Risk of cervical and other cancers after treatment of cervical intraepithelial neoplasia: retrospective cohort study

Introduction

Worldwide, cervical cancer remains one of the leading causes of death from cancer among women.1 In countries with organised screening programmes for cervical cancer, incidence rates and mortality have decreased by 60%-90%.2 All treatments for cervical intraepithelial neoplasia have excellent short term results and the differences are minimal,3 but only a few articles have studied long term outcomes after treatment. In the largest of these studies, the authors observed 2116 women for eight years after treatment and found that the incidence of cancer was reduced by 95%.4 In none of the studies, however, could follow-up data be linked to national cancer and population registries, and thus data on incidence of disease and mortality in treated women was unknown.

We assessed the incidence of cervical and other cancers long term in women treated for cervical intraepithelial cancer in Finland by linking primary data with two Finnish registries.

Methods

Our study is based on data of women treated for cervical intraepithelial neoplasia at Helsinki Central University Hospital, Finland during 1974 and 2001. Records for each patient included name, personal identifier, date and method of treatment, and diagnosis on the basis of histopathology.

The primary data consisted of 22 939 visits or treatments of 7599 women. We linked these data with the Finnish population registry and the Finnish cancer registry5 to identify cases of cancer. Follow-up was from six months after the first visit until death, emigration, or 31 December 2003. We chose a lag period of six months before diagnosing invasive cancer to exclude cancer diagnosed at the initial visit. After exclusions, 7564 patients remained for analysis.

The women were treated by knife or laser conisation, laser vaporisation, cold coagulation, or loop diathermy. At the first visit 2446 women were diagnosed as having CIN 1 precancerous lesions, 1543 as having CIN 2, 1334 as having CIN 3, and 2241 as having cervical intraepithelial neoplasia not otherwise specified.

We used cancer incidence rates in the population of southern Finland to calculate the expected numbers of cancer cases, stratified by sex, five year age groups, and five year calendar period. We present the results as standardised incidence ratios (ratio of observed to expected numbers of cases) with 95% confidence intervals (calculated on the presumption that the number of observed cases followed a Poisson distribution).

The mean number of visits per woman was 3.0 (range 1-31 visits). The mean number of visits for women with CIN 1 and CIN 2 lesions was 2.7 and 2.9, respectively, and for women with CIN 3 lesions it was 3.4. The mean age at the first treatment was 34.9 years (range 14-88 years). At the beginning of follow-up 43% of the patients were younger than 30, 52% were aged 30-59 (the group usually targeted for screening), and 5% were older than 60. The total follow-up time was 97 556 woman years. The average follow-up time was 11.9 years (range 0.5-28.0 years).

Results

We identified 448 new cases of cancer among 7564 women treated for cervical intraepithelial neoplasia—96 more cases than expected (table 1). Of these 96 excess cases, 26 were gynaecological cancers (standardised incidence ratio 1.5, 95% confidence interval 1.2 to 1.9). The risks were increased for cancers of the cervix (2.8, 1.7 to 4.2), vulva (4.1, 1.5 to 8.9), vagina (12.0, 3.9 to 28.0), lung or trachea (2.5, 1.9 to 3.5), other smoking related (1.7, 1.3 to 2.3), anus (5.7, 1.2 to 17.0), and any cancer (1.3, 1.2 to 1.4). Of the 22 cases of invasive cervical cancer 11 were diagnosed 0.5-9 years after treatment (2.7, 1.4 to 4.8), 10 after 10-19 years (3.1, 1.5 to 5.7), and one after 20 years (1.4, 0.04 to 8.0) (table 2). The standardised incidence ratios of overall cancer increased linearly with treatment of cervical intraepithelial neoplasia. We found a strong correlation between an increased risk of lung cancer and long time since treatment.

CIN 1 and CIN 2 precancerous lesions were associated with the highest risk of developing into invasive cervical cancer (3.1, 1.4 to 6.2 and 3.7, 0.8 to 10.9; table 1).

Of the eight patients with CIN 1 lesions who subsequently developed invasive cancer, five returned for one follow-up visit, two returned for two visits, and one returned for three visits. The three patients treated for CIN 2 lesions that subsequently developed into invasive disease returned for one, two, and four visits. The three patients with CIN 3 lesions who subsequently developed invasive cancer had two, three, and five visits.

Discussion

The incidence of invasive cervical cancer among women treated for cervical intraepithelial neoplasia was about 23 per 100 000 woman years. A previous large study on long term outcomes after treatment estimated the rate to be 85 per 100 000 woman years.4 In our study, follow-up of cancer incidence was based on a nationwide cancer registry, with systematic criteria for the reporting of invasive disease. We found that the relative risk of cervical cancer after treatment of preinvasive lesions was higher than that in the reference population, at least during the first and second decade of follow-up. This contrasts with that of the previous large study, which found that the risk of cancer did not increase during eight years' follow-up.

We also found that the risk of cervical cancer was increased in women diagnosed as having CIN 1 and CIN 2 lesions; the point estimates were higher than those for CIN 3 lesions. The explanation might lie in the general approach towards treatment and follow-up by grade of precancerous lesions. All lesions were treated without exception; however, patients with lower grade lesions (CIN 1 and CIN 2) are not followed-up in a similarly systematic and long lasting way as are patients with high grade lesions. Inadequate follow-up is a major factor, as suggested in the literature.6^–9

Treatment of cervical intraepithelial neoplasia is, however, effective. In an earlier study of CIN 3 lesions,10 it was estimated that 28-39% of cases without treatment would progress to invasive cancer. Our data contained 837 cases of CIN 3 lesions in women aged 30-59, thus on the basis of the previous estimate, 234-326 would develop into invasive cervical cancers, not the three observed. The treatment effect might have been nearly 100%.

Cancer coding is on bmj.com