To examine the diagnostic utility for squamous intraepithelial lesion (SIL) by cytological in situ hybridization (c-ISH) for the human papillomavirus using liquid-based cytology specimens, we investigated c-ISH signal patterns in the cases of low-grade SIL (LSIL), atypical squamous cells of undetermined significance (ASC-US), and high-grade SIL (HSIL). Episomal (E) and/or integrated (I) signals were observed. Two signal patterns ( 𝐸 𝐼 or 𝐼 > 𝐸 ) were obtained by counting the number of 𝐸 + or 𝐼 + cells. 𝐸 𝐼 was specific to LSIL and ASC-US (10/12); 𝐼 > 𝐸 , to HSIL (9/11) ( 𝑃 < 0 . 0 1 , χ2 test), suggesting significant utility of c-ISH in diagnosing SIL. In the cell fraction, 𝐸 𝐼 in large cells was dominant in LSIL. Two cases of 𝐼 > 𝐸 in large cells of LSIL showed HPV persistence and/or progression during follow-up. Thus, c-ISH is useful in routine testing for diagnosing cervical dysplastic lesions, especially for detecting LSIL suspected for progression.

1. Introduction

Human papillomavirus (HPV) is responsible for uterine cervical cancer [1]. Screening programs for cervical cancer detection have greatly reduced the incidence of cervical cancer and cancer-related mortality. However, the cytological Papanicolaou test has relatively low sensitivity as well as a high false-negative rate and high interobserver variability. Recent studies have reported that exfoliated cells sorted for liquid-based cytology (LBC) are useful for immunohistochemical (p16) and molecular analysis methods such as in situ hybridization (ISH) or polymerase chain reaction (PCR) for detecting viral DNA in these samples [2, 3]. Evaluating exfoliated cells is more convenient than evaluating biopsy specimens for screening patients at high risk of cervical cancer.

Although cervical intraepithelial neoplasia (CIN) 1 lesions are regress spontaneously [4], it is known that high risk HPV integration occurs in a subset of LSILs [5], which could be an early event in carcinogenesis. Therefore, it is important to detect a marker protein or viral genomic state (including the episomal [E] and/or integrated [I] form) before the lesion progress. Under these considerations, we evaluated cytological ISH (c-ISH) with LBC specimens by using a Ventana’s autostainer for detect the signal patterns and the results were evaluated along with the corresponding biopsy specimens.

2. Materials and Methods

The study included 33 cases who had negative for intraepithelial lesion (NILM), low-grade intraepithelial neoplasia (LSIL), atypical squamous cells of undetermined significance (ASC-US), or high-grade intraepithelial neoplasia (HSIL) (moderate dysplasia, severe dysplasia, or carcinoma in situ) and had presented at Department of Obstetrics and Gynecology of Hamada Medical Center between 2008 and 2010. The patients with LSIL were followed for 3 months to 2 years and did not undergo treatment such as laser vaporation. They underwent the biopsy procedure on the same day as or the day after the cytology specimens were obtained, and histological examination was performed as gold standard. The results of the first cytological and histological tests were shown in Table 1.

Cytological examination was performed using split samples. Briefly, the exfoliated cervical samples were directly smeared using a bloomed brush; this was followed by Papanicolaou staining. Next, the brush was immediately suspended in ThinPrep PreservCyt Solution (Hologic Corporation, Mass, USA) for LBC specimens.

We performed ISH using c-ISH and histological ISH (h-ISH) specimens. 1G-sedimented smears on slide glass using Setting Chambers (MBL, Nagoya, Japan) were prepared from ThinPrep PreservCyt Solution for c-ISH. For h-ISH, 4-μm-thick sections were prepared from paraffin blocks. ISH was performed on an autostainer (BenchMarK LT; Ventana Medical systems, Tucson, Ariz, USA) according to the manufacturer’s instructions, with an ISH iVIEW Blue Plus Research Kit and INFORM HPV III Family 16 Probes (B) (Ventana) for h-ISH, and with Probe (C) for c-ISH. The probe cocktail had an affinity with high-risk HPV genotypes such as 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 66. The nuclear signals were punctate or diffuse, corresponding to I and E signals, respectively. To determine whether the samples identified as HPV positive by INFORM ISH were true positives, we stained a positive control section in the same run. In addition, we carefully eliminated the artifact staining according to instruction manual [6].

