The Impact of Hysteroscopic Tissue Removal Systems on Histopathologic Analysis for Benign and Cancerous Endometrial Pathology: An Ex Vivo Study

Abstract

Objective

To assess the impact of hysteroscopic tissue removal systems (TRS) on histopathology tissue diagnosis.

Measurements and Methods

This is a paired-comparison ex vivo study in which 23 endometrial sections from hysterectomized uteri (13 benign and 10 hyperplasia/cancer) were analyzed in a simulation laboratory center at a university teaching hospital. After routine tissue processing, a section of endometrium was provided for ex vivo TRS with suture mounting to a uterine model (Polly, Remedy). Morcellated specimens using the Hologic^® MyoSure hysteroscopic device were processed for histopathologic analysis by two blinded pathologists (Pa and Pb) and compared to the original specimens’ tissue diagnoses.

Results

Sufficient tissue for evaluation was found in 100% (23/23) of TRS specimens by Pa and 91.3% by Pb. TRS specimen diagnoses were concordant with routine histologic diagnosis 86.9% (20/23, k = 0.76) for Pa and 80.9% (17/21, k = 0.68) for Pb. Sensitivity and specificity were 70%/100% for Pa and 80%/91% for Pb, respectively. The false-positive (overdiagnosed) and false-negative rates (underdiagnosed) were 0%/30% and 9%/20% for Pa and Pb. Both Pa and Pb underdiagnosed most specimens confirmed by routine tissue diagnosis. TRS specimen diagnoses between Pa and Pb were concordant in 76.2% (16/21, k = 0.60).

Conclusion

TRS may adversely impact the ability to provide a histologic tissue analysis. Up to 30% of samples were overdiagnosed and 20% underdiagnosed. If confirmed, pathologists may need to reassess workflows to better offset potential underdiagnosis of malignant specimens as findings may be obscured through TRS. Additionally, surgeons may need to reconsider specimen handling, so highest yield specimens are provided to pathology.

Introduction

Abnormal uterine bleeding (AUB) in perimenopausal or postmenopausal women requires diagnostic evaluation to rule out organic pathology including polyps, fibroids, endometrial hyperplasia, and endometrial cancer [1, 2]. The method of choice for histologic evaluation of endometrial tissue is visually guided hysteroscopy, where monopolar and/or bipolar high-frequency electrical current has long been a central approach to assessing endometrial pathology [3–6].

However, there are challenges with hysteroscopic energy-based resection including issues related to gas bubbles, difficulty in managing removal of tissue fragments, and not obtaining pathology [5]. Hysteroscopic TRS have emerged as a new tool for resecting focal intrauterine lesions with distinct advantages including the immediate aspiration of tissue during resection and presumed complete pathological analysis of aspirated tissue [7, 8]. Previous studies have focused on the technical and mechanical properties of TRS including operative time, fluid deficit, resection rates, intra- and postoperative complications, patient satisfaction rates, and recurrence rates. Overall, randomized clinical studies and one systematic review on three prospective studies and five retrospective studies have shown possible advantages of TRS relative to electrosurgical resection by reducing procedural time and adverse event rates [9–12].

Some studies suggest that pathological analysis of TRS tissue is comparable to electrosurgical resection. In contrast, AlHilli et al. [11] questioned the ability to accurately perform histologic reads of TRS specimens. They reported in a cohort study, comparing TRS (n = 188) to hysteroscopic resection (n = 253) for uterine polyps, that none of the cases using TRS established a histopathologic diagnosis including hyperplasia (10.1%, n = 19) and adenocarcinoma (1.6%, n = 3). They concluded that intrauterine morcellation did not affect histopathologic evaluation of removed samples despite tissue fragmentation [11]. In a randomized controlled trial (RCT) comparing TRS to electrical surgical resection, Smith et al. [10] reported that a concordant diagnosis was provided in all of their cases (n = 121) and suggested that concerns over the ability to histologically analyze morcellated tissue specimens were unfounded.

Given the limited literature focusing on uterine polyps and concerns regarding TRS blade’s excessive fragmentation of tissue during resection, we explored how morcellation affects endometrial pathology including uterine polyps, leiomyomata, and uterine cancer and the ability to adequately read TRS tissue samples in an ex vivo model.

