Cytomorphologic and molecular analyses of fallopian tube fimbrial brushings for diagnosis of serous tubal intraepithelial carcinoma

Abstract

Background:

The paradigm shift localizing the origin of ovarian high-grade serous carcinoma (HGSC) to the fimbriated end of the fallopian tube underscores the rationale for meticulous microscopic examination of salpingectomy specimens. The precursor, termed “serous intraepithelial carcinoma” (STIC), is often a focal lesion, which poses difficulties for histologic diagnosis.

Methods:

We describe a method to examine exfoliated epithelial cells from fallopian tube fimbria by gentle brushing, thereby enabling thorough sampling of the mucosal surface. Fimbrial brushings were collected from 20 fresh salpingectomy specimens from 15 patients, including 5 with pathologically-confirmed ovarian HGSC. Samples, taken only from tubes grossly negative for tumor, were processed for Papanicolaou staining, p53 immunocytochemistry, and TP53 mutation analysis.

Results:

Cells with malignant cytomorphologic features were found only in tubal brushings from patients with ovarian HGSC. In all cases, atypical/malignant cells on cytology corresponded to lesions with similar morphology and immunostaining pattern in permanent sections, demonstrating the sensitivity of the technique, while providing reassurance that specimen integrity is not disrupted by the procedure. Targeted next-generation sequencing confirmed the presence of TP53 mutations in fimbrial brushings from HGSC, but not benign cases, and demonstrated concordance with the immunostaining pattern. Identical mutations were found in matched lesions microdissected from formalin-fixed tissue sections.

Conclusions:

The described technique enables cytologic evaluation of the fallopian tube fimbria for diagnosis of STIC, serving as a complement to histology, while offering distinct advantages with respect to procurement of cellular material for ancillary testing and research.

INTRODUCTION

Histomorphologic and molecular studies have implicated the distal end of the fallopian tube as the site of origin for most ovarian high-grade serous carcinomas (HGSC)^1–3. The precursor lesion, coined “serous tubal intraepithelial carcinoma (STIC),” is characterized by marked cytologic atypia and underlying TP53 mutation. In cases without overt malignancy, and in particular for high-risk patients undergoing prophylactic surgeries (e.g. BRCA-mutation carriers), the diagnosis of STIC has important clinical implications⁴.

With increasing recognition of this entity, the “SEE-FIM” protocol (Sectioning and Extensive Examination of the FIMbriated end of the fallopian tube) has become the standard-of-practice for grossing salpingectomy specimens. As STIC lesions are most often located at the distal end of the fallopian tube, the principle of the technique is to maximize examination of the fimbria by longitudinal sectioning at 2- to 3-mm intervals, and submitting all pieces in toto for paraffin embedding⁵.

Following adoption of this protocol, STICs were reportedly found in up to 50% of pelvic HGSCs¹ and in only 3–5% of prophylactic salpingo-oophorectomy specimens^5,6. These incidence rates are much lower than expected, if we accept the fallopian tube secretory epithelial cell as the cell-of-origin of HGSC⁷. Small lesions are likely to be missed during routine histologic examination of a single level by H&E. Since a 4 μm-thick section amounts to only 0.2% of a piece of tissue 2-mm in thickness, it is easy to imagine that small STICs may be present in the 99.8% of unexamined tissue within the paraffin block. In one study, examination of multiple deeper levels revealed STIC that was negative on the initial H&E stain, and it is estimated that up to 25% of STICs are missed when only a single H&E is examined⁸. In a subsequent study, detection rates for p53 signatures and STIC doubled when extensive histologic examination was performed on prophylactic high-risk salpingectomy specimens⁹. However, such a laborious approach would be impractical to adopt into routine practice. An improved detection method is therefore needed to facilitate and enhance the quality of future clinicopathologic and molecular correlative studies, for both understanding the biology of ovarian cancer initiation and determining the clinical significance of STIC in different risk groups^10,11.

