In addition to visualization of the lesion, a second important barrier revolves around risk prediction for oral premalignant lesions (OPLs). OPLs vary widely in potential for malignant transformation with only a fraction eventually progressing into cancer. At present, the gold standard for prediction involves the determination of the presence and degree of dysplasia in a biopsy. Severe dysplasia or carcinoma in situ (CIS), grouped as high-grade premalignant lesions, are characterized by persistence, recurrence, and high likelihood of eventual progression to invasive cancer, if left untreated. Even with aggressive surgical treatment, 30-40% of the lesions still recur or progress within 30 months. Histology has a good predictive value for such lesions. In contrast, determination of prognosis for lesions with histological changes that are less than severe is more problematic. The majority of OPLs without dysplasia or with low-grade dysplasia will not progress into cancer and histology alone does not clearly differentiate between those that will progress and those that will not.
Development of approaches that will differentiate cases with a high likelihood of progression will allow more timely intervention at an earlier stage for high-risk cases while avoiding over-treatment of lesions with little risk of progression. The BC OCPP has piloted two such approaches. One involves high-resolution computer microscopy, potentially a high-throughput, cost-effective filter for outcome. The second involves a final set of sieves based on molecular alterations in the tissue. The latter uses two approaches: 1) microsatellite analysis for loss of heterozygosity (LOH); and 2) Genome Profiling. They are described below.
Histological assessment of a biopsy has as its basis a consensus by pathologists of the types of phenotypic changes associated with risk, with a classification system based on the presence of specific features (dysplasia) and the degree of epithelial involvement. High resolution computer imaging systems measure these same features as well as the frequency and distribution of cells, in a quantitative fashion. A Quantitative Tissue Phenotype (QTP) is derived. These helper tools allow the pathologist to use reproducible and repeatable measures of specific phenotype characteristics that make up the appearance of dysplastic cells in tissues and can be statistically associated with risk. The two systems are described below.
- Quantitative Pathology (QP)
The tissue analysis system used in British Columbia assesses quantitatively stained material for association with outcome in sectioned material. It examines 120 nuclear features in cell nuclei, the majority of which describe the distribution of DNA, unravelling very complex features into quantifiable algorithms. Preliminary data suggest that the QP system can identify nuclear changes that associate with the presence of high-risk molecular clones and with future outcome. For more information on the QP system, see Molecular Markers.
- Quantitative Cytometry (QC)
The second computer system is a noninvasive counterpart to the work station described above, using exfoliated cell brushings. Currently in development, the projected use for this device is to assist the clinician in identifying lesions requiring biopsy, to be used in conjunction with visualization devices. For more information on the QC system, see Molecular Markers.
Oral carcinogenesis is a multistep process requiring the accumulation of multiple genetic alterations in the epithelium. Loss of Heterozygosity (LOH) marks this genetic instability and clonal selection and is a powerful potential risk marker. Although several laboratories have previously reported an association of LOH with progression, no longitudinal studies have validated this approach. The BC OCPP has published two LOH risk models based on retrospective analyses that suggest a strong association of specific LOH patterns with outcome. The first model examines associations of LOH patterns with risk of progression for OPLs to cancer. Three ‘Molecular Risk” (MR) patterns have been identified, MR1 (lowest risk), MR2 (intermediate risk) and MR3 (highest risk). Twenty-six percent of oral premalignant lesions with MR2 patterns and 47% with MR3 patterns progressed to cancer within 5 years. In a parallel retrospective model for recurrence of cancer from OPLs developing after treatment at the former cancer site, these same markers showed the lowest risk with pattern MR1. Relative risk for MR2 and MR3 was 23.5 and 26.1, respectively. Sixty-one percent of lesions with such LOH patterns developed recurrences within 2 years and 77% within 5 years. For more information on these data, see Molecular Markers.
Within the scope of the OCPL study, the BC OCPP has created a system that can lead to rapid identification of candidate DNA markers of progression and recurrence through high resolution genome-wide profiling of biopsies of individuals with extensive follow-up. Novel genes frequently altered in oral cancers have been identified. Refinement of this platform has generated the first whole genome profile of oral dysplastic lesions and has led to the discovery of patterns of genetic change in oral dysplastic lesions that are associated with progression risk.
Building upon these early findings, the BC OCPP is developing a miniaturized microarray called the OPL Risk Prediction (OPL RP) chip. This chip is designed to analyze very small samples such as those collected in longitudinal monitoring of patients. Version 1 of the OPL RP chip concentrates on probes in the frequently altered regions identified in the >100 oral premalignant lesions analyzed to date. The OPL RP chip will evolve and reiteratively improve with feedback on how informative the individual regions are in predicting risk of progression.