Page 14 - The prevalence of the Val66Met polymorphism in musicians: Possible evidence for compensatory neuroplasticity from a pilot study

Konuma and Okada Inflammation and Regeneration (2021) 41:18 Page 3 of 5

In the calculation and validation of PRS, some meth- 898) analysis of the association of several clinical bio-
odological concerns have been argued [9, 10]. For ex- markers and revealed the association between high sys-
ample, although the construction of PRS by inclusion of tolic blood pressure PRS and a shorter lifespan in trans-
larger numbers of SNPs (including SNPs that do not ethnic individuals and the association between obesity
meet genome-wide significance) can have more predict- PRS and lifespan in Japanese and European individuals
ive accuracy, it is argued whether the inclusion of those [19]. These results showed the potential application of
SNPs with close to zero effects in PRSs makes sense PRS in improving population health by providing infor-
[11]. In another example, linkage disequilibrium (LD), mation on modifiable risk factors driving health
the correlations between nearby SNPs, which leads to outcomes.
over-representation of high LD regions in calculating From these examples, the utilities of PRS have been
PRS, potentially reduces the predictive performance of expected to be potential predictors of future risks of dis-
PRS [12]. To mitigate the effect of LD, LD pruning (ran- ease or health outcomes. Thus, they are expected for tar-
domly removing one SNP from a pair in high LD), LD get treatment application, alteration of screening
clumping (pruning by LD, while referentially retaining paradigms, and modification of non-genetic factors re-
more significantly associated SNPs), or more complex lated to predicted high-risk phenotypes.
methods that explicitly account for LD [13] have been
used. Limitations and challenges for the application of
polygenic risk score
Applications of the polygenic risk score for We focus on several limitations on the implementation
disease prediction of PRSs in clinical practice. First, PRS is highly sensitive
In the case of disease with age-dependent prevalence, to ethnic background. The variability of PRS among eth-
such as lifestyle-related diseases, it is effective to identify nic groups can be explained by the differences in allele
the population with a high risk of disease onset in ad- frequency, LD, and effect sizes of variants among ethnic
vance and implement a preventive intervention. One of groups [20]. Therefore, the performance of PRS drops if
the PRS utilities with high clinical values can be a pre- PRS developed from one ethnic group is applied to an-
dictive biomarker of disease risk. This utility of PRS has other ethnic group [21]. To overcome these ethnic
been explored in many common diseases, such as can- group-specific biases, several methods have been pro-
cer, coronary artery disease, obesity, and diabetes [14– posed. For example, the ancestry deconvolution PRS
16]. For example, in coronary artery disease, PRS, which method with consideration for an admixture of ancestry-
was developed by a GWAS of coronary artery disease specific partial sequence in individual genome demon-
from a dataset (validation dataset) of UK Biobank partic- strated improved susceptibility predictions of PRS for
ipants and applied to another dataset (test dataset) of four traits (type 2 diabetes, breast cancer, height, and
UK Biobank participants, demonstrated that participants body mass index [BMI]) [22]. In addition to the further
in the top 0.5 percentile of PRS in the test dataset had a development of the PRS method, future GWASs would
fivefold increase in the prevalence of coronary artery dis- be needed to include subjects from diverse ethnic back-
ease [16]. This result showed that PRS developed by a grounds to improve the generalizability and utility of
large-scale GWAS potentially enabled the accurate pre- PRS for all populations because the majority of GWASs
diction of disease prevalence. have been performed in European-Caucasian popula-
In another example, disease risk prediction of breast tions [23, 24]. In order to enlarge the benefit of PRS in
cancer, which had been estimated from two genes, non-European-Caucasian populations, whose amount of
BRCA1 and BRCA2 [17], was expanded by the applica- genomic data is limited compared with European-
tion of PRS. PRS of breast cancer based on 303 genetic Caucasian populations, it is important that vastly in-
variants from a GWAS of breast cancer demonstrated creasing diversity of participants is included and ana-
that women in the top 1 percentile of PRS had a fourfold lyzed in genetic studies, and open data-sharing standards
increased risk of developing estrogen receptor-positive of these results are needed for improving the accuracy of
breast cancer and a sixfold decreased risk for women in PRS in these populations [24].
the lowest 1 percentile of PRS [18]. Although this PRS Second, the distribution of PRS even in the population
had a modest AUC of 0.63, this study showed that breast group was reported to show biases according to geo-
cancer PRS potentially captured sufficient information to graphical differences. For example, geographical differ-
identify a high-risk subgroup of women who could be ences in PRSs of coronary artery disease, rheumatoid
offered preventive interventions. arthritis, schizophrenia, waist–hip ratio, BMI, and height
Application of PRS for a non-disease trait was also re- were detected in Finland [25]. Whether the cause of
ported. This analysis developed PRSs of trans-biobank these biased distributions was the geographical differ-
(BioBank Japan, UK Biobank, and FinnGen; n total = 675, ence in disease prevalence or the difference in the