Proposal for histological classification of EcPV2-associated lesions
In horses SCC is reported to be the most common neoplasm of the external male genitalia [6, 21, 22]. It is suggested that the gross description plaques and wart-like lesions (papillomas) represent precursor lesions of this malignant tumour type [10, 21, 22]. However, this study shows, that plaques as well as wart-like lesions can already represent CIS i.e. malignant lesions preceding invasive SCCs and thus requiring more aggressive therapy. Therefore the histological analysis is very important in these lesions. However, gross and histological descriptions of these precursor lesions vary considerably in the literature [6, 10, 23-25]. When reassessing H&E-stained sections of equine penile lesions in this study, it was evident that the histological diagnosis varied among pathologists of the different institutions. This is likely due to the lack of a consistent and widely approved histological classification for EcPV2-associated equine penile lesions.
In humans, diagnosis of penile precancerous lesions and invasive carcinomas is likewise challenging [11]. Given that human penile precursor lesions and SCCs closely resemble their equine counterpart [7, 26], this study attempted to categorize EcPV2-associated equine penile lesions based on human data [9] and descriptions of equine disease [6, 10, 23-25] in order to obtain well defined histological classes regardless of their clinical appearance.
Thickening of the epithelium with orderly maturation accompanied by the formation of broad rete ridges corresponds to a (benign) hyperplasia as described by Chaux et al. [9]. In the equine literature the same features are sometimes attributed to papillomas in horses [6, 27]. Lesions with crowding of basal cells or small intraepithelial keratin pearls were included, because these alterations were considered to be dysplastic rather than neoplastic. Chaux et al [9] use the term differentiated penile intraepithelial neoplasia (PIN) for such changes (with crowding of basal cells and/or intra-epithelial keratin pearls), whereby the specification "differentiated" implies a still benign, presumably non-neoplastic character of the change, which is why these two groups of lesions (hyperplasia and differentiated PIN) were combined in the present study.
The neoplastic alteration CIS can be distinguished by abnormal transepithelial maturation, increased variability of keratinocytes and frequent presence of mitotic figures in the suprabasal region. This description aligns with previous descriptions of equine penile CIS [24, 28] as well as with Chaux et al [9] who refers to this change as a dedifferentiated PIN, indicating the increasingly malignant character. In the present study two CIS lesions had very small, solitary areas with a few cells focally disrupting the basement membrane, suspected to represent areas of microinvasion. This observation does not warrant a diagnosis of SCC, which should infiltrate profusely (over a large area, broadly) and not only in a very small area, but highlights the dynamic character and progressive potential of disease. To the best of our knowledge microinvasion is not mentioned in the equine literature on penile precursor and cancerous lesions. It is controversially discussed in regard to human cancerous lesions and rarely mentioned in the literature on penile carcinomas in man [9, 29, 30]. Chaux et al. [9] describes microinvasion in a differentiated PIN without then classifying the change as SCC.
Papillomas were diagnosed when finger-like projections of thickened, usually mildly hyperkeratotic squamous epithelium with normal maturation and thin central cores of connective tissue were present. This type of benign neoplastic lesion is not frequently mentioned in the literature and most closely corresponds to the papillomas described by Suarez-Bonnet et al. [26] and the fibropapillomas described by Lange et al. [24], whereas the papillomas described by Knight et al. [27] and van den Top et al. [6] more closely fit into our category of benign hyperplasia.
Lesions classified as SCCs contained infiltrative islands and trabeculae of squamous epithelial cells with or without keratinization originating from the epithelium and infiltrating profusely the underlying dermal tissue in accordance with van den Top, 2011 [10]. This study distinguishes between early invasive forms and late invasive forms, a distinction that is not normally found in the equine literature. Early invasive tumours (with limited infiltration of the sub-epithelial tissue) most likely represent an early stage of SCC. This information can be important for a clinician when deciding whether to choose a more aggressive or less aggressive therapeutic approach. SCCs were graded according to van den Top et al. [10] because prognosis is related to tumour grade. SCC can be heterogeneous and the poorest differentiated area determines the grade [20]. Most tumours of the present study were heterogeneous as described by others [10, 31] and consist of moderately as well as poorly differentiated areas and were therefore classified as grade 3. Interestingly, the only grade 2 SCC was described clinically as wart-like lesion, while the grade 3 SCCs were described clinically as tumoural lesions of larger sizes. However, since no follow-up information is available for the SCCs of this study and the number of cases is small, no further interpretation can be made.
