Accuracy of colposcopy-directed punch biopsies



We read with interest the systematic review by Underwood et al.[1] We would like to contribute our data to strengthen the conclusions drawn by Underwood et al. and try to clear an issue raised by the authors about finding the true false-negative of colposcopy because few women with negative punch biopsies underwent loop electrosurgical excision procedures (LEEP).

Our studies, published in August 2011, described 2681 women from 16 centres in Israel who underwent LEEP between 2001 and 2007 (Table 1).[2] In all, 1683 (63.2%) women had cervical intraepithelial neoplasia 2–3 (CIN 2–3) on cervical biopsy and 545(20.5%) had CIN 1 lesions. The colposcopy agreed with the LEEP diagnosis in 80.15% of the women with CIN 2–3 and in 43.67% of CIN 1 lesions. However, what may be more important is the level of under-diagnosis: cervical cancer occurred in 40 women (2.3%) treated because of CIN2–3, and in two women(0.4%) treated because of CIN 1, and CIN 2–3 was diagnosed in 113 (20.73%) of the women treated because of CIN 1. Overall agreement between punch biopsy and LEEP histology was 83.3% (1619/2279). Diagnosis of CIN 2–3 had the highest agreement level. There was moderate agreement (K value of 0.536).

Table 1. Indications for LEEP versus the final LEEP histology
Indications for LEEPLEEP final histology 2001–07
Total (%)Cervical carcinoma (%)CIN 2–3 (%)CIN 1 (%)Normal (%)
  1. Pap, Papanicolau; PCB, postcoital bleeding.

Carcinoma51 (1.9)31 (60.7)13 (25.5)1 (1.9)4 (7.8)
CIN 2–31683 (63.2)40 (2.3)1349 (80)156 (9.3)138 (8.2)
CIN 1545 (20.5)2 (0.4)111 (20.3)238 (43.7)194 (35.6)
Abnormal Pap smear281 (10.5)8 (2.8)121 (43.1)46 (16.4)106 (37.7)
PCB/others121 (4.5)3 (2.5)21 (17.3)32 (26.5)65 (53.7)
Total2681 (100)85 (3.2)1615 (60.2)474 (17.6)507 (18.9)

The authors remark that there is a problem of verification bias caused by over-representation of women with positive biopsies. We presented data about women who underwent LEEP without having cervical biopsy or when the cervical biopsy before the LEEP was normal. In 281 women (10.5%) an abnormal Papanicolaou smear was the indication for LEEP. In our study most of the operations were performed because of pathology discrepancies (persistent abnormal Papanicolaou smears and normal cervical biopsies) and about a third because of a cytological result of high-grade squamous lesion, which led to LEEP without previous biopsy (‘see and treat’ policy). A total of 121 women (4.5%) were treated mainly because of persistent postcoital bleeding after a normal investigation, which includes colposcopy, repeat normal Papanicolaou test and, in some of the women, cervical biopsy with a normal result. Cervical carcinoma was diagnosed in 2.8% in the former group and 2.5% in the latter and CIN 2–3 in 43.1% and 17.3%, respectively (Table 1).

These numbers provide a clue about the performance of colposcopy in negative cervical biopsies. We need to collect more data about the number of colposcopies and cervical biopsies that were performed, and the time delay that passed before the LEEP to obtain the true false-negative rate of colposcopy.