The shift from primary cytology screening to HPV-screening with cytology triage is expected to influence the performance of cytology reading. We initiated an educational program prior to the start of the renewed screening program to refresh classification criteria.
Two partially overlapping cytology sets of 100 liquid-based cytology slides (ThinPrep) were collected, derived from a blinded co test pilot done previously (the DuSC study), stratified for age and anticipated percentage of abnormalities (i.e. 30%). After examination by an expert panel of 3 cytotechnologists and a pathologist a consensus diagnosis was determined. The sets were ring-studied in the 5 screening laboratories. The first set was followed by an adjudication session with each lab by the national reference official, in which results were discussed and individual classifications were aligned to classification guidelines and consensus diagnosis.Then, after a washout period, a second set of 100 cases with 50 overlapping cases was offered to the same laboratories. Cytotechnologists were asked to individually examine all slides and score them according to KOPAC and Bethesda classification. Pathologists examined only the non-NILM cases.
Discrepancies in cytology reading were examined between i) NILM (especially reactive cellular changes) and ASCUS and vice versa: cases with few atypical changes (< 5 cells per slide) were regarded as difficult for classification of NILM versus ASC-US. Notably, distinguishing between NILM and ASC-US is important because it is the threshold for referral to a gynaecologist in the new HPV-screening program; ii) cytotechnologists and pathologists: pathologists showed moderately higher scores of classifications in the non-NILM’s, iii) laboratories, iv). Missed LSIL/HSIL cases. Wide variation in classification was found in sporadic cases due to technical or obscuring factors (dark staining, thick cell groups, or low cell counts). Guidance was given on the criteria of the Bethesda classification and according to the company guidelines.
The results show considerable variation in cytology classification between cytotechnologists and pathologists and between laboratories. These results are the starting point of an ongoing external quality control (EQA) and educational program on standardization and optimal cytology classification results within the Dutch cervical screening program. The participants evaluated the learning sets as an important educational tool.