HPV prevalence, viral load and physical state of HPV-16 in cervical smears of patients with different grades of CIN

Abstract

Human papillomavirus (HPV) infection is the most important event in malignant transformation of human cervical epithelium. We analysed in cervical smears, HPV genotypes with a focus on single/multiple infections, then characteristics of HPV-16 infections (presence of other genotypes, viral load and physical state) according to the grade of histological lesions. The purpose of this study was to know if these parameters could allow to differentiate histological diagnoses. DNA was extracted from 363 cervical samples corresponding to 24 cases without lesion, 96 CIN1, 92 CIN2, 144 CIN3 and 7 cancers. Our results show that HPV-16 was predominant and its prevalence increased with the severity of lesions (CIN1: 27.1%; CIN3: 65.3%). In addition, we showed that the frequency of single infections, as compared with multiple infections, increased with the severity of the lesion (CIN1: 25.0%; CIN3: 54.8%). Among HPV-16 positive samples (n = 170), we found that viral load, determined on cervical samples by real-time PCR, did not vary significantly according to the different CIN grades. Concerning HPV-16 integration, the mixed and integrated HPV-16 forms, already present in women with normal histology, increased to the benefit of pure episomal forms with the severity of lesions (normal cervix: 28.6%; CIN3: 73.8%). Thus, our data raise the question of the viral load as a valuable clinical parameter to discriminate between lesion grades. Moreover, we emphasize integration as an early event in cervical carcinogenesis, increasing with the severity of lesions. Finally, this study underlines the importance of single versus multiple infections linked to the severity of CIN. © 2007 Wiley-Liss, Inc.

Cervical cancer is the second most frequent malignant tumour of women worldwide. The recognized association of human papillomaviruses (HPV) as the most significant etiological factor in cervical cancer is now well established.1, 2 More than 40 genotypes infect the anogenital area and are associated with a large spectrum of diseases from benign proliferation to invasive cancers. The low-risk (LR-) HPV are linked to benign lesions whereas high-risk (HR)-HPV, principally types 16 and 18, are associated with malignant lesions. The most common HR-HPV type in squamous cell cervical cancer is HPV-16, found in over 50% of the cases.3, 4

Recent studies have suggested that HR-HPV viral load, notably for HPV-16, may be a molecular biomarker of risk for developing cervical (pre-) cancerous lesions. Indeed several studies found a high correlation between high viral load and HSIL (High grade Squamous cell Intraepithelial Lesion)5, 6, 7, 8 whereas other reports concluded that there was no correlation.9, 10, 11, 12, 13 Also, despite multiple investigations, using various study designs and methodologies, the clinical utility of HR-HPV load remains unclear.

The physical status of HR-HPV in pre-neoplastic cervical lesions may be a further promising marker for the progression to cervical cancer. Molecular biological studies have shown that HPV DNA is in an episomal state during a common infection or in most pre-malignant lesions,14 but in an integrated state or in both integrated and episomal states in most cervical carcinomas and in cell lines derived from these carcinomas.15, 16 The disruption of the E2 gene during viral integration results in a loss of its function as a regulator of the viral oncogenes expression17 and a subsequent up-regulation of E6/E7 gene transcription. E6 and E7 oncoproteins deregulate cell-cycle control through interaction with different cell cycle proteins, respectively tumor suppressor gene products p53 and retinoblastoma protein (Rb), thereby initiating the transformation and immortalization of HPV-infected cells.18, 19 Thus, the physical state of HPV may be used as an early marker of cervical lesion for the follow-up of patients with chronic infection.

In the present study, our aim was to characterize the type-specific prevalence of HR-HPV types, notably HPV-16, the presence of single versus multiple HPV infections, HPV-16 viral load and HPV-16 physical status in 363 HPV positive DNA cervical smears corresponding to lesions of different histological grades.

ASC-US: atypical squamous cells of undetermined significance; ASC-H: atypical squamous cells, cannot exclude HSIL; CIN: cervical intraepithelial lesion; HPV: human papillomavirus; HR-HPV: high risk human papillomavirus; HSIL: high grade squamous intraepithelial lesion; LSIL: low grade squamous intraepithelial lesion.

