Global histone modification pattern predicts poor prognosis in organic hyperinsulinism

Here we tested whether global histone modifications predict survival in organic hyperinsulinism and whether global histone modification pattern can be used to distinguish benign from malignant primary insulinoma. A tissue microarray (TMA) was built, using samples from 63 patients with organic hyperinsulinism. The TMA was classified according to the WHO classification of 2004 [WHO 1A: benign insulinoma (wdPET); WHO 1B: unknown behavior (wdPETub); WHO 2/3: malignant insulinoma (wdPEC/pdPEC)]. The TMA consisted of tissue cores from islands of Langerhans, primary insulinomas, lymph node metastases, and hepatic metastases. Immunohistochemistry was performed on consecutive TMA slides with antibodies against H3K9Ac, H3K18Ac, H4K12Ac, H3K4diMe, and H4R3diMe. The Remmele immunoreactive scoring system was used to classify the staining. The IHC staining results were correlated to the WHO-classification of 2004 as well as to clinical follow-up data (mean: 107 months; range: 1-312 months). A nuclear staining pattern was observed for all antibodies directed against histone H3 and H4 acetylation/methylation sites. We observed significant differences in the distribution of the medians across all investigated tissue types (H3K9Ac, p=0.004; H3K18Ac, p=0.001; H4K12Ac, p=0.006; H4R3diMe, p=0.002) except for H3K4diMe (p=0.183). Correlation of the histone modification with the WHO-classification and clinical follow-up data, showed in the dichotomized groups ["low" (score 0-3), "moderate" (4-7) vs. "high" (≥8)] that patients with lower H3K18Ac levels ("low + moderate") had a significantly decreased relapse-free survival vs. patients with high H3K18Ac levels (p=0.038). The WHO classification and age were also of significant prognostic impact upon univariate analysis. A backwards Cox proportional hazards model revealed the independent prognostic effekt of H3K18Ac levels. Our data revealed low K18 acetylation levels of histone H3 as independent prognostic factor in organic hyperinsulinism. This result warrants validation with independent data sets of organic hyperinsulinism, but is in line with several previous studies in different cancer entities. The broad applicability of this potential biomarker might lead to standardized diagnostic tests in near future and may help to manage insulinoma patients more effectively.