The role of frozen section examination of conisations in the management of women with cervical intraepithelial neoplasia


*Correspondence: Dr F. Bretelle, Service de Gynécologie Obstétrique B Pr. Blanc, Centre Hospitalier Universitaire de la Conception, 147 boulevard Baille, 13385 Marseille Cedex, France.


Objective To determine whether frozen section in conisation improves the management of cervical intraepithelial neoplasia.

Design Randomised trial.

Setting Department of Obstetrics and Gynaecology, Conception Hospital, Marseille, France.

Population Patients referred for treatment or diagnosis of cervical intraepithelial neoplasia.

Methods Two hundred and three patients requiring conisation for diagnosis or treatment of cervical pathology were randomly assigned to have (n= 102) or not (n= 101) a frozen section of the endocervical margin of the cone specimen.

Main outcome measures The principal criterion was the 12 month residual or recurrent high grade disease rate.

Results At 12 months, the residual or recurrent disease rate of high grade lesions was 12.6% in the group that did not have frozen sections and 1% in the group that did (relative risk, RR, 11.9, 95% CI 1.6–89.5, P= 0.0025). The corresponding rates of margin involvement were 18.9% and 1%, respectively (RR 15.13, 95% CI 2.06–111.27, P= 0.0002). The groups did not differ as to the height of the cone (13.6 [5.6] mm vs 13.7 [4.7] mm, P= 0.75) or post-operative morbidity (6% vs 7%, RR 1.01, 95% CI 0.94–1.09, P= 0.8). The rate of patients lost to follow up was similar in both groups 17% vs 12% (RR 1.42, 95% CI 0.71–2.81, P= 0.31).

Conclusion Frozen section is effective in conisation and significantly reduces residual or recurrent high grade disease. This finding improves the management of cervical pathology achieving immediate clear margins in most of conisations. This is of great interest because high rates of patients are lost to follow up as we and others authors have found.


Conisation has two simultaneous aims: diagnosis and treatment of moderate and severe cervical intraepithelial neoplasia (CIN). It enables the seriousness of the lesion to be assessed and, in most cases, also constitutes its complete treatment. The difficulty occurs when the margins of the excision include dysplastic tissue. Some authors report a rate of margin involvement up to 50%1.

The current trend is thus to use cytology and colposcopy to monitor patients in whom margin involvement is observed; further surgery is performed only if abnormalities persist2–4. In some cases, the lesion may regress spontaneously. Nonetheless, residual disease frequently persists after cone excision with margin involvement2,5,6: it is shown in up to 90% of cases7. Moreover, margin involvement is a recognised factor in treatment failure7–9, while negative or free margins indicate a good prognosis10. A frozen section biopsy of the margins of the excised cone seems to be a useful one-step procedure for determining the margin status during surgery and enabling immediate further excision to complete it. Nonetheless, frozen section is rarely used in this way. Only one, short, retrospective series has shown its advantages11. Its use most often resulted in negative margins. In that study, it was unclear whether the frozen sections reduced the presence of residual or recurrent disease after conisation.

For this reason, we carried out a prospective, randomised trial to test the hypothesis that the use of frozen sections during conisation would reduce the 12 month rate of residual or recurrent lesion.


Between October 1995 and December 1999, we invited patients at Conception Hospital who required conisation of the cervix to participate in this randomised clinical trial. The inclusion criteria were CIN II or III, CIN I with the squamocolumnar junction not visible, ASCUS cytology or cervical smears showing dysplasia without biopsy confirmation. Patients were not eligible if they were younger than 18 years old, had a recurrence of CIN or known invasive carcinoma or refused to participate in the study. The study was approved by local ethics committee, and all patients provided written informed consent.

The women were randomly allocated to different groups of management consecutively using a computer-generated list: those whose conisation would include a frozen section as part of a one-step procedure (the experimental group) and those who would not have a frozen section (the control group). When patients were allocated to frozen section group, the nurse informed the physician that realised analysis of cone section. In both groups, colposcopic examination systematically and immediately preceded conisation so that the surgeon could determine whether loop electrosurgical excision procedure (Erbe, Tubingen, Germany) or cold-knife conisation was most appropriate to the junction area and the cervical anatomy. The surgeon choosing between loop electrosurgical excision procedure and the cold-knife procedure did not know the patient's assignment. Moreover, a pathology examination of the endocervical edge of the excised tissue was performed during surgery in the ‘frozen section’ group. If it showed CIN on the margin, additional excision took place immediately with the same procedure (loop electrosurgical excision procedure or cold knife). This second excision was not examined immediately as a frozen section. The height of the conisation in the frozen section group took the height of the second cone into account. Specimens showing CIN lesions along the endocervical edge of the excision were considered to be positive or involved cones.

The examination conducted during surgery considered only the endocervical edge of the excised tissue. We removed a transverse section, perpendicular to the cervical axis, approximately 2 mm thick. The sections were cut with a cryostat (Reichert-Jung 1800 CRYOCUT), thereby enabling us to analyse the entire surface of the disk and to examine the endocervical canal and mucus in a circumferential manner. It was stained with rapid (15–30 seconds) Sabrazes blue. Each examination took between 5 and 10 minutes. The entire surgical specimen was included (on average, 15 sections) as were any additional excisions.

