PF19: Genetics Question 4

For complex pancreatic disease VARIANT PATHOGENICITY EVALUATION, what types of evidence shall be included?  How shall the levels of evidence be defined in terms of gene dysfunction, clinical context, biomarkers, clinical disease features, and other evidences?

Chairs: Mark Lowe MD PhD, Michael Wilschanski MD, Agnieszka Rygiel PhD

Delegates: Jia-Min Chen PhD, Rup Talukard MD, Pramod Garg MD, Katarzyna Wertheim-Tysarowska, Ph.D, Vinciane Rebours, MD PhD, Andrea Párniczky MD (email request to Dr. Whitcomb or Dr. Sahin-Toth)

Considerations: How should investigators and clinicians interpret MAF, computer predictions, molecular modeling, clinical evidence, animal models, biomarker evidence, etc. Define “pathogenic” or “mutation of variable clinical significance” more accurately. How should genetic (clinical) evidence= association, effect size, p value, replications be quantify and qualified? Functional evidence (define and quantify strength of evidence different levels) i.e. predictions, modeling, biochemical, cell biological (test tube, cell culture), animal models. Should pathogenicity be quantified with a score based on the genetic and functional evidence. Should labels such as modifier or variable significance be used, or just labeling variants as pathogenic, neutral or protective and add quantitative descriptors (scores).

12 Replies to “PF19: Genetics Question 4”

  1. Consider this paper on CFTR variants by Garry Cutting’s group:

    Raraigh KS, Han ST, Davis E, Evans TA, Pellicore MJ, McCague AF, et al. Functional Assays Are Essential for Interpretation of Missense Variants Associated with Variable Expressivity. Am J Hum Genet. 2018. PMID:29805046

  2. These are my questions and thought about the subject (partly based on PMID:

    1.Should, the term “pathogenic” or “ likely pathogenic mutation” only apply to rare, well established variants associated with hereditary pancreatitis such as variants of PRSS1 and CPA1?

    2.The other variants , more common, when in heterozygous state (SPINK1 , CTRC, CFTR) should be considered as risk variants . These concern variants which were shown to be associated with CP in case-control studies and/or functional experiments. These can be divided into: “established risk alleles, likely risk alleles , uncertain risks alleles based on OR from case-control studies/genome wide association studies and functional data. The question is which cutoff of OR shell we take ?

    2. Further, complex genotypes (compound heterozygotes, diegetic interactions) could be divided into causative and contributing genotypes .

  3. Variant pathogenicity evaluation is indeed challenging, so I’m very happy to contribute to this important part of guidelines!

    First of all I think that one of the most important aims of this guidelines is unification of results interpretation, which is highly subjective now. This applies both to mutation evaluation and to genotype interpretation.


    A) Known genes

    In terms of mutation analysis we currently have few genes. This is “a basics” for genetic analyses, as these are well known causative-contributory genes. However, clinical significance of particular mutations may vary and therefore, current status of each mutation in those genes should be clearly and precisely stated in recommendations. The table for the known variants description should contain:

    · gene name,

    · refSeq,

    · list of the mutations for the certain genes or general description f.e null mutations etc,

    · OR from the case-control studies ( if available) for various populations,

    · inheritance pattern,

    · molecular/functional consequences,

    · evaluation of the risk according to pancreas genetics databse

    In this table, we would need to gather the info about CFTR mutations in the context of CP and of course, this information would require reevaluation after some time

    B) Novel genes
    There are several novel candidate genes, so herein I would also suggest similar Table summarizing current state of art.

    The question is when the genes/variants from Table 2 could be transferred to Table 1. eg. is it sufficient to have original study and one replicative cohort? This requires discussion.

    Alternatively, would it be possible to extent the content of database, so that the guidelines could refer to it directly?

    Furthermore, if VARIANT PATHOGENICITY EVALUATION is considered, it is essential to include several evidences of course (eg. MAF, computer predictions, molecular modeling, clinical evidence, animal models, biomarker evidence, etc.). For the time being, I think that ACMG classification could be useful to some extent, especially in case of known CP-causative genes. Despite some weak points, it is widely used and recognizable and pathoigenic/likely pathogenic mutations etc. are well defined.

    In my opinion it would be rather difficult to invent another system, especially that in most cases no functional evidences are available.

    Moreover, we need to classify the complex genotypes into causative, contributory and benign. Following Agnieszka suggestion, should we use the classification proposed by Masson et al in PlosOne 2013.(PMID:23951356)?
    We use this classification in diagnostics and find it rather useful and clear.

  4. Polish genetic group (Agnieszka Rygiel and Katarzyna Wertheim –Tysarowska ) have final comment to the VARIANT PATHOGENICITY EVALUATION.
    Considerations: How should investigators and clinicians interpret MAF, computer predictions, molecular modeling, clinical evidence, animal models, biomarker evidence, etc.

    Comment: Multiple body of evidence are indeed to draw conclusions about pathogenic status of certain variant. These should be at least party based on ACMG classification and should include:
    1. Clinical databases (HGMD, ClinVar, Pancreasgenetics database)
    2. Functional data about certain variants
    3. Association with CP reported in the ethnically matched case-control studies with significant OR ( if available) .
    4. Segregation of the variant with CP phenotype in the family
    In our opinion, the variants in major CP -genes which are rare or novel should be reported as VUS. We usually, in such cases, write on the report that further analysis ( f.e functional analysis) of the variant is needed to draw further conclusions.

    Considerations Define “pathogenic” or “mutation of variable clinical significance” more accurately. How should genetic (clinical) evidence= association, effect size, p value, replications be quantify and qualified?

    Comment: In our opinion, we should not score variants as high or low risk but we should score genotypes as causative or contributory.

  5. Good comments. Why not settle on risk as the term to use and define the extent of risk associated with variants? This would eliminate the need for the pathogenic and similar terms. I don’t find the causative versus contributory terminology helpful.
    We need to state that a risk variant should be primarily defined on the basis of genetic data, if available. Definition based on functional studies may be done if genetic data is inconclusive. When genetics and functional studies conflict, genetics should be considered first.

  6. We generally agree here with Miklos. However, one question is what do you exactly mean by genetic data? Are they case-control studies only ? Are the co-segregation data with phenotype in family considered to be a part of genetic data and when they can have a value ? We need to specify this.
    Coming back to the causative versus contributory terminology. We though these can be helpful for clinician to interpreted the results f.e in case of complex genotypes. Instead, however, we might suggest that the report should contain a statement that molecular results confirm or not confirm (but also not exclude if all genes were not examines) the genetic basis of CP.

  7. from our group (Phil Greer): There is no single perfect way to assess pathogenicity of a SNP in a complex disease. Databases such as SIFT, ClinVar, and Polyphen are insufficient. Some combination of Database, eqtl, epistaxis, and other info from functional studies will be required to assess pathogenicity.

  8. Expression quantitative trait locus (eQTL) analysis from GTex dataset should be done for the risk allele to look for the tissue expression of the implicated gene. This will help in understanding if the pathobiology of the risk allele is related to the pancreas directly or there is likely to be a remote indirect effect of the risk allele on the pancreatic biology.

Leave a Reply