Pregnancy outcome after cervical conisation: a retrospective cohort study in the Leuven University Hospital
Prof Willy Poppe, University Hospitals Leuven – Campus Gasthuisberg, Department of Obstetrics and Gynaecology, Herestraat 49, 3000 Leuven, Belgium. Email email@example.com
Please cite this paper as: van de Vijver A, Poppe W, Verguts J, Arbyn M. Pregnancy outcome after cervical conisation: a retrospective cohort study in the Leuven University Hospital. BJOG 2010;117:268–273.
Objective To assess pregnancy outcome after conisation.
Design Retrospective cohort study.
Setting Belgium, data from a university hospital.
Population Fifty-five pregnancies in 34 women after conisation, and 55 pregnancies in 54 women without a history of conisation or cervical intraepithelial neoplasia (CIN).
Methods Hospital data were reviewed and questionnaires were collected from 599 women who had a conisation in a 5-year period, among whom subsequent pregnancies were identified. The control group consisted of matched pregnancies of women without a history of conisation.
Main outcome measures Gestational age at delivery, neonatal biometry, neonatal condition at birth.
Results Numbers of sexual partners (4.6 ± 3.4 SD versus 2.5 ± 2.5 SD) and ex-smokers were significantly higher in the study group compared with the control group. Gestational age at delivery (266 ± 2 days versus 274 ± 9 days), neonatal head circumference (33.9 ± 2.5 cm, versus 34.6 ± 2.5 cm) and birthweight (3088 ± 754 g versus 3381 ± 430 g) were significantly lower in the study group compared with the control group. Numbers of preterm [<37 weeks; 14/55 (25%) versus 2/55 (4%); P = 0.002] and severe preterm (<34 weeks; 6/55 (11%) versus 0/55 (0%); P = 0.031] deliveries in the study group were significantly higher. There were no cases of perinatal mortality.
Conclusions Conisation affects obstetrical outcome after conisation for CIN. Babies tend to be born earlier and are smaller. It is not clear whether this is related to the procedure or to factors linked with CIN.
The incidence of cervical cancer can be reduced substantially by well-organised screening.1,2 Screening consists of the detection of cytological abnormalities and diagnostic work-up of screen-positive women, followed by treatment of high-grade cervical intraepithelial neoplasia (CIN).3,4 Most of the women with high-grade CIN are of reproductive age.5 In Belgium, the peak age at first conisation or large loop excision of the transformation zone (LLETZ) is in the range of 30–34 years, and 44% of treatments take place in women under the age of 35 years.6 As these women often show a definite desire to have children, it is important not to compromise future pregnancies by surgical interventions on the cervix. Currently used therapeutic procedures can be classified into two groups: destructive or ablative and excisional procedures or a combination of both. Conisation, first described by Lisfranc in 1815,7 is performed by cold knife, CO2 laser or electrical loop electrodes (LLETZ; but also called loop electrosurgical excisional procedure, LEEP).8 In 1938, Miller and Todd9 stated that conisation should not be performed in women of reproductive age, because of the high rate of preterm births.
A review of the literature and meta-analysis gave evidence of pregnancy-related morbidity, such as premature delivery (<37 weeks), low birthweight (<2500 g) and preterm prelabour rupture of membranes (PPROM) after cold-knife conisation, LLETZ and laser conisation, but also found substantial interstudy heterogeneity.10 In a more recent meta-analysis targeting severe pregnancy outcomes, and including larger population-based studies, it was found that only cold-knife conisation was significantly associated with perinatal mortality, severe preterm delivery (<30 weeks) and extreme low birthweight (<1500 g).11 Both meta-analyses confirmed that ablative procedures (cryotherapy or CO2 laser) are safer.
The objective of this study is to verify the strength of the association between adverse obstetrical outcomes and prior excisional treatment for CIN in the Leuven University Hospital, where yearly approximately 2200 deliveries and 150 surgical treatments for CIN occur.
