General Grading System of Malignancies: High Grade Neoplasms Express Stem Cell-Like Phenotype

Background: Grading is one of the most powerful variable of tumor prognosis, but is subjective, site dependent, and has little biologic support. We aim to identify the variables that reliably predict grade, testing stem cell features as potential discriminator.

Design: We analyzed primary and secondary growth patterns (tubulo-papillary, nested- trabecular, nodular-solid, diffuse), nuclear grade (including chromatin, nucleolus, pleomorphism and anisokaryosis), stromal reaction, and confluent necrosis in common malignancies: carcinomas (61 basal cell, 101 squamous cell, 163 adenocarcinomas, 30 urothelial, and 20 neuroendocrine), sarcomas (100), lymphomas (100) and melanomas (61). Tumors were graded according to the WHO classification (139 poorly differentiated). Representative samples were evaluated by quantitative RT-PCR and standard in situ techniques for stress-stem cell pathways (telomere PNA-FISH, TERT, TP53, ATF2, BMP4, PTCH1, FN1, CXCR3, MMP10, OCT4, SCF, MYC, JUN, and FOS), proliferation (Ki-67) and apoptosis (TUNEL assay). Appropriate controls were run. Fisher’s exact tests and analysis of variance (significant if P<0.05) were used for comparison; significant variables were then selected for discriminant analysis with cross-validation for grading groups (well-moderate vs. poorly differentiated).

Results: The variables contributing most to the poorly differentiated neoplastic phenotype were the growth patterns, necrosis presence, hemorrhage, anisokaryosis, nucleolus, and Ki67 index. Stepwise discriminant analyses correctly classified 97% of cases (96% after cross validation). Telomerase expression and telomere positive cells (%) were the added variables for carcinoma grading. Telomerase/telomere indices directly correlated with the kinetic index, being significantly higher in high-grade malignancies with upregulation of TP53, ATF2, KITLG, CXCR3, MYC and FOS. The remaining markers revealed no statistically significant differences.

Conclusions: A reliable common grading system of malignancies must include a combined evaluation of growth pattern, confluent necrosis, nuclear and proliferation features. High-grade malignancies express stem-like phenotype with emphasis on stress (ATF2, FOS, TP53), survival (MYC, TERT) and microenvironment (CXCR3, KITLG) pathways.