AutoCyte Interactive Screening System. Experience at a university hospital cytology laboratory.

OBJECTIVE

To evaluate the AutoCyte Interactive Screening System (AutoCyte-Screen) as a prescreener for the cytologic detection of cervicovaginal abnormalities.

STUDY DESIGN

Eight hundred fifty-six AutoCyte Preparation System from cervicovaginal samples were evaluated by AutoCyte-Screen. AutoCyte-Screen displayed 120 cell images and 6 low-power images for review by the cytotechnologist. The cytotechnologist classified the case as WNL, abnormal or unsatisfactory. AutoCyte-Screen then revealed its classification as either unsatisfactory, WNL, abnormal or undecided. These classifications were combined to form an interactive result that was compared to the diagnosis from previous manual review.

RESULTS

Interactive results were as follows: 251 (29.3%) abnormal, 581 (67.9%) WNL and 24 (2.8%) unsatisfactory. The abnormal interactive result contained 15 ASCUS/AGUS and 25 SILs; the WNL interactive result contained 5 ASCUS/AGUS and 2 SILs. No ASCUS/AGUS or SILs were in the unsatisfactory interactive category. The false negative proportion was 10.5% for the interactive diagnostic method vs. 15.7% for manual review for LSIL.

CONCLUSION

The interactive use of AutoCyte-Screen can effectively select those cases which are most likely to contain an epithelial abnormality and could therefore be used as a triage system to select cases for manual review. A case with an interactive result of unsatisfactory and abnormal should receive full manual review, while those classified as WNL may require a less extensive review for quality assurance purposes. This resultant decrease in the manual screening load could increase laboratory efficiency.