Objective
Conservative management of cervical intraepithelial neoplasia (CIN) grade 2 in women younger than 25 years may reduce overtreatment. However, long-term efficacy remains uncertain. This retrospective cohort study aimed to determine the rate of recurrence of high-grade abnormalities among young women with a history of CIN 2 that spontaneously regressed within 2 years and compare this with the rate of high-grade abnormality in similar women with an initial diagnosis of CIN 1.
Study Design
We identified all women aged younger than 25 years who were diagnosed with CIN 1 or CIN 2 between January 2005 and August 2009 within 2 colposcopy units. Follow-up data from the National Cervical Screening Programme were obtained to identify those women who developed recurrent high-grade lesions before October 2012. Comparisons were made using Cox proportional hazards regression.
Results
A total of 683 women were included: 106 with CIN 2 that spontaneously regressed, 299 with treated CIN 2, and 278 with conservatively managed CIN 1. Median follow-up was 4 years. There was no significant difference in the risk of development of high-grade abnormalities after 2 years between the spontaneously regressing CIN 2 and CIN 1 groups ( P = .83). Women with treated CIN 2 had a significantly lower risk of recurrence than women with untreated CIN 2 ( P = .01).
Conclusion
CIN 2 that has spontaneously regressed appears to behave as a low-grade lesion. This study contributes to the growing body of evidence that careful observation of CIN 2 is an efficacious and appropriate initial management option for women aged younger than 25 years at diagnosis.
Routine cervical screening of women under the age of 25 years is controversial and varies internationally. In New Zealand, all women are offered screening from the age of 20 years onward, whereas screening starts at age 18 years in Australia and age 25 years in the United Kingdom. The decision in the United Kingdom to start screening at age 25 years was made on the basis of an analysis that demonstrated no reduction in the incidence of cervical cancer among women screened under the age of 25 years. The US Preventive Services Task Force recommends commencing screening at age 21 years.
Given that the benefits of screening in this age group may be limited, if screening is to take place, there is a strong obligation to minimize the harms of unnecessary treatment.
Cervical intraepithelial neoplasia (CIN) grades 2 and 3 are collectively classified as high-grade squamous intraepithelial lesions (HSIL). Indeed, according to the Lower Anogenital Squamous Terminology, they are not distinguished. These lesions are generally managed similarly, using curative treatment. This generalized approach has historically been followed to ensure the safety of women; however, in the case of CIN 2, it is largely unsupported by firm evidence. Although CIN 3 has been convincingly shown to carry a high risk of subsequent cancer, CIN 2 is not so obviously precancerous.
The risk of overtreatment is highest in women under the age of 25 years because the incidence of CIN 2 peaks in this group, yet there is a low incidence of cervical cancer. Sasieni et al in 2009 estimated a less than a 1.5% risk of developing cervical cancer by age 25 years in women with untreated CIN 2 or 3 diagnosed at younger than 25 years. Given that CIN 3 has been shown to carry a higher risk of subsequent cancer than CIN 2, this risk is likely to be an overestimate when considering CIN 2 alone. Furthermore, it has been shown that 40-74% of CIN 2 in women younger than 25 years spontaneously regresses within 24 months.
In light of this fact and the harms of intervention (which may include obstetric complications ), international guidelines have shifted to recommend considering conservative management in young women with histologically proven CIN 2. The 2006 American Society of Colposcopy and Cervical Pathology consensus guidelines suggest a 6 month observation with colposcopy and cytology for 24 months for HSIL, followed by excision if the lesion is still present. Perkins et al in 2012 found that the introduction of this guideline reduced the use of loop electrosurgical excision procedures to treat moderate dysplasia in women aged 18-23 years from 55% to 18%. The 2012 American Society of Colposcopy and Cervical Pathology guidelines also agree with 24 months of follow-up.
Prior research has shown no progression to cancer among conservatively managed women with CIN 2 and a high rate of regression in both the adolescent and 20-25 year old age groups. However, these studies have encapsulated only short-term outcomes (up to 24 months of follow-up). There is therefore an unexplored possibility that, following apparent spontaneous regression, women with conservatively managed CIN2 may be at increased risk of recurrent high-grade abnormalities. This would suggest diminished efficacy of conservative management (with treatment being delayed rather than prevented), create a need for close observation beyond 2 years, and raise questions of safety. Investigation is therefore necessitated.
CIN 1 is generally accepted to be a manifestation of human papillomavirus (HPV) infection only, carrying a very low risk for subsequent malignancy. This group therefore provides a good measure of what an acceptable degree of recurrence is.
This retrospective cohort study aimed to determine the long-term rate of recurrence of high-grade abnormalities of the cervix among women with a history of spontaneously regressing CIN 2 and compared this with the rate of high-grade abnormality in similar women with an initial diagnosis of CIN 1.
