Human papillomavirus 16 load in normal and abnormal cervical scrapes: An indicator of CIN II/III and viral clearance

Authors


  • This work is dedicated to the memory of Jan M.M. Walboomers, who passed away on 2 February 2000.

Abstract

The relation between human papillomavirus type 16 (HPV 16) viral load in cervical scrapes and development of high-grade cervical intraepithelial neoplasia (CIN II or III) was studied in a nested case-control study of women with normal cytology (group A) and in a cohort of women with abnormal cytology (group B). HPV 16 DNA load was determined using a quantitative real-time PCR assay. In group A, case women (women with CIN II/III, n = 12) had a significantly higher viral load than control women (women with CIN ≤ I, n = 47). This resulted in an increased relative risk of women with the 50% highest viral load for development of CIN II/III (OR 7.7; CI 1.6–33). In group B, women with CIN II/III (n = 38) had a significantly higher viral load than women with CIN ≤ I (n = 25). Women with the 50% highest viral load had an increased relative risk of CIN II/III (OR 3.2; CI 1.1–9.3) and a decreased chance of both viral clearance and cytologic regression. Our data suggest that in women with normal cytology an increased HPV 16 load confers an increased risk of developing a CIN lesion. A sustained high viral load is subsequently informative for progression to a high-grade CIN lesion. © 2002 Wiley-Liss, Inc.

A multitude of studies have established infection with high-risk human papillomavirus (HPV) types as a necessary cause of cervical cancer.1, 2 On the basis of these studies, it has been proposed that testing for the presence of high-risk HPV DNA in cervical scrapes as an adjunct of cervical cytology can be of great value in a variety of clinical settings.3 Persistence of high-risk HPV DNA appeared to be a prerequisite for the development of high-grade CIN lesions.4, 5 Nevertheless, the far majority of high-risk HPV infections are transient and will either not give rise to a lesion or at most lead to spontaneously regressing, low-grade cervical intraepithelial neoplasia (CIN).6, 7 Hence, there is a strong need for additional markers that predict the outcome of high-risk HPV infection with increased specificity. Using semiquantitative approaches, several studies have implicated that the amount of high-risk HPV DNA present in the cervical scrape, the viral load, can be of value to predict prevalent or incident high-grade CIN.8–16 This has recently been substantiated by a fully quantitative, real-time PCR-based retrospective study on cytomorphologically normal archival smears of women who developed cervical carcinoma in situ (CIS). It appeared that women with the highest HPV 16 load were at increased risk of developing CIS.17

However, with the aid of a fully quantitative method, no studies so far have addressed whether viral load could also be informative for women with abnormal cervical scrapes or to predict future clearance of high-risk HPV infections. Using frozen, freshly collected cervical scrapes collected from well-defined cohorts of women with normal18, 19 and abnormal cytology,5 we addressed the following questions. First, does an increased viral load confer an increased risk of developing CIN II/III in women with normal cytology? Second, is the viral load a good indicator of progression or regression of CIN lesions in women with abnormal cytology?

MATERIAL AND METHODS

Study groups

Group A: Women with normal cytology.

From 1988 to 1991, cohort studies of women with normal cytology took place in the Amsterdam area. Inclusion criteria are described in detail elsewhere.18, 19 A total of 3,932 women were included in these studies, 175 of whom had a single HPV 16 infection at baseline. Of 125 women with HPV 16 at baseline, frozen cervical scrape material was left. These included 12 women who developed CIN II/III (median age 34 years; range 21–43 years). For each of these 12 case women, we took 4 control women with CIN ≤ I lesion at the end of follow-up. One control baseline scrape was excluded because of a negative quantitative β-globin PCR result. This was not replaced. Therefore, 59 baseline scrapes were used for HPV 16 quantification. Upon revision by an experienced pathologist, all these baseline smears appeared unambiguously normal. The median age of control women at baseline was 36 years (range 22–54 years) and not significantly different from the age of case women (p = 0.12, Mann-Whitney U test). The median follow-up time of control women was 34 months (range 13–97 months) and this was not significantly different from that of the case women (median 32 months; range 19–85 months; p = 1.0, Mann-Whitney U test). For HPV 16 quantification, the scrapes taken at end of follow-up from 9 case women and 10 control women were also used. Study groups and scrapes included in the analysis are shown in Figure 1.

Figure 1.

Flow chart of group A women (normal cytology).

