since blood flows through the entire body it can pick up localised tumour cells (for instance) and the few circulating cells with mutations can be detected with pcr or FISH techniques. Similarly disorders of the fetus ( trisomy for instance) can also be detected in the circulating maternal blood stream

Yes, It can be detected from the blood. Extract DNA and perform Sanger sequencing of desired genes or whole exome sequencing can be used to detect mutations.

@Paul Rutland

How reliable would be the result? Somatic mutation occurs in DNA of specific region. More precisely it can be detected using normal and disease sample.

@Muhammad Umair 

Your answer is not clear. I am talking about the somatic mutation which can occur in the body locally. Since blood is circulating in the entire body how detection of somatic mutation is possible?

@Snijesh

The results can be very reliable if you know what you are looking for. Nested pcr will detect single cell amounts of DNA and normal pcr amplifies the amount of dna present 1000 times for every 10 cycles so with well designed primers and stringently set up pcr the results should be very good and specific. I do not see it as very different from detecting virus/bacterial dna in the circulation. Specific primers will amplify well even in the presence of large amounts of normal dna

@Paul Rutland

I am going for next generation sequence analysis with the blood sample to detect the somatic mutation. I don't have any specific target list genes with me. I want to see all somatic mutations that can occur for a patient for a particular disease (say, breast cancer).

Once the get the fastq file from NGS machine, I will be using variant calling pipeline to detect the somatic mutation. So in this case, how reliable will be the result?

@Snijesh

you may find the attached of interest Snijesh but there are more recent papers online. I cannot help with pipeline as this type of analysis is outside of my area of competence

@Paul Rutland

Thank you. I have already gone through papers like this and that is why my doubt came out. When we are analyzing a tumor tissue sample in comparison with a normal tissue sample, it is quite easy to identify somatic variants. More importantly, almost all variants can be identified here using this method with high accuracy. But, when we are going for a blood sample, all the mutated genes might not be present in the sample compared to tissue? And sometimes we may loose most important genetic variants? Am I right?

@Snijesh

There is a risk that low incidence variants will be missed but I think that this effect will be minimised as the sample size ( amount of DNA analysed) gets larger so an important factor will be the method of CTDNA  isolation. The DNA will be small and broken so methods suitable for viral dna isolation may be best  but again there is much information about which kits are best online. Tumours often have very good vasculature so you can expect ctDNA to be present in blood and probably at amounts representing the amount of each variant present in the tumour(s). You will probably never know if you missed one variant of many but the technique is good and improving but perfection is probably too high a target to aim for until more targeted techniques become commonplace

I believe you must have read a lot of paper about somatic variant calling from blood DNA. It is possible but the accuracy cannot be promised.

Ideally, the alternate allele fraction should be 0% (homozygous) or 50% (heterozygous). If you see 10%, it is possible to be a somatic variant. Again, it needs to be confirmed by Sanger sequencing.

I have doubt about how to define somatic variants as well. I often see variant allele fraction around 30%. I do not know if it can be explained by a somatic variant. I feel the fraction is too high. I would like to know how the other people define them as well.