FC 01-07TEN YEARS EXPERIENCE IN 541.000 CASES: LIQUID BASED CYTOLOGY AND COMPUTER-ASSISTANCE COMPARED TO CONVENTIONAL CYTOLOGY

15. Automation in cytology
H. Ikenberg 1, B. Pittel 1, M. Faber 1, A. Bernhardt 1, C. Börsch 1, A. Xhaja 1.
1Cytomol Laboratory for Cytology and Molecular Diagnostics, D-60437 Frankfurt (Germany)

Background / Objectives

Major studies showed inconsistent results for the comparison of liquid-based cytology (LBC) with conventional cytology (CC). However, some trials found a significantly higher sensitivity for HSIL (high grade intraepithelial lesions) with the computer-assisted ThinPrep-Imaging-System (TIS) compared to conventional cytology (CC)  and even manually read LBC. Here we report the performance of TIS compared with CC in women who participate in the German cervical cancer screening program.


Methods

At Cytomol, a commercial lab specialized in cervical cancer prevention, since 2007 all LBC specimens have been processed by TIS. In Germany LBC is reserved to privately insured and self-paying patients while public healthcare only reimburses CC. To avoid bias we split this analysis between privately insured and self-paying patients. Finding rates of cytologic abnormalities with TIS and CC were compared. Cytologic diagnoses originally reported in the Munich Nomenclature II (MN; with the use of the inofficial Pap IIW category) until 30.6.2014, from then in the MN III (which is still the reporting standard in Germany) were translated to TBS (The Bethesda System).


Results

From 2007 to 2016 463.966 slides of privately insured patients have been analyzed among them 320.416 by TIS and 143.550 with CC. Except of extremely bloody and very cell-rich probes 97.4% of the smears were accepted for analysis by TIS. TIS had a rate of LSIL (low grade intraepithelial lesions; MN III: Pap IIID1) of 2.03% compared to 0.54% for CC, an increase of 276%. HSIL (MN III: Pap IIID2 + Pap IVa/b) was found in 1.10% with TIS vs 0.33% with CC (+233%). The ASC-US rate (MN III: Pap II-p/g + III-p/g) was 2.54% with TIS and 1.21% with CC, an increase of 110% which is much lower than the rise in LSIL and HSIL. This points to a higher sensitivity of TIS without decreasing specificity. Among 77.282 self-paying patient cases (all TIS) we found almost the same rates for ASC-US and LSIL but 51% more for HSIL compared to private patients. All these results remained stable over the 10 years analyzed. With TIS 20.4 slides/h were screened, compared to 12.2 for manually read TPs and 8.0 with CC. However, the technical expenditure for TIS was much higher.


Conclusion

In long-time routine use of a commercial lab computer-assisted LBC with the ThinPrep-Imaging-System provided higher sensitivity and higher productivity without lower specificity at the cost of higher technical expenditure.


References