Group B: Women with cytomorphologically abnormal scrapes.

From 1990 to 1996, a cohort study of nontreated women with mild to severe dyskaryosis (n = 353) was performed in the Amsterdam area to investigate the natural history of CIN disease in relation to HPV.5, 20 At baseline and during follow-up, we performed cytologic and colposcopic examinations as well as HPV testing. No biopsies were taken during follow-up to avoid any interference with the natural course of the disease. Biopsies were taken when the primary end point, clinical progression, was reached or at the end of follow-up (median 33 months; range 2–72 months). Clinical progression was defined as CIN III, covering 3 or more cervical quadrants on colposcopy or a cytologic diagnosis suspected of cervical cancer. The secondary end point was end histology. The intermediate end point was viral clearance.

All women were referred to the gynecologist because of abnormal cytology (Fig. 2). During the examination by the gynecologist at study entrance, another cervical scrape was taken for cytology and HPV detection. This particular scrape is designated as baseline scrape. In 102 women, the baseline scrape tested positive for single HPV 16 infection. In 19 of them, this scrape was cytomorphologically normal. Although the viral load was also determined in these women, they were excluded from group B. Twenty additional women were excluded because no cervical scrape material was available anymore (n = 17) or quantitative β-globin PCR was negative (n = 3). Ultimately, group B consisted of 63 women with an HPV 16-positive abnormal baseline scrape. At the end of follow-up, histomorphologically CIN II/III was found in 38 women (age 32 years, range 22–54 years; CIN II, n = 2; CIN III, n = 36) and CIN ≤ I in the remaining 25 women (36 years, range 24–58 years; CIN I, n = 9; no lesion, n = 16). Since no biopsies were taken at study entrance or during follow-up for histomorphologic assessment of the underlying lesion, the CIN lesions detected at the end of follow-up in group B women may represent both incident and prevalent lesions.

Figure 2.

Flow chart of group B women (abnormal cytology).

HPV 16 clearance was defined as no HPV 16 in the last scrape or absence of HPV 16 DNA in at least 2 consecutive scrapes during follow-up, whereas HPV 16 persistence was defined as HPV 16 DNA presence in all scrapes analysed. Twenty-one women showed HPV 16 clearance, and 42 women showed HPV 16 persistence. In the group of women with CIN II/III lesions, 1 showed HPV 16 clearance and 37 showed HPV 16 persistence. In the group of women with CIN ≤ I, 20 showed HPV 16 clearance and 5 showed HPV 16 persistence.

Women with cytologic regression were defined as having at least 2 normal consecutive smears during follow-up. Cytologic regression was found in 29 women. These included 21 women with CIN ≤ I, 18 of whom showed HPV 16 clearance and 3 HPV 16 persistence, and 8 women with CIN II/III, 1 of whom showed HPV 16 clearance and 7 HPV 16 persistence (the total number of women with cytologic regression and the number of women with cytologic regression within CIN categories are not given in Figure 2). Thirty-four women showed persistence of abnormal cytology during follow-up. These included 30 with CIN II/III, all with HPV 16 persistence, and 4 women with CIN ≤ I. Of the latter, 2 showed HPV 16 clearance and 2 HPV 16 persistence (the numbers are not given in Figure 2).

In addition to baseline scrapes, follow-up scrapes of about half of the women with available follow-up material were randomly chosen for quantification (30 of 57). These included 19 of 38 women with CIN II/III (total number of scrapes tested: 46), all of whom had a persistent HPV 16 infection, and 11 of 19 women with CIN ≤ I (total number of scrapes tested: 30), 8 of whom showed HPV 16 clearance and 3 HPV 16 persistence.

The study protocols were approved by the ethics committee of the hospital, and enrollment of women in both normal and abnormal cytology cohorts took place only after having obtained informed consent.

Preparation of cervical scrapes for amplification

Cervical scrapes were collected in 1 ml Tris-HCl (pH 7.5) as described previously.21 DNA was isolated using the high pure PCR template preparation kit according to the recommendations of the manufacturer (Roche, Mannheim, Germany). Briefly, 100 μl PBS was added to 100 μl of cervical scrape suspension. This mixture was subjected to Proteinase K digestion (1.8 mg/ml Proteinase K; 2.7 M guanidine-HCl; 4.5 mM Tris-HCl, 4.5 mM urea, 9.1% triton X-100 [v/v]) at 72°C for 10 min. Isopropanol was added (18% v/v) and the DNA was extracted using a special glass-fiber-containing filter tube. The DNA was washed first with guanidine-ethanol (5 M guanine-HCl; 20 mM Tris-HCl, pH 6.6, 37.7% ethanol) and subsequently twice with NaCl-ethanol (20 mM NaCl, 2 mM Tris-HCl, pH 7.5, 88% ethanol). The DNA was eluted using prewarmed Tris buffer (10 mM Tris, pH 8.5, 70°C). DNA was subsequently stored at −20°C until use.

