Clonality analysis of combined Brenner and mucinous tumours of the ovary reveals their monoclonal origin

Abstract

The derivation of ovarian intestinal-type mucinous tumours is not well established. Some are derived from teratomas but the origin of the majority is not clear. It has been recently proposed that the non-germ cell group may be derived from Brenner tumours, as the association of a mucinous tumour with a Brenner tumour is frequently observed. In order to explore the histogenesis of these neoplasms, we undertook a clonality analysis of the two components of ten combined Brenner and mucinous tumours using a human androgen receptor gene (HUMARA) assay. All eight informative cases of ten showed a concordant X-chromosome inactivation pattern between the two tumour components, indicative of a shared clonal origin (p = 0.0039). Microsatellite genotyping in five of the combined tumours displayed an identical heterozygous pattern with paired Fallopian tube tissue, indicative of a somatic cell origin. In addition, paired box protein 8, a highly sensitive Müllerian epithelial marker, was not detected by immunohistochemistry in either tumour component in any of the ten tumours, suggesting that this subset of mucinous tumours does not originate from Müllerian-derived epithelium. In conclusion, this study demonstrates that in combined mucinous and Brenner tumours, there is a shared clonal relationship between the two different tumour components and suggests that some pure mucinous tumours may develop from a Brenner tumour in which the Brenner tumour component becomes compressed and obliterated by an expanding mucinous neoplasm.

Introduction

In recent years, studies have shown that most epithelial ovarian carcinomas develop from Müllerian-derived precursors. The origin of the majority of mucinous tumours remains a mystery, although a small proportion of intestinal-type mucinous tumours develop from teratomas [1–4]. The origin of Brenner tumours is also not well established but in view of their resemblance to Walthard cell nests, it is believed that these are the precursor lesions. Brenner tumours are composed of nests of transitional-type epithelium, which frequently contain a central cavity lined by mucinous epithelium, and it has been proposed that outgrowth of the mucinous epithelium could result in the formation of a mucinous tumour. To test this hypothesis, we analysed the X-chromosome inactivation pattern of the two separate tumour components [5–11] in combined Brenner and mucinous tumours to see if they were clonally related.

Materials and methods

Tissue specimen collection

We retrieved formalin-fixed, paraffin-embedded (FFPE) tissue blocks from ten ovarian tumours that contained both a mucinous neoplasm and a Brenner tumour from The Johns Hopkins Hospital archives. Slides were reviewed by two pathologists (LES and RJK). The mucinous component of the ten tumours was an atypical proliferative mucinous tumour (APMT) in three instances and a mucinous cystadenoma (MC) in seven. Five ovarian mature teratomas were also analysed. The study was approved by the Institutional Review Board of The Johns Hopkins Hospital (NA_00089844, 08/2013-05/8/2015).

Laser capture microdissection, macrodissection, and DNA extraction

Twelve-micrometre-thick sections were obtained from FFPE tissue blocks of the tumours, placed on membrane slides (Carl Zeiss MicroImaging, Göttingen, Germany), and neoplastic epithelial cells were microdissected (Leica LMD7000; Leica Microsystems, Wetzlar, Germany). Ten-micrometre-thick sections of Fallopian tube tissue were obtained from the FFPE blocks as normal control. After 24 h of proteinase K digestion, genomic DNA was extracted using a QIAamp DNA FFPE Kit (Qiagen, Valencia, CA, USA).

HUMARA assay

The PCR thermocycling conditions and primers used for the HUMARA assay may be found in the Supplementary methods.

Five to ten microlitres of DNA from each tumour and control tissue was incubated for 16 h at 37 °C in a 25 μl reaction containing 10 U of the methylation-sensitive restriction enzyme HpaII (New England Biolabs, Ipswich, MA, USA). Each sample was simultaneously subjected to mock digestion in buffer lacking restriction enzyme. The reaction was terminated by incubation at 80 °C for 20 min, followed by purification using a QIAquick PCR Purification Kit (Qiagen). Both digested and mock-digested DNAs were PCR-amplified (Applied Biosystems, Foster City, CA, USA). Capillary electrophoresis of the PCR products was performed on an ABI 3130 XL Genetic Analyzer Sequencer (Applied Biosystems) and analysed using GeneMapper Software vs 5 (Applied Biosystems). Samples were informative when the undigested control tissue DNA showed two major peaks, representing two parental alleles, respectively. To compensate for possible PCR bias and constitutional X-chromosome inactivation (XCI) skewing, we calculated the cleavage ratio of control tissue and tumour (CR_ctrl, CR_tumour) and the corrected cleavage ratio of tumour (CR_cor) as previously described [6,8,10]. A case with CR_ctrl value that deviated significantly from 1 was considered as constitutional XCI skewing and excluded from the HUMARA analysis. A reproducible loss or marked reduction in the intensity of one of the two alleles in the digested tumour DNA, represented by CR_cor > 2 or < 0.5, was determined as predominance of one single allele and thus a monoclonal pattern. A matching XCI pattern in each pair indicates a 1 in 2 chance of a shared clone of two different tumour components, while a discrepancy represents different clone origins.

