Clinical or neuroimaging profiles in the assessment of genetic variants associated with neurodegenerative diseases

Genome‐wide association studies (GWAS) have led to the discovery of many genetic variants associated with diseases, with more than 50 000 associations described over the past 10 years.1 However, these variants explain only part of the risk of presenting these conditions, since neurodegenerative diseases may be related to a large number of variants whose effects are too subtle to be detected; therefore, they can only be observed in studies with large sample sizes or in meta‐analyses. Furthermore, many of the variants identified by GWAS occur frequently in the general population,2 giving way to doubts regarding their relationship with the disease in question: in principle, if a variant is frequent, it is unlikely to have a differentiated functional impact causing the disease, since a large portion of the population does not present the pathology. In recent years, many articles have discussed potential biases affecting GWAS findings, especially methodological biases resulting from the advancement of detection techniques or the statistical analysis of the data obtained,3 as well as the influence of environmental factors.4

GWAS are case‐control studies, and may therefore be subject to biases in terms of their design and performance, as well as to confounding factors hindering attempts to establish causation; many such factors are inherent to the populations analyzed.5 However, to a lesser extent, the literature has also reported that if a variant is related to the disease, this may in some way influence its clinical manifestation, leading to a different progression or specific profile, as we would expect if the variant is related to a functional effect. Demonstrating that a variant causes a functional alteration is one of the criteria for establishing its pathogenicity.6

In this issue of CNS Neuroscience & Therapeutics, Shen et al7 report a study aiming to observe the functional effect of a variant identified in GWAS in patients with Parkinson's disease (PD). The study belongs to the field of neuroimaging genetics, which has grown rapidly in recent years and whose application to neurodegenerative diseases may be a key to improving our knowledge of the function of certain genetic variants. The rs4698412 variant consists of a G‐to‐A substitution in the gene encoding the bone marrow stromal cell antigen 1 (BST1) protein, located on chromosome 4. BST1 is a glycosyl‐phosphatidylinositol‐anchored glycoprotein of bone marrow stromal cell lines that facilitates pre‐B‐cell growth.8 This variant is frequent in the general population, with an allele frequency of 40%. Furthermore, this is an intronic variant which is unlikely to cause functional effects, since it should not produce an anomalous transcript, and is not considered a variant of uncertain significance according to genomic criteria.

Several GWAS studies have shown an association between BTS1 and PD, although other studies have found no clear association (Table 1). Likewise, this is not the only variant of the BST1 gene observed in GWAS studies (Table 2); however, given the noncoding character of these variants and the hypothesis that they act on a regulatory network, a shared mechanism may be involved.9 Shen et al7 analyze the pathogenicity of this variant, attempting to identify a specific clinical and functional neuroimaging profile in patients, according to the allelic variant of the gene. Other authors have previously attempted to establish a relationship between the disease and presence of the variant using the Unified Parkinson's Disease Rating Scale and the Hoehn‐Yahr Stage, but rs4698412 was excluded from the analysis for methodological reasons.10

Table 1.

GWAS studies between variants of BTSI gene and Parkinson disease

Author, year	Country	Genetic variant (rs)	Number of patients with PD/controls	Association observed with PD
Satake et al, 2009	Japan	11931532	1078/2628	A statistical association is observed
Tan et al, 2010	Japan	11931532 12645693 4698412 4538475	433/916	None observed
Saad et al, 2010	France	4698412	1039/1984	A statistical association is observed
Simon‐Sanchez et al, 2011	The Netherlands	11931532 12502586 12645693 4698412 4538475 12646913	772/2024	No association is observed; only a weak association with the rs12502586 variant is found
Chang et al, 2011	China	1191532	636/510	None observed
Spencer et al, 2011	UK	4698412	1705/5175	A weak statistical association is observed
Liu et al, 2011	Ashkenazi Jews from New York, US	11931532, 12645693, 4698412 12502586	268/178 and replication study on another data set (2050/1836)	No association is observed; only a weak association with the rs12502586 variant is found
Zhu et al, 2012	China	4538475	215/212	None observed
Sharma et al, 2012	19 countries	11724635	8750/8955	A statistical association is observed
Hernandez et al, 2012	Finland	11724635	387/496 (early‐onset PD)	A weak statistical association is observed
Miyake et al, 2012	Japan	11931532 12645693, 11724635	229/357	None observed
Liu et al, 2013	China	11724635	1737 including patients with PD and controls	None observed
Wang et al, 2013	China	Whole exome	524/527	None observed
Soto‐Ortolaza et al, 2013	Poland	11724635	345/234	An association only observed depending on the statistical model (recessive)
Soto‐Ortolaza et al, 2013	USA	11724635	674/724	An association only observed depending on the statistical model (additive and recessive)
Soto‐Ortolaza et al, 2013	Ireland	11724635	362/370	An association is only observed depending on the statistical model (additive and dominant)
Chen et al, 2014	China	11724635	468/487	None observed
Guo et al, 2015	China	4698412	1061/1066	A statistical association is observed
Chang et al, 2015	China	11724635	596/597	None observed

Table 2.

Variants of BTSI gene observed in GWAS studies

Variant (rs)	Location	Type	Nucleotide modification	Frequency	Clinical significance
4538475	chr4:15736314	Single nucleotide variation, intronic	A > G	23%‐30%	Not established
4698412	chr4:15735725	Single nucleotide variation, intronic	G > A/G > T	40%‐43%	Not established
11724635	chr4:15735478	Single nucleotide variation, intronic	C > A/C > G	40%‐43%	Not established
11931532	chr4:15724143	Single nucleotide variation, intronic	T > C	15%‐23%	Not established
12502586	chr4:15724941	Single nucleotide variation, intronic	G > A	4%‐6%	Not established
12645693	chr4:15727911	Single nucleotide variation, intronic	G > A	9%‐18%	Not established

The study by Shen et al7 is one of few studies that attempt to validate the detected variants through profiles and the first to include rs4698412. The authors observe a clinical profile associated with gait deficits and a functional magnetic resonance imaging profile that suggests that BST1 variants may affect stability in patients with PD. Regardless of whether these data should be reproduced, the findings of Shen et al7 support the idea that the variants identified in genetic association studies should be systematically validated by studying clinical profiles and, in the case of such neurodegenerative diseases as PD, neuroimaging profiles; data from these profiles should therefore be considered when analyzing the pathogenicity of these variants. Therefore, observing a higher frequency among patients of a variant whose incidence is high in the general population must be accompanied by other criteria if we are to consider that it has a role in the development of a disease, whether in its incidence or its progression. Therefore, observing a deleterious effect on the gene or gene product6 and differences between patients who do and who do not present the variant, may be a factor associated with its pathogenicity.