HPV 16 clone, human placental DNA and cell lines

The pHPV 16 plasmid clone containing full-length HPV 16 DNA was kindly provided by H. zur Hausen and L. Gissman (Heidelberg, Germany). Human placental DNA was purchased from Sigma (St. Louis, MO). The HPV 16-containing human cervical cancer cell lines SiHa and CaSki were obtained from the American Type Culture Collection (Manassas, VA). These cell lines with known HPV 16 copy number were used to validate the HPV 16 quantification by constructing a series of cell mixtures with the identical absolute number of HPV 16 copies but an increasing amount of HPV 16 copies per genome equivalent.

Primer selection and quantitative PCR for HPV 16 and β-globin

On the basis of DNA sequences obtained from the GenBank database, β-globin and HPV 16 oligonucleotides to be used for quantitative PCR were selected using the Primer Express software (ABI; Perkin Elmer, Foster City, CA), according to the TaqMan specifications. The nucleotide sequence of the selected HPV 16 oligonucleotides were forward PCR primer HPV16 L1-278F: 5′-CAGATACACAGCGGCTGGTTT-3′ (nt position 278–299 of HPV 16 L1 open-reading frame [ORF]); reverse PCR primer HPV16 L1-417R: 5′-TGCATTTGCTGCATAAGCACTA-3′ (nt position 396–417 of HPV 16 L1 ORF) and the TaqMan probe HPV16 L1-302pr: 5′-TGACCACGACCTACCTCAACACCTACACAGG-3′ (nt position 302–333 of the HPV 16 L1 ORF). The nucleotide sequence of the selected β-globin oligonucleotides were forward PCR primer beta-403f: 5′-TGGGTTTCTGATAGGCACTGACT-3′ (nt position 403–426 of the β-globin gene); reverse PCR primer beta-532r: 5′-AACAGCATCAGGAGTGGACAGAT-3′ (nt position 532–555 of the β-globin gene) and the TaqMan probe beta-471pr: 5′-TCTACCCTTGGACCCAGAGGTTCTTTGAGT-3′ (nt position 471–501 of the β-globin gene).

Amplification of the β-globin gene was used as a means to quantify the amount of cellular DNA in the sample. TaqMan probes were labeled, according to the TaqMan principle, with 6-carboxyfluorescein (FAM) at the 5′ end and 6-carboxytetramethylrhodamine (TAMRA) at the 3′ end.

Amplification and detection were performed with an ABI Prism 7700 sequence detection system (Applied Biosystems, Perkin Elmer, Foster City, CA). The PCR was initiated by a preincubation step at 50°C for 2 min to activate the uracil N′-glycosylase (UNG), followed by an incubation at 95°C for 10 min to inactivate UNG and release the DNA polymerase (AmpliTaq Gold, Perkin Elmer). Subsequently, 50 2-step PCR cycles were run consisting of a denaturation step at 95°C for 15 sec and an annealing step at 50°C for 1 min. All samples were tested in duplicate and the results averaged. The threshold cycle number (Ct) was calculated with Sequence Detection System software (Applied Biosystems, Perkin Elmer). Standard curves were generated by plotting Ct values against known concentrations of input DNA; human placental DNA was used for the β-globin assay (100, 10 and 1 ng dilutions) and purified pHPV 16 DNA for the HPV 16 assay (1 pg, 100 fg and 10 fg dilutions). The amount of β-globin DNA (ng) was calculated from the Ct values of sample DNA, using the standard curve of human placental DNA. Similarly, the amount of HPV 16 DNA (fg) was calculated from the Ct values obtained for sample DNA, using the standard curve of pHPV 16. For calculating the number of copies per scrape (c/s), the amount of HPV 16 DNA in fg was multiplied by the dilution factor (200) and divided by 2.2 × 10−2 fg, the weight of pHPV 16 viral genome with vector (i.e., number of copies per scrape = amount of HPV 16 DNA (fg) × 200 / 2.2 × 10−2). For calculation of the number of genome equivalents, the amount of β-globin DNA in ng was multiplied by the dilution factor of 200 and divided by 6 × 10−3 ng, the weight of 1 genome equivalent and a factor 2 to compensate for the 2 β-globin DNA copies per genome equivalent (i.e., number of genome equivalents per scrape = amount of β-globin DNA × 200 / 6 × 10−3 × 2).

