Associations between cervical intraepithelial neoplasia during pregnancy, previous excisional treatment, cone-length and preterm delivery: a register-based study from western Sweden

This study was reported according to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines (Additional file 2: STROBE Checklist). The study was prospectively planned in 2015, the dataset was retrieved in 2019, and analyses were conducted until July 2021. The analysis plan has not been published but overall exposures, outcomes, confounders, and analyses were planned in 2015 by the research team, based on hypotheses drawn from previous studies. After obtaining and reviewing the database contents, but before starting the analyses, the final definition of study groups and outcomes was determined based on the available data and the quality of the dataset.

All women with singleton births between January 1, 2008, and December 31, 2016, registered in the MBR were identified. Women with a history of chronic inflammatory disease, organ transplantation, or human immunodeficiency virus infection were excluded (see Additional file 3: Table S1 for the International Statistical Classification of Diseases and Related Health Problems -10th revision (ICD-10) codes leading to exclusion).

Forty-six thousand eight hundred seventy women had at least one delivery fulfilling the exposure criteria for inclusion in one of the study groups (Fig. 1). Together with women that did not fulfill the exposure criteria for study group inclusion, and that did have a singleton delivery recorded in western Sweden in the MBR during this time period, they comprised a population of 145,529 women with 210,126 singleton deliveries.

A woman with a singleton delivery registered in the MBR 2008–2016 was eligible for inclusion in one of the study groups if the exposure criteria were fulfilled, according to NKCx and Swedish Cancer Register data. Women with treatment recorded in NKCx/Process before 2008 were excluded from all study groups. Furthermore, women with histological diagnoses of CIN2 or of more severe lesions (CIN2+) before 2008 in the NKCx/Analysis and the Swedish Cancer Register were excluded, since they might have undergone excisional treatment (Fig. 1). For an exact description of study groups, see Additional file 3: Table S2 [37].

The study groups were defined as follows:

There were 158 women with a delivery included in the CIN during the pregnancy group who also had a subsequent delivery included in the treated group. One hundred fifteen of these had recorded cone-lengths. These women were excluded from the treated group when it was compared to the CIN during the pregnancy group.

The primary outcome was PTD at 22+0–36+6 weeks (154–258 days) of gestation, with subanalyses for early PTD (22–33 weeks (154–237 days) of gestation) and very early PTD (22–27 weeks (154–195 days) of gestation). Gestational age was retrieved from the MBR based on the best estimate, i.e., ultrasound determination when available and last menstrual period or estimation of gestational age at the delivery ward in the remaining cases.

Secondary outcomes were pPROM (determined according to ICD-10 codes in the MBR) and spontaneous PTD (a delivery starting with pPROM or preterm labor, excluding preterm deliveries that started with induction or cesarean section).

Additional outcomes comprised prelabor rupture of membranes (PROM) in term pregnancies (≥37 weeks of gestation), chorioamnionitis, and neonatal sepsis. Furthermore, comparisons were also made between the study groups concerning intrauterine fetal death, neonatal mortality (1–28 days), Apgar score <7 at 5 min, small for gestational age (SGA) (birthweight less than − 2 standard deviations (SD) according to Swedish reference curves) [38] and intrapartum fever (see Additional file 3: Table S3 for outcome definitions).

Multivariate analyses were adjusted for the following variables retrieved from the prenatal care records in the MBR and SCB registers: year of delivery (2008–2010, 2011–2013, 2014–2016), maternal age at delivery (<23, 23–30, 31–38, >38), body mass index (BMI) (underweight (<18.5), normal-weight (18.5–24.9), overweight (25–29.9), obese (≥30), missing), parity (0, 1–3, >3), marital status (cohabiting, single, other, missing), country of birth (Sweden, Europe, Asia, America/Oceania, Africa, other/unknown), infant’s sex (boy/girl), smoking (never, before pregnancy, in early pregnancy only, in the third trimester, missing), highest disposable household income during the 3 years preceding delivery (population divided into tertiles for every year), education level at delivery (primary, secondary, post-secondary <3 years, post-secondary ≥3 years, missing), and assisted reproduction (yes/no). Adjustments were based on a priori knowledge of risk factors for PTD [39‐41]. Furthermore, data on employment at time of delivery, chronic renal disease, diabetes, epilepsy, and chronic hypertension were also collected.

