Patient and tissue samples

This retrospective study has been approved by the Review Board at Faculdade Fátima on the advice number 183.756 and all patients have given written informed consent. Women seen at the Lower Genital Tract Pathology Service of the Central Outpatient Clinic (UCS) and at the Diagnosis Laboratory were included in this study. This outpatient clinic is a secondary referral unit for women sent by primary health care units in the region due to Pap smear abnormal results.

The study included patients with abnormal conventional Pap smear and subsequent cervical biopsy under colposcopic guidance on suspicion of viral infection, changed colposcopic examination or molecular biology examination positive for HPV DNA.

Cervicovaginal samples were obtained from all patients and frozen at –20°C in TE buffer for molecular biology studies. Subsequently, based on a flowchart prepared by the outpatient department of the Lower Genital Tract Pathology service, patients were submitted to a colposcopy and to a guided biopsy if a lesion was found.

A transversal analysis of 146 formalin-fixed cervical biopsies obtained from women who underwent biopsy between 2003 and 2008 was performed. The study included patients sexually active, non-pregnant patients who were all but negative for other sexually transmitted diseases. The women were aged between 20 and 50 years old (average age of 35 years old).

Cytopathological and histopathological analysis

Histopathological analyses were processed by the paraffin embedding technique with 3μm thick histological sections, using staining with hematoxylin-eosin (HE), and a subsequent analysis by a pathologist using an optical microscope [16].

Classification was oriented as follows: negative when no significant cytological changes were observed; LSIL when cellular changes consisted of mild dysplasia or CIN 1 and HPV infection were present; HSIL when cellular changes consisted of moderate dysplasia (CIN 2) or severe dysplasia (CIN 3). The control group was composed of samples negative for LSIL or HSIL, all patients in control group were evaluated, including for multiple levels of H&E, in order to exclude focal squamous intraepithelial lesions.

HPV typing

Cervical-vaginal swabs for HPV-DNA detection by the PCR method were collected at the time of the gynecological examination. The determination of HPV DNA presence and viral typing was performed by the polymerase chain reaction method (PCR) [17]. The DNA samples obtained from cervical smears were isolated and purified by using GFX Genomic DNA Purification Kit Blood® kit (Amersham Biosciences, Piscataway, NJ, USA), according to the manufacturer's instructions.

After the DNA extraction, the samples were subjected to PCR using a set of generic primers for HPV, PGMY 09/11 [17], which are capable of amplifying a 450 base pair (bp) from L1 gene of different types of genital HPV. GH20 and PCO4 primers are added to the same PCR. They amplify 268 bp of the human β-globin gene and serve as internal control to assess the integrity and completeness of DNA from each sample.

The samples were amplified in the first PCR reaction using degenerate primers GP-E6-3F (GGG WGK KAC TGA AAT CGG T), GP-E6-5B (CTG AGC TGT CAR NTA ATT GCT CA) and GP-E6-6B (TCC TCT GAG TYG YCT AAT TGC TC), and W, A/T; K, G/T; R, A/G; Y, C/T and C, A/C/G/T. These primers amplify E6 and E7 regions of the 38 most common HPV types. Nested PCR reaction was performed specifically with the types 45, 11, 16, 6, 18, 42, 52, 33, 58, 53, 39, 54, 66, 51, 31, 67, 59, 62, 69, 35 e 68. All samples were analyzed by both the first reaction (PCR) and the second reaction (nested PCR). Some patients infected with HPV had the type classified as not defined due to the fact that HPV type was not identified by any specific PCR technique used.

TG2 immunohistochemistry

Immunohistochemistry (IHC) to TG2 was carried out according to the manufacturer's instructions. IHC was conducted in an automated manner in an Autosteiner-Link 48 Dako equipment.