Seven of 141 scrapes (5%), originally tested positive with the GP 5+/6+ PCR, were negative in the quantitative HPV 16 PCR. For calculations, the HPV 16 load of these women was set to 5 c/s (threshold value).

Statistical analysis

Age, follow-up time, levels of β-globin and HPV 16 DNA were compared using the Mann-Whitney U test. If groups were compared with more than 1 result per woman in the same time group, 1 result was included in the statistics (Fig. 3).

Figure 3.

Viral load in box plots in different time groups before end of follow-up. (a) Median HPV 16 loads in scrapes of group A women with CIN ≤ I (controls) vs. those with CIN II/III (cases). (b) Median HPV 16 loads in scrapes of group B women with CIN ≤ I vs. those with CIN II/III. (c) Median HPV 16 loads in scrapes of group B women with HPV 16 clearance vs. those with persistence. Filled boxes indicate CIN II/III (a,b) and viral persistence (c); the open boxes CIN ≤ I (a,b) and viral clearance (c). The upper boundary of the box represents the 75th percentile and the lower boundary the 25th percentile. The centerline in the box represents the median and the whiskers represent the minimum and maximum values. Asterisks denote differences with statistical significance. Dashed lines separate time groups. Note that in (c) for women with viral clearance no values are given at time point zero since no HPV 16 DNA was present anymore in the corresponding scrapes.

The results of HPV 16 load comparisons are given together with those of β-globin comparisons in the samples. In group A, Cox regression analysis with adjustment for β-globin levels was used for risk assessment of CIN II/III development. For analysis of viral load in relation to time, baseline scrapes were dichotomized into 2 time groups: scrapes taken 13–34 months (median 25 months; 7 cases and 24 controls) and 35–97 months (median 48 months; 5 cases and 23 controls), respectively, before end point (see below).

In group B, Cox regression analysis with adjustment for β-globin levels was used for risk assessment of HPV 16 clearance and cytologic regression.

Logistic regression analysis with adjustment for β-globin levels was used to study viral load in relation to development of CIN II/III. Time of viral clearance was defined as the midpoint between the time of the last HPV 16-positive scrape and the subsequent first HPV-negative scrape. If HPV was not cleared, the women were censored at end of follow-up.

For analysis of viral load in relation to time, baseline and follow-up scrapes were divided into 3 time groups: 3–12 months (median 6 months), 13–24 months (median 16 months) and 25–59 months (median 37 months) before end of follow-up. In case women had more than 1 scrape per time group, the value of a randomly chosen scrape was used for statistical analysis. P-values of less than 0.05 were considered significant. All analyses were performed using SPSS 9.0 software.

RESULTS

Validation of the HPV 16 quantification

Validation of the HPV 16 quantification was performed as described previously22 by mixing DNA from the cell lines SiHa and CaSki in a set of 6 known proportions. These cell lines contain approximately 1–10 and 400–600 copies of HPV 16 per cell, respectively. A series of cell mixtures were constructed with an identical absolute number of HPV 16 copies but an increasing amount of HPV 16 DNA copies per genome equivalent (from 1–10 to 400–600 HPV 16 copies per genome equivalent). Each of the cell mixtures was tested in duplicate. The results are shown in Table I. The mean numbers of HPV 16 copies per genome equivalent in the different cell mixtures were in the expected range. CaSki DNA and SiHa DNA were found to contain on average 660 (±14.8) copies and 1.1 (±0.1) copy of HPV 16 per genome equivalent, respectively.

