Premium
An improved algorithm using a Health Canada‐approved DNA‐image cytometry system for non‐invasive screening of high‐grade oral lesions
Author(s) -
Parfenova Ekaterina,
Liu Kelly Y. P.,
Harrison Alan,
MacAulay Calum,
Guillaud Martial,
Poh Catherine F.
Publication year - 2021
Publication title -
journal of oral pathology and medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.887
H-Index - 83
eISSN - 1600-0714
pISSN - 0904-2512
DOI - 10.1111/jop.13149
Subject(s) - feulgen stain , malignancy , medicine , cytometry , predictive value , biomarker , dysplasia , pathology , aneuploidy , nuclear medicine , algorithm , biology , staining , flow cytometry , mathematics , immunology , genetics , gene , chromosome
Background DNA‐image cytometry (DNA‐ICM) is able to detect gross alterations of cellular DNA‐content representing aneuploidy, a biomarker of malignancy. A Health Canada‐approved DNA‐ICM system, ClearCyte ® in combination with a cytopathologist's review, has demonstrated high sensitivity (89%) and specificity (97%) in identifying high‐grade oral lesions. The study objective was to create an improved automated algorithm (iClearcyte) and test its robustness in differentiating high grade from benign reactive oral lesions without a cytopathologist's input. Methods A set of 214 oral brushing samples of oral cancer (n = 92), severe dysplasia (n = 20), reactive lesions (n = 52), and normal samples (n = 50) were spun down onto slides and stained using Feulgen‐Thionin reaction. Following ClearCyte ® scan, nuclear features were calculated, and nuclei categorized into “diploid,” “hyperdiploid,” “tetraploid,” and “aneuploid” DNA ploidy groups by the ClearCyte ® software. The samples were randomized into training and test sets (70:30) based on patient's age, sex, tobacco use, and lesion site risk. The training set was used to create a new algorithm which was then validated using the remaining samples in the test set, where sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated. Results The proposed iClearCyte algorithm (>1 “aneuploid” cell or ≥ 1.7% combined “hyperdiploid” and “tetraploid” nuclei frequency) identified high‐grade samples with sensitivity, specificity, PPV, and NPV of 100.0%, 86.7%, 89.7%, and 100.0%, respectively, in the test set. Conclusion The iClearCyte test has potential to serve as a robust non‐invasive automated oral cancer screening tool promoting early oral cancer detection and decreasing the number of unnecessary invasive biopsies.