Persistent infection with human papillomavirus following the successful treatment of high grade cervical intraepithelial neoplasia

Authors


*Dr M. E. Cruickshank, Wellbeing Centre for the Prevention of Cervical Cancer, Gynaecology Oncology Unit, Ward 43, Aberdeen Royal Infirmary, Aberdeen AB25 7ZD, UK.

Abstract

In a nested case–control study, we identified women who were successfully treated for CIN 3. Cases had biopsy proven recurrence, whilst controls remained disease free for at least five years. One hundred and seventy-two women were β-globin positive at diagnostic biopsy and at six-month post-treatment smear (90 controls and 82 cases). Thirty-nine cases (47.6%) were HPV16/18 positive at biopsy or follow up smear and 14 (17.1%) of 82 were positive at both. Of the controls, 37 (41.1%) were HPV positive at biopsy or smear with 3 (3.3%) positive at both. The unadjusted OR associated with being HPV positive at both points compared to being HPV negative at both was 8.0 (95% CI 2.13–30.37). The persistence of HPV 16/18 infection following the confirmed eradication of CIN is a highly significant risk factor for recurrence.

Introduction

Human papillomavirus (HPV) is the most important factor in the development cervical cancer. Women treated for cervical intraepithelial neoplasia (CIN) remain at increased risk of developing invasive cancer and currently annual smears are recommended for five years post-treatment surveillance to detect any new disease1. Recognition of this risk has promoted the suggestion that annual surveillance be extended to 10 years2. Previous small studies have demonstrated HPV persistence following treatment of genital warts and CIN3–5 in only a small proportion of women. Although it was concluded that HPV was often eradicated by treatment, the rate of HPV persistence matches that of recurrence. An alternative explanation is that oncogenic HPV persistence promotes the development of new disease. This would mean that HPV persistence should be a risk factor for recurrence even when CIN lesions are successfully treated. However, studies to date have included women with residual disease and information on women completing five-year follow up has not been reported. This study aimed to test the hypothesis that HPV status at treatment and six-month follow up is a predictor of disease recurrence in women successfully treated for high grade CIN.

We conducted a case–control study nested in a cohort of 3127 Grampian residents treated for CIN between 1982 and 1992 by either laser ablation or diathermy loop excision. We identified cases and controls from women who were treated for CIN 3 and who had negative smear and normal colposcopic examination at six months post treatment. This was to exclude residual disease from inadequate treatment. All biopsies were independently reviewed to confirm the original diagnosis of CIN 3. The presence of amplifiable DNA was assessed using beta-globin in the biopsies and in the six-month post-treatment smears. We then identified women who subsequently developed recurrent high grade CIN or invasive cancer which had been confirmed histologically on biopsy. This resulted in 108 women who formed the cases. Controls included 101 women who remained free of disease for a minimum of five years after treatment and who were matched to cases by year of treatment and treatment type. HPV types 16 and 18 were detected using polymerase chain reaction and type-specific primers6. Logistic regression methods were used to compute odds ratios and 95% confidence intervals, using STATA; goodness-of-fit was assessed by the Hosmer and Lemeshow statistic7. Ethical approval was obtained from the Grampian Joint Ethical Committee.

Results

The number of cases and controls included in the study are shown by treatment method in Table 1. The median age of both cases and controls was 31 years (case range 19–49; controls range 18–61). Table 2 shows unadjusted and adjusted odds ratios and 95% confidence intervals associated with HPV status at diagnostic biopsy and of the six-month post-treatment smear. The inclusion of age and type of treatment in the regression models did not alter the odds ratios significantly. None of the adjusted models was an inadequate fit to the data. At diagnostic biopsy, 176 women had detectable beta-globin. Forty-four of 83 cases were positive for HPV16 or 18 (53.0%), compared with 35 of 93 controls (37.6%). The unadjusted odds ratio for HPV positive at biopsy compared with HPV negative was 1.9 (95% CI 1.02–3.41).

Table 1.  Number of cases and controls included in the study by treatment method.
 LoopLaserTotal
Cases5454108
Controls5150101
Total105104209
Table 2.  Numbers of cases and controls by HPV status, unadjusted and adjusted odds ratios (OR) with 95% confidence intervals (CI), at diagnostic biopsy and six month post-treatment smear.
 No. of casesNo. of controlsUnadjusted OR95% CIAdjusted OR195% CI
  1. 1Adjusted for treatment (loop or laser) and age at treatment (<25 years, 25–29, 30–34, 35–39, 40 and older).

