Effect of cellularity on the sensitivity of detecting squamous lesions in liquid-based cervical cytology

To evaluate the effect of cellularity on the sensitivity of both screening and diagnosis in a liquid-based cervical sample.

SurePath samples (TriPath Imaging Inc., Burlington, North Carolina, U.S.A.) with known diagnoses were selected, including 18 negative, 16 low grade squamous intraepithelial lesion (LSIL) and 12 high grade squamous intraepithelial lesion (HSIL) cases. Through a serial dilution technique, samples of varying cellularity were prepared. The 275 slides were assigned random numbers and were routinely screened by 1 of 2 senior cytotechnologists, blinded to the reference diagnosis. Specimens with a screening diagnosis of atypical squamous cells of undetermined significance (ASCUS) or higher were reviewed by two pathologists, resulting in a final consensus diagnosis. Using a grid counting system, cellularity was determined for each slide.

There was a clear demarcation in sensitivity between specimens with a cellularity of < 5,000 or > or = 5,000 squamous cells. This applied to both the sensitivity for screening and to the final consensus diagnosis. For cases with a reference diagnosis of LSIL+, at a cytotechnologist screening level of ASCUS or greater, sensitivity increased from 72.8% (< 5,000 cells) to 98.1% (> or = 5,000 cells) and for a reference diagnosis of HSIL from 85.7% to 100%, respectively. Similarly, for the consensus diagnosis, sensitivity rose from 78.5% (< 5,000 cells) to 96.6% (> or = 5,000 cells) for LSIL+ and from 82.9% to 100%, respectively, for HSIL. These differences were statistically significant (P < .001).

A minimum cellularity of 5,000 squamous cells is recommended for SurePath liquid-based cervical preparations.