Early Cancer Detection Using mRNA Profiles of Tumor-Educated Platelets
1.
2. Background
• Current research on early diagnosis of cancer
• Molecular profiling of tumor tissue samples as
a potential cancer classifying method
• The use of blood-based liquid biopsies have
been suggested
• Plasma DNA
• Circulating tumor cells
• Non-specificity of these liquid biopsies makes it
difficult to target primary tumors
3. Tumor-Educated Platelets
• Tumor-educated platelets (TEPs) emerged as
promising biosource for cancer diagnostics
• Recently known to be central players in systemic
and local responses to tumor growth
• External stimuli on TEPs induce specific
splicing of pre-mRNAs
4. mRNA Profiles of TEPs are Distinct from
Platelets of Healthy Individuals
• 283 platelet samples in total
• 55 healthy donors
• 228 cancer patients
• Cancer types
• Hepatobiliary Cancer (HBC)
• Pancreatic Cancer (PAAD)
• Glioblastoma (GBM)
• Breast Cancer (BrCa)
• Colorectal Cancer (CRC)
• Non-Small-Cell Lung Carcinoma (NSCLC)
• 1,453 increased and 793 decreased
mRNAs in TEPs as compared to
healthy donors
5. mRNA Profiles of TEPs are Distinct from
Platelets of Healthy Individuals
• Heatmap
• Up-regulated and down-
regulated mRNAs
• Increased TEP mRNAs
• Vesicle-mediated transport
• Cytoskeletal protein binding
• Decreased mRNAs
• RNA processing
• RNA splicing
6. mRNA Profiles of TEPs are Distinct from
Platelets of Healthy Individuals
• Figure F
• Training set
• To test the algorithm
• n = 175
• Sensitivity – 96%
• Specificity – 92%
• Figure G
• Validation set
• To validate the algorithm
• n = 108
• Sensitivity – 97%
• Specificity – 94%
• Designed algorithm for pan-
cancer analysis
• 1,072 RNAs selected
• Total samples split into training
and validation set
7. mRNA Profiles of TEPs are Distinct from
Platelets of Healthy Individuals
• Receiver Operating Characteristic
(ROC) Curve
• Area Under the Curve (AUC) values
• Training = 0.988
• Validation = 0.986
• Random classifiers ≈ 0.5
8. Tumor-Specific Educational Program of Blood
Platelets Allows for Multiclass Cancer Diagnostics
• Tumor-specific algorithms
classifications compared to healthy
controls
• One-by-one
• High sensitivity and specificity for
each training and validation test
• Unable to differentiate non-
metastasized and metastasized
tumors
9. Tumor-Specific Educational Program of Blood
Platelets Allows for Multiclass Cancer Diagnostics
• Multiclass algorithm to differentiate all
different cancers at once
• Figure D – Training
• Figure E – Validation
• Overall accuracy = 71%
• Not very sufficient for introduction to cancer
diagnostics
• Quite low, could be due to small sample size
for each tumor type
• Around 10-35 samples per type
10. Companion Diagnostics Tumor Tissue Biomarkers
are Reflected by Surrogate TEP mRNA Onco-
Signatures
• Distinguish gene mutant
tumors with its wild-type
tumors in various tumor types
• KRAS
• EGFR
• MET
• HER2
• Some ability to differentiate
subtypes of cancers
11. Implications
• Current blood-based biosources are being researched
for their diagnostic capabilities
• Plasma DNA
• Exosomes
• Circulating tumor cells
• Platelet RNAs profile were shown to be affected
almost all cancer patients, regardless of type of
tumor
12. Advantages of TEPs as Biosource
• Only need small amount of platelet RNA (100 –
500 pg)
• Can be stored up to 48 hours in room temp.
• Still have high quality RNA
• Easy for daily clinical analysis
• Ship to others for research
• Can determine tumor type or subtype by just
looking at platelet RNA profile
13. Limitations
• Not able to differentiate non-metastasized
and metastasized tumors
• Could be due to small sample size
• The multiclass detection of some tumor
types were not very accurate
• E.g., CRC, PAAD, HBC: accuracy < 76%
• Not much known/researched on TEPs
• Needs further proof for validity
14. Conclusion
• Using blood platelets as biosource shows
promising clinical application in detecting
cancer
• The self-learning algorithm may be more
accurate and precise with the addition of more
platelet samples
• Further studies are required to address factors
that may influence platelet mRNA profile
• Chronic or transient inflammatory diseases
• Cardiovascular events & other non-cancerous diseases