Material and methods

Clinical samples

We selected a population of 363 women with a HR-HPV infection presenting lesions from ASC-US to HSIL at cytology according to the Bethesda classification (26 ASC-US, 29 ASC-H, 3 AGC, 98 LSIL, 178 HSIL) and 29 cases without any detectable lesion but with a prolonged recurrent HR-HPV infection (from 9 to 35 months, mean = 19 months). All these women had a histological control from multiple histological biopsies to cone biopsies showing as normal cervix (24 cases), CIN1 (96 cases), CIN2 (92 cases), CIN3 (144 cases) and invasive carcinoma (7 cases). The histopathologist had no access to initial cytology interpretations nor to HPV testing results.

All molecular analyses were performed on the liquid-based cervical samples collected in PreservCyt Medium (ThinPrep, Cytyc Corp, Marlborough, Mass) and tested initially positive with the HCII assay.

HPV testing with hybrid capture 2 assay

After slide preparation for cytological examination, 4 mL of the remaining samples were centrifuged and the cell pellet was resuspended in 200 μL of phosphate-buffered saline for HPV testing. HPV DNA detection was performed by the commercially available hybrid capture 2 (HCII) System (Digene, Gaithersburg, MD). All scrapes were analysed for the global presence of HR-HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59 and 68. Samples were classified as positive for HR-HPV DNA if the relative light unit (RLU) reading obtained from the luminometer was equal or greater than the mean value of a triplicate of positive control supplied with the HCII kit. As some authors have reported that increasing HPV DNA levels of the HR-HPV types were the principal predictors of CIN,20 we used, as proposed, the ratio RLU/positive controls values to quantify HR-HPV DNA in our samples.

Extraction of DNA from cervical smears

DNA extraction was performed with the EZ1 DNA tissue kit (Qiagen Inc., Courtaboeuf, France) and the BioRobot EZ1 (Qiagen Inc., Courtaboeuf, France) according to the manufacturer's instructions. Briefly, samples were lysed using proteinase K at 56°C for 3 hr and extractions were realized in an elution volume of 200 μL.

Linear array HPV genotyping test

The linear array (LA) HPV genotyping test (Roche Molecular Systems, Inc., Branchburg, NJ) is a qualitative in vitro test for the determination of 37 HPV DNA genotypes (15 HR-HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 68, 73 and 82, 3 potentially HR-HPV, 26,53,66 15 LR-HPV 6, 11, 40, 42, 54, 55, 61, 64, 67, 69, 70, 72, 81, IS39 and CP6108) and 4 HPV at indeterminate risk. 62,71,83,84

PCR amplification of HPV DNA: The LA test uses biotinylated PGMY primers to amplify a 450 bp fragment within the polymorphic L1 region of the HPV genome. The PGMY amplification system has been described previously.21 An additional primer pair targets the human β-globin gene (268 bp amplicon) to provide a control for cell adequacy, extraction and amplification. PCR was performed in a final reaction volume of 100 μL, containing 50 μL HPV master mix, 40 μL DNase-RNase-free water, and 10 μL isolated DNA. The mixture was incubated as described by the manufacturer. DNase-RNase-free water as negative control and DNA extracted from baculoviruses having integrated different known HPV types as positive controls were included in each PCR run of 24 samples to assess the validity of the reaction.

Hybridization and detection: Following amplification, the HPV and human β-globin amplicons were denatured by immediately adding 100 μL of denaturation solution to each PCR tube. The strips were manually interpreted using the LA HPV reference guide, by reading the individual types down the length of the strip.

Real-time PCR

The determination of HPV-16 physical status and quantification of albumin gene using Taqman probes, were performed on the iCycler iQ™ (BIO-RAD, Marne-la-Coquette, France), using the qPCR™ Mastermix-No ROX (EUROGENTEC, Seraing, Belgium) in a 50 μL final volume. The data collected with the iCycler iQ™ were analyzed with the iCycler iQ™ Real-time PCR Detection System Software version 3.0A (BIO-RAD).

Detection of HPV-16 physical status.