As a secondary aim, we also intended to compare with recurrence in women who have received treatment for CIN 2.
Materials and Methods
Our study cohort was derived from 2 colposcopy units in New Zealand (Christchurch and Dunedin).
Data were extracted retrospectively; therefore, the study investigators had no influence on management decisions made regarding cohort members. All cohort members received a new histological diagnosis of CIN 1 or CIN 2 between January 2005 and August 2009. At this time, local management guidelines were in favor of a conservative approach to managing CIN 1, consisting of 12 month cytological surveillance for 24 months and in favor of immediate treatment for CIN 2.
A selection of cohort members with a diagnosis of CIN 2 were managed conservatively with 6 month cytological and colposcopic surveillance for 24 months; however, this decision was dependent on individual patient and clinician preferences.
Local colposcopy records at each unit were searched to identify all women who received a new histological diagnosis of CIN 1 or CIN 2 between January 2005 and August 2009 and who were aged younger than 25 years at the time of diagnosis. These records were also used to determine age at diagnosis, ethnicity, smoking status at diagnosis, parity at diagnosis, and to determine whether any woman had undergone hysterectomy.
Full cervical screening records for these women were obtained from the National Cervical Screening Programme register. This provided records of all smear tests and colposcopies received by the women, across all centers in New Zealand, in both the public and private health care sector, along with cytological and histological results, and a record of any treatments received, prior to October 2012.
All screening records were compared against inclusion and exclusion criteria for the cohort. Women were excluded if they had fewer than 2 years of available follow-up data, if they opted out of the screening register, or if they received treatment for an initial diagnosis of CIN 1. Women were also excluded if both their last result prior to 24 months and their first result after 24 months indicated a high-grade abnormality, with no documented treatment between these 2 results (ie, if there was no documented treatment or regression) because this would violate both conventional treatment and conservative management protocols. In practice this means that all women in the study cohort were free of known high-grade disease at 2 years of follow-up.
Women with an initial diagnosis of CIN 2 were divided into 2 groups: spontaneous regression and treatment.
The spontaneous regression group included all women in whom there was evidence of regression within 2 years of initial diagnosis and no evidence of treatment. The treatment group included all women who received some form of treatment for cervical dysplasia within 2 years.
Evidence of regression was defined as at least 1 histology or cytology result showing normal cervix or low-grade changes only (CIN 1, HPV infection only, low-grade squamous intraepithelial lesion, atypical squamous cells of undetermined significance), in the absence of a second sample taken concomitantly showing high-grade changes (CIN2, CIN 3, HSIL, atypical squamous cells: high grade squamous intraepithelial lesion cannot be excluded [ASC-H], adenocarcinoma in situ, and atypical glandular cells).
Treatments included local loop excision of the transition zone, cold knife cone biopsy, and laser ablation.
In accordance with our cohort selection criteria, all included women with CIN 1 did not receive any treatment within 2 years of initial diagnosis so were comparable in management with the CIN 2 spontaneous regression group.
We anticipated the total number of women with CIN 1 being too large to be logistically manageable. Accordingly, we obtained full screening records for only a proportion of the women with CIN 1 and excluded the remainder.
The selection of women with CIN 1 to be excluded was random but with a modification to create comparability between the CIN 2 spontaneous regression and CIN 1 cohorts in the duration of follow-up.
All women were stratified by year of initial diagnosis. For a given year, if the number of women in the CIN 1 group was less than 3 times the number of women in the CIN 2 spontaneous regression group, all of the women with an initial diagnosis of CIN 1 were included. If not, then women with an initial diagnosis of CIN 1 were randomly excluded by a researcher blinded to their long-term outcome, such that the total number of women in the CIN 1 group was 3 times the number in the CIN 2 spontaneous regression group. This process was carried out after biostatistical consultation.
Baseline characteristics of the cohort across the 3 groups were compared using χ 2 tests for categorical variables, and 1-way analysis of variance for continuous variables. Independent-sample Student t tests were used when comparing 2 groups only. When more than 1 follow-up event (defined as a colposcopy or community smear test) had occurred beyond 2 years, the average time between each event was calculated and compared between groups.
Outcomes were determined from cytological and histological results. Where both cytology and histology were available from 1 colposcopy visit, the histology result was used unless the histology result was normal and the cytology result abnormal, in which case the cytology result was used.
The primary outcome was development of high-grade changes (defined as cytological or histological evidence of CIN2/3, HSIL, ASC-H, adenocarcinoma in situ, atypical glandular cells, or worse) at least 2 years after the initial diagnosis. Accordingly, Kaplan-Meier curves were constructed starting at 2 years. A Cox proportional hazards model was used to calculate hazard ratios and confidence intervals and to adjust for potential confounders.
