Thread by @amit_sud

Will polygenic risk scores (PRSs) for #cancer ever be #clinically useful?
Our (@clare__turnbull @HoulstonLab_ICR @ICR_London) commentary in @Nature_NPJ is out: nature.com/articles/s4169…
A thread:
1/n

We would really like to hear your thoughts (including but not exclusively @paulpharoah @ewanbirney @GENES_PK @antonis02 @DJHunter_EPI @EricTopol @nilanjan10c @cecilejanssens @VPrasadMDMPH @DrSidMukherjee @genandgenes @hannah_wand @chris_wigley @CharlesSwanton @d_spiegel)
2/n

Genome-wide associations studies (GWAS) have identified associations between many #genetic variants and cancer #risk:
nature.com/articles/nrc.2…
3/n

There is enthusiasm for PRS testing in cancer, with clinics now offering this and @GOVUK integrating PRS in #healthcare #policy: gov.uk/government/pub….
4/n[...]

We suggest the risk discrimination offered by PRS testing in cancer is not sufficient to make it a useful #screening #test. It may never be the case. This has been articulated by Nick Wald @Oxford_NDPH. nature.com/articles/s4143…
5/n[...]

For example, in prostate cancer, a detection rate of 16% will result in a false positive rate of 5%.
6/n

For cancers that are uncommon in the population, the absolute increase in cancer risk against baseline is small, even at the extreme upper tail of PRS.
7/n[...]

PRS can be used with non-genetic risk factors in risk models. These non-genetic risk factors may be correlated with the PRS (or components) and thus far, improvements in discrimination are modest: academic.oup.com/jnci/article-a…
8/n[...]

PRS may be integrated with existing cancer screening. Care needs to be taken to: i) protect current participation, ii) consider impact of withdrawing existing screening from people, iii) consider cost and complexity, iv) assess the efficacy of PRS+existing screening.
9/n[...]

Communicating risk in the absence of counselling may cause harm, and evidence for promoting risk-reducing behaviour is currently lacking:
bmj.com/content/352/bm…
10/n

We should strive for #equitable access to #healthcare resources. The majority of cancer PRSs are derived from European populations. The predictive performance of these PRSs in non-European populations is poorer: nature.com/articles/s4146…
11/n

Implementing programmes without care may compromise the application of PRSs for niches, and indeed, of #genomic #medicine as a whole.
12/12