Follow up involved a Pap smear and colposcopy at 2, 6 and 12 months after surgery. Criteria for re-operation were the same at 2, 6 and 12 months. A new operation was performed if any high grade lesion confirmed histologically on biopsy was observed after colposcopic and/or histologic evaluation during the three post-operative visits. A new operation before the two-month visit took place if the permanent section showed invasive lesions or micro-invasion. Margin involvement of the permanent section was not an indication for re-intervention. Thus, all the patients that underwent a re-intervention presented histologically proven high grade lesion. Recurrence was defined as disease detected at the 6 or 12 month check-up, when the cone had been considered to be healthy tissue or when an intervening follow up after conisation had been negative. All the patients considered as presenting recurrent disease had at least a normal two month evaluation. All of them had histologically proven lesion that determines the recurrent disease.

Any decision to perform a hysterectomy took into consideration parity and menopausal status. If low grade lesions were observed, follow up continued without re-operation. Patients with these invasive lesions received radiation treatment and then follow up surgery. All patients with low grade recurrence were followed, and patients with CIN II or III recurrence underwent re-operation.

The principal criterion was the 12-month residual or recurrent disease high grade lesion, and the secondary outcome criteria were the percentage of specimens with uninvolved margins after excluding re-operations before the 2-month visit, the total height of the cone (including any addition excisions after the frozen section at the initial surgery), the intra-operative and post-operative complication rates and the percentage of patients lost to follow up.

Our re-operation rate after conisation with involved margins was estimated to be 31%12. Anticipating that the use of frozen sections would reduce the percentage of patients requiring repeat surgery for high grade residual or recurrent disease from 60%, and assuming that 20% of patients would be lost to follow up, we estimated that the study required 96 patients in each group to obtain a power of 80% and a type I error (α) of 5%. The statistical analysis use of χ2 test or Fisher's Exact Test, as appropriate, for binary variables, and the analysis of variance for continuous variables.


In all, 383 patients required cervical conisation during the study period: 302 met the inclusion criteria. One surgeon in the department did not wish to participate and did not ask any of the 99 patients he treated to participate. We enrolled 203 patients between October 1, 1995, and December 22, 1999 (Fig. 1). In the group allocated to undergo frozen section, one patient did not have an intra-operative pathology examination. Two patients, one in each group, were lost to follow up immediately after the conisation and thus excluded from the analysis.

Figure 1.

Trial profile.

The indications for conisation and demographic characteristics of both groups are reported in Table 1. Loop electrosurgical excision procedure was used significantly more often in the group without frozen 53% and 37%, respectively (relative risk, RR, 1.42, 95% CI 1.04–1.96, P= 0.023). The mean height of the cone did not differ significantly between the two groups: 13.6 [5.6] and 13.7 [4.7] mm, respectively (P= 0.75). The lost to follow up rate at 12 months was 17% in the control group and 12% in the other (RR 1.42, 95% CI 0.71–2.81, P= 0.31). The final histopathologic findings of cone biopsy are reported in Table 2.

Table 1.  Clinical and preconisation findings for patients without frozen sections and group with frozen sections of the endocervical margin. Values are given as mean [SD] or n (%).
 Group without frozen sections (n= 100)Group with frozen sections (n= 100) 
  1. HPV = human papilloma virus; CIN I ± HPV = CIN I with or without HPV; SCJ = visibility of squamocolumnar junction.

Age38.6 [11.1]39.3 [11.6]NS
Parity1.3 [1.3]1.5 [1.2]NS
CIN II–III8889177 (88.5)
CIN I ± HPV549 (4.5)
Conisation for diagnosis6612 (6)
Other112 (1)
Table 2.  Cone biopsy final histopathologic findings.
Histopathologic findingsGroup with frozen section (n= 100)Group without frozen section (n= 100)Total
No disease or metaplasia325
CIN I12719
CIN II–III7885163
Invasive squamous cell carcinoma101
Invasive adenocarcinoma011

At 12 months residual or recurrent disease rate in high grade lesion was 12.6% (12/95) in the group that did not have frozen sections and 1.06% (1/94) in the group that did after exclusion of patients that underwent immediate re-intervention for micro-invasion or invasion (RR 11.9, 95% CI 1.6–89.5, P= 0.0025).

In the group without frozen section, 17 patients underwent re-operation. Five patients had hysterectomies before the first follow up visit (one carcinoma in situ, three micro-invasions at the 2-month check-up and one hysterectomy after radiation therapy for invasive carcinoma on the permanent section) (Fig. 2a). Eleven had a re-operation because of persistent histologically proven CIN III lesions on biopsy results at the 2-month visit. The last patient had a CIN III recurrence at 12 months and underwent a second conisation, this patient had no residual disease at the two month visit. In the group with frozen sections, five patients underwent hysterectomy before the first check-up (two with invasive squamous cell carcinoma, three with micro-invasions) and one another conisation because of a suspected micro-invasion (Fig. 2b). The frozen section group included one patient whose permanent section endocervical margin showed CIN III. She underwent secondary further excision because of an abnormal Pap smear at two months and histologically proven CIN III lesion on biopsy results.

