Guidance on the use of liquid-based cytology for cervical screening

The Department of Health commissioned an independent evaluation of the English pilot study, which compared LBC with the Pap smear test at three sites (Norfolk and Norwich University Hospital; Southmead Hospital, North Bristol NHS Trust; and Royal Victoria Infirmary, Newcastle upon Tyne). Since the publication of the Assessment Report for the earlier appraisal, evidence has also become available from the following new studies: 6 studies comparing LBC and Pap smears with a reference method (histology/pathologist diagnosis); 8 split-sample studies; 6 two-cohort studies; a Scottish implementation study; a New Zealand Health Technology Assessment of LBC; and a cross sectional study by Coste et al.

A meta-analysis of 14 studies (comprising all new studies, and studies contained in the previous assessment where data were available) comparing the sensitivity of LBC and the Pap smear in the detection of abnormalities of low-grade squamous intraepithelial lesions or greater demonstrated that sensitivity may be up to 12% better with LBC compared with the Pap smear. When the results of the Coste study are included in the meta-analysis, the total sensitivity improvement for LBC is 4.9% for the ordinary population and 2.8% for the high-risk and ordinary populations combined. Split-sample studies and two-cohort studies supported increased sensitivity with LBC.

A meta-analysis of 6 studies that reported specificity found no difference between the specificity of LBC and Pap smear.

The English pilot study showed a statistically significant decrease in the number of inadequate samples, from 9.1% with Pap slides to an average of 1.6% with LBC (87% reduction, p<0.0001). The majority of 34 studies reporting the rate of inadequate samples noted that the rate was reduced with LBC.

The English pilot study reported a statistically significant reduction in the detection of glandular neoplasm, from an average of 0.08% with the Pap smear to 0.04% with LBC (RR 0.496, 95% CI 0.292 to 0.807). Follow-up data from the pilot sites demonstrated that although there was a reduction in the cytological detection of glandular neoplasm during the pilot period with LBC, the number of histologically confirmed cases of adenocarcinoma remained unchanged. Pilot study data on the performance of LBC in the post-pilot period demonstrated that the cytological detection of glandular neoplasm with the LBC test is similar to the pre-pilot rate using the Pap smear.

A literature review identified four new economic evaluations of LBC compared with the Pap smear in the US population. These are of limited application to the UK because of differences in US incidence rates and costs, and differences in the Bethesda (US) and BSCC (UK) classification systems for cervical smears.

PathLore Limited provided a cost analysis of the SurePath test. Increased capital costs of £50,000 and consumables costs of £2.50 per test may be offset by savings from the reduction in the number of inadequate samples and a quicker diagnosis, to give a gross saving of £0.89 per LBC test compared with the Pap smear. This is consistent with the costs reported in the pilot studies.

The Assessment Group updated the economic model in the previous Assessment Report with the data from the English pilot study and literature to estimate the incidence of, and mortality from, cervical cancer among women who had had cervical screening using LBC and Pap smear technologies. The model simulated a cohort of 100,000 15-year-old women enrolled in the cervical screening programme (screened between the ages of 21 and 64 years), who were followed throughout their lifetime using a state transition model. Key outcomes of the English pilot study used to update the economic model were the rate of inadequate specimens, and the cost per test (incorporating capital, consumables and the amount of staff time required for smear taking, slide preparation and smear diagnosis). For LBC, the economic evaluation was based on the costs of the T3000 device, which represented the average cost across the 3 methods.

In the English pilot study, laboratory report forms indicated a 5-minute reduction in the time required for smear taking and consultation with LBC (average of 8 minutes and 35 seconds compared with 13 minutes and 20 seconds for the Pap smear). Staff questionnaires estimating the time required for smear taking suggested that the LBC method may be 1 minute quicker than the Pap smear. The extent of the increase in slide preparation time with LBC depended on the labour requirements of different LBC methods. Slide preparation with LBC took 4 minutes and 15 seconds (ThinPrep T2000), 38 seconds (ThinPrep T3000), or 1 minute and 52 seconds (SurePath system) compared with an average of 15 seconds for the conventional Pap smear. The average aggregate cost of LBC was £22.30 (£22.99 for T3000, £23.15 for T2000, and £20.76 for PrepStain) compared with £21.68 for the conventional Pap smear. Overall, there was an increase in the throughput of slides at the screening stage with LBC compared with the Pap smear. At primary screening, 9.04 slides were read per hour with LBC compared with 8.3 slides read per hour with the Pap smear. At rapid review, 44.1 slides were read per hour with LBC compared with 46.7 slides read per hour with Pap, and at slide checking, 12.4 slides were read per hour with LBC compared with 9.5 slides read per hour with the Pap smear.