This TG2 IHC assay utilized a monoclonal mouse antibody (Diagnostic Biosystems, Diluition 1:50) directed against the human TG2 domain. The negative control reagent is a monoclonal rabbit IgG isotype control (DA1E; Cell Signaling Technology, Danvers, MA). The deparaffinization, rehydration, and target retrieval were performed by a PT Link (Dako PT100). Slides were then processed by an Autostainer Link 48 (Dako AS480) using an automated staining protocol validated for TG2 IHC assay. IHC staining protocol includes sequential application of a peroxidase-blocking reagent, TG2 primary antibody or negative control reagent, mouse anti-rabbit IgG linker, visualization reagent consisting of secondary antibody molecules and horseradish peroxidase coupled to a polymer backbone, 3,3′-diaminobenzidine tetrahydrochloride (DAB) chromogen reagent with hydrogen peroxide substrate, and a DAB enhancer which modifies the color of precipitated chromogen. Reagents utilized in addition to TG2 assay components included a wash buffer specially formulated for automated IHC staining and a hematoxylin counterstain. IHC-stained slides were mounted in a nonaqueous permanent mounting medium.

TG2 expression by staining patterns is remarkable. Cells showing TG2 expression are shown stained in brown, which could be seen both in the nucleus and in the cytoplasm.

Interpretation and quantification of the staining

The extent of immunoreactivity in the samples was assessed by two different pathologists, using the same microscope with a 40x objective lens with a field diameter of 0.52 mm. The official histopathology classification of the samples was not disclosed before the scoring was performed.

Analysis of protein expression by immunohistochemistry was performed in two ways. The first immunohistochemical results were evaluated considering the overall proportion of positive cells: samples considered positive for TG2 had their fields photographed. Sections were chosen to give a representative selection of tissue morphologies and stain intensities. Samples were photographed and individually dysplastic and not dysplastic cells were counted, thereby establishing a percentage of positivity. Cells with both cytoplasmic and nuclear staining were considered to be positive. TG2 relative quantification was categorized into three different intervals markings: > 50%, ≤50% or negative. This classification was based on a previous study [3]. Uncategorized relative expression were also evaluated.

In the second evaluation mode images were scanned and transferred to a computer, and the intensity of staining was defined using NIH Image J 1.36b software (National Institutes of Health, Maryland, USA). Image J is a public-domain image-analyzing software on Java platform inspired by NIH Image software for Apple's Macintosh. Therefore, it can be run in different operating systems provided they have a suitable Java virtual machine. The repertoire of the software's functions can be extended through different off-the-shelf plug-ins, available on the web (18). NIH Image J 1.36b software was downloaded from their website (http://rsb.info.nih.gov/ij). The color selection and classification of positively stained points was done by using a distribution diagram for the colors red, green and blue (RGB), which shows the changes in intensity and color saturation. This distribution provides information on pixel quantity of the analyzed image. This was obtained through the Threshold Colour plug-in, in which the color range of interest was defined. The software colors the areas that meet this standard white and the remaining areas black [18–20].

Squamous epithelial cells were included in the evaluation and stromal cells were excluded. Initially, the intensity values were grouped into bands of 10 and the corresponding regions in the image were confirmed by using the threshold feature. Thus, initially all intensities from 0 to 10 were turned white on an image with a known pathological score of high positive. Similarly, the zone containing the lightest color shade of pixel intensities was also determined by using an image with a known pathological score of negative. This was because once the highest intensity (high positive) and the lowest intensity (negative) zones were determined, it would help a better determination of the size of the intermediate (positive and low positive) zones. The region between 0 and 60 contained pixels of the high positive stained images. Similarly, samples with known pathological lower scores were used to optimize the correct range.

Immunohistochemical quantification carried out by Image J were compared and evaluated in order to verify the effectiveness and acceptance of methods.

Statistical analysis

Statistical analysis was performed using SPSS - Statistical Package for Social Sciences (version 20.0). For comparison of qualitative variables as frequencies and proportions, χ2 test was used for independent samples with a possible analysis of adjusted residuals. Ordinal qualitative and quantitative data were compared with nonparametric Mann-Whitney test with the level of significance set at 5%.

The variable HPV was dichotomized into "high risk" and "low risk". According to the literature [21] the following types were considered high-risk HPVs: 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 26, 53, 66, 67, 68, 69, 70, 73 and 82. The low-risk types are classified as 6, 11, 40, 42, 43, 44, 55, 54 and 62. If patients presented any type of high and low risk concomitantly, they were classified as high risk HPV. To evaluate the possible association of TG2 expression with different types of lesions, and to assess the percentage of positivity, relative quantification was categorized into intervals of ≤50%, > 50% and 0 (negative).