Materials and Methods

The study was designed as an ex vivo blinded paired-comparison study that was reviewed and approved by the Wright State Physicians, Boonshoft School of Medicine, Institutional Review Board (Dayton, Ohio), and the study was in compliance with Privacy Act guidelines.

Subjects

Subjects were recruited from patients undergoing abdominal, vaginal, or laparoscopic (robotic-assisted or traditional) hysterectomy for pre-operative diagnosed benign intrauterine pathology including polyps and/or uterine fibroids and pre-operative diagnosed endometrial hyperplasia and/or stage 1–2 endometrial cancer at Miami Valley Hospital in Dayton, Ohio. Exclusion criteria were contraindications to surgical management of carcinoma, clinical stage 3–4 endometrial cancer, and patients with limited English potentially hindering informed consent. Informed consent was obtained from study subjects at the time of study recruitment. In total, 23 subjects were recruited for the study: 13 with benign uteri and 10 with endometrial hyperplasia/cancer.

Tissue Preparation

After surgical removal of the uterus, each specimen in its entirety was submitted to the pathology department. The uterus was opened using a scalpel for gross inspection, and a bloc specimen was processed for routine or frozen section histopathologic diagnosis. A representative section of the bloc specimen was used for ex vivo TRS (Fig. 1).

Fig. 1.

Flowchart for recruitment and processing

Ex Vivo TRS

An ex vivo TRS model using the Hologic^® MyoSure TRS (Hologic Inc., Marlborough, MA) was created to resemble the clinical setting of in vivo operative hysteroscopy. Each section of endometrium was suture-mounted to a uterine model (Polly, Remedy) (Fig. 2). A 7.25-mm (22-Fr) MyoSure XL hysteroscope with a MyoSure XL device was inserted through the 4-mm (12-Fr) working channel. TRS was then performed completely in total on each specimen using a new device for each case. All procedures were performed by two investigators (KL or SL). Inflow pressures of 110 mmHg with normal saline and vacuum settings of 180 mmHg were used to mimic the in vivo setting for each specimen. The morcellated tissue was then preserved in formalin for histopathologic assessment.

Fig. 2.

Ex vivo TRS with uterine model (Polly, Remedy)

Histopathologic Tissue Diagnosis

Histopathologic tissue diagnosis for TRS samples was not performed until subject recruitment was completed when all tissue samples were morcellated sequentially. Two pathologists with > 10 years of experience were blinded to paired de-identified slides (routine tissue processing and TRS samples). Slides were read twice to account for intraobserver variability. Benign tissue was designated as negative with a score of “1.” Hyperplasia or malignant diagnoses were assigned a score based on severity of disease: “2” = hyperplasia, “3” = grade 1 adenocarcinoma, “4” = grade 2 adenocarcinoma, and “5” = grade 3 adenocarcinoma. In order to account for inter-observer variability, histologic diagnoses within one “degree” of each other were considered concordant. (That is, if the routine tissue specimen was scored “4” for grade 2 adenocarcinoma, but the TRS tissue was scored a “3” for grade 1 adenocarcinoma, this was considered concordant, while if a TRS sample tissue was scored a “2” for hyperplasia, then the specimen diagnoses were discordant.) On the other hand, if a specimen was read as benign, but the other tissue diagnosis was different in any way, it was considered as discordant.

Statistical Analysis

Statistical analyses were performed using the SAS, version 9.4 (SAS Institute Inc., Cary, NC, USA), and R, version 3.4 (R Development Core Team, 2018). Percent agreement and weighted Cohen’s kappa (κ) statistic, sensitivity, specificity, and positive (PPV) and negative (NPV) predictive values of TRS histopathologic tissue diagnosis against findings based on the routine processed tissue diagnosis (considered the gold standard) were calculated for Pa and Pb, respectively. 95% confidence intervals for sensitivity, specificity, PPV, and NPV were estimated. Inter-rater reliability between Pa and Pb was measured using the kappa (κ) statistic. The κ-value was interpreted with regard to reporting the reliability/strength of agreement as follows: poor, < 0.20; fair, 0.21–0.40; moderate, 0.41–0.60; substantial/good, 0.61–0.80; and almost perfect/very good, 0.81–1.00. Sample size was determined as a convenience sample.