Aetiological association of EcPV2 with equine penile SCCs and precursor lesions based on PCR and RISH
As anticipated on the basis of previous reports reviewed by Sykora et al [4] all lesions tested positive for EcPV2 DNA as revealed by virus type-specific PCR. Since there are other papillomaviruses described to be associated with equine genital lesions, and co-infections with other PVs were reported[32-34], also broad-range PV PCR using CP4/5 [35] and Fap 59/64 [36] primer sets was done to detect other potentially involved PVs. These primers are designed to also detect the equine PVs type 3-8 [32-34, 37] and the sarcoid associated bovine PVs type 1+2 [38]. Using these two broad-range PV primer sets, no other PVs were detectable, hence it is not likely that other PVs were involved in the development of the lesions assessed in this study. However, due to the lower sensitivity of broad-range primers compared to specific primers and the potential involvement of papillomaviruses which are not detectable by this primer sets, it cannot be excluded that other PVs play a role in some of these lesions. Nevertheless, it appears that EcPV2 is one of the main factors involved in the development of equine genital lesions, as it was detectable in all equine genital lesions assessed in this study.
Lesions were assessed also by EcPV2 E6/E7 RNA in situ hybridization (RISH) [39] to detect and localize viral nucleic acids. Of note, EcPV2 RISH recognizes both E6/E7 DNA and mRNA, possibly explaining the two different types of signal (GS and DNS) yielded by the method. Interestingly, RISH signal patterns analogous to those observed here in the EcPV2-associated equine genital lesions have been detected with hrHPV-associated neoplasms as well [40, 41]. Previous studies support the concept that DNS can be attributed to the presence of both papillomaviral RNA and DNA [12, 41]. One study proposed that DNS may result from probe hybridization with ‘unzipped’ single-stranded HPV, which occurs during episome synthesis in the productive phase of the HPV life-cycle. In that study, the authors proposed that RISH could be used for grading HPV-associated cervical intraepithelial neoplasia (CIN1-3) lesions, as CIN1 lesions exhibit abundant DNS, while CIN3 lesions exhibit little to no DNS in the transformative phase [12]. Similarly, in the present equine study benign lesions had a high level of DNS, suggestive of a productive EcPV2 life cycle, while CIS and SCC had a low level of DNS, as expected for transformative processes.
Importantly, histological reassessment of some lesions, which have previously been macroscopically interpreted as (benign) plaques, led to a diagnosis of CIS. The abundance of DNS helped distinguishing these malignant lesions from benign hyperplasia and is thus proposed as an adjuvant diagnostic marker. These findings also make clear that plaques do not necessarily represent early benign lesions and thus require careful diagnosis and corresponding treatment.
In this study 100% of lesions exhibited GS in accordance with EcPV2 PCR results. The perilesional normal skin surrounding the assessed lesions, if present on the same slide, was negative for EcPV2 RISH. It is known that EcPV2-DNA can also be detected in healthy horses with a prevalence between 1-18% in different studies [22, 42, 43], and that subclinical infections also might get cleared by the immune system, as up to 15-36% of healthy horses have antibodies against EcPV2 as well [42, 44]. However, looking at the results obtained in this study, it seems that if the horses develop lesions, this occurs exclusively in EcPV2 infected and transcriptionally active areas but not in EcPV2 negative areas. This finding further supports the concept that EcPV2 infection has an active role in tumour onset and progression [4].
In summary, RISH can be used to detect and localize EcPV2 RNA/DNA in FFPE tissue samples. The different signal distribution patterns and amount of DNS might help to distinguish between early proliferative and late progressive tumour stages, as the amount of DNS significantly differed between benign and malignant lesions.
IHC-based marker gene expression in different lesions
P53 has a critical role in cell cycle regulation and is involved in cell cycle arrest, DNA repair and apoptosis [45]. In approximately 50% of human tumours, the p53 gene is mutated. Mutations within the p53 gene frequently compromise the tumour-suppressive activity of the corresponding protein, thus promoting tumour onset and progression [13]. In contrast to wild-type p53, which is an unstable protein with a short half-life, mutant p53 can accumulate within tumour cells, thus representing a potential tumour marker [46]. P53 immunopositivity has also been shown to strongly correlate with p53 mutational status [47]. However, anti-p53 antibody usually recognizes both accumulated wild-type and mutated forms of the protein [46]. This fact explains why the use of p53 as human cervical cancer marker is controversial [13]. In human penile SCCs however, strong nuclear p53 immunostaining has been consistently detected, in contrast to adjacent normal perilesional skin, which has a moderate to weak signal mainly confined to the basal layer [48]. In addition, p53 immunostaining can be used to distinguish human penile intraepithelial lesions from squamous hyperplasia [49].