The aim of this study is to overcome the technical limitations inherent to the current histopathologic method of diagnosing STICs by applying an exfoliative sampling strategy analogous to the cervical Pap test. We propose a novel and simple method for random sampling of epithelial cells lining the tubal fimbria for liquid-based cytologic evaluation (Figure 1) and describe the cytomorphologic features of normal and neoplastic tubal epithelial cells. Importantly, the technique does not cause structural alterations to the specimen, provides abundant material for ancillary testing (i.e. immunostaining and molecular analyses) and thus can complement conventional histology to improve detection of precursor lesions in salpingectomy specimens.

Figure 1:

Schematic illustrating the rationale for brush sampling to detect occult tubal precursor lesions.

MATERIALS AND METHODS

Case Selection

Human investigations were performed after approval by the institutional review board and informed consent was obtained from all study participants. In this feasibility study, samples were collected from patients undergoing gynecologic surgeries that included salpingectomy as part of the procedure. High-risk patients (e.g. BRCA mutation carriers) undergoing prophylactic surgeries were not eligible. Fresh surgical specimens were examined to locate the fallopian tube fimbria. Cases with gross tumor involvement at or near the tubal fimbria were excluded.

Specimen Collection and Liquid-Based Cytology

Epithelial cells were collected from the fimbria (in the fresh-state) using an electronic toothbrush, repurposed as a Tubebrush©. Single-use disposable brush heads were sterilized by autoclaving prior to the procedure. After rinsing the distal portion of the fallopian tube in phosphate-buffered saline to wash off blood, the rotating brush was guided across the surface of the fimbria, while stabilizing the specimen at the ampullary region using forceps. The brush was then submerged in a collection vial containing fixative solution (SurePath™) to release the collected cells. Cells were centrifuged (x300 g for 5 min) and re-suspended in 1.5 mL of fixative, from which 200 uL was aliquoted for DNA extraction. The remaining sample was used to prepare two slides by SurePath liquid-based methodology; one slide for Papanicolaou stain and one for p53 immunoperoxidase staining.

Slide Review

Cytologic slides were initially reviewed blindly by a board-certified cytopathologist (C.J.V.) and then re-reviewed with knowledge of the histologic diagnosis (M.H.C., I-M.S., C.J.V.) under a multi-headed microscope. Surgical pathology slides were also reviewed for each case to correlate cytology with histologic findings. Because fallopian tube exfoliative cytology is not well-established, this initial cohort serves as an exploratory “training set”, which we used to describe the pertinent cytomorphologic features associated with each histologic diagnosis, and can serve as a reference in future studies.

p53 Immunoperoxidase Staining

Immunostaining was performed on unstained cytology slides and formalin-fixed paraffin-embedded (FFPE) tissue sections with the automated XT iVIEW DAB V.1 procedure on the BenchMark XT IHC/ISH Staining Module, Ventana with anti-p53 (clone Bp53–11, prediluted, Ventana). Antigen retrieval was carried out with CC1 (Ventana). Sections were incubated with primary antibodies for 16 min at 37°C. Staining was detected with I-View DAB detection system.

TP53 Mutation Analysis

DNA was extracted from cytology samples fixed in SurePath fixative and corresponding tissues manually microdissected from unstained 10 μm-thick FFPE tissue sections. DNA was extracted using the Qiagen DNeasy kit. The Accel-Amplicon Comprehensive TP53 Panel (Swift Biosciences), which covers the full-length coding region of the TP53 gene, was used for library preparation, following the manufacturers’ instructions, with 10 ng of qPCR-quantified DNA as input. Paired-end sequencing (150 bp) of pooled libraries (4nM) was performed using the MiSeq (Illumina) using v2 chemistry, targeting an average coverage of 5000X for a lower limit of detection of 1%. Bioinformatics analysis was performed using an automated pipeline developed specifically for this assay (Genialis, Inc).