Real-time PCR for HPV-16 E2 and E6 was performed as described elsewhere22 except that the E6 and E2 probes were labelled with 6-FAM and Texas-Red at the 5′end, respectively (EUROGENTEC, Seraing, Belgium). The primers' sequences are provided in Table I. Briefly, a PCR was carried out in a 50 μL reaction volume containing 1X qPCR™ Mastermix-No ROX (EUROGENTEC, Seraing, Belgium) and 50 ng of target DNA from cervical smears. The amplification conditions were 2 min at 50°C, 10 min at 95°C, and a two-steps cycle of 95°C for 15 sec and 60°C for 60 sec for a total of 45 cycles. The final primers and probes concentrations, in a total volume of 50 μL, were 0.3 and 0.1 μM, respectively. Two standard curves were obtained by amplification of a dilution series of 5 million to 500 copies of an HPV-16 plasmid provided by the BIO-RAD Company. At least 2 no-template control reaction mixtures (water) and 2 controls of SiHa cell line were included in each run. All experiments were performed twice in duplicate. The results were recorded as copy numbers in 50 ng of total DNA. Assignment of integrated, mixed or episomal physical status was calculated for each clinical sample as proposed elsewhere.22 This calculation assumed that E2 and E6 gene segments are present in equivalent proportions within each episomal HPV genome and that integrated HPV genome forms would have the E2 target deleted or absent. Thus, integration was determined by subtracting the copy numbers of E2 (episomal) from the total copy numbers of E6 (episomal and integrated). The ratio of episomal E2 gene target to the integrated E6 gene target represents the amount of the HPV episomal form in relation to the integrated form. Integration was defined by absence of the E2 signal or ratios of 0.001–0.003. Ratios of less than 1 (range, 0.004–0.99) indicated the presence of both integrated and episomal forms, and ratios of greater than 1 (range, 1.00–2.01) indicated a predominance of episomal forms.

Table I. Sequences of Primers and Probes Used for Real-Time PCR to Determine the HPV16 Physical State

Primer name	Sequence 5′-3′	Nb of bases
For the episomal state, E2 primers and a probe for the episomal viral load were used. For the integrated and episomal states together, E6 primers and a probe for the total viral load were used. F, forward primer; R, reverse primer; BHQ, Dark Quencher, 6-FAM, 6-carboxyfluorescein.
16E2F	AAC GAA GTA TCC TCT CCT GAA ATT ATT AG	29
16E2R	CCA AGG CGA CGG CTT TG	17
16E6F	GAG AAC TGC AAT GTT TCA GGA GC	23
16E6R	TGT ATA GTT GTT TGC AGC TCT GTG C	25
16E2-PRO	Texas Red-CAC CCC GCC GCG ACC CAT A-BHQ-2	19
16E6-PRO	6-FAM-CAG GAG CGA CCC AGA AAG TTA CCA CAG TT-BHQ-1	29

Determination of viral load

Each DNA sample was submitted to an albumin real-time PCR, twice in duplicate, in order to determine the number of cells. Quantification of human albumin gene was performed as previously described.23 The reference human genomic DNA provided by Roche Diagnostics (Meylan, France) was serially diluted and run in parallel with the DNA from cervical samples. Dilutions corresponding to 152,000; 15,200; 1,520; 152; 15.2 genomic DNA copies were used to plot the standard curve and to determine the number of cells present in the sample. Primers and probe were purchased from EUROGENTEC (Seraing, Belgium) (Table II). The albumin reaction was carried out in a final volume of 50 μL containing 1X qPCR™ Mastermix-No ROX (EUROGENTEC, Seraing, Belgium), 0.1 μM each primer, 0.1 μM Taqman probe and 5 μL of DNA corresponding to 50 ng of cellular DNA. Thermal cycling consisted in a step of 10 min at 95°C in order to activate the DNA polymerase, followed by 45 two-steps cycles of 15 sec at 95°C and 60 sec at 65°C.

Table II. Sequences of Primers and Probe Used for Real-Time PCR to Amplify the Albumin Gene

Primer name	Sequence 5′-3′	Nb of bases
F, forward primer; R, reverse primer; BHQ, Dark Quencher, 6-FAM, 6-carboxyfluorescein.
Alb-F	GCT GTC ATC TCT TGT GGG CTG T	22
Alb-R	ACT CAT GGG AGC TGC TGG TTC	21
Alb-PRO	6-FAM-GGA GAG ATT TGT GTG GGC ATG ACA GG-BHQ-1	26

Data analysis

Quantitative variables were presented as mean values ± standard errors. To assess the relation between the different histological lesions and their potential determinants, Pearson χ² tests and Fisher's exact tests were used for qualitative variables (age group, multiple infections, HPV-16 presence, HPV-16 physical status). For quantitative variables (age, HC2 RLU and HPV-16 viral load), comparisons were assessed with Student's t-tests, two-samples Wilcoxon rank-sum tests and one-way variance analysis. All analyses were carried out using the STATA Statistical Software (Release 5.0; StataCorp. 1997, College Station, TX).