Data registered in western Sweden concerning histological diagnosis and type of treatment were retrieved from NKCx/Process (Additional file 3: Table S4). If several diagnoses had been recorded, the most serious was included.

Analyses were performed with the “R” data analysis tool (version 4.0.0, The R Foundation for Statistical Computing, Vienna, Austria, https://​www.​r-project.​org/​) and SPSS software (version 26.0, IBM, https://​www.​ibm.​com/​analytics/​spss-statistics-software). A significance level of 0.05 was applied throughout. Descriptive data are presented with numbers and percentages for categorical variables and with mean and standard deviation (SD) and median and interquartile range (IQR) for continuous variables.

The normal cytology group, the treated group, and the CIN during pregnancy group were compared, regarding obstetric and neonatal outcomes, by unadjusted and adjusted logistic regression analysis.

In a subgroup analysis, the treated group with benign histology was compared to the normal cytology group regarding risk of spontaneous PTD and pPROM. Furthermore, stratified analyses were used to compare PTD risk, within the CIN during pregnancy group, in women with high-grade lesions to women with low-grade lesions.

Women in the CIN during the pregnancy group were younger than those in the other groups. There were fewer nulliparas in the normal cytology group (48%) than in the CIN during the pregnancy group (58%) and the treated group (60%). The treated group had given birth more frequently in 2014-2016 than the other groups. Women in the normal cytology group were more frequently non-smokers. There were fewer pregnancies after assisted reproduction in the CIN during the pregnancy group than in the other groups (Table 1).

In the cone-length group, the majority (n=2055 (85%)) were treated with LEEP. Cone-lengths varied between 3 and 31 mm (mean 9.02, SD 2.99) The women treated with laser had longer cones (mean 9.53 mm, SD 3.36) than those treated with LEEP (mean 9.02, SD 2.92) (p=0.003).

In the cone-length group, histology in the excised specimen was classified as follows; cancer (1.0%), high-grade lesions (74.5%), low-grade lesions (12.5%), dysplasia not further specified, signs of HPV infection (2.5%), benign (7.8%), and not classified/missing (1.7%).

In the total dataset of 210,126 singleton deliveries in 145,529 women in western Sweden in 2008–2016, 9920 (4.7%) were PTD, 6699 (3.2 %) were spontaneous PTD, 2782 (1.3%) were pPROM, and 10,389 were deliveries after PROM at term (4.9% of all deliveries and 5.2% of term pregnancies). Of these deliveries, 537 (0.3%) were diagnosed with chorioamnionitis and 2525 (1.2%) with neonatal sepsis.

Mean (SD) and median (IQR) gestational age were 279.5 (12.4) and 281 (274–287) in the normal cytology group, 279.1 (13.3) and 281 (274–287) in the CIN during pregnancy group, and 277.3 (15.2) and 280 (272–286) in the treated group. The CIN during pregnancy group had an increased risk of PTD, compared to the normal cytology group; this increase was, however, no longer significant in the adjusted analyses (Table 2). When women with low-grade lesions were compared to women with high-grade lesions in the CIN during the pregnancy group, there was no significant difference in risk of PTD, (aOR 0.90, 95% CI 0.55–1.47, p=0.67).

The treated group had an increased risk of PTD, spontaneous PTD, early PTD, pPROM, and PROM, compared to the normal cytology group (Table 2), as well as compared to the CIN during pregnancy group (Table 3). Compared to the normal cytology group, the treated group also had an increased risk of chorioamnionitis and neonatal sepsis; the increases were, however, no longer significant in the adjusted analyses (Table 2). Neonatal mortality and intrauterine fetal death were too rare to yield conclusive results.

When only women with benign histology in the treated group (n=271) were included, there was still an increased risk of spontaneous PTD (aOR 1.79, 95% CI 1.06 − 3.03, p=0.03), as well as of pPROM (aOR 3.04, 95% CI 1.60–5.79, p=0.001), compared to the normal cytology group. The corresponding increased risk of PTD (aOR 1.51, 95% CI 0.93–2.45, p=0.09) was, however, not significant. The risk of spontaneous PTD was similar in women treated with laser (n=24, 6.5%) and in women treated with LEEP (n=126, 6.1%) in the cone-length group.