Microsatellite genotyping analysis

The PCR thermocycling conditions used for microsatellite genotyping may be found in the Supplementary methods.

Nine microsatellite loci (Figure 3) were tested in the five combined tumours and five mature cystic teratomas by using the AmpFlSTR Profiler kit (Applied Biosystems). After multiplex PCR amplification, 1 μl of PCR product was mixed with 9 μl of HiDi formamide/GeneScan 500 ROX size standard mixture and analysed on the ABI 3130xl Genetic Analyzer (Applied Biosystems). The heterozygous loci in the control tissue were considered informative. The homozygous loci in the control tissue or loci with insufficient amplicons were considered otherwise. The genotype of tumour was compared with that of normal control tissue at each locus.

Figure 3.

Microsatellite genotyping of combined mucinous and Brenner tumours at nine loci. BT = Brenner tumour; MT = mucinous tumour. Allelotypes of each tumour at nine tested short tandem repeat loci are indicated as coloured blocks, with white indicating uninformative loci, orange indicating heterozygous loci, and green indicating homozygous loci. Loci were considered homozygous when one allele was seen in tumour tissue, while two alleles were seen in control tissue. In contrast, loci were considered heterozygous when the two alleles in the tumour tissue matched two alleles in the control tissue. Uninformative loci included those with only one allele present in the normal control tissue or those that failed to amplify in PCR. No loci with one allele in normal control tissue but two alleles in tumour tissue were detected.

Immunohistochemistry

For paired box protein 8 (PAX8, rabbit polyclonal; Proteintech, Chicago, IL, USA), slides were stained as previously described [12]. Positive nuclear staining of secretory cells in Fallopian tube epithelium was used as a positive control.

Statistical analysis

Statistical analysis was performed using the paired t-test in GraphPad Prisma 5.0.

Results

Patient mean age was 57.6 years (range 50–90 years) (Table 1). The mucinous tumours (three APMTs and seven MCs) were all of intestinal type, one associated with pseudomyxoma ovarii (Figure 1). Mucinous epithelium lining the cystic space within the transitional nests of the Brenner tumour was present in all ten cases. The mean size of the mucinous tumours was 19.85 cm (range 8.5–41 cm) and the mean size of the Brenner tumours was 3.04cm (range 0.6–5.5 cm). The difference in size between the two tumour components was statistically significant (p < 0.01). None of the tumours was PAX8-positive (Figure 1).

Table 1. Clinical, histopathological, and clonal XCI findings of combined mucinous and Brenner tumours.

Case	Age (years)	Diagnosis	Site	Tumour size, cm (MT/BT)	CRcor/BT	CRcor/MT	Active allele
1	60	BT with APMT	LO	41/2	▲	▲	Short allele
2	68	BT with APMT	LO	20.5/3.5	▼	▼	Long allele
3	69	BT with APMT	LO	29/6	▼	▼	Long allele
4	85	BT with MC	LO	9/3	▲	▲	Short allele
5	55	BT with MC	LO	20/2	▲	▲	Short allele
6	59	BT with MC	RO	16/1.8	▼	▼	Long allele
7	52	BT with MC	LO	17.5/5.5	▼	▼	Long allele
8	50	BT with MC	LO	8.5/2.2	▲	▲	Short allele
9	57	BT with MC	LO	19/0.6	N/A	▼	Long allele
10	90	BT with MC	LO	18/3.8	N/A	N/A	N/A

MT = mucinous tumour; BT = Brenner tumour; APMT = atypical proliferative mucinous tumour; MC = mucinous cystadenoma; LO = left ovary; RO = right ovary; XCI = X-chromosome inactivation; CR_cor/MT = corrected cleavage ratio of mucinous tumour; CR_cor/BT = corrected cleavage ratio of Brenner tumour; N/A = not applicable. A triangle represents a CR_cor higher than 2, suggesting that the short allele was active; an inverted triangle represents a CR_cor lower than 0.5, suggesting the opposite. In case 9, CR_cor/MT was lower than the cut-off, revealing the monoclonal characteristic of the mucinous cystadenoma, while the co-existing Brenner tumour did not yield suffcient DNA to analyse. Constitutional skewed XCI found in the control tissue of case 10 suggested its unsuitability for HUMARA assay.

Figure 1.

Combined mucinous borderline tumour/atypical proliferative mucinous tumour and Brenner tumour. (A) A small Brenner tumour containing nests of transitional-type cells within the wall of the mucinous neoplasm. (B) Gastrointestinal-type mucinous epithelium lining the central cavity within the nests of transitional-type cells of the Brenner tumour. (D) Atypical mucinous glands. (E) Ruptured mucinous glands and extravasated acellular mucin in the ovarian stroma (pseudomyxoma ovarii). (C, F) Absence of paired box protein 8 (PAX8) immunostaining of a combined atypical proliferative mucinous (APMT) and Brenner tumour. Inset: Fallopian tube control showing positive PAX8 in secretory cells.