Table I. Validation of Quantitative PCR, Using Mixtures of Cervical Cancer Cell Line Cells With Known Copy Number/Cell
Number of cells (total 105 HPV 16 copies)Mean copy number of HPV 16 per genome equivalent
Expected (min-max)Quantified (mean ± SD)
200 CaSki + 0 SiHa400–600660 ± 14.8
190 CaSki + 1,000 SiHa65–104127 ± 17.4
175 CaSki + 2,500 SiHa27–4833 ± 8.2
150 CaSki + 5,000 SiHa13–2728 ± 1.3
100 CaSki + 10,000 SiHa5–168 ± 0.2
0 CaSki + 20,000 SiHa1–101.1 ± 0.1

HPV 16 load in women of group A

At baseline the median HPV 16 load of all women was 2.4 × 104 c/s. The median HPV 16 load in case women (5.9 × 105 c/s, range 1.3 × 102–4.5 107 c/s) was significantly higher (p = 0.04) than in control women (8.9 × 103 c/s, range 5–3.9 × 108 c/s). The median β-globin values were similar for case women (1.1 × 105 genome equivalents per scrape, range 1.3 × 103–2.3 × 106) and control women (1.3 × 105 genome equivalents per scrape, range 2.5 × 103–7.9 × 105; p = 0.3).

Ten of the case women (83%) vs. 19 of the control women (40%) belonged to the group of women with the 50% highest amount of HPV 16 DNA (>2.4 × 104 c/s). Using the β-globin values for adjustment of differences in amount of input DNA, this results in a 7 times increased relative risk (OR 7.7; CI 1.6–33;Table II) of women with the 50% highest load to develop CIN II/III.

Table II. Development of CIN II/III in Group A and B Women and Viral Clearance and Cytologic Regression in Group B Women
HPV 16 load (c/s)Group A CIN II/III OR (95% CI)1Group B CIN II/III OR (95% CI)HPV 16 clearance OR (95% CI)Cytologic regression OR (95% CI)
  • 1

    Odds ratio with 95% confidence interval.

≤2.4 × 1041 (reference)
>2.4 × 1047.7 (1.6–33)
≤4.3 × 1061 (reference)5.0 (1.7–15)2.6 (1.2–5.6)
>4.3 × 1063.1 (1.1–9.3)1 (reference)1 (reference)

For assessment of the relation between HPV 16 load and time to end of follow-up, baseline scrapes were stratified into 2 time groups of 13–34 months (median 25 months) and 35–97 months (median 48 months) before end of follow-up. The median HPV 16 load in relation to time before end of follow-up is shown in Figure 3a. At the end of follow-up, the viral load in 9 case women (1.0 × 106 c/s, range 1.0 × 103–1.2 × 108 c/s) was significantly higher (p = 0.02) than in 10 control women (4.1 × 104 c/s, range 3.6 × 102–1.6 × 106 c/s). In the 2 time groups, case women did not reveal a significantly higher median HPV 16 load than control women (13–34 months group, 6.3 × 105vs. 2.3 × 104 c/s, p = 0.25; 35–97 months group, 5.6 × 105vs. 3.5 × 103 c/s, p = 0.09).

HPV 16 load at baseline in women of group B (abnormal cytology)

Excluded from group B were 19 women who, despite an abnormal referral smear, had an HPV 16-positive but cytomorphologically normal baseline scrape at study entrance. The median viral load in their baseline scrapes was 3.8 × 105 c/s, which is not significantly different from scrapes of group A women (p = 0.5).

The median HPV 16 load in baseline scrapes of group B women was 4.3 × 106 c/s (range 5–4.0 × 109 c/s). This was almost 200-fold higher than the median HPV 16 copy number in baseline scrapes of group A (p < 0.005). On the other hand, this value was not significantly different from the median load in the abnormal scrape taken at the end of follow-up of group A women who developed abnormal cytology (4.3 × 106vs. 9.1 × 105 c/s; p = 0.7). In group B women, the median viral load at baseline was 7-fold higher (p = 0.03) in women with CIN II/III (6.9 × 106 c/s, range 5–4.0 × 109) than in those with CIN ≤ I (9.6 × 105 c/s, range 5–8.6 × 107). The median β-globin values in both groups were similar: 2.4 × 105 genome equivalents per scrape (range 3.8 × 103–4.2 × 106) in women with CIN II/III and 4.4 × 105 genome equivalents per scrape (range 9.7 × 103–6.6 × 106) in women with CIN ≤ I (p = 0.5).

Thirty-one women belonged to the group with the 50% highest viral load (i.e., >4.3 × 106 copies in the baseline scrape), 23 of whom had CIN II/III and 8 CIN ≤ I at the end of follow-up. Of the 32 women with the 50% lowest viral load, 15 had CIN II/III and 17 CIN ≤ I. Using the β-globin values for adjustment of differences in amount of input DNA, women with the 50% highest HPV 16 load had a 3 times increased relative risk (OR 3.1; CI 1.1–9.3) for CIN II/III (Table II).

