Development of a diagnostic tool for the evaluation of progression risk of early oral premalignant lesions Free

A51

Introduction: Forty percent of patients with oral squamous cell carcinoma (OSCC) do not survive past 5 years. Identification of oral premalignant lesions (OPLs) at high-risk for progression would allow for earlier intervention thus improving prognosis. Genetic alterations have been recognized as predictors of progression risk in OPLs. Here, we describe an “oral pre-cancer risk assessment array” (OPRA) dedicated to the evaluation of cancer risk of early premalignancies.
 >Methodologies: To select regions for OPRA, we analyzed 94 oral lesions by tiling-path array comparative genomic hybridization (CGH) to globally identify genetic signatures at a resolution of 50 kb. This whole genome array allowed the interrogation of 26,819 overlapping bacterial artificial chromosome (BAC) clones with complete coverage of the human genome. These lesions included 47 high-grade dysplasias (severe dysplasia and carcinoma in situ), 24 low-grade dysplasias (hyperplasia, mild and moderate dysplasia) with clinical outcome (follow-up ≥5 years), and 23 OSCCs. Genetic changes were detected by breakpoint algorithm aCGH-Smooth and were filtered for copy number polymorphisms previously identified in 95 non-cancer subjects.
 >Results: We first focused on 47 high-grade dysplasias, the premalignant stage that is associated with the strongest risk of progression. Genetic regions identified in high-grade dysplasias that are important for progression are likely to be maintained in OSCCs. Thus, regions recurrent in ≥20% high-grade dysplasias were compared with 23 OSCCs to identify the common minimal regions of alteration. Next, genetic alterations were assessed in early stage OPLs including 24 low-grade dysplasias (9 progressing and 15 non-progressing cases). The level of genomic instability was low in these lesions, however, alterations frequent in progressing low-grade dysplasias but infrequent in non-progressing cases were identified (p<0.05). To account for regions of high-level amplification despite their low frequency of occurrence, gene amplifications detected in ≥2 oral lesions were also listed. As controls, cancer genes, genes previously identified in OSCC, regions harboring microsatellite markers previously shown to be predictive of progression risk, and 4 regions of copy number polymorphisms found in normal oral mucosas were also included in this first version of mini-chip.
 >In total, 384 BACs representing 159 regions were selected from the RPCI-11 human BAC library. Additionally, 1152 random BACs were also incorporated in the array for normalization. All BACs were spotted in quadruplicates in this mini-chip. We demonstrate the detection of copy number alterations in clinical specimens in this mini-chip. These include both copy number gains and losses, with changes validated by the whole genome tiling-path array.
 >Conclusion: By whole genome profiling of OPLs and OSCCs, we obtained a priori knowledge of regions of interest that would minimize the cost and increase the robustness to target specific regions for the development of OPRA. This mini-chip will serve as a valuable diagnostic tool, and the use of DNA-based technology has practical advantages in clinical setting due to its inherent stability and ease of handling. This work represents the first attempt at the construction of a clinical tool that is well-targeted towards predicting outcome in early stage OPLs.