Stream: genomics
Topic: driver/passenger mutation
Patrick Werner (Apr 02 2020 at 08:55):
we have the need to capture if a variant is a passenger or driver mutation. What do you think about using Variant.interpretation for that purpose?
Jamie Jones (Apr 02 2020 at 10:42):
Interpretation is locked to a specific value set and I don't think it's right even if it weren't... This is a derived field from functional annotation, yes?
Jamie Jones (Apr 02 2020 at 10:43):
Is it not in the SO subtree we selected for that?
Patrick Werner (Apr 02 2020 at 10:55):
(nvm)
Patrick Werner (Apr 02 2020 at 10:58):
No it isn't derived from functional annotation (at least not directly or mappable)
Patrick Werner (Apr 02 2020 at 10:59):
Driver mutation has the meaning, that this variant is one of the causes why the tumor started existing
Patrick Werner (Apr 02 2020 at 11:00):
variants which are there, but don't have an influence on the formation of the tumor are called passengers
Jamie Jones (Apr 02 2020 at 11:00):
https://www.nature.com/articles/s41598-019-53454-1
Patrick Werner (Apr 02 2020 at 11:01):
Identifying and distinguishing cancer driver genes among thousands of candidate mutations remains a major challenge.
I'm glad the lab does this :-D
Jamie Jones (Apr 02 2020 at 11:02):
Sounds like a tricky enough space that it should be a separate observation rather than a component on variant
Patrick Werner (Apr 02 2020 at 11:02):
Technically interpretation is an extensible binding, but i don't want to push too hard for this as it is very specific
Jamie Jones (Apr 02 2020 at 11:03):
It seems closest to our implication structure
Patrick Werner (Apr 02 2020 at 11:03):
Jamie Jones said:
Sounds like a tricky enough space that it should be a separate observation rather than a component on variant
that would have been my next question. Do we need a new Profile for that?
Patrick Werner (Apr 02 2020 at 11:03):
or is it just an implication
Jamie Jones (Apr 02 2020 at 11:04):
Since it is basically saying "we got a hit on this external knowledge resource/algorithm on this variant
Patrick Werner (Apr 02 2020 at 11:07):
So this would somehow fit better into the diagnostic-implication?
Jamie Jones (Apr 02 2020 at 11:10):
What is the clinical relevance for identifying driver mutations?
Patrick Werner (Apr 02 2020 at 11:11):
component:clinical-significance would be a good place to capture this.
Jamie Jones (Apr 02 2020 at 11:11):
Yes similar to calling something pathogenic / oncogenic
Patrick Werner (Apr 02 2020 at 11:12):
you look at the pathways of driver mutations to hopefully identify a place in the pathway where you can influence the pathway by medications
Patrick Werner (Apr 02 2020 at 11:13):
i think (don't know) that oncogenic and driver could be names for the same concept. I'm going to ask our onco-genetic expert about that.
Jamie Jones (Apr 02 2020 at 11:18):
I'm not an expert here either but it sounds like pointing out the clinical significance of a somatic variant could be viewed as classifying a subtype of cancer, which would be a diagnostic implication
Patrick Werner (Apr 02 2020 at 11:18):
I think the AL of https://loinc.org/53037-8/ could need an update. But i wait for the responses first.
Jamie Jones (Apr 02 2020 at 11:18):
That's been a standing item for a while...
Patrick Werner (Apr 02 2020 at 11:19):
happy to move it forward :smiling_devil:
Patrick Werner (Apr 02 2020 at 13:50):
ok, got the first feedback: driver is equivalent to pathogenic/cancerogenic, but a more specific concept
Patrick Werner (Apr 02 2020 at 13:52):
passenger could be expressed as bening. They don't like the terms, but i think it is mappable. Or double codeable: pathogenic and benign due to the extensible coding, adding a second Coding to express driver/passenger explicitely
Patrick Werner (Apr 02 2020 at 13:53):
@May Terry @Alexander Mankovich @Halina Labikova what do you think?
Kevin Power (Apr 02 2020 at 14:32):
FWIW - This does feel like something that fits with Diagnostic Implication. However, some questions from me. If we created this as a new component:
- What would we call it?
- Does LOINC happen to have a code for it? Maybe SO?
- Are the answers just Driver and Passenger? Does it need an 'Unknown'?
- I assume it would be optional?
Bob Milius (Apr 02 2020 at 14:47):
is driver vs passenger something found in a lab report, or is it downstream interpretation of the lab report? If the latter, maybe start thinking about a new IG?
Patrick Werner (Apr 02 2020 at 14:48):
It is part of the lab report
Patrick Werner (Apr 02 2020 at 14:48):
@Kevin Power 's point: i don't think that's a new component, it is an Implication.
Patrick Werner (Apr 02 2020 at 14:49):
couldn't find a loinc code yet. Looking at NCIT now.
Kevin Power (Apr 02 2020 at 14:49):
So, a new profile? What other components/attributes are important to fully describe it?
Patrick Werner (Apr 02 2020 at 14:50):
answers: driver/passenger/unknown
Patrick Werner (Apr 02 2020 at 14:50):
just a simple profile with 3 value binding: driver/passenger/unknown
Patrick Werner (Apr 02 2020 at 14:51):
could also just use diagnostic implication as mentioned before.