TG2 and Histopathological analysis of samples for medical pathologist

Association between LSIL, HSIL, control goup and TG2 was statistically significant (p <0.05). Histopathological analysis was performed with χ2 test complemented by Adjusted Residual Analysis at a significance level of 5%. Results showed that patients with normal samples and LSIL were locally associated with expression levels of TG2 > 50% (p <0.05), and patients with HSIL were associated with no TG2 expression (p <0.05) (Fig. 1a, b). A significant difference related to HSIL and LSIL relative quantification (p <0.05) was observed.

The analyses of samples by using Image J software show a significant (p <0.001) decrease in immunostaining of TG2 in HSIL if compared to normal and LSIL samples (Fig. 1d), immunostaining ranged from zero to 198.17.. This depicts a correlation between relative analysis by the pathologist and Image J.

When analyzing quantification of Image J, in relation to relative quantification a significant difference (p <0.001) was observed (Fig. 1g, e). This demonstrates a relationship between the two analytical methods (Fig. 1f).

Distribution of HPV in cervical lesions

All evaluated patients were tested for HPV through PCR, all 146 cases showed amplification of the human ß globin gene (100%), confirming the adequacy of the specimen regarding the quantity and integrity of the genetic material. Patients were classified as high and low risk depending on the type of HPV identified. Patients who had both low and high risk viral types were classified as high risk. Some patients infected with HPV had the type classified as not defined due to the fact that the HPV type was not identified by any specific PCR technique used.

The evaluated control group consisted of patients without malignant precursor lesions, including cases of squamous metaplasia, endocervicitis and cervicitis. In this group all patients had squamous metaplasia, and 76.3% of these presented HPV infection. It is important to note that even in the presence of HPV, the collected samples showed no related precursor lesions of high or low degree. Among patients in the control group, 42.1% were infected with high-risk genotypes, 2.6% with low-risk HPV, 31.6% with non-defined types and 23.7% were negative for HPV.

LSIL cases show the presence of high-risk HPV in 54.5% of the cases, low-risk HPV in 6.8%, not defined HPV in 29.5% and negative in 30.1%. HSIL cases showed the presence of high-risk HPV in 50% of the cases, low-risk HPV in 9.4%, not defined HPV in 28.1%, and negative HPV in 9.4%. These results demonstrate the highest prevalence of high-risk genotypes in LSIL (Table 1).

TG2 and HPV typing

TG2 expression was observed in all groups, with an average of 72.5% for low-risk HPV types, 59% for high-risk types, and 60.7% for not defined types (Fig. 1c).

There was no significant statistical association between the relative quantification of TG2 and different classifications of HPV viral types.

A significant similarity between the expression of not defined types and high-risk types can be observed, thus, demonstrating the possible presence of high-risk HPV in the not defined group.

The most frequently identified virus types were 16 with 29 cases, 6 with 20 cases, 11 with 18 cases, and 41 cases belonged to the not defined group. When individual HPV types were analyzed in relation to TG2 expression levels a statistical significance (p = 0.031) was demonstrated for HPV 11 indicating that patients with type 11 had higher values of TG2 if compared to patients with a different type. These results may represent a trend as the evaluation is based on a limited number of patients.

When analyzing TG2 expression according to HPV type, one notices that generally the expression decreases in high-risk types and increases in low-risk (Fig. 1c). Particularly, when types of HPV which are not defined are observed by relative quantification and Image J, TG2 behaves similarly to the high-risk types. This shows the importance of analyzing other viral types, since it is more likely they become malignant tissues (Fig. 1e).

Viral types investigated in this study were defined according to their prevalence in cervical lesions. High and low risk are the most common types identified, however, considering the obtained results, it became relevant to delineate the types which are not defined, owing to the fact that probably among them there are prevalent high-risk viral types. When the association between TG2 and viral HPV types was verified through Image J analysis, no significant association was found (Fig. 1d).

Results suggest that TG2 involving the presence of LSIL and non-malignant alterations (Fig. 2) of the cervix are significantly associated to low-risk viral types. Therefore, TG2 variable expression is associated with cancerous changes of the cervix.