Epigenetic modifications of histone h4 in lung neuroendocrine tumors.

Global profiling of histone changes in some human cancers demonstrated that loss of histone H4 acetylation at lysine16 (H4KA16) and trimethylation at lysine 20 (H4KM20) was a common hallmark of cancer. It is not clear whether these epigenetic changes also exist in neuroendocrine carcinomas. We semiquantitatively analyzed 32 cases of lung neuroendocrine tumors (LNETs) immunohistochemically stained with H4KA16, H4KM20, and Ki67 antibodies by calculating cumulative scores based on the sum of the product of nuclear stain intensity (1-3) and percentages of positive cells in each category. H4KA16 and H4KM20 levels were compared among typical carcinoid (TC, 11), atypical carcinoid (AC, 6), large cell neuroendocrine carcinoma (LCNEC, 8), and small cell lung cancer (SCLC, 7) and correlated with histologic types and Ki67 labeling. Data were presented as mean±standard error of the mean and statistically analyzed by 1-way analysis of variance and Holm-Sidak method. Normal bronchiolar epithelium had relatively uniform and strong +3 positivity of H4KM20 and H4KA16, which was considered as internal positive controls. This uniformity, however, was gradually lost from low to high grades of LNETs. Semiquantitative analysis revealed that there were significant differences in cumulative scores of H4KA16 (TC, 2.36±0.03; AC, 2.04±0.08; LCNEC, 1.58±0.07; SCLC 1.32±0.05) among LNETs. For H4KM20, significant differences were only observed between low grade (TC, 2.49±0.05 and AC, 2.24±0.09) and high grade (LCNEC, 1.58±0.10 and SCLC 1.68±0.11) LNETs, but not within low or high grade LNETs. The Ki67 cumulative scores (TC, 0.06±0.02; AC, 0.41±0.08; LCNEC, 1.29±0.09; SCLC 1.83±0.06) were inversely correlated with both cumulative H4KA16 and H4KM20 scores by Pearson correlation. We conclude that progressive loss of H4KA16 and H4KM20 from low to high grade LNETs reflects the degree of differentiation and proliferative activity. These histone modifications may serve as tumor biomarkers that may aid in diagnosis and prediction of clinical outcome.