Article type
Year
Abstract
Background: A meta-analysis was performed in the framework of the preparation of the new European Guidelines for Quality Assurance in Cervical Cancer Screening and was funded by the Europe Against Cancer Programme.
Objective: To compare the test performance and quality of liquid and conventional cervical smears.
Methods: Multiple meta-analyses including studies comparing the Pap smear with liquid-based cytology (LBC) were performed using low-level and progressively higher-level inclusion criteria, and separating 'split-sample' from 'direct-to-vial' studies. The lowest level outcomes were differences in cytological detection rates and quality judgement. The highest considered outcome was diagnostic validity for histologically confirmed CIN2+ (cervical intra-epithelial neoplasia, grade II or worse) where all cases, including cytological negative cases, were submitted to colposcopy, and biopsy if suspect colposcopy. In studies with histological verification for at least 85% of screen-positives and for a random sample of screen-negatives, a Beggs-corrected sensitivity and specificity was computed using bootstrapping to simulate confidence intervals.
Results: Results pooled from split-sample studies showed neither increased detection rates of high-grade lesions, nor higher positive predictive value, sensitivity or specificity in the few studies where all tested subjects were submitted to gold standard verification. However, in direct-to-vial studies significantly higher detection rates of high-grade cytological abnormalities were observed, whereas the positive predictive value for histologically confirmed CIN2+ was not lower than for conventional cytology. These findings may suggest increased sensitivity without significant loss in specificity. However, the level of evidence for this statement is rather low because of insufficiently controlled and verified study designs. In general the quality of LBC preparations has improved and their interpretation requires less time.
Conclusion: An impressive amount of literature exists comparing experiences with conventional and liquid-based cytology. Nevertheless, randomised controlled trials comparing LBC versus conventional cytology, respecting the rules of good diagnostic research and using biopsy proven outcomes, are still needed before superior performance of LBC can be considered as evidence-based. Such population-based trials are currently being conducted in the Netherlands and Italy and are planned in Germany and New-Zealand.
Objective: To compare the test performance and quality of liquid and conventional cervical smears.
Methods: Multiple meta-analyses including studies comparing the Pap smear with liquid-based cytology (LBC) were performed using low-level and progressively higher-level inclusion criteria, and separating 'split-sample' from 'direct-to-vial' studies. The lowest level outcomes were differences in cytological detection rates and quality judgement. The highest considered outcome was diagnostic validity for histologically confirmed CIN2+ (cervical intra-epithelial neoplasia, grade II or worse) where all cases, including cytological negative cases, were submitted to colposcopy, and biopsy if suspect colposcopy. In studies with histological verification for at least 85% of screen-positives and for a random sample of screen-negatives, a Beggs-corrected sensitivity and specificity was computed using bootstrapping to simulate confidence intervals.
Results: Results pooled from split-sample studies showed neither increased detection rates of high-grade lesions, nor higher positive predictive value, sensitivity or specificity in the few studies where all tested subjects were submitted to gold standard verification. However, in direct-to-vial studies significantly higher detection rates of high-grade cytological abnormalities were observed, whereas the positive predictive value for histologically confirmed CIN2+ was not lower than for conventional cytology. These findings may suggest increased sensitivity without significant loss in specificity. However, the level of evidence for this statement is rather low because of insufficiently controlled and verified study designs. In general the quality of LBC preparations has improved and their interpretation requires less time.
Conclusion: An impressive amount of literature exists comparing experiences with conventional and liquid-based cytology. Nevertheless, randomised controlled trials comparing LBC versus conventional cytology, respecting the rules of good diagnostic research and using biopsy proven outcomes, are still needed before superior performance of LBC can be considered as evidence-based. Such population-based trials are currently being conducted in the Netherlands and Italy and are planned in Germany and New-Zealand.