Increased risk of preterm birth after treatment for CIN

Meta-analyses published a few years ago in the Lancet and the BMJ showed an increased risk of adverse obstetric outcomes after treatment for cervical intraepithelial neoplasia (CIN).1 2 Since then many studies, from large national linkage to small cohort studies, have confirmed the association. The cause remains unclear; however, potential mechanisms include anatomical changes,3 cicatrisation of the cervix,4 immunological factors, and alterations of the cervicovaginal flora. In a linked research study (doi:10.1136/bmj.e5174), Castanon and colleagues suggest that the increased risk of adverse pregnancy outcomes may not be attributable solely to the treatment itself but to common risk factors that also predispose to precancerous cervical conditions.5 This retrospective linkage study from 12 quality assured British colposcopy centres evaluated outcomes derived from a sample population of 18 441 births. With a comprehensive analysis aimed at optimising control for possible confounders, the authors again found an association between treatment for CIN and preterm birth, but to a smaller degree than previous estimates.

The differences in the size of the treatment effect across studies may be partly explained by the choice of control population, because women with CIN may have characteristics or even background immunological imbalances that place them at higher baseline risk of preterm birth. As discussed by Castanon and colleagues,5 a recent meta-analysis showed that the use of historical external controls might produce inherent biases that could inflate the contribution of cervical excision treatment to adverse outcomes.6 The use of internal controls (pregnancies in the index woman before treatment) is an attractive alternative approach, but even this might be inadequate for confounders that are liable to change with time. Women with mild precancerous lesions who do not warrant excision treatment probably provide the best, although still imperfect, comparator.6 Perhaps the higher rates of premature birth seen in studies from the Nordic countries result from the use of historical external comparison groups. However, inability to extract separate data for individual excision techniques could also affect associations.7

Different approaches to the surgical excision of CIN lesions are associated with varying degrees of perinatal morbidity; more serious outcomes have been noted for knife excision, followed by laser conisation, large loop excision of the transformation zone (LLETZ), and lastly laser ablation.1 2 8 The first three techniques remove a cone of tissue from the cervix and the last ablates the lesion. As listed they represent a progressive reduction in the average amount of tissue removed (or destroyed). This suggests that there may be a dose-effect relation between the amount of cervical tissue removed and the severity of adverse birth outcome.

Accumulating evidence suggests that the depth of the cone of tissue removed predicts the degree of prematurity at birth, particularly if it exceeds 10-12 mm.1 9 10 A large Danish linkage study stratified the progressive increase in risk (6% per mm),9 and for excisions of less than 11 mm it reported an odds ratio similar to that documented for the untreated population attending colposcopy in a Welsh linkage study.11 The pronounced increase in risk for deeper (>10-12 mm) or repeat cones probably reflects the true aggregate of risk arising from both confounders and excision factors.9 By virtue of varying loop sizes, LLETZ treatment can remove a variety of cone volumes.9 Therefore, an analysis that merges data on all excisions without adjustment for depth of excision may mask the true effect of deep excisions.

The current evidence base consists, almost exclusively, of retrospective cohorts with poor or absent documentation of the depth of excision; this limitation impedes statistical pooling of data. Commonly, valuable patient characteristics, the treatment type, and, more importantly, excision depth are inadequately reported. Conversely, studies that use data from self selected centres of excellence, such as that of Castanon and colleagues,5 are more likely to have modified practice in light of recent research findings and to have offered antenatal prophylactic interventions to women in subsequent pregnancies. Such studies may underestimate the effect of treatment. Data from linkage studies should eliminate selection bias and reflect national practice rather than centre specific specialist practice. Because cervical volume and length vary, it is biologically plausible that the relative proportion of the cervix excised may more accurately reflect the level of cervical damage than the absolute dimensions of the tissue removed. It might be feasible to calculate a cut-off value for the proportion of cervical tissue excised above which a woman is at risk of premature delivery. Preliminary prospective data suggest that cervical deficit and prematurity may correlate with the proportion of cervical tissue excised.3

Adequately powered prospective studies using appropriate controls and stratification according to cone depth, or preferably proportion of excision, and other crucial risk factors are needed. Alternatively, meta-analyses using individual patient data from existing high quality databases might help to determine which parameters influence birth outcomes the most. Development of a risk algorithm to help individualise treatment for women with a precancerous cervical lesion, and models that explore optimum antenatal management in subsequent pregnancies, might improve outcomes in the future.

Hopefully, human papillomavirus vaccination programmes will greatly reduce the incidence of precancerous lesions of the cervix. Until then, colposcopists should carefully select women who need treatment and tailor the intervention to minimise morbidity. Formal audit of colposcopy practice should be routine, and the development of treatment standards against which to benchmark practice would be ideal.12