Results

Correlation between cytology and histology findings

The study comprised 363 cases, all positive with the HCII assay. The relationships between cytological and histological diagnoses are summarized in Table III. There was 74.9% of good correlation between cytology and histology findings for CIN2+ diagnosis (CIN2,-3 and cervical cancer) and a rate of 76.7% of good correlation for Normal-CIN1 diagnosis. In 25.1% of CIN2+ cases, cytological diagnosis underestimated the severity of lesions. By contrast, in 23.3% of Normal-CIN1 cases, the cytological diagnosis overrated the severity of lesions. In all these cases, multiple biopsies and cone biopsies were performed excluding a CIN2+ lesion. Moreover, in the follow-up of these women, no CIN2+ appeared and most of the cytological lesions disappeared. Thus, for the subsequent analysis, assuming that histology is accurate, only histological diagnoses were considered.

Table III. Correlation Between Cytology and Histology Findings

		Histological diagnosis
		Normal-CIN1 (n=120)	CIN2+ (n=243)	Total (n=363)
Dark grey: overrated diagnoses; Pale grey: underestimated diagnoses.
	Normal	14	15	29
	ASC-US	14	12	26
	ASC-H	8	21	29
Cytological diagnosis	AGC	0	3	3
	LSIL	64	34	98
	HSIL	20	158	178
	Good correlation (%)	76.7	74.9	75.5
	Underestimated cytological diagnosis (%)	0	25.1	16.8
	Overrated cytological diagnosis (%)	23.3	0	7.7

Relationship between age and cervical lesions

The mean age was 38.5 ± 3.3 years in 24 women with normal histology, 32.3 ± 1.2 in 96 women with CIN1, 34.8 ± 1.1 in 92 women with CIN2, 36.0 ± 0.9 in 144 women with CIN3 and 48.4 ± 3.7 in 7 women with cervical cancer. To be noticed, our population was older than the average ages of acute HPV infection. There was a significant association between the age of patients and increasing severity of cervical lesions (p = 0.007).

HPV genotyping

The frequencies of main types of HR-HPV in normal-CIN1 group and CIN2+ group, are summarized in Figure 1. The most prevalent HR-HPV types in this study were: HPV-16 (46.8%), HPV-31 (18.7%), HPV-51 (14.9%), HPV-58 (8.5%), HPV-56 (6.6%) and HPV-18 (6.3%). The prevalence for the other types was low. As expected, HPV-16 predominated and its prevalence increased statistically with the stages of histological lesions: 29.2% for patients with normal histology, 27.1% for patients with CIN1, 42.4% for patients with CIN2 and 65.3% for patients with CIN3 (p = 0.001).

Figure 1. Distribution of HR-HPV types in 363 women with different histological diagnosis. The prevalence of HR-HPV was estimated as percentage of the total number of HPV infections observed in this sample set (n = 363).

Concerning the LR-HPV, the distribution was HPV-42 (6.3%), HPV-61 (6.1%) and HPV-54 (5.2%), other types being much less common.

In 18 samples (5%), only LR-HPV could be detected with the LA genotyping test while HCII detected HR-HPV. These discrepancies are in accordance with the general data of the literature and are attributed to cross-reactions.

Single/multiple HPV infections

To determine single/multiple infection, we did not consider HPV-52 because the genotyping kit of Roche Diagnostic does not permit to discriminate HPV-52 only. Furthermore, we ranged HPV potentially at high risk and HPV at indeterminate risk in the LR-HPV category.

As shown in Table IV, out of the 363 samples, 151 (41.6%) were defined as single infections (with one HR-HPV only). Interestingly, the frequency of single infections with HR-HPV significantly increased with the severity of lesions: 22 (22.9%) of 96 women with CIN1, 39 (42.4%) of 92 with CIN2, 79 (54.8%) of 144 with CIN3 and 5 (71.4%) of 7 with cervical cancer (p = 0.001). By contrast, the frequency of LR-HPV decreased with the severity of lesions. In the same way, the prevalence of multiple infections whatever the genotypes (LR- and HR-HPV) decreased with the severity of lesions.

