It is becoming increasingly apparent that there are subgroups of patients with potentially malignant oral lesions that are at extremely high-risk for cancer development despite repeated excision. Histology alone is a relatively poor indicator for this aggressive clinical behavior. Quantitative Tissue Pathology (QTP) assessment of nuclear phenotypes using a computer microscope imaging platform has shown promising results in risk prediction of oral premalignant lesions (OPLs).

Objective: The objective of this study was to examine the predictive value of high throughput QTP to assess nuclear phenotype scores (NPS) of samples on a unique oral premalignant tissue in situ microarray (OPTiMA) (all samples have known outcome) and to link this data with 2 other indicators of progression risk for OPLs: 1) loss of heterozygosity analysis (LOH) (done on tissue dissected from parallel sections to those used to construct the microarray) and 2) in-situ analysis of chromosomal number alteration as determined with fluorescence in situ hybridization.

Method: Nuclear Phenotypic Scores (NPS) were determined for each sample on OPTiMA: 19 normal cases and 35 primary dysplastic lesions (no head/neck cancer history) with known outcome. Twenty of the dysplasia were selected from an on-going longitudinal study and were refractory to treatment (D-RTT), characterized clinically by repeated recurrence or progression after resection. The remaining 15 dysplasias were randomly chosen from the population-based archives of the BC Oral Biopsy Service to represent the general population of dysplasias (D-GP), with outcome identified from records after completion of the study.

Results: Although there was no significant difference in demographics and degree of dysplasia for D-RTT and D-GP, the groups differed significantly in molecular pattern as determined by loss of heterozygosity analysis and in FISH profiles. Seventeen (89%) of 19 D-RTT showed loss of 3p and/or 9p and 11 (58%) showed additional loss at 17p. Similarly, D-RTT samples showed significant elevations in abnormal FISH profiles including increases in frequency of samples with CEP7 and CEP11 trisomy (> 10% nuclei: 60% vs. 13% of cases; 50% vs. 7%) and polysomy (> 10% nuclei: 20% vs. 7% of cases and 40% vs. 20%). Strikingly, NPS scores showed a significant association with resistance to treatment, with median NPS values for D-RTT and D-GP samples of 5.56 and 3.89 respectively (P = 0.0003).

Conclusion: The data support for the first time, the use of tissue microarray as a tool to integrate the data from the high throughput QTP approach and other in-situ analyses. More importantly, the study supports the potential application of nuclear phenotypic scores as a useful marker in identifying oral lesions requiring more aggressive treatment and/or novel interventions. (Sponsored by MOP-77663, CIHR and R01 DE13124 and R01 DE 17013, NIDCR).

98th AACR Annual Meeting-- Apr 14-18, 2007; Los Angeles, CA