Results

From May 2016 until March 2017, 23 subjects were recruited for the study. In total, 13 subjects were included with benign indications for hysterectomy for abnormal uterine bleeding (n = 4), pelvic pain (n = 2), cervical dysplasia (n = 2), endometriosis (n = 1), pelvic organ prolapse (n = 3), and family history-related prophylaxis (n = 1), and 10 subjects with malignant indications for hysterectomy including endometrial hyperplasia (n = 1), grade 1 adenocarcinoma (n = 5), grade 2 adenocarcinoma (n = 2), grade 3 adenocarcinoma (n = 1), and endometrial cancer pathology denoted as “other” (n = 1). Demographic characteristics are depicted in Table 1. The mean age, gravity, and parity of women with benign indications were 44.1 ± 11.5 (standard deviation: SD) years, 3.8 ± 2.0, and 3.2 ± 1.7 compared to 62.5 ± 4.9 years, 1.6 ± 1.3, and 1.4 ± 1.0 in the cancer group, respectively.

Table 1.

Patient demographics

Characteristic	Benign cases (n = 13)	Hyperplasia—malignant cases (n = 10)
Age (mean)	44.1 ± 11.5	62.5 ± 4.9
Race
White (n = 16)	56.3% (9)	43.8% (7)
Black (n = 6)	50% (3)	50% (3)
Hispanic (n = 1)	100% (1)	0% (0)
Gravity	3.8 ± 2.0	1.6 ± 1.3
Parity	3.2 ± 1.7	1.4 ± 1.0

± SD

A histopathologic diagnosis of TRS specimens was able to be given in 100% (23/23) by Pa and 91.3% (21/23; two with no tissue identified) by Pb. Depicted in Tables 2 and 3 are the comparisons of actual histologic diagnoses for both routine and TRS samples. The weighted kappa statistics for TRS specimen diagnoses by Pa and Pb, in comparison with actual histologic diagnoses, were 0.76 and 0.68, respectively. The sensitivity and specificity for Pa and Pb were 70%/100% and 80%/91%, depicted in Tables 2 and 3. The false-positive (overdiagnosed) and false-negative error rates (underdiagnosed) were 0%/30% for Pa and 9%/20% Pb with PPV and NPV of 100%/81% and 89%/83%, respectively. Specifically, Pa underdiagnosed one G1 legion as benign, and two G3 malignant as benign and G1 on TRS specimens, while Pb overdiagnosed one benign as hyperplasia and underdiagnosed all three G3 malignant lesions as hyperplasia and two G1 lesions on TRS specimens. Approximately 76.2% (κ = 0. 60) TRS specimen diagnoses were concordant between Pa and Pb.

Table 2.

Sensitivity, specificity, and degree of agreement between routine histologic diagnosis and TRS for Pa

TRS	Disease diagnosed by routine tissue diagnosis
	Benign	Hyperplasia	G1	G2	G3	Total
Benign	13	0	1	0	1	15
Hyperplasia	0	0	1	0	0	1
G1	0	0	3	2	1	6
G2	0	0	0	0	0	0
G3	0	0	0	0	1	1
Total	13	0	5	2	3	23

TRS consistent	Hyperplasia/cancer diagnosed by routine tissue diagnosis
	Yes	No	Total
Yes	7	0	7
No	3	13	16
Total	10	13	23

Sensitivity 0.70 (95% CI 0.35–0.93)

Specificity 1.0 (95% CI 0.75–1.0)

PPV 1.0 (95% CI 0.59–1.0)

NPV 0.81 (95% CI 0.54–0.96)

False-positive error rate 0.00

False-negative error rate 0.30

Table 3.

