INTRODUCTION: To evaluate predicting clinical factors for regression and progression of cervical intraepithelial neoplasia (CIN) grade 2 (CIN2) in young women during two years of active surveillance. MATERIAL AND METHODS: This was a single-center prospective observational cohort study. Women under 31 years of age giving written informed consent with histologically confirmed CIN2 were followed with colposcopy, cytology, and biopsies every 6 months up to 24 months. At baseline, HPV genotyping was performed on cervical samples. The rates of regression and progression were recorded for every timepoint and at the end of study overall and stratified according to clinical factors and HPV genotypes at baseline. Risk ratio (RR) was used to estimate the relative risks for regression and progression. The study was registered in the ISRCTN registry (ISRCTN91953024). RESULTS: In total, 205/243 (84.4%) women completed the study. Complete regression (normal histology and/or normal or atypical squamous cells of undetermined significance (ASC-US) cytology) was detected in 64.4.% (n = 132) while 16.1% (n = 33) of the lesions progressed to CIN grade 3 (CIN3) or worse including 31 CIN3 cases, one adenocarcinoma in situ and one cervical cancer case. Factors associated with progression were initial large (>
50% of the transformation zone) lesion size, risk ratio (RR) 3.06 (95% confidence interval (CI) 1.40-6.69), and high-grade referral cytology RR 4.73 (95% CI 1.18-19.03). Compared with baseline HPV negativity or having only low-risk HPV genotypes present, high-risk HPV (hrHPV) positivity was associated with lower likelihood of regression RR 0.74 (95% CI 0.60-0.91). Age, cigarette smoking, use of combined oral contraceptives or baseline high-risk HPV genotype, including HPV16, were not associated with the outcomes. CONCLUSIONS: The majority of CIN2 lesions regress in young women. Women with large lesions and/or high-grade referral cytology should perhaps more often be treated instead of active surveillance. Initial hrHPV genotype does not appear to predict outcomes while not harboring hrHPV favors regression.