A retrospective matched cohort study was designed, starting from a list of all 599 women who underwent a conisation of the cervix (LLETZ, laser or cold knife) between 1 January 1999 and 31 December 2003 in the University Hospital in Leuven. In this group, 47 women could be identified who became pregnant after the intervention and delivered before 1 January 2007.
The study concerned 72 pregnancies. Seventeen (23.6%) resulted in a miscarriage and were excluded. Finally, the study group (SG) consisted of 55 evolutive pregnancies (delivery after 22 weeks of gestation) in 43 women. There were nine women with two and two women with three subsequent pregnancies. A control group was composed of 55 evolutive pregnancies in 54 women with equal age (same birth year ± 2 years), parity and year of delivery (±2 years), who had had no intervention on the cervix or were never diagnosed with CIN.
Data were collected from electronic and paper patient files at the Leuven University Hospital, and from a questionnaire sent to women of the study and control groups.
Different risk factors for preterm birth (smoking behaviour, socio-economic status, education level and number of sexual partners) were recorded to control for possible residual confounding factors.
The following obstetrical pregnancy outcomes were assessed: duration of pregnancy (in postmenstrual days), proportion of preterm deliveries (before 37 weeks of gestation) and proportion of deliveries with PPROM. The use of oxytocin and prostaglandins to induce or to augment labour, aspects of the amniotic fluid, and mode of delivery were recorded.
Neonatal outcomes were birthweight (g), length (cm), head circumference (cm), sex of the newborn, Apgar scores after 1 and 5 minutes, pH of the umbilical artery and admission on the neonatal intensive care unit (NICU). The volume of the cone and the interval between conisation and delivery date were correlated with the pregnancy outcome.
Differences in proportions between the study group and control group were assessed by Pearson’s χ2 test and Fisher’s exact testing when one or more of the numbers in a 2 × 2 contingency was lower than 5. For differences in the continuous variables, we used the Student’s t-test. A P-value smaller than 0.05 was regarded as statistically significant. Correlation between continuous variables (cone height, time between treatment and pregnancy, duration of pregnancy) were assessed by the correlation coefficient. Statistical tests were two-sided.
The 17 miscarriages excluded in the study group consisted of five complete spontaneous abortions, two incomplete spontaneous abortions, four missed abortions, two ectopic pregnancies and two provoked abortions, all of them before 11 weeks of gestation. There were no ‘late’ or second-trimester miscarriages. No data could be found for two of the miscarriages.
In the study group (43 women), three patients underwent a laser conisation (five pregnancies) and 40 patients were treated with LLETZ (50 pregnancies). Two patients underwent a reconisation and delivered at term (four pregnancies). The histopathological diagnoses (Table 1) were chronic inflammation (5), squamous metaplasia (5), CIN (43) and glandular intraepithelial neoplasia (GIN) (2). In the study group there were three twin pregnancies. These women delivered at 36, 28 and 36 weeks of gestation. In the control group there was one twin pregnancy with delivery at 37 weeks of gestation.
Table 1. Histopathological diagnosis of the cone
|Cervical intraepithelial neoplasia (CIN)|| |
|Glandular intraepithelial neoplasia (GIN)|| |
Women in the control group were matched for age at the start of the pregnancy, parity and year of delivery. In 52 pairs of women, the year of delivery was the same, and for three pairs, there was a difference of 2 years.
Between the study and control groups there were no significant differences concerning maternal height (166.1 ± 6.2 cm versus 167.0 ± 5.6 cm), age at menarche (12.9 ± 1.5 years versus 13.2 ± 1.5 years) and age of first sexual contact (17.7 ± 1.9 years versus 18.6 ± 2.6 years) (Table 2).