However, p53 expression levels in human penile carcinomas have no prognostic value [50]. In an equine study on penile lesions it was already shown, that expression of p53 increases with decreased differentiation of the tumour, but could not be used as prognostic factor for metastatic disease [20]. In the present equine study, lesional and histologically normal perilesional tissue were positive for p53. Given that, to date, equine p53 has not been assessed for mutations in the context of penile tumour disease, it is unclear whether wild-type and/or mutant p53 are detected by IHC. Nonetheless, p53 expression by basal cells of most SCCs, CIS and papillomas (>90%), markedly differed from that of most hyperplasia and normal perilesional tissue (<50%). Within the non-basal layers, <10% of cells were positive for p53 in most perilesional, hyperplastic and papilloma tissues, while distinctly more non-basal cells expressed p53 in CIS and SCCs (10-90%). Consequently, assessing the number and locations of p53-positive cells within equine penile lesions might help to distinguish between benign and malignant lesions.
Ki67 is widely used in routine tumour assessment as a prognostic and predictive indicator [51]. In human cervical lesions, Ki67-staining is used for the grading of progressive cervical tumours [13]. In human penile cancer, Ki67 expression is correlated with tumour grade, although it has no prognostic value [52]. Based on this, the hypothesis of this study was that Ki67 may represent a potential diagnostic marker in horses. However, Ki67 staining of equine penile lesions yielded inconsistent results, with two SCCs appearing almost negative for this proliferation marker. Standardized tissue sample handling may be an important prerequisite for accurate assessment of the ki67 index. Using the same ki67 antibody, it was shown that insufficient, delayed and prolonged fixation can have a negative effect on ki67 antibody binding [53]. The two negative/weak SCC samples were received from Germany and had both a prolonged fixation time, which might explain these results. When collecting horse samples from different regions, these procedures cannot always be standardized. Consequently, Ki67-staining cannot be recommended for diagnosis and lesion assignment of EcPV2-associated equine penile lesions. This is also supported by another study using this antibody in equine penile and preputial SCCs whereby they could also not find expression differences between different subtypes [20]. The proliferation marker MCM7 is more sensitive in detecting cycling cells than ki67 in humans and dogs and therefor might be superior to ki67 as a proliferation marker [54-56].
High expression of MCM7 can serve as a predictive biomarker for poor prognosis in human cancers [57]. In hrHPV-induced high-grade lesions, MCMs are over-expressed and serve as established surrogate markers for papillomaviral E6/E7 oncoprotein activity [14]. In particular MCM7 constitutes a reliable human cervical cancer marker. In normal and benign hyperplastic cervical tissue, MCM7 expression is restricted to basal and immediate para-basal layers, while full thickness immunostaining is observed in high grade cervical dysplasia and invasive cancers [58]. In the present study, lesional as well as normal perilesional basal keratinocytes expressed MCM7, as expected on the basis of human data. In non-basal keratinocyte layers, MCM7 expression was most often increased in papillomas, CIS and SCCs, depending on the degree of differentiation within the lesions. In non-differentiated parts of the lesions, cells appeared to be still proliferating and express MCM7 in higher levels, than in more differentiated parts of the lesion. In human cervical cancer, MCM expression is thought to depend on hrHPV E7 expression, which results in the inactivation of pRB and the related proteins p107 and p130. This in turn activates the E2F family of transcription factors and thereby induces expression of E2F-responsive genes such as MCMs [58]. Interestingly, the EcPV2 E7 open reading frame lacks a pRB binding domain [3, 4]. Hence, the mechanisms by which EcPV2 infection promotes MCM7 expression by non-basal cells remains unclear. Importantly, it has been shown that HPV16 can likewise induce E2F responsive genes - especially MCM7 - in the absence of E7 in reproductive epithelia and tumours [59]. In analogy, it can be speculated that MCM7 expression in upper keratinocyte layers of progressive penile lesions may be driven by EcPV2 E6. The RISH data generated in the present equine study do not fully support this possibility, as MCM7 expression and signals obtained by E6/E7 RISH did not always colocalize. In humans, MCM7 expression is also seen in HPV-negative penile cancers, suggesting that MCM7 expression may depend on the amount of proliferating cells within the lesions and not entirely on the papillomaviral oncoproteins E6 and E7 [60]. Gene expression profiling of EcPV2-associated equine genital cancer has shown a strong up-regulation of MCM transcripts [8]. In the present study, this finding has been confirmed on the protein level. Evaluation of MCM7 expression in upper epithelial layers may therefore help to distinguish well-differentiated equine penile lesions from less-differentiated, more proliferative lesions.
MMP1 transcription was recently shown to be upregulated in EcPV2-associated genital SCCs [8], and over-expression of this proteinase has been previously demonstrated to correlate with progression of bovine PV-induced equine sarcoids [17, 18]. However, in the present study MMP1-immunostaining of equine penile lesions yielded a diffuse cytoplasmic signal throughout the different cell layers of all samples, disqualifying MMP1 IHC or the staining protocol applied in the present study as a useful technique for investigation of EcPV2-associated penile lesions.