The percentage of PTD, spontaneous PTD, and pPROM increased with increasing cone-length (Fig. 2a–c). The number of cones >15 mm was low (n=68), but this group had a high rate of adverse obstetric outcomes.

The risk of PTD, spontaneous PTD, and pPROM increased with increasing cone-length, compared to women with normal cytology. Treated women with cone-lengths up to 10 mm also had an increased risk of PTD, spontaneous PTD, and pPROM, compared to the normal cytology group (Tables 4 and 5). For unadjusted analyses, see Additional file 3: Tables S5a-b.

When treated women with cone-lengths up to 10 mm were compared to the CIN during the pregnancy group, we found an increased risk of PTD (aOR 1.41, 95% CI 1.02–1.94, p=0.038), spontaneous PTD (aOR 1.73, 95% CI 1.18–2.54, p=0.005), and pPROM (aOR 2.44, 95% CI 1.40–4.28, p=0.002), (Additional file 3: Table S6); this was also found in the subgroup analyses including only high-grade lesions (Additional file 3: Table S7).

The risk of PTD and spontaneous PTD was similar for cone-lengths up to 10 mm and increased with cone-length (Fig. 2a, b and Tables 4 and 5). Cone-length was associated with an increased risk of PTD, spontaneous PTD, pPROM, PROM, and neonatal sepsis in the logistic regression analyses (Additional file 3: Table S8). When only cone-lengths 3–10 mm were included in the analyses (n=1805) no risk increase with increasing cone-length was found for these outcomes (Additional file 3: Table S9). Therefore, in a truncated analysis, all cone-lengths up to 10 mm were grouped (n=1805) into 10 mm and the risk increase for every mm above 10 mm was analyzed. The aOR for PTD increased by 15% with every mm above 10 mm (Table 6).

Additional adjustment for the treatment method (LEEP or laser) did not change these findings.

The risk increases for PTD, spontaneous PTD, and pPROM with cone-length are also illustrated in Fig. 3, showing risk calculation in the truncated model using adjusted multivariable logistic regression.

Our findings—increased risk of PTD, spontaneous PTD, and pPROM with increasing cone-length, measured before fixation—are consistent with previous studies of cone-length measured after fixation [3, 20, 21, 23]. To the best of our knowledge, this is the first study to compare cone-length at previous treatment to women with untreated CIN during pregnancy. Novel findings of this study are that cone-lengths ≤ 10 mm were associated with increased risk of PTD, spontaneous PTD, and pPROM, including in comparison with women with high-grade CIN, and thus HPV, during pregnancy. Another novel finding was that increasing cone-length was associated with increased risk of PROM at term and increased risk of neonatal sepsis.

In this study population, we could not confirm the increased risks of neonatal mortality, chorioamnionitis, and neonatal sepsis after treatment that were previously found in a larger national Swedish cohort (1999–2016) [8]. However, since our cohort was much smaller this might be due to a lack of power for such uncommon outcomes. Nevertheless, there was an increased risk of chorioamnionitis and neonatal sepsis after treatment in our unadjusted analyses.

This study included treatments from 2008 and onwards, since colposcopists in western Sweden started to record cone-lengths that year, in response to increasing evidence that more extensive treatment might be associated with worse obstetric outcomes [1]. When treating a woman of reproductive age, clinicians try to perform as minor an excision as possible and to avoid unnecessary diathermy after excision. However, treatment must achieve removal of the CIN in order to prevent the development of cancer. Swedish guidelines prescribe a cone-length of 6–9 mm in the most common clinical situation, i.e. when the transformation zone (TZ) is entirely visible on the portio (TZ 1), in order to include the cervical crypts. The mean cone-length in this study, 9.1 mm, reflects adherence to these recommendations. The risk of spontaneous PTD (aOR 2.00, CI 1.70–2.34) after treatment, compared to women with normal cytology, in this cohort, was in the same range as in the previously published national cohort [8] (aOR 2.06, 95% CI 1.95–2.17). The corresponding risk increase of pPROM in this study (aOR 2.63, CI 2.11–3.28) corresponded to aOR 2.36 (95% CI 2.19–2.54), although the national cohort included earlier treatments (60% treated before 2007), undertaken when more radical methods were applied and awareness of obstetric risks was lower. Since detection and treatment of CIN began, excisions have continuously become smaller, preserving cervical tissue. However, this development may possibly have occurred at a cost of diminished reduction in the development of cervical cancer among treated women [28]. This highlights the need for expertise, when it comes to colposcopy, to selection of women for treatment, and to excisional treatment.

