20 (44.4%)
Re-Operation
No	122(80.3%)	103(96.3%)	19(42.2%)	<0.001
Yes	30(19.7%)	4(3.7%)	26(57.8%)
Post Op Treatment				1
No	104(68.4%)	73(68.2%)	31(68.9%)
Yes	48(31.6%)	34(31.8%)	14(31.1%)

Note: p-values that appear in red are significant.

BMI, Body Mass Index; EMB, Endometrial ablation; D&C, Dilation and curettage; MIS, Minimally invasive surgery; MSK, Memorial Sloan Kettering Cancer Center; OSH, Outside hospital; TAH, Total abdominal hysterectomy; SCH, Supracervical hysterectomy; TLH/TVH, Laparoscopic hysterectomy/Total vaginal hysterectomy; RA-TLH, Robotic-assisted total laparoscopic hysterectomy; LA-SCH, Laparoscopic supracervical hysterectomy; Abd, Abdominal

Among the 45 patients in the morcellation group, 23 (56.1%) were fragmented manually with scalpel or scissors, 6 (14.6%) with a power morcellator; in 12 patients (29.3%) the specific type of morcellation (i.e. power vs. manual) was not described in the operative notes (Table S1). Thirty patients (73.2%) had morcellation in the peritoneal cavity, while in 10 patients (24.4%) morcellation was performed vaginally. In 27 patients (65.9%) morcellation was uncontained. Of the 45 patients in the morcellation group, 26 (58%) underwent re-operation after morcellation: 7 patients (27%) underwent completion hysterectomy, 10 (38%) completion trachelectomy, and 9 (35%) had other procedures (i.e. laparoscopy, bilateral salpingo-oophorectomy, biopsies). Of the 26 patients who underwent re-operation, 9 (35%) were found to have no residual disease, 4 (15%) had uterine-confined disease, and 13 (50%) had disseminated disease at the time of surgery (Table S2).

In total, 104 of 152 patients (68.4%) did not receive adjuvant postoperative therapy, while 48 (31.6%) did receive postoperative therapy. Of the 48 patients who received postoperative treatment, 31 (64.6%) received chemotherapy alone, 7 (14.6%) received radiation therapy alone, 1 (2%) received chemotherapy and radiation therapy, 1 (2%) received radiation and hormone therapy, and 8 (16.7%) received hormone therapy alone. Fourteen of the 45 patients (31%) in the morcellation group received adjuvant therapy: 8 (57.1%) received chemotherapy alone, 1 (7%) received chemotherapy and radiation therapy, 1 (7%) received radiation therapy and hormone therapy, and 4 (28.6%) received hormone therapy alone. (Table 1)

Median follow-up of the entire cohort was 42.1 months (range 1.1-197.8 months). Median PFS for the entire cohort was 10.7 months (95% CI 8-16.2). Median PFS for the non-morcellated group was 13.8 months (95% CI 9.2-20.2) compared with 7.3 months (95% CI 313.1) for the morcellation group; p=0.036, HR 1.52 (95% CI 1.02-2.24). (Figure 1). Univariate and multivariate analysis for PFS are shown in Table S2. Race and LVSI were independently associated with poorer PFS, while receipt of postoperative chemotherapy was independently associated with improved PFS.

Figure 1.

Progression-Free Survival based on specimen manipulation type

Median OS for the entire cohort was 77.6 months (95% CI 50.5-115.9). Median OS for the non-morcellated group was 82.1 months (95% CI 52.4-122) compared with 47.8 months (95% CI 28.5-129.6) for the morcellation group; p=0.7, HR 1.1 (95% CI 0.67-1.82) (Figure 2). Univariate and multivariate analyses are shown in Table S3. Mitotic index was independently associated with poorer OS.

Figure 2.

Overall Survival based on specimen manipulation type

In a sub-analysis of morcellated patients alone, neither location of the morcellation (i.e vaginal or peritoneal cavity) nor specimen containment (i.e. within an endocatch bag) were associated with PFS or OS. Median PFS for patients who underwent vaginal morcellation was 16.1 months (range 0.6-36.9), compared with 5 months (range 2.1-9.4) in patients who underwent morcellation within the peritoneal cavity, P=0.3. Median PFS for patients whose specimen was not contained was 8.7 months (range 2.3-35.1) compared with 11.2 months (range 5-16.2) for those who were contained, P=0.5. Median OS was 71.8 months (11.9-129.6) for patients who underwent vaginal morcellation compared with 77.6 months (range, 34-NE) for those who underwent morcellation within the peritoneal cavity, P=0.8. Median OS was 129.6 months (range, 28.3-NE) for those whose specimen was not contained. Median OS was not reached for those whose specimen was contained.

In total, 110 of 152 patients (72.4%) had a recurrence at the time of the analysis. Recurrences were observed in 72 of 107 patients (67.2%) in the non-morcellated group and 38 of 45 patients (84%) in the morcellation group. (Table 2) In the non-morcellated group, 50 of 72 patients (69.4%) recurred with a hematogenous pattern only; 13 (18.1%) with a peritoneal pattern only; 2 (2.8%) with a retroperitoneal pattern only. In the morcellation group, 24 of 38 patients (63.2%) recurred only within the peritoneum; 11 (28.9%) recurred only hematogenously. There were no retroperitoneal recurrences in the morcellation group.

