The Ability of Serum miRNAs to Act as Biomarkers for the Oral Squamous Cell
Carcinoma.
Christopher DickmanCathie Garnis
Experimental Medicine Research Day
Background
• Oral Squamous Cell Carcinoma (OSCC) is the most common form of head and neck cancer
• Poor survival rate due to rapid progression and frequent recurrence
• Understanding molecular alterations in OSCC may lead to drug targets and biomarkers
miRNAs
• Short 22 base pair molecules of RNA• Responsible for post-transcriptional genetic
regulation• Known to play a role in many cancer processes
Aims• We aim to find miRNAs that are significantly
different between the serum of individuals with oral cancer or carcinoma in situ (CIS) and healthy individuals that can be used as a biomarker for high risk lesions
Methods• Serum from 51 healthy individuals and 48
individuals with oral cancer or CIS were profiled for 742 miRNAs using RT-PCR
• 47 additional control samples and 32 additional cancer/CIS samples were included in a validation set
Haemolysis
• Analysis of serum miRNAs is made more difficult by the varying contribution of blood cell RNA to the serum
• 2 Options– Exclude samples with haemolysis– Exclude miRNAs affected by haemolysis
Data Analysis
• miRNAs normalized to the expression of miR-23b
• Creation of a subset using general discriminant analysis
• Subset analysis was performed using a generalized linear model (logistic)
• Best subsets of 4,3 and 2 miRNAs were discovered
Validation
• Samples were run 8 per plate as opposed to one sample per two plates at ~1/8th the price
• Samples are run in triplicate– Variation among replicate increases in miRNAs
with low expression– There is difficulty in analyzing the data where all
replicates have a different CT value
Sources of Down-Regulated miRNAs
• Many of theses miRNAs are down regulated in cancer patients’ serum (miR-23a, 342-3p, 33a) what is their source?
• Often it’s assumed cancer related miRNAs are released from the tumor
• Immune system is known to be a large contributor to miRNA excretion
Conclusions
• We have identified candidates for a validation test of serum miRNA biomarkers for Oral Cancer
• Ongoing work must be preformed to determine what role overtraining plays in data analysis
Model Training AUC Validation AUC
23a, 205, 33a 342-3p .91 .77
23a, 205, 33a .90 .62
23a, 205 .83 .66
23a, 346 .85 NA