Table 2. Maternal characteristics in the study and control groups
|Maternal heighth (cm)||166.1||6.2||167.0||5.6||0.44|
|Maternal weight at start pregnancy (kg)||63.9||12.2||67.5||14.8||0.18|
|Age at menarche (year)||12.9||1.5||13.2||1.5||0.41|
|Age at first sexual intercourse (year)||17.7||1.9||18.6||2.6||0.12|
|Number of sexual partners||4.6||3.4||2.5||2.5||0.01|
|Smoked during pregnancy||12/51||23.5||5/54||9.3||0.057|
The number of sexual partners was significantly higher in the study group (4.6 ± 3.4 SD) compared with the control group (2.5 ± 2.5 SD). For smoking behaviour, the results were different before and during pregnancy. Before pregnancy we found a significantly higher proportion of smokers in the study group, compared with the control group (50.0% versus 20.4%; P = 0.002); during pregnancy smoking behaviour was not significantly different between the groups (23.5% versus 9.3%; P = 0.057). Education level (P = 0.74) and socio-economic status (P = 0.66) were not different between the groups.
When evaluating pregnancy outcome, there were no significant differences between the study and control groups for the following variables: maternal weight at both the start [63.9 kg (95% CI 60.6–67.3 kg) versus 67.5 kg (95% CI 63.4–71.7 kg); P = 0.18] and end of pregnancy [78.6 kg (95% CI 75.3–81.9 kg) versus 79.8 kg (95% CI 76.1–83.5 kg); P = 0.62]; neonatal height [49.1 cm (95% CI 48.1–50.2 cm) versus 50.2 cm (95% CI 49.7–50.7 cm); P = 0.07]; pH of the umbilical artery [7.3 (95% CI 7.3–7.3) versus 7.3 (95% CI 7.3–7.3); P = 0.82]; Apgar score after 1 minute [8.6 (95% CI 8.3–8.9) versus 8.6 (95% CI 8.3–8.9); P = 0.98] and after 5 minutes [9.4 (95% CI 9.2–9.7) versus 9.5 (95% CI 9.3–9.7); P = 0.56]; use of oxytocin [16/44 (36.4%) versus 17/55 (31.1%); P = 0.62] and prostaglandins during labour [10/44 (22.7%) versus 15/55 (27.3%); P = 0.64]; presence of meconial amniotic fluid [3/43 (7.0%) versus 6/55 (11.1%); P = 0.75]; use of a vacuum extractor [1/58 (1.7%) versus 5/56 (8.9%); P = 0.11]; primary caesarean section [11/58 (19.0%) versus 13/56 (23.2%); P = 0.65]; secondary caesarean section [2/58 (3.5%) versus 3/56 (5.4%); P = 0.68]; admission of the neonate to NICU [15/58 (25.9%) versus 9/58 (15.5%); P = 0.25]; and number of PPROM [5/55 (9.1%) versus 1/55 (1.8%); P = 0.20] (Tables 3 and 4). A significant difference was observed for the duration of pregnancy [266.0 days (95% CI 260.3–271.7 days) versus 273.9 days (95% CI 271.4–276.4 days); P = 0.01]. The duration of pregnancy was on average 7.9 (95% CI 1.8–14.0) days shorter in the study group, compared with the control group.
Table 3. Pregnancy outcome (continuous variables)
| Maternal weight at end of pregnancy (kg)||78.6||11.8||79.8||13.2||0.62|
| Length of gestation (days)||266.0||21.2||273.9||9.3||0.01|
| Birthweight (g)||3087.9||753.7||3380.5||430.0||0.01|
| Length (cm)||49.1||3.8||50.2||2.0||0.07|
| Head circumference (cm)||33.9||2.5||34.6||1.2||0.04|
| pH of umbilical artery||7.3||0.1||7.3||0.1||0.82|
| Apgar after 1 minute||8.6||1.0||8.6||1.1||0.98|
| Apgar after 5 minutes||9.4||0.8||9.5||0.8||0.56|
Table 4. Pregnancy outcome (dichotomic variables)
|Preterm delivery (<37 weeks)||14/55||25.5||2/55||3.6||0.002|
|Severe preterm delivery (<34 weeks)||6/55||10.9||0/55||0||0.031|
|Complications during pregnancy||22/55||40||14/55||25.5||0.15|
|Use of oxytocin||16/44||36.4||17/55||30.9||0.67|
|Use of prostaglandins||10/44||22.7||15/55||27.3||0.65|
|Clear amniotic fluid||40/43||93.0||49/55||89.1||0.75|
|Admission on NICU||15/58||25.9||9/58||15.5||0.25|
|Sex of the neonate|
The frequency of preterm delivery (delivery before 37 weeks of gestation) in the study group was high [14/55 (25.5%)], and was low in the control group [2/55 (3.6%)], resulting in a relative risk of 7.0 (95% CI 1.8–28.0; P = 0.002). There were six cases of severe preterm birth (birth before 34 weeks of gestation) in the study group, and none in the control group (P = 0.031). Birthweight in the study group [3088 g (95% CI 2895–3281 g)] was found to be significantly lower than in the control group [3381 g (95% CI 3266–3495 g); P = 0.01]. Elimination of the twin pregnancies in both groups still resulted in a lower birthweight in the study group [3215 g (95% CI 3034–3396)] compared with the control group [3396 g (95% CI 3275–3517 g)], but the difference was not statistically significant (P = 0.089). Four low birthweights (<2500 g) were observed in the study group [4/52 (7.7%)] and none were observed in the control group [0/54 (0.0%)]. However, this difference was not significant (P = 0.125).