Ethics approval was obtained from the University of Otago Human Ethics Committee and regional authorization to use clinical data gained in both Christchurch and Dunedin.
Analyses were performed using SPSS 19.0 for windows (SPSS Inc, Chicago, IL).
Results
The final study cohort included 683 women: 106 (16%) with spontaneously regressing CIN 2, 299 (44%) with treated CIN 2, and 278 (41%) with conservatively managed CIN 1. Exclusions are outlined in Figure 1 .
Baseline data of the cohort are presented in Table 1 . Some of these data were missing because of individual clinicians not recording full demographic or background clinical features in the colposcopy database. The denominator for percentages presented is women with available data.
| Variable | CIN 1 (conservative management) | CIN 2 spontaneous regression | CIN 2 treatment | Sig ( P value) a |
|---|---|---|---|---|
| Age at diagnosis, y | ||||
| <20 | 46 (18%) | 32 (31%) | 45 (15%) | .002 |
| ≥20 | 214 (82%) | 72 (69%) | 253 (85%) | |
| Age range, y (mean) | 16–24 (21.4) | 16–24 (20.7) | 17–24 (21.7) | < .001 |
| Nulliparous at diagnosis | ||||
| Yes | 194 (83%) | 67 (80%) | 189 (76%) | .16 |
| No | 40 (17%) | 17 (20%) | 60 (24%) | |
| Hysterectomy during follow-up | ||||
| Yes | 0 (0%) | 0 (0%) | 0 (0%) | N/A |
| No | 178 (100%) | 75 (100%) | 165 (100%) | |
| Smoker at diagnosis | ||||
| Yes | 79 (42%) | 31 (36%) | 82 (46%) | .37 |
| No | 110 (58%) | 54 (64%) | 98 (54%) | |
| Ethnicity | ||||
| NZ European | 200 (81%) | 84 (88%) | 235 (82%) | .07 |
| NZ Maori | 19 (8%) | 8 (8%) | 27 (9%) | |
| Pacific Islander | 1 (0.4%) | 0 (0%) | 6 (2%) | |
| Other | 28 (11%) | 4 (4%) | 18 (6%) | |
| Average gap between follow-up visits, y | 1.07 | 1.02 | 1.20 | .020 |
| Mean duration of follow-up, y (range) | 4.08 (2.00–7.36) | 3.96 (2.02–6.92) | 4.55 (2.03–7.73) | < .001 |
| Median duration of follow-up, y | 3.86 | 3.69 | 4.45 | |
| Five year follow-up data available | 58 (21%) | 20 (19%) | 97 (32%) | .001 |
| Year of initial diagnosis | ||||
| 2005 | 40 (14%) | 12 (11%) | 63 (21%) | < .001 |
| 2006 | 30 (11%) | 16 (15%) | 69 (23%) | |
| 2007 | 78 (28%) | 23 (22%) | 76 (25%) | |
| 2008 | 83 (30%) | 29 (27%) | 61 (20%) | |
| 2009 | 47 (17%) | 26 (25%) | 30 (10%) |
a Determined from the Χ 2 distribution for categorical variables, and using one-way analysis of variance for continuous variables.
Women in the CIN 2 spontaneous regression group were significantly more likely than women in the other 2 groups to be aged less than 20 years at the time of diagnosis, but there was no significant difference between groups in ethnicity or rates of smoking and nulliparity. No women were documented as having undergone a hysterectomy at any point in time.
The median length of follow-up for the overall cohort was 4.1 years from time of initial diagnosis, with 26% of the women having at least 5 years of follow-up data available.
Significant differences in follow-up were found between the 3 study groups. However, as expected under the study design, there was no difference between the CIN 1 and CIN 2 spontaneous regression group with regard to percentage of women with 5 year follow-up data ( P = .76), in average length of follow-up ( P = .43), or year of initial diagnosis ( P = .25). There was also no difference between these 2 groups in frequency of follow-up ( P = .54).
Kaplan-Meier curves for the 3 groups are presented in Figure 2 . Hazard ratios for development of high-grade abnormalities (calculated from Cox proportional hazards regression) are presented in Table 2 . Adjustments were made for age at diagnosis, smoking status at diagnosis, ethnicity, and initial treatment center. After adjustment, no significant difference was found between the CIN 2 spontaneous regression and CIN 1 groups ( P = .83). However, the CIN 2 treatment group was found to have a significantly lower risk of high-grade abnormality than both the CIN 2 spontaneous regression and CIN 1 groups ( P = .01 and P = .001, respectively). The treatment center was found to have a significant effect on the risk of high-grade abnormality ( P = .007); no other covariate in the model was significantly associated with this outcome.