Table IV. Frequency of Single and Multiple Infections According to the Histological Diagnosis

Histological diagnosis	A (n)	%	B (n)	%	C (n)	%	D (n)	%	E (n)	%	F (n)	%	Total
A: Samples with one HR-HPV; B: More than one HPV-HR; C: One LR-HPV; D: More than one LR-HPV; E: One HR-HPV and one or more than one LR-HPV; F: More than one HPV-HR and more than one HPV-LR.
Normal	6	25.0	2	8.3	2	8.3	1	4.2	9	37.5	4	16.7	24
CIN1	22	22.9	17	17.7	5	5.2	3	3.1	29	30.2	20	20.9	96
CIN2	39	42.4	11	11.9	4	4.3	1	1.1	18	19.6	19	20.7	92
CIN3	79	54.8	23	16.0	2	1.4	0	0	21	14.6	19	13.2	144
Cancer	5	71.4	1	14.3	0	0	0	0	0	0	1	14.3	7
Total	151	41.6	54	14.9	13	3.6	5	1.4	77	21.2	63	17.3	363

Since HPV-16 infections predominated and increased with histological grades, we analyzed subjects with specific HPV-16 infection (Table V). As previously shown, the rate of single HR-HPV infection increased with the importance of lesions: 5 (19.2%) of 26 women with CIN1, 15 (38.5%) of 39 with CIN2, 49 (52.1%) of 94 with CIN3 and 4 (100%) of 4 with cervical cancer (p = 0.0001).

Table V. Frequency of Single and Multiple Infections in HPV-16 Sample, According to the Histological Diagnosis

Histological diagnosis	A (n)	%	B (n)	%	C (n)	%	D (n)	%	E (n)	%	Total
A: Samples with HPV-16 only; B: HPV-16 with an other HR-HPV; C: HPV-16 with more than one other HR-HPV; D: HPV-16 with one or more LR-HPV; E: HPV-16 with more than one HR-HPV and more than one LR-HPV.
Normal	4	57.1	0	0	0	0	2	28.6	1	14.3	7
CIN1	5	19.2	2	7.7	7	26.9	6	23.1	6	23.1	26
CIN2	15	38.5	4	10.3	1	2.6	6	15.4	13	33.3	39
CIN3	49	52.1	4	4.3	12	12.8	13	13.8	16	17.0	94
Cancer	4	100	0	0	0	0	0	0	0	0	4
Total	77	45.3	10	5.9	20	11.8	27	15.9	36	21.2	170

HR-HPV DNA amounts measured by the HCII assay

In our study, HR-HPV DNA levels did not correlate positively with the severity of CIN. By contrast, we obtained a mean value of 429 ± 136 RLU/CO for patients without histological lesions, an increase of this value for patients with CIN1 lesions (806 ± 99 RLU/CO) and then a decrease for patients with high grade lesions (CIN2: 536 ± 76, CIN3: 463 ± 52 RLU/CO). The value obtained for CIN1 lesions was significantly higher than for the other groups (p = 0.025). When considering single versus multiple infections, we could not find any significant difference.

HPV-16 viral load estimated by quantitative real-time PCR

Among the 170 HPV-16 positive samples, 122 (72%) were available for viral load measurement. As expected, the lowest viral load values were observed for patients with normal histology with a mean viral load of 2,114 ± 1,615 copies/100 cells (n = 7). On the other hand, there was no important change in the mean viral loads according to the severity of lesions: 113,168 ± 70,878 (n = 15) for CIN1; 122,023 ± 80,770 (n = 32) for CIN2; 122,764 ± 41,365 (n = 65) for CIN3. However, there was a slight decrease in viral load value for women with cervical cancer (76,519 ± 50,057 copies/100 cells, n = 3). Overall, women without any lesion at histology or with CIN1 were found to have lower HPV-16 viral loads (77,833 ± 49,103, n = 22) than women with CIN2+ (121,140 ± 37,079, n = 100) (p = 0.011). There was a slight difference in HPV-16 viral load values when HPV-16 was found as a single HPV type (95,028 ± 39,151, n = 50) or was associated with other HPV types (126,040 ± 46,334, n = 72) (p = 0.06).