All ten tumours were heterozygous at the HUMARA locus. For cases 1–8, CR_ctrl was close to 1 (range 0.83–1.22), representing a balanced XCI pattern. All tumours showed a monoclonal pattern of XCI. Unanimous matching XCI patterns were identified when the XCI patterns were compared between two tumours in each pair (Figure 2). As a matching XCI pattern in one single pair indicates a 1 in 2 chance of a shared clone in the two different tumour components, eight pairs of matching XCI patterns would occur with a probability of p = (½)⁸ = 0.0039, indicating that the chance of two tumours being clonally unrelated is extremely unlikely. For case 9, DNA extracted from the Brenner tumour was insufficient for the analysis. For case 10, skewed XCI was found in the control tissue (CR_ctrl = 0.16) and this case was therefore excluded from the analysis (Table 1).

Figure 2.

HUMARA assay showing concordant XCI patterns in the two tumour components in a combined atypical proliferative mucinous tumour (APMT) and Brenner tumour. Fallopian tube control (A1), Brenner tumour (B1), and mucinous tumour (C1) were dissected to extract DNA. The electropherogram shows the fragment analysis of PCR products amplified from undigested and digested DNA of control tissue (A2, A3), Brenner tumour (B2, B3), and APMT (C2, C3). Two major peaks represent two alleles with different numbers of short tandem repeats at the HUMARA locus. After digestion, the control DNA retains a balanced XCI pattern (A3), while Brenner and mucinous tumour DNA (B3 and C3) display a matching pattern of preferential loss of the long alleles (black arrows).

Microsatellite genotyping of five combined tumours (Figure 3) showed no homozygous loci, and no discordant allelotypes between the two tumour components were detected. In contrast, pure isodisomy was detected in three out of five teratomas, among which teratoma 4 displayed isodisomy at all nine loci tested. In all the teratomas tested, 89% (25/28) of informative loci were homozygous; the mean number of homozygous loci was 5 (range 4–9). The number of homozygous loci between these two cohorts was significantly different (p < 0.01).

Discussion

By investigating the pattern of X-chromosome inactivation using the HUMARA assay [6–11], we found that the two distinct components in ten combined mucinous and Brenner tumours are clonally related. We speculate that the transitional epithelium of the Brenner tumour undergoes cell lineage reprogramming to mucinous epithelium through metaplasia [13]. The latter then proliferates and gives rise to a mucinous tumour. In our ten cases, the mucinous tumours always formed the dominant mass, which is consistent with the findings of Seidman and Khedmati [14]. Thus, it is conceivable that, in some instances, what appears to be a pure mucinous tumour may have arisen from a Brenner tumour that was compressed and obliterated by the expanding mucinous tumour. Our microsatellite genotyping confirmed that the combined tumours lacking a teratomatous component were of somatic and not germ cell origin by showing a heterozygous pattern in a variety of short tandem repeat loci, which is consistent with their paired Fallopian tube tissue somatic cell genotypes.

The pathogenesis of Brenner tumours is not well established, although it is generally believed that they arise from Walthard cell nests, which typically are found at the tuboperitoneal junction [14]. The link between Walthard cell nests and Brenner tumours is further supported by their shared immunoprofile, including expression of GATA3 and p63, and absence of germ cell markers [14–16].

The literature indicates that PAX8, a Müllerian marker, is present in up to 50% of mucinous tumours [12,17], but in our ten cases, as well as in eight cases of combined Brenner and mucinous tumours reported by Roma and Masand [18], PAX8 expression was not detected. Since mucinous tumours of germ cell origin are uncommon, PAX8 expression in roughly half of mucinous tumours implies that a significant proportion are of Müllerian origin. As noted above, it is our view that the precursor of Brenner and Brenner tumour-related mucinous tumours is not of Müllerian origin but rather from transitional metaplasia (Walthard cell nests). As transitional metaplasia is located at the tuboperitoneal junction, it is conceivable that, in this region, where Müllerian-derived epithelium (fimbria of the Fallopian tube) is in close, intimate contact with mesothelium (serosa of the Fallopian tube and ovarian surface epithelium), PAX8 expression could overlap in non-Müllerian as well as Müllerian-derived epithelium [19,20].

Can a mucinous tumour develop directly from transitional metaplasia? We suspect that a Brenner tumour is a requisite intermediate step in progression since mucinous epithelium is rarely associated with transitional metaplasia, whereas it is commonly found in Brenner tumours. In conclusion, this study, by demonstrating a clonal relationship between Brenner and mucinous tumours, supports the hypothesis that in addition to origin from a teratoma, at least some intestinal-type mucinous tumours may be derived from Brenner tumours. This conclusion could be further supported by future studies of tumour-specific genetic markers in this subset.