HPV 16 load in group B women in relation to HPV 16 clearance and cytologic regression

Twenty-one women showed HPV 16 clearance and 42 women showed HPV 16 persistence. At baseline, the median viral load in women with viral clearance (2.8 × 105 c/s, range 5–7.1 107) was significantly lower (p < 0.005) than in women with viral persistence (9.1 × 106 c/s, range 2.7 102–4.0 109). The median β-globin values were similar in both groups (p = 0.3). Seventeen of 21 women (81%) with HPV 16 clearance vs. 15 of 42 women (36%) with HPV 16 persistence belonged to the women with the 50% lowest viral load (<4.3 × 106 c/s). This results in a 5 times higher chance for HPV 16 clearance (OR 5.0; CI 1.7–15) for women with the 50% lowest viral load (Table II).

Twenty-nine women showed cytologic regression, and in 34 women abnormal cytology persisted. At baseline, the median viral load in women with cytologic regression (9.6 × 105 c/s, range 5–4.3 × 108 c/s) was significantly lower (p < 0.005) than in women with persistence of abnormal cytology (1.0 × 107 c/s, range 4.9 × 103–4.0 × 109). The median β-globin values were similar in both groups (p = 0.7). Twenty of 29 women (69%) with cytologic regression vs. 12 of 34 women (35%) with persistence of abnormal cytology belonged to the women with the 50% lowest viral load (<4.3 × 106 c/s). This results in a 2 times higher chance for cytologic regression (OR 2.6; CI 1.2–5.6) for women with the 50% lowest viral load (Table II).

The course of HPV 16 load in time determined on baseline and follow-up scrapes of group B women

For assessment of the relation between HPV 16 load and time to end of follow-up, baseline scrapes were stratified into 3 time groups of 3–12 months (median 6 months), 13–24 months (median 16 months) and 35–97 months (median 59 months) before end of follow-up. Viral load levels for time groups are shown in Figure 3b and c. The median viral load of women with CIN II/III (3.6 × 106 c/s, range 4.2× 10 3–3.0 × 108) was higher (p = 0.05) than in women with CIN ≤ I (3.9 × 104 c/s, range 6.5 × 103–3.6 × 105) in the scrapes taken at the end of the study. Also in the time group representing scrapes taken 3–12 months before the end of the study, the median viral load in women with CIN II/III (1.2 × 107 c/s, range 5.8 × 104–4.0 × 109 c/s) was significantly higher (p < 0.005) than in women with CIN ≤ I (5.8 × 104 c/s, range 5–5.6 × 107 c/s). Conversely, the median viral load in the earlier time groups was not significantly different between women with CIN II/III and CIN ≤ I (2.4 × 106vs. 9.9 × 105 c/s, respectively, at 13–24 months before the end of the study, p = 0.5 and 5.0 × 106vs. 3.1 × 106 c/s, respectively, at 25–59 months, p = 1.0).

The median viral load of women with HPV 16 clearance (4.2 × 104 c/s, range 5–5.6 × 107) was significantly lower (p < 0.005) than in women with HPV 16 persistence (4.8 × 106 c/s, range 9.1 × 103–4.0 × 109) in the time group representing scrapes taken 3–12 months before the end of follow-up. The median viral load in the earlier time groups was not significantly different between women with viral clearance and persistence (1.0 × 106vs. 2.3 × 106 c/s, respectively, at 13–24 months before the end of the study, p = 0.5 and 1.1 × 106vs. 4.7 × 106 c/s, respectively, at 25–59 months, p = 0.9).