Exfoliated cells were recovered from the residue of the Thin Prep PreservCyt Solution. Approximately 1 mL of PreservCyt Solution was washed in phosphate-buffered saline, followed by genomic DNA isolation with MagCoreR HF16 System (RBC Bioscience Corp., Taipei, Taiwan). The presence of HPV DNA was determined by PCR and reverse-hybridization on a dot array by using HPV GenoArray Test Kit (HybriBio Limited, Hong Kong). HPV DNA was amplified with the L1 consensus HPV PGMY09/PGMY11 primer set as described previously [7]. Detectable HPV genotypes by type-specific oligonucleotides immobilized on a nylon membrane are as follows: high risk (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68), low risk (6, 11, 42, 43, and 44), and unknown risk (HPV53- and CP8304). Dot signals visible on the membrane were considered positive; however, quantitative estimation was not possible.

The comparison between the two groups was analyzed using Chi-square test. 𝑃 value of <0.05 were considered statistically significant.

3. Results

3.1. Results of Cytology, Histology, In Situ Hybridization (ISH) and Polymerase Chain Reaction (PCR)

For 33 patients, we performed cytological and histological examinations in combination with ISH and PCR as listed in Table 1.

HPV DNA was detected in 27 patients by PCR in NILM (2/3, 66.7%), LSIL (12/13, 92.3%), ASC-US (2/5, 40.0%), and HSIL (11/12, 91.7%). Of 27 cases, 19 were found to have genotypes 16, 52, 18, 58 [8].

In h-ISH, an E-signal (diffuse signal) and an I-signal (punctated signal) were detected in each cell layer as previously reported [9]. Dysplastic cells showing the I-signal were detected in the intermediate and superficial layers as well as in the basal layer in CIN1 cases (Figure 1). In c-ISH, the positive cells were easily detectable (Figures 2(a) to 2(f)). Among the PCR positive cases of NILM, LSIL, ASC-US, and HSIL, a positive signal of c-ISH was detected in 0% (0/2), 84.6% (11/13), 50.0% (1/2), and 90.9% (10/11) cases, respectively. However, the morphology of the positive cells (E-signal, Figures 2(a) to 2(c); I-signal, Figures 2(d) to 2(f)), was not exactly the same as the morphology observed in the Papanicolaou smear. We classified the cells on the basis of the nuclear signal and cell size into categories such as large-sized cells (L-cell; Figures 2(a) and 2(d)), medium-sized cells (M-cells; Figures 2(b) and 2(e)), and small-sized cells (S-cells; Figures 2(c) and 2(f)). It is likely that these sizes correspond to the superficial, intermediate, and basal cells observed in the epithelial layer in biopsy specimens.

3.2. Evaluation of c-ISH Signals

To evaluate the signal in c-ISH, we counted the E- and I-positive cells in each cell population (Table 2). The positive cells on each glass slide were counted. The pattern with E signals greater than I signals is expressed as E > I, the pattern with an equal number of E signals and I signals is expressed as E = I, and the pattern with more I signals than E signals as I > E. In the total cell (L-cell + M-cell + S-cell) fraction, the positive cell pattern of EI was observed in 10 out of 12 cases in LSIL and ASC-US. On the other hand, the pattern I > E was observed in 9 out of 11 cases in HSIL. There was a significant difference between the frequency of the EI pattern in LSIL and that of the I > E pattern in HSIL ( 𝑃 < 0 . 0 1 in the Chi test). This indicates the usefulness of c-ISH in diagnosing the SIL according to the pattern of signals.