Physical status of HPV-16

As for HPV-16 viral load, physical status has only been assessed in 122/170 HPV-16 positive samples, and results were considered as pure episomal forms versus mixed and integrated forms. In women with normal histology, pure episomal forms predominated (71.4%, 5/7), whereas they accounted for only 26.2% (17/65) in women with CIN3 (Figure 2). Moreover, only 3 pure integrated forms were found and they were limited to CIN2 and CIN3 lesions. Overall, we found a decrease of the pure episomal forms to the benefit of mixed and pure integrated forms, as the grade of the lesion increased. This association between physical status and lesion severity was not found to be statistically significant (p = 0.093), but one can argue that this was due to a lack of power due to the very small size of the normal group (n = 7).

Figure 2. Distribution of HPV-16 physical status according to histological diagnosis.

Discussion

Our study is a transversal approach addressing the interest of HPV genotyping, HPV-16 presence, its viral load and its physical status related to histological grade lesions. We considered only the final histological diagnosis even if there were some discrepancies between cytology and histology (16.8% of underestimation and 7.7% of overestimation) in our series. The histological controls (multiple biopsies and cone biopsies) and the subsequent follow-up strengthen our choice to evaluate the various parameters for predicting a CIN2+.

For each cervical lesion, except for cervical cancer, the most prevalent HR-HPV types were −16, −31 and −51. Our findings are in agreement with those reported in the meta-analyses of Clifford et al.24, 25 In another hand, the prevalence of HPV-18 in our series was relatively low as compared to other studies,26 maybe because of the limited number of cancer cases. Whatever the lesions observed, the differences in the relative prevalence of HPV types might be also related to the complex geographical and biological interplay between different HPV types or variants and host immunogenetic factors (HLA polymorphisms).27 Our genotyping results revealed a single HR-HPV type infection in 41.6% of cases. Currently, there is a lack of consensus within the literature about the extent and implications of single and multiple HPV infections, with a reported prevalence of multiple infections varying from 17.5% in a gynaecological referral clinic population28 to 46% in cancer biopsies29 and 58.9% in women with low grade lesions.30 In our study, multiple HPV infections were more frequently found in women with CIN1, these women being also the youngest. This observation suggests that a greater sexual activity of younger women may be associated with sexual transmission of multiple HPV types. Moreover, our data plead in favour of the transient nature of the majority of multiple HPV infections, since we observed a significantly higher prevalence of single HR-HPV infections in the most severe cervical lesions (54.8% in CIN3 and 71.4% in carcinomas). In a similar manner, in women with HPV-16 infection, the number of single HPV-16 infection does not cease augmenting with the worsening of lesions (19.2% in CIN1; 52.1% and 100% in CIN3 and carcinomas, respectively). Thus, a single infection, especially with HPV-16, may select a population already having CIN2+.

Several studies have reported the association of a high viral load with the risk for cervical cancer and its precursors. A large bunch of studies used the HCII assay to measure viral load, and while some found viral load to be positively associated with an increased risk for prevalent or incident disease,31, 5, 6 others did not.9, 32 Studies using quantitative PCR (q-PCR) to determine HPV load have more consistently demonstrated an association between viral load and evolution of lesions.33, 34, 35 In our study, we explored the relationship between HR-HPV viral load and CIN severity using 2 different methods : HCII assay and q-PCR for HPV-16. In agreement with other investigators,10, 36, 11 we did not find any significant correlation between RLU ratio and severity of CIN lesions but, like Tsai et al.37 and Sherman et al.,11 we observed a slight decrease of HR-HPV DNA load with progression of CIN lesion. This fact could be explained by a productive infection with high levels of viral replication in the early development of CIN lesions. So, in our experience, the HCII assay for the assessment of the viral load could not clearly distinguish among cases with CIN2+. One major bias of the HCII viral load is that this test does not provide an evaluation of the cell numbers, which can vary substantially from 1 sample to another. Moreover, the global high viral load detected by HCII is rather confusing, since it may represent single or multiple HPV types among the 13 high-risk types detected by the kit. However, even when we considered only the single infections, we failed to detect any significant differences. Thus, in our hands, HR-HPV DNA load levels determined by HCII in PreservCyt samples cannot be used as a discriminating parameter for the diagnosis. Using an HPV-16 specific q-PCR, we found that the average of HPV-16 copies per 100 cells were not significantly different in the CIN1, -2 and -3 cases. However, the viral load estimated in women with normal histology, presented a significant lower value than women with cervical neoplasia. Thus, in our present series, the HPV-16 viral load presented a limited value for diagnostic purpose. Supporting our conclusion, in an elegant study, Sherman et al.11 underlined 2 limits at the viral load determination on liquid-based cytology samples: (i) variability in the severity, extent and HPV DNA content of lower grade lesions surrounding the lesion diagnosed as the most severe pathology present, (ii) the intrinsic characteristics of exfoliated specimens obtained by scraping, which necessarily consist mainly of surface epithelium.