Jamie Jones (Apr 02 2020 at 14:53):
I think this is a sub concept of clinical significance on implications
Patrick Werner (Apr 02 2020 at 14:54):
?
Patrick Werner (Apr 02 2020 at 14:54):
NCIT doesn't have codes for driver/passenger ... but they have a code for:
Passenger in Car for Hour without Break (Code C106687)
Patrick Werner (Apr 02 2020 at 14:54):
:rolling_on_the_floor_laughing:
Jamie Jones (Apr 02 2020 at 14:55):
I don't want to be in a car without brakes
Patrick Werner (Apr 02 2020 at 14:57):
Jamie Jones said:
I think this is a sub concept of clinical significance on implications
so it is a diagnostic implication? It is representing a specific "flavour" of pathogenic and benign.
Kevin Power (Apr 02 2020 at 14:57):
Jamie Jones said:
I think this is a sub concept of clinical significance on implications
I am hesitant to go there just yet, mostly because I don't know that we have the best answer/guidance on 'clinical significance' for cancer testing as of right now.
Jamie Jones (Apr 02 2020 at 14:58):
We don't, I agree. Splitting it off into it's own may help a lot
Patrick Werner (Apr 02 2020 at 15:01):
ok :+1:
Patrick Werner (Apr 02 2020 at 15:02):
searching for loinc codes i found this: image.png
Patrick Werner (Apr 02 2020 at 15:02):
:rolling_on_the_floor_laughing:
Patrick Werner (Apr 02 2020 at 15:09):
Tumorigenesis (Code C18121) could be the code for this Observation.
Patrick Werner (Apr 02 2020 at 15:12):
Jamie Jones said:
I think this is a sub concept of clinical significance on implications
i still didn't quite get it. Did you mean, this should be a new profile derived from the implication profile and be named clincal significance?
Jamie Jones (Apr 02 2020 at 15:14):
It seems like what we might recommend to use for values to report clinical significance in the component we currently have on implications
Patrick Werner (Apr 02 2020 at 15:21):
ah ok. Back to the start. Yes, i think recommending clinical significance for driver/passenger is also ok. And avoids "another Box"
Patrick Werner (Apr 02 2020 at 15:21):
driver=pathogenic, passenger=benign
Kevin Power (Apr 02 2020 at 16:54):
Patrick Werner said:
driver=pathogenic, passenger=benign
How confident are we in this mapping? And as we have discussed, does Pathogenic really work for cancer testing?
Patrick Werner (Apr 02 2020 at 17:57):
As written above, two of our biologists agreed to this. disclaimer: driver is a narrower concept than pathogenic, but "lives" inside of pathogenic. So i would double code it. First coding bening/pathogenic. Second slice (not part of the ig) CodeableConcept: driver-passenger.
Patrick Werner (Apr 02 2020 at 17:58):
Which also allows to search for all pathogenic Implications with just on search code, but also enables specific search for driver/passenger
Patrick Werner (Apr 02 2020 at 17:59):
still waiting for a response on this topic from our geneticist and @May Terry @Alexander Mankovich
Patrick Werner (Apr 02 2020 at 18:00):
I don't want to imply that if our staff agrees on something this is the truth or the way we should handle it, just contributing opinions and searching for the best solution.
Bret H (Apr 04 2020 at 17:34):
@Patrick Werner I think you've done what you can, and great job of it. I agree with your use of the components that exist. It is not uncommon for synonyms to develop in a space. E.g. E.coli 'positive' and E.coli. 'detected' do not have a meaningful difference for a physician but a lab technologist sees them as different. There is no functional decision making difference between reporting oncogenic and tumor driver. IN fact, ask an oncologist how they would use a report of Pathogenic variant and associated phenotype as tumor type VS Tumor driver variant. If we were a solely data driven organization, we would carryout a survey to see if there is really a difference. AND we would absolutely note the 1) role of the respondent, 2) level of training/experience with molecular lab work. But we have to work with the information that is brought to us or that we can find.
Alexander Mankovich (Apr 06 2020 at 15:01):
I think we should be looking to other efforts for insight here. For example, ACMG guidelines are a bit more nuanced for inherited disorders (see https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4544753/), with five possible categories: pathogenic, likely pathogenic, uncertain significance, likely benign, benign. For somatic variants I've never run into the need to report variants as drivers or passengers per se - clinical significance is king and AMP guidelines are what I tend to see here.
Kevin Power (Apr 07 2020 at 14:30):
I have been following some of the work around Variant Annotation work group in GA4GH, and I thought this ongoing thread might be good for this group to review for ideas: https://github.com/ga4gh/va-spec/issues/22
Patrick Werner (Apr 08 2020 at 07:50):
Alexander Mankovich said:
I think we should be looking to other efforts for insight here. For example, ACMG guidelines are a bit more nuanced for inherited disorders (see https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4544753/), with five possible categories: pathogenic, likely pathogenic, uncertain significance, likely benign, benign. For somatic variants I've never run into the need to report variants as drivers or passengers per se - clinical significance is king and AMP guidelines are what I tend to see here.
That was my approach. Driver = pathogenic, passenger = benign. In our use-case i would add another explicit driver passenger coding.
Patrick Werner (Apr 08 2020 at 07:50):
(deleted)
Last updated: Apr 12 2022 at 19:14 UTC