  2. 2Hosmer and Lemeshow goodness-of-fit χ2 (8df) = 12.76; P= 0.121.

  3. 3Hosmer and Lemeshow goodness-of-fit χ2 (7df) = 5.97; P= 0.544.

  4. 4Hosmer and Lemeshow goodness-of-fit χ2 (8df) = 9.84; P= 0.277.

HPV status at diagnostic biopsy
−ve3958112
+ve44351.91.02–3.411.81.00–3.41
HPV status at 6-month post-treatment smear
−ve7686113
+ve31122.91.40–6.093.11.44–6.84
HPV status at diagnostic biopsy and 6-month post-treatment smear
−ve on both2950114
+ve at biopsy or smear39371.80.96–3.451.80.94–3.50
+ at biopsy and smear1438.02.13–30.378.72.18–34.62

At the six-month post-treatment smear, 31 of 107 cases (29.0%) were HPV16/18 positive compared with 12 of 98 controls (12.2%). The unadjusted odds ratio for HPV positive compared with HPV negative was statistically significantly higher (OR = .9 (95% CI 1.40–6.09).

One hundred and seventy-two women were beta-globin positive at diagnostic biopsy and at the six-month post-treatment smear (90 controls and 82 cases). Thirty-nine cases (47.6%) were HPV16/18 positive at diagnostic biopsy or six-month smear and 14 (17.1%) of 82 were positive at both. Of the controls, 37 (41.1%) were HPV positive at biopsy or smear with three (3.3%) positive at both. The unadjusted odds ratio associated with being HPV positive at both points compared with being HPV negative at both was 8.0 (95% CI 2.13–30.37).

Discussion

The persistence of HPV 16/18 cervical infection following the confirmed eradication of CIN is a highly significant risk factor for recurrence. In contrast to previous follow up studies, we did not find that age was a significant factor. However, the majority of our participants were premenopausal and inadequately treated disease, which is likely to be harboured in the endocervical canal in older women, was excluded. Margin status is also a risk factor for treatment failure. We included data for those women treated by LLETZ, but there were insufficient numbers with categorical reporting of disease-free resection margins for a meaningful analysis. Again, the involvement of resection margins following excisional treatment is more likely to predict residual disease than a new recurrence.

Screening cytology may not be the most effective means of monitoring women after treatment not least because of the recognised problem of false negative results. In population screening, this balance between sensitivity and specificity is accepted to reduce the number of false positives and unnecessary interventions. However, women treated for high grade CIN are a high risk group for invasive cancer. They have already demonstrated exposure and susceptibility to HPV and other possible co-factors for the development of the pre-invasive stage and must be of clinical concern. Persistence of HPV infection will continue to promote disease development and may initially be undetected by conventional cytology or colposcopy follow up when there are no apparent residual foci of CIN. Indeed, the high success rate of local treatments for CIN may result from clearance of oncogenic HPV rather than solely the planned removal of more localised CIN lesions. The means for HPV eradication are not fully understood but could result from triggering a local immune response.

These results suggest that post treatment surveillance could be targeted at women who remain HPV positive. A likely strategy would incorporate dual testing with conventional cytology and HPV status at six and 12 months post-treatment. Women, who remain negative for both tests, would be considered as very low risk of recurrence and could then return to routine recall. Based on our results, this could over a five-year period reduce the number of smears taken by 37% and, where colposcopy is routinely performed at first follow up, a 74% reduction in colposcopies. Additional savings are likely as default increases for each additional follow up8.

Conclusion

Our findings suggest that surveillance could be targeted at women who remain HPV 16 or 18 positive after treatment, while allowing low risk women who are HPV negative to return to routine surveillance at an earlier stage. A randomised controlled trial would provide robust evidence on this issue.

Acknowledgements

The authors would like to thank Dr I. Miller and Dr S. Payne for reviewing the pathology specimens used in this study. This research was funded by a grant from the Acute Healthcare Committee of the Scottish Office Department of Health.

Ancillary