In most cervical immortalized cells, HR-HPV DNA often integrates into the host cell genome. This integration is processed by disrupting some part of E2 gene16, 18 causing overexpression of the E6 and E7 proteins.38, 39 Most authors agree with the hypothesis that the integration of the HPV genome takes place very early in the development of cancer. Indeed, Gallo et al.40 and Do Horto et al.41 reported early integration of HPV-16 DNA in 54% and 43% of LSIL cases, respectively. Andersson et al.42 also found mixed forms of HPV-16 in CIN1. By contrast, Hudeslist et al.43 reported that in normal epithelium and in CIN1 and CIN2, HPV-16 and HPV-18 were exclusively found in the episomal form. In our experience, we also observed an early viral integration in normal epithelium and CIN1 but a near-to-significance correlation between the physical state of viral DNA and the increasing grade of cervical neoplasia (71.4% of episomal forms in normal histology vs. 26.2% in CIN3 lesions). Moreover, the rare fully integrated forms were limited to CIN2 and CIN3. Contrary to many studies, we found a low proportion of cases with complete integration (only 2 cases out of 65 CIN3 and 1 case out of 32 CIN2). The frequency of viral integration varies between the reports and may be explained by differences in the technical approaches employed. For example, we cannot exclude that HPV-16 integration, detected by the method of Peitsaro et al. that we used, may be underestimated when it is associated with deletions outside an intact E2 ORF. However, our present data show that if integration of HPV-16 is an early event in cervical carcinogenesis, even present in normal cervical samples, the prevalence of mixed and integrated forms is higher in CIN3 and in carcinomas (respectively 73.8% and 100%). The role of viral DNA integration into the host genome as a major event leading to malignant transformation of dysplastic cervical epithelium is still discussed. In fact, this phenomenon seems to be HPV type-dependant. Indeed, Ho et al.44 showed that in patients with cervical cancer, the mixed and integrated forms of HPV-52 and -58 was found in 25.0% and 12.5% of swabs, respectively, while 0HPV-16 and -18 was found in 82.6% and 100% of swabs, respectively. In addition, Van Tine et al.45 demonstrated that numerous integrated viral DNA copies are transcriptionally inactive, the DNA methylation being partially responsible for transcription silencing. At last, although an accumulation of integrated HPV genome was observed at few loci, a general integration hot spot could not be identified except for the common fragile sites. The cause for HPV integration seems to be rather related to the accessibility of these fragile genomic areas than due to selection of clones that harbour integrated HPV in regions with tumour relevant genes.46, 47 So, the viral integration of HR-HPV types does not seem to be a critical prerequisite in the development of cervical cancer.

In conclusion, our study, performed on a large number of cervical smears with a systematic histological control, shows that the RLU/CO value seems not to be an adequate marker for discriminating cervical lesions. Also, the viral load of HPV-16, determined by quantitative real-time PCR is of limited value for the diagnosis. HPV-16 genome integration occurs at an early step and we observed an increase of mixed and integrated forms with the severity of lesions. Nevertheless, the lack of significance for the discrimination of lesion grades with this parameter suggests that it could not be adequate for triage in a cost-effective fashion. At last, we emphasize the importance of a single HR-HPV infection associated with CIN2+. The interest of this parameter for the follow-up of infected women has to be evaluated in longitudinal studies.

Acknowledgements

We thank Pr. Christian Quéreux and the gynaecologists of his group and all the women who participated in this study. We thank Marie Martin, Isabelle Puteaud, Christelle Mangeonjean and Claire Kiletzsky, for excellent technical assistance. We thank the Bio-RAD company for assistance with HPV-16 plasmids.