RESULTS

Specimen Characteristics

Fimbrial brushings were collected from fresh salpingectomy specimens from 15 patients, 7 of which had either a known history of HGSC or a clinically-suspected malignant ovarian neoplasm at the time of surgery (Table 1). All 20 brush samples were taken from tubal fimbriae grossly uninvolved by tumor. The 7 cases of suspected ovarian malignancy were confirmed by histology to be HGSC (n = 5) and metastatic mucinous carcinoma (n = 2).

Table 1:

Summary of cytologic and histologic findings

Case ID	MAIN DIAGNOSIS	FALLOPIAN TUBE FINDINGS
		Laterality	Brush Cytology	Tissue Diagnosis
1	Endometrial hyperplasia	Left	Benign	Neg
2	Endometrial carcinoma	Left	Benign	Neg
		Right	Benign	Neg
3	Ovarian serous cystadenofibromas	Right	Benign, mild atypia	Neg
4	Ovarian serous cystadenomas	#1	Benign	Neg
		#2	Benign	Neg
5	Endometrial carcinoma (serous)	Left	Atypical cells present	Neg (reactive)
		Right	Atypical cells present	Neg (reactive)
6	Ovarian APST (left)	Left	Benign	Neg
7	Ovarian fibroma (left)	Left	Benign	Neg
8	Ovarian fibroma (right)	Right	Benign	Neg
9	Metastatic mucinous carcinoma		Benign, reactive atypia	Neg (reactive)
10	Metastatic mucinous carcinoma		Benign, reactive atypia	Neg (reactive)
11	Ovarian HGSC, bilateral*	Right	Malignant cells	HGSC (serosal)
12	Ovarian HGSC (post-chemo)	Left	Malignant cells	HGSC (mucosal)
		Right	Neg	Neg
13	Ovarian HGSC, bilateral	Left	Malignant cells	STIC/HGSC
		Right	Malignant cells	HGSC (serosal)
14	Ovarian HGSC –postchemo	Left	Neg	Neg
15	Ovarian HGSC	Left	Malignant cells	STIC

^*Brush sampling performed only on right fallopian tube, as left tube was grossly involved by tumor.

Cytologic preparations of the specimens showed variable cellularity. Highly cellular samples were typically associated with a granular background with abundant naked nuclei, likely a result of cell lysis from over-vigorous brushing. For all cases, review of H&E-stained fallopian tube tissue sections confirmed the presence of intact epithelium, indicating that the brushing method did not interfere with routine histologic diagnosis.

Cytomorphologic Features

Specimens were comprised of a mixture of cells morphologically compatible with ciliated and non-ciliated (secretory) tubal epithelial cells, mesothelial cells and lymphocytes in variable proportions. Cells were distributed in small clusters, large groups and single cells. Normal tubal epithelial cells were typically cuboidal to tall columnar, with round or oval nuclei showing finely dispersed chromatin, and round small nucleoli (Figure 2A–C). It was not uncommon to find clusters of cells with enlarged nuclei and small, but prominent, cherry-red nucleoli - features interpreted as reactive atypia - in specimens from patients without ovarian malignancy (Figure 3A,B).

Figure 2:

Tubal brushings from benign fallopian tubes. A. Tissue fragment containing crowded cells with round-to-oval shaped nuclei, regular nuclear contours, and bland chromatin. While cilia are not detected, the fragment appears to contain a terminal bar (left side of field). B. Fragment of ciliated cells with minimal cytoplasm and a mild amount of nuclear border irregularities and anisonucleosis. The chromatin is bland and contains a few small chromocenters. C. A loosely-cohesive fragment of ciliated cells with similar nuclear characteristics as seen in A and B. Several cells have a columnar shape. (A-C. 60X).

Figure 3:

Tubal epithelial cells with reactive atypia. A. Fimbrial brushing specimen showing a group of cells with crowded, monotonous, round nuclei with small prominent nucleoli. Note scattered “stretched-out” nuclei, likely derived from “intercalated cells” (60X). B. Histologic section of the same case of an area showing similar cytologic features (60X).

