Abstract
PURPOSE: New insights in prognostic predictions are urgently needed for papillary thyroid carcinoma (PTC). The present study aimed to investigate whether computerized analysis of chromatin texture allows the identification of PTC patients with a poor prognosis. METHODS: We randomnly selected paraffin-embedded blocks from surgical specimens of 103 classic cases of PTC. During follow-up, 68 of the patients were classified as free of disease, whereas 35 presented with recurrences. Characteristics of chromatin were obtained from digitized images of at least 100 randomly selected tumor nuclei per patient. An independent series of 30 goiters was used to validate our observations. RESULTS: Stage, age and distant metastases were found to serve as independent prognostic factors for survival. In addition, multivariate Cox regression confirmed variable cluster prominence as an independent prognostic factor. By comparing malignant and benign nodules, we found that the PTC lesions presented with higher nuclear perimeters, nuclear areas, Minkowski fractal dimensions, optical densities and nuclear longest chords. CONCLUSION: From our results we conclude that, in conjunction with clinical and histopathological data, morphometric data may provide relevant prognostic information in PTC patients.