Table 2:

Recurrence Patterns

	Total Cohort	Non-morcellated	Morcellation
Recurrence Type	110	72	38
Hematogenous	61(55.5%)	50(69.4%)	11(28.9%)
Peritoneal	37(33.6%)	13(18.1%)	24(63.2%)
Retroperitoneal	2(1.8%)	2(2.8%)	0(0%)
Hematogenous + Peritoneal	9(8.2%)	7(9.7%)	2(5.3%)
Hematogenous + Peritoneal + Retroperitoneal	1(0.9%)	0(0%)	1(2.6%)

The cumulative incidence failure (CIF) rate at 3 years for hematogenous recurrences for the total cohort was 40.8% (95% CI 32.2-49.5). In the non-morcellated group, the CIF at 3 years for hematogenous recurrences was 49.4% (95%CI 38.6%-60.1%), compared with 21.9% (95%CI 8.9%-34.9%) in the morcellation group, p=0.002. (Table 3). The CIF rate at 3 years for peritoneal recurrences, for the entire cohort, was 27.8% (95%CI 20%-35.7%). The CIF at 3 years for peritoneal recurrences in the non-morcellated group was 15% (95%CI 7.4%-22.6%) compared with 56.3% (95%CI 40.7%-72%) in the morcellation group, p<0.001. (Figure 3).

Table 3.

Recurrence Pattern Analysis Using Competing Risk Methodology. A. Hematogenous Recurrences; B. Peritoneal Recurrences

A. Hematogenous Recurrences
Variable	Pt#	Hematogenous Recurrences	Death#	CIF at 3Yr (95%CI)	p-value
Total Cohort	140	61	2	40.8% (32.2-49.5%)
Specimen Manipulation
Non-morcellated	98	50	2	49.4% (38.6-60.1%)	0.002
Morcellation	42	11	0	21.9% (8.9-34.9%)
B. Peritoneal Recurrences
Variable	Pt#	Peritoneal Recurrences	Death#	CIF at 3Yr (95%CI)	p-value
Total Cohort	140	37	2	27.8% (20-35.7%)
Specimen Manipulation
Non-morcellated	98	13	2	15% (7.4-22.6%)	<0.001
Morcellation	42	24	0	56.3% (40.7-72%)

Note: p-values that appear in red are significant.

CIF, Cumulative incidence of failure

Figure 3.

Cumulative Incidence of Failure (CIF) Curve

DISCUSSION

In this study, we analyzed a homogenous group of patients with stage I uLMS. The pattern of recurrence after complete hysterectomy without fragmentation confirms the natural history of early-stage disease, with 70% of recurrences being hematogenous and only 18% occurring in the peritoneum. Our data confirms that tumor fragmentation is associated with inferior PFS and a higher risk of the 3-year cumulative peritoneal recurrence rate (4-fold increase), suggesting that the type of surgical procedure changes the natural history of the disease. Only 29% of patients who underwent tumor fragmentation experienced a hematogenous recurrence. Patients who had tumor fragmentation were younger and had less LVSI, suggesting a selection bias in favor of lower-risk disease in the fragmented group, highlighting the detrimental impact of tumor fragmentation in a group of patients with an otherwise favorable prognosis. We tried to reduce recall bias, referral bias and selection bias by excluding patients who presented with recurrent or suspected recurrent disease as well as patients who presented at our institution three months or longer after their initial diagnosis was made. Although our study shows that morcellation does not independently affect PFS or OS, the difference in median OS between the non-morcellated and morcellation groups is 35 months. A larger number of patients and longer follow-up time may have an impact on the results, and multivariate analysis suggests that other prognostic factors such as mitotic index and age remain important when predicting outcomes, irrespective of tumor fragmentation.

The impact of uterine morcellation on oncologic outcomes continues to be debated [6,11–18]. Our data confirms the findings of other retrospective studies that tumor fragmentation is associated with worse recurrence-free survival, especially with peritoneal recurrence pattern. It is unclear how to manage patients who underwent fragmentation of uLMS. Some clinicians are in favor of second-look surgery, others prefer to treat with systemic therapy, while others choose to observe. In our study most patients were taken back for a surgical assessment. In over 50% of patients we found residual peritoneal disease after tumor fragmentation. The limited number of patients in the current study does not allow for a comparison between different types of tumor fragmentation, such as inadvertent rupture of the uterus during manipulation or power morcellation of a large specimen. The small sample size of morcellated patients limits our comparison between contained (e.g. in an endocatch bag) versus exposed fragmentation, as well as morcellation location of the specimen. A larger cohort of patients would be required to show a possible significant difference between these cohorts. Differences in outcomes depending on type of tumor fragmentation would improve risk stratification.

In conclusion, morcellation of uLMS has an impact on recurrence pattern and risk of recurrence. Our group reiterates that tumor fragmentation should be avoided in patients with advanced age, as the risk of incidence of uLMS increases with age. However, if a severe contraindication to total hysterectomy or en bloc resection exists, other prognostic biomarkers remain important and should be considered when counseling patients who underwent tumor fragmentation as part of their initial surgery.

Highlights.

• Tumor fragmentation impacted oncologic outcomes and recurrence patterns in patients with Stage I uterine leiomyosarcoma

• Tumor fragmentation was associated with significantly higher risk of recurrence

• Patients with tumor fragmentation had nearly 4-fold increase in peritoneal recurrence vs. those without tumor fragmentation