The risk estimates for spontaneous PTD after excisional treatment, compared to women with normal cytology, were similar to the results of a cohort study from Denmark (1997–2005) (OR 2.07, 95%CI 1.88–2.27) [5] and to those of a recent Dutch study (2005–2015) (aOR 2.07, 95% CI 1.85–2.33) [13]. However, the risk of spontaneous PTD among previously treated women, compared to women with untreated CIN (aOR 1.95, 95% CI 1.40–2.72), was higher in our study than in the Dutch study (aOR 1.51, 95% CI 1.29–1.76) [13]. Untreated CIN in that study (n=5940) included CIN diagnosed in non-pregnant women and was associated with increased risk of spontaneous PTD (aOR 1.38, 95% CI 1.19–1.60), compared to women with normal cytology [13]. In our study, the results pointed in the same direction, i.e., aOR 1.13 (95% CI 0.84–1.51) for spontaneous PTD in women with CIN, compared to women with normal cytology. This was, however, not significant in our smaller cohort.

The Dutch study found an increased risk of spontaneous PTD related to excisions of cervical tissue volumes of 0.50–0.99 cc, measured after fixation, compared to women with CIN. However, information about cone-length was missing and the results are therefore difficult to compare to ours. In contrast to our study, a case-control study from England [20] found no increased risk of PTD when treated women with cone-length <10 mm were compared with untreated women undergoing punch biopsy at colposcopy before or after delivery. Consistent with our result, a Danish study found an increased risk for PTD after cone-length of 10 mm (aOR 1.46 (95% CI 1.11–1.92) compared to untreated women [23]. Other previous studies published on cone-length up to 10 mm have included minimal data [21, 22]. The Danish study of cone-length (LEEP or laser excision, n=3605), measured after fixation, found a 6% increase in risk per mm (aOR 1.06, 95% CI 1.03–1.09) [23], lower than in our non-truncated analyses of cone-length (aOR 1.10, 95% CI 1.05–1.15) but within the same CI as in our study. Our truncated analysis, however, showed a higher risk increase, i.e., a 15% increase for each millimeter exceeding 10 mm.

To the best of our knowledge, this study is the first to investigate cone-length before fixation, measured in a standardized manner. Thus, our results can be related to the clinical situation. Treatments were performed after 2008, with a mean cone-length of 9.1 mm, and can be assumed to represent modern treatment undertaken by colposcopists aware of the increased risk of subsequent PTD. Our study compared treated women both to women with a history of normal cytology as well as to women with untreated CIN during pregnancy. Adjustment for a priori identified confounding factors was performed. For example, assisted reproduction, smoking, and parity had an uneven distribution among the different study groups. However, as this is an observational study, causality cannot be established, and there might still be residual confounding affecting the results. Another limitation is that the small size of the exposure groups, especially the CIN during pregnancy group, limited the power to detect a significantly increased risk for some adverse outcomes.

The risk of PTD increased with cone-length, but this study shows that small excisions are also associated with an increased risk of spontaneous PTD. The effect of excision size on cervical length measures during pregnancy and, more critical, on the risk of PTD might differ between individual women and possibly depends on the excised volume and/or on the ratio of excised tissue to the total pre-operative cervical volume [16, 24, 42].

Prospective studies are needed in which the cervix is measured before treatment, cone-length and volume are determined before fixation, and women are followed up during pregnancy, including surveillance with standardized ultrasound-measured cervical length [42].

We also suggest more extensive studies comparing women with CIN diagnosed during pregnancy to previously treated women and to women with a history of normal cytology.

Information on excised cone-length, measured at treatment, should be available for risk estimation in subsequent pregnancies, in order to plan surveillance for those at the highest risk of PTD. Risk in the individual after treatment might depend on if other risk factors for PTD also are present. Spontaneous PTD is a multifactorial condition, and only a comprehensive model including major risk factors may present adequate risk estimation for a specific patient [43]. We suggest that future models for individual prediction of spontaneous PTD include cone-length in the excised specimen, together with other risk factors for PTD.