We estimated the sensitivity, specificity, PPV, and NPV of cytology exams (Papanicolaou) in combination with c-ISH signals to detect CIN1 or CIN2/3 using histological examination as the gold standard (Table 3). As shown in Tables 3(2) and 3(3), the combination of Papanicolaou test HSIL with the c-ISH signal pattern EI increased the specificity (90.0% to 100%), PPV (83.3% to 100%), and NPV (85.7% to 90.9%), and the I > E pattern increased the sensitivity (76.9% to 90.0%) and PPV (83.3% to 90.0%) for the detection of CIN2/3. In LSIL, the combination with EI increased the sensitivity and PPV (69.2% (9/13) to 88.9% (8/9)) and I > E increased the specificity and NPV (80% (16/20) to 88.9% (8/9)) for CIN1.

The numbers of E or I cells in each cell population (Table 2) are shown in scatter plots (Figures 3 and 4). As shown in Figure 3(a), cases of the I > E pattern of LSIL (*1 and *2) and E > I pattern of HSIL (*3 and *4) were found to be minor cases of each lesion. Comments on these cases are provided in the Discussion section. Positive large-sized cells were present in all 11 (100%) patients with LSIL and 7 of 11 (63%) patients with HSIL. On the other hand, M + S cells were observed in 6 of 11 patients (54%) with LSIL and all 11 (100%) patients with HSIL (Figure 4). The major cellular component of LSIL was revealed to be large-sized cells in c-ISH. The results obtained for the large-sized cells (Figure 3(b)) with the I > E pattern cases (*1 and *2) were appear to be unusual in comparison with those for the 5 EI cases (*6, *7, *8, *11, and *12), where regression was observed in all cases (NILM, PCR negative, and ISH negative). In *1 (case  4), the I > E pattern of HPV was detected (Figure 5). Although regression to NILM was observed, HPV was consistently detected (PCR positive (type 58) and ISH positive) during a period of at least 6 months; *2 (case  15) was CIN2 with a negative ISH; *5 (case  13) progressed to HSIL (moderate dysplasia) after 22 months. Therefore, LSIL cases showing the I > E pattern in a large-sized cell fraction appear to indicate that careful diagnosis and follow-up may be necessary.

4. Discussion

It is important to morphological, viral, oncological, and immunological markers for the early stage of uterine cervical dysplasia. The viral-infected condition, such as the episomal or integrated form, is believed to be important because the HPV integration is known to represent a key step towards the progression of the disease [10]. Therefore, we investigated HPV-infected cells to search for a diagnostic and/or prognostic marker by using the c-ISH in liquid-based specimens. PCR was also performed for the comparison.

Although HPV type 52 was not the most commonly detected genotype in this study, which does not agree with the results obtained in a previous report [8], this difference is probably due to the selected LSIL patients who were followed up without any treatment.

We used Ventana autostainer for the ISH examination using the Inform HPV III probe. Guo et al. [11] reported an ISH method, which uses the Inform HPV III probe. This method has significantly improved sensitivity compared to HPV II, and this sensitivity is comparable to PCR for detecting HPV DNA in tissue sections. Although higher specificity was reported [12, 13], Alameda et al. [14] pointed out that the Ventana Inform has lower sensitivity. However, the version used is not described in the report. In this study, we detected c-ISH positive signals, which were similar to the results provided by PCR in both LSIL and HSIL. We also detected more positive signals of c-ISH than of h-ISH in PCR-positive LSIL cases (91.6% (11/12) versus 58.3%(7/12)) and PCR-positive HSIL cases (90% (10/11) versus 81.8% (9/11)). Thus, c-ISH is revealed to be more sensitive than h-ISH, especially in low-grade lesions. As discussed by Guo et al. [11], the heterogenous distribution of HPV in low-grade CIN might cause signal absence in tissue sections, resulting in false-negative results. On the other hand, in case  10 (CIN3), weak integrated signals were detected in h-ISH but not in c-ISH. Because the punctate signal pattern is known to be more frequent in CIN3 and because it becomes more difficult to recognize or interpret the ISH signals, it is possible that we did not recognize the weak signals in c-ISH.

