Gene Expression Profiling of Lung Neuroendocrine (NE) Tumors Reveals Gene Clusters Correlated with Central Versus Peripheral Location for Carcinoids
Hangjun Wang, Mee Sook Roh, Ronglai Shen, Junting Zheng, Gabriel Sica, Cameron Stock, Inderpal Sarkaria, Maria Pietanza, Natasha Rekhtman, Akira Iyoda, Valerie Rusch, William Travis. MSKCC, New York; Dong-A University College of Medicine, Busan, Korea; Emory University, Georgia; Kitasato University, Kanagawa, Japan; Weill Cornell Medical College, New York, NY
Background: Primary lung NE tumors include a spectrum of neoplasms from low grade typical carcinoid (TC), intermediate grade atypical carcinoid (AC) to high grade small cell carcinoma (SCLC) & large cell NE carcinoma (LCNEC). This study uses gene expression profiling to provide molecular signatures of the tumor subtypes and their correlation to clinical features.
Design: Gene expression profiling was performed with the Affymetrix U133 Plus 2.0 array in 120 primary lung NE tumors including 64 TC, 13 AC, 20 SCLC & 23 LCNEC. Hierarchical clustering analysis was performed based on the gene expression of the most variable genes among 1286 genes that were analyzed. Carcinoids were divided into central & peripheral locations. The association of the clusters with clinical and histological characteristics were tested using Fisher's exact test or the Wilcoxon rank-sum test. Differential gene expression (limma package in R) and gene set enrichment analyses were performed. Survival was evaluated using competing risk analysis.
Results: Cluster analysis identified three clusters (1, 2 & 3). Cluster 1 included all of the high grade NE neoplasms and 4 TC (6.2%). Cluster 2 included both TC (37.5%) & AC (38.5%). However, the majority of TC (56.3%) & AC (61.5%) grouped into cluster 3. For carcinoids, cluster 2 was associated with a central location (82.8%) while cluster 3 was associated with peripheral (70.5%) location (P<0.01). Supervised clustering analysis revealed many genes in cluster 1 that discriminate SCLC (i.e. ZIC2, ID4) from LCNEC (P<0.001). Cluster 1 was highly associated with smoking history and pack/years compared with clusters 2 and 3. Patients with cluster 2 tumors were much younger than patients in cluster 1 & cluster 3 (mean age 55 vs 67 vs 63 yrs, P<0.01).The 30-month cumulative incidence of death due to disease was much higher for cluster 1 (27%) than cluster 2 (0%) & cluster 3 (4%) (P=0.012).
Conclusions: Gene expression profiling reveals three clusters of genes in lung NE tumors. The high grade NE neoplasms, SCLC and LCNEC share the same cluster while TC and AC are divided between two clusters that correlate with central vs peripheral location. Although SCLC and LCNEC share the same cluster, these tumors are distinguished by expression profiles of many genes. Further study on these genes such as ZIC2 and ID4 may provide important information about these tumors.
Monday, March 19, 2012 2:30 PM
Platform Session: Section D, Monday Afternoon