Figure 2.

(a) Outcome schema of conisation group without frozen section. (b) Outcome schema of conisation with frozen section group.

In the group with frozen section, three patients had a low grade recurrent disease after a normal two month visit. In the group without frozen section, three patients had low grade recurrent disease and one high grade recurrent lesion, all these patients had a normal two month visit. In the frozen section group, all three recurrences occurred in clear margins cone resection.

The rate of involved margins was 18.9% (18/95) in the group without frozen sections and only 1% (1/94) in the other (RR 15.13, 95% CI 2.06–111.27, P= 0.0002). In the group without frozen sections, the positive margins involved CIN III in 16 cases and CIN I in 2 cases. Seven patients in that group were monitored: five whose endocervical margin showed CIN III involvement and two with CIN I involvement. Their follow up proved normal except for two patients who had CIN I and III recurrence at 6 and 12 months, respectively, after a normal two month follow up visit.

In the frozen section group, the intra-operative pathology examination could not be performed for two cases that used loop electrosurgical excision procedure: one case with a conisation of 6 mm and one case in which the specimen was badly fragmented.

Table 3 reports the predictive value of the frozen section compared with that of the final histological examination. In six cases, the final examination showed dysplasia within 2 mm of the endocervical edge, the follow up of these patients remained normal at 12 months. One patient had a second conisation with the presence of micro-invasion and refused the hysterectomy. Her final check-up was normal.

Table 3.  Predictive values of frozen sections of the endocervical edges of resected cones. Predicative values are %.
 Presence of CINAbsence of CINPredictive values
  1. Presence or absence of CIN defines existing lesions on permanent conisation specimen. Se = sensitivity; Sp = specificity; PPV = predictive positive; NPV = predictive negative.

Positive frozen section110Se = 57.9
Negative frozen section879Sp = 100
Total1979PPV = 100
   NPV = 90.8

There were no intra-operative complications in either group. The rate of post-operative complications was 7% in the frozen section group (one fever, five haemorrhages, one cervical stenosis) and 6% in the control group (five haemorrhages, one cervical stenosis) (RR 1.01, 95% CI 0.94–1.09, P= 0.8).

No further surgery was required for any of the patients in either group whose initial histological examination showed no involved margins.


This is the first randomised trial to evaluate the value of frozen sections on residual or recurrent disease rates after conisation. Only a short series has shown the benefits of this frozen section. It was retrospective, however, and included few patients (n= 60). The authors found that both the height of excised cones and the rate of involved margins were lower in the group with frozen sections. We have shown in the present study that the use of frozen sections results in a significant diminution in rate of residual or recurrent high grade disease at 12 months. This decrease is due to the higher rate of uninvolved or free margins; it did not entail an increase in the cone height, which would affect the obstetric prognosis13. This is consistent with other studies in which the failure of CIN treatment was due incomplete excision1–8,14–16. This important diminution in residual or recurrent disease leads to decrease in re-operation rates. This is of great interest, regardless of its possible cost effectiveness17. Frozen section may avoid a new hospitalisation and also could reduce the overall cost of conisation. This is due to the absence of duration extent of conisation in our study, as observed by Neiger et al.18. Therefore, frozen section may improve the treatment of CIN especially if the extemporaneous analysis can be realised rapidly.

The study of the use of frozen sections in conisation has primarily focussed on grading dysplasia to ensure the appropriate extent of surgery18–22. Our aim in using frozen sections was to ensure the most conservative treatment possible. Although squamocolumnar junction was considered in the majority of the cases as visible, the exact extent in the endocervix of the lesion is not always possible, especially for untrained physicians. Frozen section may help to obtain one step treatment even if the surgeon is not well trained to colposcopic examination.

The sensitivity of the frozen sections in our study was only 57.9%. This relatively low sensitivity has no clinical consequences, however, because these patients had no residual lesions and because the lesions were very localised. Moreover, frozen section has limitations: it was impossible in the case of one very small cone and in another case with a fragmented specimen.

The risk of residual lesions after conisation with involved margins cannot be easily managed23–25, which is why it is important to obtain the lowest possible rate of margin involvement. This is especially true because negative margins, although they do not guarantee the absence of residue, are positive prognostic indicators11.

The high rates of patients lost to follow up, also found by other authors, show some of the limitations of conservative treatment3. The likelihood that a patient will not comply scrupulously with strict follow up is one of the major disadvantages of conservative treatment after conisation with margin involvement. Immediate surgical intervention should follow such findings except when regular follow up is guaranteed. On the whole, the reduction in the rates of margin involvement and follow up surgery, combined with the high rate of patients who fail to return, should encourage physicians to use frozen sections more often in conisation.


The authors thank to Dr Patrick Rozenberg for reviewing the manuscript.