Premium
ERCP‐directed brush cytology prepared by the Thinprep ® method: test performance and morphology of 149 cases
Author(s) -
Duggan M. A.,
Brasher P.,
Medlicott S. A.
Publication year - 2004
Publication title -
cytopathology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.512
H-Index - 48
eISSN - 1365-2303
pISSN - 0956-5507
DOI - 10.1111/j.1365-2303.2004.00129.x
Subject(s) - medicine , malignancy , cytology , false positive paradox , positive predicative value , liquid based cytology , pathology , nuclear medicine , cancer , predictive value , machine learning , computer science , cervical cancer
Conventionally prepared endobiliary brushings are moderately (42%) sensitive and highly (98%) specific in detecting malignancy. The performance and morphological features of brushings prepared by Thinprep ® , a liquid‐based method are mostly unknown. All brushings were retrieved from the laboratory files. Disease was classified as benign or malignant by linkage with the provincial cancer registry and sensitivity, specificity, positive (PPV) and negative predictive values (NPV) calculated. True positives and negatives were reviewed and predictive morphological features analysed by regression tree analysis. Out of 149 brushings, 55 (37%) were positive and 94 (63%) negative. Malignancy was identified in 86 (58%) and benign disease in 63 (42%) of the cases. The sensitivity was 51%, specificity 83%, PPV 80% and NPV 55%. Absolute discriminants of positive and negative brushings were not found, but nuclear cytoplasmic ratio was a useful feature. The performance of Thinprep ® ‐prepared brushings from this anatomical site was comparable with conventional preparations.