RE: pharmacogenomics evidence/clinical relevance schemes (was Re: Minutes from Clinical Decision Support teleconference)

Thanks for such a comprehensive overview.  This was very helpful.  In
full disclosure, I am a member of CPIC.
 
I agree that it would be better to reference each scheme/scale
independently rather than trying to merge them into a common one.
First, as was mentioned, the scales are still quite arbitrary.  Second,
the field is advancing very rapidly and I would expect significant
revision to the schemes over time.  Specifically, I think those that are
implementing PGx clinically will need to determine what factors are most
informative to the process through experience.
 
In response to your aside, I would like to get a better understanding of
SO-PHARM and how it relates to other ontologies (SO, GO, TMO, etc).
There are plenty of use cases that we could use to evaluate (and update,
if needed) ontology(ies) for PGx, and I think such an effort would be a
timely contribution to the field (and potentially a very nice
manuscript).

Thanks, 
Bob 

 


________________________________

	From: public-semweb-lifesci-request@listhub.w3.org
[mailto:public-semweb-lifesci-request@listhub.w3.org] On Behalf Of
Richard Boyce
	Sent: Saturday, May 05, 2012 9:35 AM
	To: Matthias Samwald
	Cc: public-semweb-lifesci@w3.org
	Subject: pharmacogenomics evidence/clinical relevance schemes
(was Re: Minutes from Clinical Decision Support teleconference)
	
	
	On 04/27/2012 06:40 AM, Matthias Samwald wrote: 

		There exist some evidence classification schemes in the
pharmacogenomics domain (e.g. from PharmGKB, of course). I documented
the PharmGKB schemes at
https://docs.google.com/spreadsheet/pub?hl=en_US&key=0AiGT-vnkGcoLdFFVME
dqcFdYaDFqS0xHTnlUT0N3cEE&hl=en_US&gid=1 (values in the decision table
are auto-completed based on the entities there). I guess for this
project we should standardize to using a single scale, rather than
mixing different scales. 
		 
		TODO: Would you (or someone else) be interested in
looking into these classification schemes to see what would fit best?
Are some of these schemes available as ontologies, e.g. via BioPortal?
However, I guess we should at least retain some compatibility with
PharmGKB, CPIC and other such initiatives!

	I searched for the DPWG evidence/clinical relevance scheme
<http://www.pharmgkb.org/home/dutch_pharmacogenetics_working_group.jsp>
<http://www.pharmgkb.org/home/dutch_pharmacogenetics_working_group.jsp>
in Bioportal and did not find any results. 
	
	The CPIC scheme covers level of evidence and the strength of a
consensus recommendation on dosage adjustment (see below, and
<http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3098762/?tool=pubmed>
<http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3098762/?tool=pubmed> ) .
This scheme appears to be used in all of the guidelines here
<http://www.pharmgkb.org/cpic.jsp> <http://www.pharmgkb.org/cpic.jsp> .
I also do not see anything in Bioportal. 
	
	In terms of "best fit", I would argue that it would be best to
*not* try and map all recommendations to one scheme because the
recommendations are to some extent subjective to the party that assigns
them. For example, CPIC appears to assign a high level of evidence to in
vitro  experiments in its clopidogrel guidelines (see supplemental table
S6 in
http://www.pharmgkb.org/download.action?filename=cpic-cyp2c19-clopidogre
l-supplement.pdf). However, DPWG requires in vivo trials for its highest
ratings. 
	
	It might be best to keep the ratings as assigned by the group
developing the guidelines but just be explicit about the source and
meaning, perhaps by developing "evidence rating scheme", "clinical
significance rating scheme", and "strength of recommendation rating
scheme" types. Each type having a predicate that identifies the expert
consensus source of the rating scheme, the codes for the rating scheme,
which code is considered the lowest rating, and descriptions/definitions
for each rating level. Anyways, more to discuss than can be included in
this email.....
	
	Another issue to always keep in mind is mentioned in the
following quote from
<http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3098762/?tool=pubmed>
<http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3098762/?tool=pubmed> :
	
	"Although some pharmacogenetic cases may have level 3 evidence
[the highest], other considerations may affect the recommendations, such
as the  potential preventable burden of disease or morbidity, potential
harm of intervention, and current practice,10
<http://www.ncbi.nlm.nih.gov/pubmed/19189910>  as exemplified by the use
of warfarin in the presence of CYP2D9 and VKORC1."
	
	As an aside, it might make sense for these evidence and clinical
relevance schemes to eventually be integrated into SO-PHARM
<http://bioportal.bioontology.org/ontologies/1061>
<http://bioportal.bioontology.org/ontologies/1061> .  
	
	hope it helps -Rich
	
	*CPIC pharmacogenomics evidence/clinical relevance scheme:*
	
	"Levels of Evidence Linking Genotype to Phenotype
	Based on previously published criteria (42), a simple scale of
high, moderate, or weak to grade
	the levels of evidence was chosen:
	* High: Evidence includes consistent results from well-designed,
well-conducted studies.
	* Moderate: Evidence is sufficient to determine effects, but the
strength of the evidence is
	limited by the number, quality, or consistency of the individual
studies; generalizability
	to routine practice; or indirect nature of the evidence.
	* Weak: Evidence is insufficient to assess the effects on health
outcomes because of
	limited number or power of studies, important flaws in their
design or conduct, gaps in
	the chain of evidence, or lack of information."
	
	"Overall, the therapeutic recommendations are simplified to
allow rapid interpretation by
	clinicians. They have been adopted from the rating scale for
evidence-based therapeutic
	recommendations on the use of retroviral agents.
	A: Strong recommendation for the statement
	B: Moderate recommendation for the statement
	C: Optional recommendation for the statement"
	
	
	
	-- 
	Richard Boyce, PhD
	Assistant Professor of Biomedical Informatics
	Scholar, Comparative Effectiveness Research Program
	University of Pittsburgh
	rdb20@pitt.edu
	412-648-6768