In evaluating the ISH signals, diffuse, punctate + diffuse, and punctate patterns were described previously [2, 14]. As the coexistent pattern (punctate + diffuse) was found to be abundant in both h-ISH and c-ISH, we classified the c-ISH signal patterns into EI and I > E categories according to the number of E+ or I+ cells in each case. Therefore, we found that the EI pattern corresponds to LSIL, and the I > E pattern corresponds to HSIL. The differences were found to be statistically significant ( 𝑃 < 0 . 0 1 ). The signal patterns of c-ISH seem to be very useful for diagnosing the SIL. In the combination of the Papanicolaou test with the c-ISH signal patterns, HSIL with the EI pattern increased the specificity, PPV and NPV, and the I > E pattern increased the sensitivity and PPV for detection of CIN2 or more advanced lesions. And also LSIL with EI pattern increased the sensitivity and PPV, I > E pattern increased the specificity and NPV to detect the CIN1. So, the combination of Papanicolaou test with c-ISH signal patters appears to be useful in the diagnosis of CIN lesion. However, the specificity and NPV of HSIL with I > E to detect CIN2/3 was decreased to 50% (1/2). LSIL with E ≧ I pattern decreased the specificity and NPV (66.7%) to detect CIN1. Therefore, the results of signal patterns of c-ISH are useful if they are separately thought about the sensitivity and specificity in the cases of LSIL or HSIL compared with the diagnosis of CIN.

The E > I pattern cases of HSIL (*3 and *4) shown in Figure 3 are discussed. In *3 (case  25), many E+ S-cells were detected. It is difficult to demonstrate a diffuse integrated pattern in ISH using Ventana system [12]. Therefore, there is a possibility that c-ISH-positive S-cells undergo diffuse integrated staining. In the case of *4 (case  26), both CIN1 and CIN3 (severe dysplasia) lesions were observed in the same histological specimen. Because the cytological specimens were composed of superficial and intermediate layered cells according to the histological analysis, the E+ cells of CIN1 appear to be easily detected in c-ISH.

We focused on the LSIL patients with an I > E pattern indicating L-sized positive cells, because an atypical persistent detection of HPV in NILM or progression to HSIL were observed during the followup of these patient. Although persistent detection of HPV is reported to be common among CIN patients at followup even in cases where cytology and histology results are obtained [2], the integration of high-risk HPV is generally a key event in cervical carcinogenesis [10]. De Marchi Triglia et al. [9] reported that the presence of a punctate signal in the superficial layer in histological analysis is associated with cases without progression. Although we did not detect the I-positive S-sized cells in c-ISH in both the case  4 and 13, I-positive basal cells were detected in h-ISH. We do not know the correlation between I-positive basal and superficial cells as well as the destiny of I-positive superficial cells now. To reveal the significance of I-positive L-cell (c-ISH) and superficial cells (h-ISH), we will investigate such cases in the future studies involving a greater number of patients.

5. Conclusion

We investigated c-ISH (Ventana method) for liquid-based cytology specimens and detected c-ISH-positive cells that showed E- or I-signals. From these cases, the EI pattern was found to be in LSIL, and the I > E pattern in HSIL, and the differences were revealed to be statistically significant ( 𝑃 < 0 . 0 1 ). In follow-up examination for the LSIL cases, we found the 2 patients who had large-sized cells with the I > E pattern. They showed HPV persistence and/or progression during follow-up. Therefore, presence of the I > E pattern of c-ISH large-sized cells in LSIL may be a promising marker for follow-up studies.


This work was financially supported in parts by the National Hospital Organization of Japan.