Liquid based cytology in cervical cancer screeningBMJ 2007; 335 doi: https://doi.org/10.1136/bmj.39262.506528.47 (Published 05 July 2007) Cite this as: BMJ 2007;335:1
Cervical screening has been shown to reduce the incidence of cervical cancer, but only in the setting of well organised, high quality programmes. In the United Kingdom the NHS cervical screening programme has been estimated to prevent around 80% of deaths from cervical cancer.1
Liquid based cytology represents the first major change in preparation method for cervical screening samples for over 50 years. Instead of cells being smeared onto a glass slide, they are washed into a vial of liquid and filtered, and a random sample is presented in a thin layer on a glass slide. These slides can then either be screened by skilled staff or subjected to partially automated imaging. The process is being widely used in the United States, many European countries, and elsewhere.
In this week's BMJ two studies compare the accuracy of liquid based cytology with conventional cytology.2 3 The randomised trial by Ronco and colleagues found no significant difference in sensitivity for cervical intraepithelial neoplasia of grade 2 or more with liquid based cytology using ThinPrep (Cytyc, Boxborough, MA, USA) compared with conventional cytology.2 However, more positive results were found with liquid based cytology, leading to a lower positive predictive value. The observational study by Davey and colleagues compared the accuracy of the automated ThinPrep imaging system to that of conventional cytology, using split sample pairs (the ThinPrep sample was obtained after the conventional one in a single collection).3 The ThinPrep Imager detected 1.29 more cases of histological high grade squamous disease per 1000 women screened than conventional cytology, where cervical intraepithelial neoplasia grade 1 was the threshold for referral to colposcopy.
What do these results add to what is already known? Several manufacturers have developed liquid based cytology systems. The literature evaluating them is extensive and varied. Most are observational studies, either split sample studies or studies comparing outcome with previous outcomes at the same laboratory.4 Several national screening programmes and bodies such as the Food and Drug Administration have evaluated the evidence for liquid based cytology and come to different conclusions,5 6 although most evaluations have led to implementation. Of the few randomised controlled trials, those considered to be of high quality have tended to show no difference in sensitivity with liquid based cytology.7
This wealth of conflicting data presents difficulties for screening programmes worldwide, trying to make decisions on implementation in their own setting. Cervical screening has had varying levels of success in different countries. Rates of inadequate conventional smears vary from 9.5% in the United Kingdom to <1% elsewhere. Detection rates for abnormalities vary as well (even after differences in terminology are taken into account) and are influenced by sensitivity of screening, incidence of disease in the population, coverage of the population, age of starting screening, and screening interval—all factors that vary between countries. The UK has one of the highest sensitivities for detecting abnormality in a single conventional smear,8 one of the longest screening intervals (three or five years, depending on age), and high population coverage (over 80%). In other countries women may have smears taken every six months, and screening every 12-24 months is common. The UK has rigorous training and quality assurance and is the only country with a policy of rapidly rescreening 100% of negative and inadequate samples.
These variables mean that results of the most rigorous study in one setting may not be directly applicable in another. Factors such as training of laboratory staff and the people who take samples, which may have a major effect, have been overlooked in several otherwise well designed trials. The manufacturers of liquid based cytology systems have mandatory training courses, but in the UK a much more lengthy and controlled training process is undertaken by training centres approved by the NHS cervical screening programme. Trainers in the UK have noted that performance improves during and after training, suggesting that sensitivity will be increased, but this has not been formally evaluated. The quality and duration of training specific to liquid based cytology may have an impact on detection rate, and this could contribute to the differing outcomes in reported studies.
The imager study by Davey3 is exciting because a significant increase in sensitivity has been robustly shown, and the authors also found a dramatic increase in screening productivity (the number of slides screened per hour by a single member of staff).9 However, this must be interpreted with caution, because the outcome could be different elsewhere. Controlled trials of automated systems compared with liquid based cytology alone are under way in other national settings.10
Introduction of liquid based cytology to the UK will be complete during 2008. This follows an evaluation of the available evidence by the National Institute of Health and Clinical Excellence, which concluded that liquid based cytology was as sensitive as conventional cytology, and commissioned an implementation pilot.11 Increased sensitivity was not the aim of implementing liquid based cytology in the UK, but in laboratories that have converted to liquid based cytology, specificity has been maintained,12 and detection rate may have increased.13 The desired end points in the UK—to reduce the rate of inadequate samples and increase screening capacity—have been achieved, and at least some of the additional cost of liquid based cytology has been offset by fewer repeat tests. Liquid based cytology also gives a platform for human papillomavirus testing, automation, and other new technologies, and is popular with staff. Women benefit from faster reports and less anxiety.
Is liquid based cytology superior to conventional cytology? The answer is yes, but sensitivity is not the reason for this superiority, or at least not the only one. The addition of automation may make liquid based cytology even better.
Competing interests: None declared.
Provenance and peer review: Commissioned, not externally peer reviewed.