Johns Hopkins School of Medicine
Data Analysis in Cancer Management: Risk Assessment. Looking for germ line variants in families.
In cancer management: on tissue. Prognosis, classification and response.
IN absence of tissue: are there still cancer cells present? are there actionable changes? Are there signs of cancer?
Use ctDNA to answer some of these questions. ctDNA is minority of DNA that’s present. half life is 30 min – one hour.
- Technical – specificity and sensitivity
- Biological: is there detectable amount of DNA in fluid
- Interpretation: How do you do it?
Favourite markers: somatic mutations, indels and rearrangements. Very good specificity.
Technical challenges – finding the needle in the haystack. How do you do it? Used to be that you would search the hay one by one… digital PCR for instance. Now, we marry the digital signal to NGS. Disadvantages: error rare is high for the clinical applications. Early detection.
- assignment of unique identifier to each molecule
- a UID is used to group reads from a common template.
- Use a PCR based approach. (Allows detection at 0.01%.)
- divide results into mutant, non-mutant and artifact.
Which tumour types can we detect in cancer? sensitivity depends on cancer type. Best for bladder, colorectal, ovarian, gastroesophageal, etc. Bad for Thyroid, Glioma, prostate, renal cell, etc.
ctDNA is a dynamic biomarker. Can be used to follow tumour.
Stage II: colorectal cancer. 80% of patients cures by surgery alone. Good setting for ctDNA. 230 patients. Plasma for 4 to 10 weeks. >1000 plasmas analyzed. Physicians blinded to ctDNA.
Panel used to probe small number of variants. ctDNA is a strong marker for recurrence. If it’s not there, patients fare very well.
ctDNA detects minimal residual disease and predicts recurrence. Provides real-time measurement of disease burden.
Detection of Resistance.
Liquid biopsy resistance mutations. Pre-treatment, no mutations, post treatment, many did. 0-12 mutations.
Early detection: the holy grail
The most difficult and the longest to get out of the lab. We know it works. People will be missed if not part of screening. Localized disease are frequent.
For early detection, the source makes a difference. Stool works better than plasma for colorectal cancer. CNS tumours are not detectable in plasma, pap smear tests work best for ovarian and endometrial cancer.
Doing both plasma and pap smear, improves over either one alone.
Pancreatic cancers are tough – However, a lot of cysts are being found. Cyst types have different genes mutated. Cyst fluid gathered in a test – SNP analysis done, 1026 resected cysts.
Goal: reduce surgeries, without leaving anything behind.
Summary: It can be done, it can be done well.
Vision: Prevent advance tumours by integrating such testes into routine physical exams.