The English pilot study estimated that a one-off transition cost of £10.27 million (see implications for the NHS for more detail) would be required for the national implementation of LBC. The one-off transition cost of implementing LBC was incorporated into the economic model as a cost of £0.13 per smear test (discounted over a 20-year lifetime of LBC). If the transition cost were discounted over a 10-year lifetime of the LBC technology, this would equate to £0.34 per smear test.

Assumptions of the base-case economic analysis were based on data from the English pilot study and included a laboratory processing capacity of 60,000 tests per annum, sensitivity improvements with LBC relative to Pap of 13.4% for the detection of CIN1 and CIN2 combined and 4% for the detection of CIN3, and a reduction in the rate of inadequate samples from 9% with the Pap smear to 1.4% with LBC. The results of the base-case economic analysis demonstrated the following.

• At each screening interval LBC dominated the Pap smear as it was less costly and more effective.

• Three-yearly screening with LBC was found to be a cost-effective alternative to 5-yearly Pap screening, with an incremental cost effectiveness ratio below £8,000 per life-year gained.

• The cost effectiveness of LBC screening at different intervals was compared. The incremental cost-effectiveness ratio of moving from 5-yearly to 3-yearly screening with LBC was £9,621 per life-year gained.

• Conventional screening with the Pap smear at 5-yearly intervals is extremely cost effective compared with no screening, at a cost of £372 per life-year gained.

Sensitivity analysis for differences in the natural history of cervical cancer (cancer incidence, progression and regression), sensitivity of LBC and the Pap smear, the rate of inadequate samples and the marginal cost of LBC demonstrated that under most conditions, 5- and 3-yearly screening with LBC is a cost-effective alternative to 5-yearly screening with the Pap test.

The assumptions of the base-case analysis were changed to a scenario where there was decreased processing capacity for LBC of 30,000 tests per annum, a 20% increase in capital costs and 50% increase in consumable costs, and a time saving of 1 minute per test at smear taking. The resulting increase of £6.50 in the marginal cost of LBC compared with the Pap smear did not greatly affect the cost effectiveness of screening using the LBC method.

Another sensitivity analysis combined an increase in the unit cost of LBC to £25.88 per test that is, £4.21 more than the Pap test, with various improvements in sensitivity with LBC relative to Pap (2.8%, 4.9% and 12%). At 2.8% improved sensitivity relative to the Pap test, the cost per life-year gained of LBC compared with 5-yearly screening with the Pap test was £5,500 and £38,250 for 5- and 3-yearly screening respectively. At 4.9% improved sensitivity the cost per life-year gained of LBC compared with 5-yearly screening with the Pap test was £3,250 and £22,500 for 5- and 3-yearly screening respectively. At 12% improved sensitivity, the cost per life-year gained of LBC compared with 5-yearly screening with the Pap test was £1,500 and £10,250 for 5- and 3-yearly screening respectively.

No studies were identified that compared the difference in the quality of life between women who had LBC smears and those who had Pap smears, and thus a cost per quality-adjusted life year for different screening scenarios could not be reliably determined. The Assessment Group's model incorporated the assumption that the decrease in utility associated with living with invasive cancer, undergoing a colposcopy and receiving a borderline test would have an adverse effect on a woman's quality of life because of anxiety. When quality of life is taken into consideration, LBC still dominates conventional screening if the baseline 12% improvement in sensitivity is assumed.

The report of the English pilot study also contained an economic evaluation, which was consistent with the results of the model generated by the Assessment Group.

The increased capital, consumable and implementation costs of LBC may be offset by savings in the reduction of inadequate samples and time savings in sample collection and time to diagnosis. It was not possible to distinguish between specific LBC technologies on the basis of the available data.

The Committee reviewed the data available on the clinical and cost effectiveness of LBC. It carefully considered results from the pilot site studies undertaken since the original guidance was issued, and the opinions of clinical experts, and took into account the likely effect LBC would have on women taking part in the NHSCSP and Cervical Screening Wales. It was also mindful of the need to take account of the effective use of NHS resources.

