Question: NGS on cancer tumor
gravatar for patrkuenzi
13 months ago by
patrkuenzi20 wrote:

NGS was done on MSI-H tumor. Tumor has 30 percent purity and was reported seperately by pathologist. Many point mutations identified-20 out of 125. One frameshift mutation mutant fraction is 35 percent. The rest of the mutations are 20-28 percent. Would the mutation with a mutant fraction of 35 percent be a homozygous mutation?

next-gen • 706 views
ADD COMMENTlink modified 13 months ago • written 13 months ago by patrkuenzi20

Was the whole sample used for DNA extraction? Be noted that although the tumor was reported with 30% purity, maybe only a part of the sample with much higher purity was used for DNA extraction.

ADD REPLYlink modified 13 months ago • written 13 months ago by chen1.7k

So you are saying that somatic mutations in tumor are rarely homozygous? I am unsure if the whole sample was used for DNA extraction. Would it be common to only use the part of the sample with higher purity? The contact person at the lab (PHD) stated that the mutant fraction of 30 percent would be related to the tumor purity of 30 percent. Does this make sense? Any explantation is greatly appreciated.

ADD REPLYlink written 13 months ago by patrkuenzi20

You said all mutations are 20% or higher, that's uncommon for a 30% purity tumor.

ADD REPLYlink written 13 months ago by chen1.7k

It's rare that a mutation in tumor is homozygous.

That's emphatically not true, as chromosomal deletions (or copy-number neutral LOH) lead to homozygosity quite often in tumors.

ADD REPLYlink written 13 months ago by Chris Miller20k

You're right, I agree.

ADD REPLYlink written 13 months ago by chen1.7k

Depends on the definition of homozygous--two hits to the same base (both parental alleles), or ~100% of the cells carrying a mutation (one mutant allele, selected for, somehow)? The definition is important when thinking about a patient and familial risk implications, etc.

ADD REPLYlink written 13 months ago by Sean Davis25k

Yes, all mutations are around 20 percent, except for on with 35, and one with 28. This is a MSI-H tumor with what seems to be a very high mutational load as there is a POLD 1 mutation at 20 percent, many DSB repair gene mutations, and loss of msh2 and msh6. The tumor purity was read by pathologist separately. Any other thoughts?

ADD REPLYlink written 13 months ago by patrkuenzi20

I am a medical professional reading this input for a family member. The report specifies that there are no translocations or amplifications. I do not have any of the other information. So the tumor purity would be double the average of the median of VAF? This would make a big difference. If the median VAF were around 24 percent, as you say, which is pretty accurate. Then to identify a homozygous SNV you would expect a VAF of 48 percent? You can assume most of the other mutations are heterozygous mutations, and it is true this one is close to that median.

I was confused because of her (NGS PHD explanation): MAF is calculated as follows: (number of distinct mutant reads/number of distinct total reads)*100 It is a reflection of how much of a clonal population is present in a tumor and it is difficult to be certain whether a mutation is heterogeneous or homozygous based on MAF alone. The MAF will correlate with tumor purity. If the tumor purity is calculated as 30% and the MAF is also around 30%, this suggests that the mutation is present in almost all the cells within the tumor. A lower MAF would suggest it is present in a subset of cells. In this report, the tumor purity scored by the pathologist is set at ~30%. The average MAF of the mutations is around 30%, suggesting that the mutations reported are present in almost all the cells within the tumor.

Thanks so much for you response. These answers are very important..

ADD REPLYlink written 13 months ago by patrkuenzi20

I am also a medical professional and would like to strongly discourage such clinical conversations in an online bioinformatics forum. I don't mean to discourage academic conversation here, but having a discussion about interpreting a variant for a family member on an online forum seems problematic. I sincerely DO understand the desire to help a family member, but I would very strongly recommend that the results be discussed with the care team directly.

ADD REPLYlink modified 13 months ago • written 13 months ago by Sean Davis25k

Agreed, with the addendum that I believe it's fair game here to ask questions about terminology or how sequencing data is processed and analyzed. Doctors are busy and patients seeking to understand results isn't a bad thing, as long as we're careful not to cross the line into making treatment recommendations here.