Finally, within the study group the correlation between cone volume and duration of pregnancy was calculated. The correlation coefficient was −0.038, but this was not significant (P = 0.82). The correlation coefficient of the time interval between conisation and delivery versus pregnancy duration was 0.15 (P = 0.30).
Despite the small numbers, our study shows a significantly smaller head circumference of the neonate and a shorter duration of pregnancy, with a seven-fold relative risk for preterm delivery (95% CI 1.8–28.0 days) in women with a history of cervical conisation. The relative risk of 7 is the highest of all studies found in the meta-analysis of Kyrgiou,10 excepting the investigation of Crane12 in 2006. Also, severe preterm births (<34 weeks) were significantly more frequent in the study group.
A lower birthweight was observed in the study group, but after elimination of twin pregnancies, the difference became statistically insignificant. However, the risk of preterm and severe preterm delivery remained statistically significantly higher in the treated group after removal of twin pregnancies (P = 0.006 and 0.020, respectively). Although this study involved only a small population, the results largely confirm the findings of previous meta-analyses.10,11
The mechanism that explains the effect of conisation on future pregnancies has not been clarified until now. Kristensen raised several hypotheses.13 A sizeable part of the connective tissue is removed during conisation, which can lead to a weakening of the cervix and inability to support a full-term pregnancy. Also, glandular tissue is removed, which may cause a reduction of the mucus plug. The mucus, which is composed of mucin, secretory immunoglobulin A, and a lysosome-like substance, prevents ascending bacterial colonisation. Some of the bacteria associated with preterm birth, such as Bacteroides fragilis and group-B Streptococcus, can release phospholipase A2 or proteolytic enzymes. Phospholipase A2 may initiate the arachidonic acid cascade, resulting in a locally elevated concentration of prostaglandins E and F. Elevated concentrations of prostaglandins may lead to cervical ripening and uterine contractions. The release of proteolytic enzymes may lead to the premature rupture of the membranes.
The presence of vaginal lactobacils may be less prominent in women after conisation, which can indicate changes in vaginal flora leading to a higher risk of preterm birth.14 A prospective study and a detailed follow-up of vaginal bacteriology is required to confirm this hypothesis.
Whether pregnancy outcome is improved by performing a cerclage during a pregnancy after conisation, cannot been concluded from our study. Only one patient in the study group underwent a preventive cerclage (McDonald) at 14 weeks of gestation. She delivered at 279 days of gestation. Most authors agree that a history of conisation as such is not an indication to perform a preventive cerclage.15–17
This investigation is a retrospective cohort study, and therefore has the limitations of this type of research: the possibility of confounding bias, which cannot always be corrected for (e.g. the borderline significant results for smoking during pregnancy); the possible information bias, as a result of reporting errors (incomplete information, registration errors); and selection bias in finding controls (e.g. the low number of preterm deliveries in the control group).