The Committee considered evidence from cytopathologists and representatives of cervical screeners that LBC produced a more homogeneous cellular preparation that was free from exudates, and that this would be likely to decrease the number of smears classified as inadequate (that is, considered unreadable and in need of repeat smear). However, there were concerns that, with the LBC method, there are no formal criteria defining the adequacy of a slide preparation in terms of the cell numbers required. Because the LBC sample is a homogenate there is no way of verifying that a sufficient number of cervical cells have been harvested by the smear taker. The Committee considered this to be an important issue that must be addressed as part of the implementation of LBC. Poor sampling technique, resulting in the collection of too few cells, could mean that a sample might not adequately represent cells on the surface of the cervix. Consequently abnormalities may be missed, resulting in some false-negative results. However, the Committee concluded that this potential risk of false-negatives should be balanced against the likelihood of abnormalities being detected at a subsequent screen because of the regular screening frequency of the cervical screening programme, and the increased detection of high-grade lesions (severe dyskaryosis) with the LBC technique. Overall, the Committee was persuaded that LBC was likely to be an improvement over the currently used technique, and in particular that the reduction of inadequate smears would be an important benefit to women in the NHSCSP and Cervical Screening Wales because it would reduce the requirement for repeat smears.

Experts expressed concerns that sensitivity improvements with LBC may not be as great as 12%, and the Committee was made aware of various confounding factors that may have contributed to the perceived increase in sensitivity of LBC, such as increased training and experience of the smear screeners and the type of sampling device used by the smear takers. However, experts and the Committee agreed that the overall sensitivity of LBC was at least as good as, and may be better than, the Pap smear.

The Committee reviewed the recent paper by Coste et al. (2003) and expressed concerns regarding the robustness of the conclusions of the study, which were not in favour of LBC. In the light of the sensitivity analysis that included the results of the Coste study, the Committee concluded that LBC is likely to be a cost-effective alternative to the Pap smear test. In addition, the Committee understood that an important factor in the assessment of the increased sensitivity of LBC was its enhanced ability to detect high-grade lesions (severe dyskaryosis), which was confirmed by the results from the pilot study.

The Committee considered the difference in the detection of glandular neoplasm in the English pilot study report, and the potential this may have for differences in the detection of adenocarcinoma between the LBC method and the Pap smear. The Committee reviewed in detail the results from the pilot sites on the rate of histologically proven adenocarcinoma and the evidence from the post-pilot results of detection of glandular abnormalities. They were satisfied on the basis of this evidence that LBC is at least as good at detecting these abnormalities as the Pap smear.

The Committee reviewed the report from the UK English pilot study and noted a lack of consistency between the results of the pilot site using the SurePath device and 2 sites using the ThinPrep device. However, the Committee considered that, taking into account the advice from experts, the possible confounding factors of indirect comparisons, and the fact that the pilot studies were not designed to evaluate clinical effectiveness, there was currently insufficient evidence to suggest that one technique should be recommended over another.

The Committee heard from experts that LBC was likely to result in increased productivity in cytology laboratories because of a rise in the number of slides that can be read per hour and a reduced workload as a result of a lower incidence of inadequate samples and a reduced number of repeat smears. Although the screening of LBC thin layers is quicker because of the ease of reading a 'cleaner' slide preparation and screening a smaller area of the slide, it was mentioned that screening LBC slides is more tiring for staff, who consequently may require more breaks. However, the experts involved with the pilot sites said that, overall, smear takers and readers favoured the LBC method above the Pap smear.

The Committee considered that, taking into account a number of factors – including the potential for increased sensitivity, reduction of inadequate smears and probable improvements in laboratory efficiency – the LBC method was likely to be cost effective compared with the Pap smear, despite its higher associated cost. The Committee discussed whether the use of LBC would affect the cost effectiveness of population screening at different screening intervals. However, this was considered to be beyond the remit of this review and to be the responsibility of the NHSCSP and Cervical Screening Wales.

The Committee was aware that a number of manufacturers have LBC-related products. Evidence received during the appraisal related only to the SurePath and ThinPrep devices – no evidence was provided by the other manufacturers.