Sensitivity, specificity, and degree of agreement between routine histologic diagnosis and TRS for Pb

TRS	Disease diagnosed by routine tissue diagnosis
	Benign	Hyperplasia	G1	G2	G3	Total
Benign	10	0	0	0	0	10
Hyperplasia	1	0	1	0	1	3
G1	0	0	4	2	2	8
G2	0	0	0	0	0	0
G3	0	0	0	0	0	0
Total	11	0	5	2	3	21

TRS consistent	Hyperplasia/cancer diagnosed by routine tissue diagnosis
	Yes	No	Total
Yes	7	1	8
No	3	10	13
Total	10	11	21

Sensitivity 0.70 (95% CI 0.35–0.93)

Specificity 0.91 (95% CI 0.59–1.0)

PPV 0.88 (95% CI 0.47–1.0)

NPV 0.77 (95% CI 0.46–0.95)

False-positive error rate 0.09

False-negative error rate 0.30

Discussion

In summary, a histopathologic diagnosis of TRS specimens was confirmed in over 90% of samples by the reviewing pathologists. False-positive and false-negative error rates ranged between 0 and 30 and 9 and 20% for both pathologists. Overall, TRS specimen diagnoses were concordant between 86.9 and 80.9% for each pathologist. The agreement (κ) between Pa and routine histologic diagnosis was 0.76, and between Pb and routine histologic diagnosis was 0.68. Both kappa statistics indicate modest agreement between TRS and routine diagnosis.

Hysteroscopic TRS have increasingly been utilized for treating intrauterine pathology as they provide a simple, non-electrosurgical resection method where pathology can be resected and removed in a single step [8]. Furthermore, in randomized trials, TRS appear to be quicker to perform, more effective at completing polypectomy, less painful, and more acceptable to women than traditional electrosurgical resection [10].

Concerns have been raised with respect to histopathologic examination of morcellated specimens. To our knowledge, this is the first report of using an ex vivo model comparing standard tissue processing to TRS tissue specimen processing “of the same” benign and malignant intrauterine pathology. It appears that TRS may meaningfully affect the pathologist’s ability to read the tissue, though two samples had no tissue identified. Up to 30% of malignant specimens were overdiagnosed and up to 20% underdiagnosed after TRS, where in the latter, high-grade malignancies were read as benign and low-grade lesions. This supports the concern that TRS may alter tissue in such a way that a malignancy may be missed after being morcellated. Though it is possible that the tissue processing protocol may have resulted in specimens not representative of a higher-grade lesion, this would not explain why there were both under- and overdiagnosis of TRS processed tissue, which is suggestive of increased variability in interpretation.

We acknowledge that our study has limitations including the small sample size, and the comparison of TRS specimen may not necessarily be representative of the initial sample used for standard tissue processing and may underrepresent real pathology that may have been sampled in in vivo conditions. Furthermore, processing of samples was done more than 10 months after initiation of the study and possibly could have impacted on tissue integrity.

Further study is warranted to confirm our findings. Alerting pathologists of the potential to under- and overdiagnose morcellated specimens may be important for their workflows. Additionally, though this did not apply to this simulation, clinicians morcellating more than just the primary area of interest and consolidating it with the main specimen may further dilute pathologic samples. Hysteroscopists should consider subdividing specimens in a way that separates primary from secondary pathology.

Conclusion

Our findings reveal that up to 30% malignant specimens were underdiagnosed after TRS. Though further study is needed to confirm these findings, pathologists may need to reassess workflows so as to better offset potential underdiagnosis of malignant specimens where findings may be obscured through morcellation. Though not performed in this study, if clinicians are morcellating more than the primary area of interest, specimens may need to be subdivided in a manner distinguishing primary from secondary pathology.

Dr. Steven R. Lindheim

is a Professor of Obstetrics and Gynecology at Wright State University, Boonshoft School of Medicine, and Section Chief of Reproductive Endocrine and Infertility. His areas of interest include assisted reproduction, reproductive surgery, and oncofertility.

Compliance with Ethical Standards

Conflict of interest

The authors declare that they have no conflicts of interest and nothing to disclose. Hologic^® supported this study by providing disposable MyoSure hysteroscopic tissue removal systems and uterine model (Polly, Remedy©).

Disclosure

The authors have no financial disclosure to make.

Research Involves

Human subjects.

Informed Consent

Informed consent was obtained from all recruited subjects.