Brushings taken from cases with STIC or microscopic tubal involvement by HGSC (confirmed on histologic sections) were characterized by cytomorphologic features of malignancy (Figures 4A–F). Compared to the orderly arrangement typical of benign cell clusters, hyperchromatic crowded groups lacked cilia and were composed of a haphazard arrangement of nuclei. Tumor cells were also present as dispersed single cells. Classic morphologic features of high-grade carcinoma were appreciated, including enlarged cell size, increased nuclear-cytoplasmic ratio, marked nuclear pleomorphism, clumped chromatin, and large, prominent nucleoli. Mitotic figures and apoptotic bodies were features supportive of a malignant diagnosis (Figure 4C). Rare cell clusters comprised of malignant cells juxtaposed with normal epithelial cells, was interpreted as evidence of a mucosal-based lesion, as seen in Case #15, in which a STIC lesion was confirmed histologically (Figures 4F).

Figure 4:

Fimbrial brush cytology specimens from patients with ovarian HGSC (Papanicolaou stain). A. Hyperchromatic crowded group with haphazardly arranged nuclei. Compare with adjacent mesothelial-like sheet of normal cells (Case #13, 20X); B. In another area of the same case, enlarged cells with crowded, pleomorphic nuclei and prominent nucleoli (60X). C. Relatively smaller, densely packed cells with high nuclear-cytoplasmic ratio, and coarse chromatin. Note the mitotic figure with a lagging chromosome (arrowhead) (Case #11, 40X). D. Small clusters of markedly atypical cells. Compare with adjacent normal ciliated cells (Case #12, 60X). E. Elongated, hyperchromatic nuclei with multiple nucleoli, with adjacent benign with features of transitional metaplasia (Case #15, 60X). F. Admixture of transformed and normal cells, suggesting a mucosal based lesion (Case #15, 40X). STIC was found in corresponding H&E - see Figure 5D.

p53 Immunoperoxidase Staining

Immunocytochemistry for p53 performed directly on cytospin specimens demonstrated concordance with staining patterns seen on corresponding histologic sections (Figure 5A–F, Table 2). Normal tubal epithelial cells showed weak heterogeneous nuclear staining for p53. In one case, a single small cluster of p53-positive cells were identified (Figure 5A). Immunohistochemical staining on the matching tissue section revealed a single linear stretch of morphologically unremarkable cells with aberrant p53 expression, consistent with the so-called “p53-signature” (Figure 5B). In STIC/HGSC cases, diffuse nuclear staining for p53 was observed in 2 cases, while two other cases exhibited the p53-null pattern (Figure 5C–F). Of note, in Case #13, malignant cells, barely visible on initial H&E, emerged on the p53 stain (Figure 5F), while they were readily identified in all cytologic preparations (Figures 4A,B, 5E).

Figure 5:

p53 Immunocytochemistry of brush cytology specimens correlates with the p53 immunohistochemical staining pattern seen on corresponding tissue sections. A, B. Nuclear p53 staining in an isolated cluster of normal tubal epithelial cells, consistent with the histologic finding of a focus of “p53-signature” (Case #9, A. brush cytology, p53 stain, 60X; B. matched tissue, p53 stain, 20X). C, D. STIC with p53-null staining pattern, consistent with a nonsense mutation confirmed by sequencing (Case #15, C. brush cytology, p53 stain, 60X; D. matched tissue, H&E stain, 20X, inset - p53 stain, 20X). E, F. Microscopic tubal involvement by STIC/HGSC with diffuse nuclear staining, consistent with a missense mutation confirmed by sequencing (Case #13, E. brush cytology, p53 stain, 20X; F. matched tissue, H&E stain, 10X, inset - p53 stain, 10X).