DISCUSSION

Data obtained in our study by quantitative PCR for HPV 16 are in favour of the following concept: Women with normal cervical smears and an increased HPV 16 load have a higher risk of developing a CIN lesion. Subsequently, once a lesion has developed, a sustained or increased viral load is predictive of a greater chance of viral persistence and progression to a CIN II/III lesion, whereas a decrease in load increases the chance of HPV 16 clearance and regression of the CIN lesion (Fig. 4). This concept is supported by the following results from our study. First, in women with originally normal cytology who develop abnormal cytology (group A), the median HPV 16 load in scrapes taken at the end of the study was increased and not significantly different from that of women with abnormal cytology at baseline (group B). Second, we found that once a lesion has developed (baseline group B), a decrease in HPV 16 load precedes viral clearance by 6 to 16 months. This phenomenon is associated with regression of the lesion, which is supported by the fact that lower HPV 16 loads in women with abnormal cytology also gave rise to a higher chance of cytologic regression. Third, women with initially abnormal cytology but with normal cytology at baseline who were excluded from group B had a median viral load that was significantly lower than the load found in baseline scrapes of group B women (3.8 × 105 c/s vs. 4.3 × 106 c/s, p = 0.007). Conversely, this load was not significantly different from the viral load at baseline found in group A women. In addition, 19 of the 21 women (90%) with abnormal cytology who showed viral clearance had a normal Pap smear at the end of the study (data not shown). This tight link between viral clearance and cytologic regression has recently been demonstrated.23, 24 Viral clearance has been found to become manifest at a median of 3–4 months before cytologic regression.23, 24 The data obtained in our study suggest that women who will undergo regression could be identified at an earlier stage when including a quantitative HPV test.

Figure 4.

Concept of HPV 16 load and progression of CIN lesions in women with normal and abnormal cytology.

In our study, we used a real-time PCR assay for quantification of HPV 16 DNA and β-globin DNA. The quantitative character of these assays was validated using mixtures of 2 HPV 16-positive cervical carcinoma cell lines (SiHa and CaSki) of which the approximate number of HPV 16 copies per cell is known. The mean copy numbers of HPV 16 per cell calculated on the basis of the real-time PCR results were similar to the predicted copy numbers for these cell lines. Recently, we obtained similar results with an alternative, though more laborious, competitive PCR-EIA (Q-PCR-EIA) technique for quantification of HPV 16 load.22

The lack of differences in β-globin values between the various categories of women in our study makes it unlikely that statistically significant differences in HPV 16 load are the result of differences in the amount of amplifiable DNA in the scrapes. In addition, our data obtained from cases and controls amongst women with normal cytology are in good agreement with those collected by Josefsson et al.17 and Ylitalo et al.,25 who used a similar real-time PCR methodology for quantification.17, 25 The results obtained in our study do not only confirm but also reinforce previous data on women with normal cytology17 since our analyses were performed on fresh-frozen, prospectively collected scrapings, stored at −80°C, rather than archival Pap smears. Therefore, a possible confounding effect of the use of archival smears can be excluded.

In group A women, after dichotomization of baseline scrapes into 2 time groups, viral load was higher in case women than in control women, although this difference was not significant. The latter is probably due to small sample size. In agreement with this, Ylitalo et al.25 have found, using a larger sample size, that women who developed CIS had a significantly increased HPV 16 load already 13 years or more before diagnosis. It is of note, however, that the study of Ylitalo et al.25 did not include women with regressing CIN lesions. Here, we showed that the HPV 16 load only declines up to 16 months before regression of the lesion becomes manifest. In fact, an increased chance of cytologic regression in case of a low load was accompanied by an increased chance of viral clearance and decreased risk of CIN II/III. This is fully consistent with the finding that HPV persistence is a prerequisite for the development and maintenance of CIN II/III.4, 5

At the moment, we are evaluating the potential role for HPV testing as an adjunct to cytology for primary screening of cervical cancer. In particular, we are interested to see whether viral quantification may be useful as an indicator of progressive lesions to add to current HPV testing protocols. Viral quantification may also be of value for the management of women with borderline or mild dyskaryosis (BMD). The current policy in the Netherlands for women with BMD is to repeat the smear after 6 months. On the basis of our data it could be recommended to refer women with >4.3 × 106 HPV 16 c/s directly for colposcopy-directed biopsy since these women are at increased risk of high-grade CIN. Still, it is of note that, at the individual level, a large variation in viral load was evident amongst women of all categories, indicating that a lower viral load does not necessarily exclude progressive CIN disease. Consequently, a repeat smear is recommended for women with ≤4.3 × 106 HPV 16 c/s, rather than referring them back to the cervical cancer screening program. Prospective studies are necessary to determine whether other (semi-) quantitative methods for determination of viral load, such as Hybrid Capture II, could be of help in limiting referral of HPV-positive women with BMD for colposcopy-directed biopsy.

Acknowledgements

For clinical data, we are indebted to Dr. A. Remmink. We are grateful to Mrs. N. Fransen-Daalmeijer and Mr. R. Pol for excellent technical assistance.

Ancillary