ADD REPLYlink written 13 months ago by Chris Miller20k

I really appreciate this response, because I ordered the test, for personal use. When I asked for clarification, I really couldn't understand some basics. The care team cannot answer my questions, they don't know the answer, so you all are my best chance for clarification.

ADD REPLYlink written 13 months ago by patrkuenzi20

I now understand the context better and am glad that you are getting the answers you need.

ADD REPLYlink written 13 months ago by Sean Davis25k

The description you provide isn't internally consistent. If a heterozygous mutation is in every cell of the tumor, then the VAF/MAF will be at 50% of the tumor purity. Think about a pure population of 100 cells. Each cell has 2 copies of each gene, but only one that's mutated (for a het event). This means that 50% of your reads will be from the mutant allele, and your VAF will be 50%. That scales for purity, so if your tumor cellularity/purity really is 30%, then the founding clone will be at 15% VAF. Copy number events may shift some values up or down, and subclonal populations can also result in downward shifts.

ADD REPLYlink written 13 months ago by Chris Miller20k

Three things: 1) I was assuming there was a heterozygous mutation in every cell of the tumor based on my limited knowledge of VAF. I know nothing about if they really are heterozygous mutations. 2) The sequencing coverage is 125x, and this was for 125 genes.

3) What I am trying to figure out is if the somatic mutation with MAF of 35% would be considered a homozygous mutation. I see many medical articles stating that VAF over 60% (adjusted for tumor purity) are homozygous mutations. I am actually hoping that it is not a homozygous mutation, as this would negatively effect treatment. What I was doing was thinking was that if 35 percent VAF of a 30 percent tumor purity then this would be around 100 percent VAF, and as such, a homozygous mutation.

I really, really appreciated your help and any clarification of my understanding. This is far from the field I practice in, but a very important question that I cannot seem to have clarified thus far.

ADD REPLYlink written 13 months ago by patrkuenzi20

patrkuenzi : Please use ADD COMMENT/ADD REPLY when responding to existing posts to keep threads logically organized.

ADD REPLYlink modified 13 months ago • written 13 months ago by genomax59k

Technically, due to tumour heterogeneity and clonality, a mutation found at a frequency of 1% in a bulk tumour biopsy may be homozygous in the clonal population of cells in which it's located. No analysis of a bulk biopsy can give a definitive answer, though, and for an important clinical decision I would neither come to any conclusion based on tumour purity, as many 'normal' cells surrounding a bulk tumour may themselves carry somatic mutations. I believe that they determine purity based on the expression of certain proteins on the cell surface, but neoplastic cells may exist long before these markers appear.

Take a look at my small study on breast cancer whole genome sequencing from a couple of years back: I focussed specifically on mutation frequencies and how they changes between the primary and metastatic tumour biopsies.

ADD REPLYlink modified 13 months ago • written 13 months ago by Kevin Blighe33k
gravatar for Chris Miller
13 months ago by
Chris Miller20k
Washington University in St. Louis, MO
Chris Miller20k wrote:

1) Your ability to infer this depends greatly on the depth of sequencing. It helps to picture each VAF with error bars on it, that grow smaller as your depth increases. In 30x sequence, 35% is absolutely not significantly different from 20-28%. In a 500x tumor, it probably is.

2) Without seeing your data, I'll assume that your tumor purity is somewhere around the median of the VAFs you describe. (20-28, so we'll call it 24%). That means you'd expect tumor purity to be double that (48%), and homozygous SNVs should appear around 48%. That doesn't seem to fit your frameshift VAF, so we can explore other hypotheses. What if there is copy number amplification of the mutant allele (3x CN). That would lead to a VAF of 1.5 times the baseline, which would be 36%. That seems to fit your point nicely, but you should go looking for corroborating evidence. What does CN inference tell you? What are the b-allele frequencies of germline SNPs in that region? etc.

ADD COMMENTlink modified 13 months ago • written 13 months ago by Chris Miller20k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 2142 users visited in the last hour