Table 2:

Results of p53 immunostaining and sequencing analysis

Case ID	p53 STAINING PATTERNa		TP53 MUTATION (ALLELIC FREQUENCY)b
	Brush Cyto	Tissue	Brush Cyto	Tissue
1	wt	wt	No Mutation	-
2	wt	wt	No Mutation	No Mutation
	wt	wt	No Mutation	-
3	wt	wt	-	-
4	wt	wt	No Mutation	No Mutation
	wt	wt	No Mutation	No Mutation
5	wt	wt	-	-
	wt	wt	-	-
6	wt	wt	-	-
7	wt	wt	-	-
8	wt	wt	-	-
9	Focal MutD	Focal MutD	No Mutation	No Mutation
10	wt	wt	No Mutation	-
11	MutD	MutD	p.R175H (0.81)	p.R175H (0.45)
12	MutN	MutN	c.559+1G>T, p.? (0.11)c	c.559+1G>T, p.? (0.25)e
	wt	wt	No Mutation	No Mutation
13	MutD	MutD	p.D281N (0.32)	p.D281N (0.44)
	MutD	MutD	p.D281N (0.17)	p.D281N (0.16)
14	wt	wt	-	-
15	MutN	MutN	p.E271* (0.34)	p.E271* (0.56)

^a.Staining pattern is reported for the lesion, if present; otherwise, refers to background normal epithelial cells. Patterns of p53 immunoreactivity: wt – wildype, MutD – Mutated, diffuse nuclear pattern, MutN – Mutated, null pattern.

^b.Only non-synonymous variants shown.‘ – ’ indicates sequencing was not performed.

^c.This mutation has been reported to affect splicing, resulting in loss of gene expression.

TP53 Mutational Analysis

Targeted deep sequencing of all coding exons of TP53 was performed on all samples with malignant cytology and a random subset of benign tubal brushings. Focal lesions identified histologically were manual microdissected from unstained sections, while no further enrichment was performed for brush cytology specimens. With the only exception being the case with the “p53-signature,” in which no mutation was detected, sequencing results were consistent with the immunostaining pattern in all cases, with known pathogenic missense mutations corresponding with diffuse and intense nuclear p53 staining, and deleterious mutations (nonsense/frameshift) or splice site mutations corresponding with the p53-null staining pattern. For matched cytology and tissue specimens, identical mutations were observed in both sample types, at comparable allelic frequencies (Table 2).

DISCUSSION

Since the widespread implementation of the SEE-FIM protocol, meticulous histologic examination of the fallopian tube submitted in toto to identify STIC lesions has become standard routine for all gynecologic resection specimens. Even so, histologic confirmation of a tubal precursor in cases of ovarian HGSC is lower than expected, which raises concern that lesions may potentially be missed in salpingectomies that are not associated with overt malignancy. In this study, we developed a novel technique for collecting epithelial cells from the tubal fimbria, demonstrating for the first time, the feasibility of cytologic diagnosis of STIC or tubal involvement by HGSC, aided by p53 immunocytochemistry and targeted next-generation sequencing.

Dudkiewicz provided original morphologic descriptions of tubal cytology in 1968, from the evaluation of luminal aspirates obtained after salpingectomy, and proposed up to nine different types of cells¹². For practical purposes, we believe that classification of tubal epithelial cells as either secretory or ciliated cells may suffice. Presently, the most likely scenario in which tubal epithelial cells are encountered by the cytopathologist is in the setting of peritoneal fluid/pelvic washing specimens. While the same cell types are identified in endosalpingiosis, their architectural arrangement differs from samples directly obtained from the fallopian tube. Whereas the latter are typically composed of single cells and cell clusters, the presence of endosalpingiosis in peritoneal fluid specimens is characterized by orderly arrangement of epithelial cells within small fragments, occasionally forming tubular and small papillary structures with associated psammoma bodies¹³.

In the recent literature, there are only a few published reports of tubal brush cytology. Rodriguez et al obtained tubal brushings laparoscopically from patients undergoing surgery for benign conditions, with the goal of developing a screening method for HGSC¹⁴. Photomicrographs from their study illustrate typical clusters of epithelial cells with varying degrees of nuclear overlapping, pleomorphism and hyperchromasia, and cilia at the periphery, which are similar to our observations.

