Outcome-Related Genome Loci in T-Cell Non-Hodgkin's Lymphoma Identified by High-Resolution SNP-Typing Arrays.

Abstract

Peripheral T-cell lymphoma, unspecified (PTCL) and angioimmunoblastic T-cell lymphoma (AITL) are non-Hodgkin's lymphoma of T-cell origin. Despite their relatively poor outcome, little is yet understood for the molecular pathogenesis of these disorders. To screen for transforming/tumor-suppressor genes among PTCL/AITL, we here tried to examine copy number alterations (CNAs) among HTLV-I-negative lymph node specimens of PTCL (n = 32) and AITL (n = 42). Utilization of high-density SNP-typing arrays (GeneChip Mapping 100K Hind III, Affymetrix) has allowed us to measure CNA at a mean resolution of 47 kbp among the samples. Interestingly, both of PTCL and AITL cells have frequent CNAs throughout the genome. Among the 55,700 SNP sites examined for PTCL, for instance, chromosome copy number at 20,860 sites was shown to be ≥ 3 in at least 2 specimens, while loss of copy number (≤ 1) was found at 2,287 sites in ≥ 2 samples. Similarly, among the AITL specimens, copy number gain (≥ 3) in ≥ 2 samples was observed at 9,897 SNP sites, and copy number loss in ≥ 2 samples was identified at 43 sites. We further searched for frequent CNAs among the whole dataset, revealing that copy number gain of ≥ 4 was observed in ≥ 10 cases at 1q, 2p, 2q, 8q, 9p, 13q, 15q, and 19q. On the other hand, copy number of ≤ 1 was found among ≥ 10 cases at 14q. Further, the copy number of a genomic region was shown to be linked to a differential diagnosis of PTCL/AITL. From the same dataset, we could also reveal a vast number of loss-of-heterozygosity (LOH) regions with a divergent length. Frequent LOH (≥ 5 cases) was mapped to 2q, 5q, 8p and 9p. By using a Cox proportional hazard analysis, we then investigated whether chromosome copy number or LOH likelihood is statistically linked to prognosis of the patients enrolled in this study. Interestingly, the copy number of narrow regions at 2q, 5p, 5q and 17q was shown to be statistically associated with the clinical outcome (P <0.01) while the LOH-likelihood at 8q was linked to it (P <0.05). The chromosomal regions identified in our screening are the good candidates where transforming- or/and clinical character-related genes are mapped. Especially, the narrow regions thus identified contain a few genes per region, and provide us with a provocative list of PTCL/AITL-related genes.