Ideally, a prospective randomised controlled trial should be designed for women with the diagnosis of treated premalignant cervical lesions. These patients should be divided in two groups, one with ablative treatment and the other with excisional treatment, and ideally they should be compared with a group of women with untreated CIN.
The analysis for possible confounding factors in the study group showed a significantly higher number of sexual partners and smoking before pregnancy. These differences can be expected, because these are independent risk factors in the development of CIN. Although a significant difference could not be observed concerning smoking behaviour during pregnancy, smoking as an independent confounding factor for pregnancy outcome cannot be excluded because of the borderline significance (P = 0.057).
Within the study group we tried to perform an analysis of the effect of the dimensions of the removed cervical fragment. This is very difficult because the exact measurements are not always registered, and the specimens are not measured according to a pre-specified procedure. Some specimens were excised in parts. We assume that the larger the conus, the more functional impact on cervical continence may be expected. According to Leiman et al.15, the cut-off with a significant higher risk for preterm delivery is a cone height of more than 2 cm (prematurity risk of 9.2% versus 30.4% before and after cold-knife conisation, respectively) or a volume of more than 4 cm3 (3.2% versus 31.7%, respectively). A literature review concluded that the risk of preterm delivery increased when the cone height is more than 10 mm, but this conclusion was based on only three studies.10 The importance of cervical length to predict premature birth is not generally accepted.18 A study by Gentry et al.19 showed that after adequate healing of the cervix after LLETZ, cervical length is almost identical to the initial length before the procedure. Immediately after surgery, a shortening of 0.6 cm (0.3 cm SD) was observed, but 3 months after conisation the difference in cervical length, measured by transvaginal ultrasound, was only 0.1 cm (0.4 cm SD). We might hypothesise that similar cervical lengths may not necessarily reflect intact function.
In addition, it seems plausible that the proportion of excised tissue in relation to the initial size of the cervix determines obstetrical prognosis. A cone of given size excised from a larger cervix will probably have less functional consequences than the resection from a smaller cervix. A prospective study with follow-up of these parameters by ultrasound measurements is needed. Furthermore, the technique of LLETZ may differ and is not at all standardised for the volume or shape of resected tissue (depending on the shape and topography of the transformation zone), or for the duration of coagulation of the cervical crater and margins after the resection.
The correlation between time lag from conisation to delivery and duration of pregnancy was not significant (P = 0.30). We could expect that a greater time interval between the procedure and delivery would reduce the effect on pregnancy duration. Our findings did not support this hypothesis.
Because of adverse pregnancy outcomes after conisation, expectant management of all lesions in young women and of low-grade lesions in older women is recommended.20–22 In women of fertile age, we try to find the best balance between maximal treatment of the lesions and minimal disturbance of cervical anatomy. Because of the adverse effects associated with excision of CIN lesions, appropriate triage strategies are required to reduce the risk of over-diagnosis and over-treatment.23–26
The clinical implications of this study are that women with a definite desire to have children should be informed about the possible risks associated with excisional treatment of CIN. If treated by LLETZ or cold-knife conisation, women need careful follow-up of their pregnancies. More research is needed to assess the quantitative relationship between the dimension of the excised cone and the clinical and obstetrical outcomes, and the impact of CIN as such on pregnancy.
Disclosure of interests
Contribution to authorship
AvdV collected all of the data and calculated the differences between the study and control groups. WP initiated the study, assisted in data collection and coordinated the writing of the manuscript between the authors. MA performed the statistical analysis and helped in discussing the study results. JV helped in discussing the study results.
Details of ethics approval
This is a retrospective cohort study of hospital data and questionnaires. No approval of the local ethics committee was needed to perform the study.
We kindly acknowledge Chantal Vanden Bosch for secretarial assistance. This study was not funded, but the work of MA was supported by the Fonds national de la Recherche scientifique ‘TELEVIE’, Brussels, Belgium (ref. 7.4.628.07.F), through the Department of Obstetrics and Gynaecology of the University of Liège (Belgium), and by the European Commission (DG Sanco, Luxembourg) through the ECCG project (European cooperation on development and implementation of cancer screening and prevention guidelines, IARC, Lyon, France).