In a subsequent report, Chen et al. described a method using a cytobrush for sample collection, which was applied in a study cohort that included specimens with STIC¹⁵. In all cases with histologically-confirmed STIC or carcinoma, cells with malignant nuclear features were detected by cytology. The presence of 3-dimensional cell clusters and prominent cherry-red nucleoli were considered the most diagnostic features.

We have initially attempted to use a cytobrush, as previously described^15,16, but found the method causes avulsion of large tissue fragments. In the previous study, two cases had malignant cells in the brush specimen but not on histology, likely due to the removal of all lesional tissue from the specimen by the destructive procedure. Recognizing that a commonly observed feature of STIC is the exfoliation of cells from the mucosal surface¹⁷, presumably due to weakened cell-matrix interactions in transformed cells, we found that gentle brushing of the fimbria is sufficient to preferentially detach STIC/HGSC cells, which were often enriched in the collected sample over normal cells. With this method, sufficient normal and lesional epithelium remained on the surgical specimen to allow accurate tissue diagnosis. In all cases, atypical cells found on cytology correlated with lesions exhibiting similar morphologic features on histology.

In surgical pathology, a dilemma that is not uncommonly encountered in the diagnostic workup of focal epithelial atypia identified on H&E is depletion of the lesion in subsequent immunohistochemical stains. With brush sampling, lesional and normal cells are dispersed throughout the specimen, therefore, allowing for better distribution across multiple slides. While this improves the ability to perform ancillary studies on focal lesions, an inherent limitation of cytologic evaluation is the inability to assess tissue architecture. As such, it is unable to differentiate between STIC and tubal involvement by HGSC. However, this only becomes a relevant issue in cases presenting with an ovarian mass, for which the finding of STIC carries less clinical importance.

Since our main objective was to demonstrate feasibility of the tubal brushing technique, the small sample size and the assessment of cytology slides with a priori knowledge of the histologic diagnosis should be acknowledged as limitations of this study. Using standardized protocols described in the current work, future studies will involve larger cohorts and address issues of inter-observer reproducibility and performance characteristics (i.e. sensitivity and specificity) of tubal brushing cytology as a diagnostic tool for detecting STIC.

Exfoliative brush sampling will likely have an important role in procuring material for molecular genetic testing and research. Isolating STIC from tissue sections is an expensive and time-consuming process, necessitating careful laser-capture microdissection. Brushing specimens circumvents these issues, as there is preferential enrichment of transformed cells given their propensity to detach, and immediate preservation in ethanol-based fixatives is more compatible with downstream molecular analysis compared to formalin. We demonstrated the feasibility of performing next-generation sequencing to detect TP53 mutations in cytology samples, which were observed at high allelic frequencies, without having to perform special isolation procedures to separate tumor cells from normal cells.

Tubal cytology will likely be most impactful in screening prophylactic specimens from high-risk patients. With a more reliable method in diagnosis of STIC using tubal brushings, investigators will be able to determine the clinical and biological significance of STIC and predict the development of peritoneal serous carcinoma in the future. The potential advantages of the technique are summarized in Table 2.

By analogy, in screening for cervical carcinoma, cases of discrepant diagnoses rendered by cytologic evaluation of cells obtained by cervical brushing compared to tissue biopsy are not uncommon. One major explanation accounting for the discordance is the limited tissue sampling of biopsy taken during colposcopic procedures.¹⁸ The fact that the lesion is found upon repeat biopsy demonstrates the superior sensitivity of cervical cytology as a screening tool. Moreover, increased sensitivity is achieved by combining cytology with molecular methods to detect HPV. Co-testing identified 82.6% of women preceding the diagnosis of cervical malignancy, compared to 59.1% when relying on cytology alone¹⁹.

Similar to cervical screening, the optimal workflow for STIC detection should incorporate complementary techniques to maximize the sensitivity of detection and appropriate classification of the lesion. Future studies are planned to evaluate the effectiveness of brush cytology in identifying STIC in prophylactic salpingectomies from high-risk women.