Predictors of recurrence in high-grade cervical lesions and a plan of management.

2010 
Abstract Precis Positive endocervical margins are an important predictor of recurrence in high-grade cervical lesions, and though they do not always warrant retreatment, closer surveillance is recommended. Objective To identify predictors of recurrence and persistence of high-grade cervical dysplasia and to determine appropriate follow-up. Design: prospective pilot study. Setting: Gynaecological surgical center. Population Three hundred fifty-two patients were treated between 1999 and 2002 for high-grade lesions. Methods According to the accessibility of the transformation zone and the degree of dysplasia, patients were treated either by conization or by loop electrosurgical excision procedure (LEEP). Follow-up comprised colposcopy and Pap-smear screening 4–6 months after treatment as well as high-risk human papillomavirus (HR-HPV) testing before and after treatment. Main outcome measures underscore predictors of recurrence and propose a treatment flowchart for both management and follow-up. Results Of the 352 patients, 37 (10.5%) had true recurrence 6 months after initial surgical treatment and 6 patients (1.7%) had persistent lesions. Overall, 43 patients (12.2%) were considered as having recurrent disease. Patients were followed up for 5 years with a mean of 73 months. The most important predictor of recurrence was a positive HR-HPV test at 6 months postoperatively (odds ratio 38.8, 95% confidence interval 14.09, 107.05). The second significant predictor was positive endocervical margins and the third was positive pre-treatment HPV typing. A positive post-treatment HPV test had a more significant influence on risk than a positive test before treatment. Conclusion In agreement with recent findings, our study supports the usefulness of the HR-HPV test in the follow-up of treated high-grade lesions, especially when excision margins were positive.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    37
    References
    42
    Citations
    NaN
    KQI
    []