When Polygenic Scores Fail: A Family Case Study That Reframes Genetic Risk in Multiple Sclerosis
Multiple sclerosis (MS) is a prototypical multifactorial neuroinflammatory disease in which inherited susceptibility and non-genetic exposures jointly shape risk. Large genome-wide association studies (GWAS) have identified more than 100 MS-associated loci, yet most individual non-HLA variants confer only modest effects, motivating aggregation approaches such as weighted genetic risk scores (wGRS) that sum risk alleles while weighting each by its reported odds ratio. Akkad and colleagues interrogate a key translational question: whether a wGRS derived from established MS loci can meaningfully explain—or even predict—familial clustering of MS, where shared genetics and shared environment are both plausible drivers.
A High-Information Pedigree: Affected Siblings Including Monozygotic Triplets
The report centers on a nuclear family without prior MS in earlier generations, yet with four affected offspring: an older brother diagnosed with relapsing–remitting MS at 25 years and three female triplets diagnosed at 21 years. Clinical ascertainment is carefully described, including typical MS features (relapses, MRI lesions) and cerebrospinal fluid oligoclonal bands, with systematic exclusion of important differential diagnoses (e.g., lupus, sarcoidosis, infectious and metabolic mimics). Particularly noteworthy is the presence of a monozygotic triplet set, which creates an unusually informative natural experiment: shared genome, shared early-life milieu, and closely aligned developmental timing—yet with meaningful differences in symptom onset and disease activity.
Disease Course and Treatment Response as a Clue to Shared Pathobiology
The clinical trajectories of the triplets highlight both commonality and heterogeneity. All three began on high-dose, high-frequency interferon beta-1a, but each demonstrated inadequate disease control, requiring escalation to higher-efficacy agents. One triplet developed early symptoms in childhood and experienced substantial inflammatory activity and disability progression despite sequential exposure to daclizumab (trial participation), fingolimod, and natalizumab, with subsequent clinical stabilization and partial functional improvement after alemtuzumab. The other two triplets also required therapy switches (to fingolimod and/or natalizumab) based on ongoing clinical or MRI activity, though they achieved more durable stability. The authors emphasize that shared poor responsiveness to first-line therapy and generally severe phenotypes may indicate common mechanisms influencing not only susceptibility but also disease course—an important distinction when evaluating the utility of genetic risk scores built primarily for susceptibility.
Genetic Strategy: Risk-Locus Genotyping, Risk-Score Construction, and Genome-Wide Structural Screening
Methodologically, the study applies three complementary steps. First, monozygosity of the triplets is confirmed using short tandem repeat profiling. Second, the family is genotyped for 57 established non-MHC MS risk loci plus the HLA-DRB1*1501 tagging SNP (rs3135388), yielding 58 markers for wGRS construction and pedigree concordance checks. The wGRS is computed as a weighted sum across loci using published odds ratios as weights and allele dosage as the multiplier (0, 0.5, or 1 depending on the number of risk alleles). Third, to move beyond common-variant burden, the authors perform SNP-array microarray analysis (CytoScan HD) across all family members to identify copy-number changes and large regions of loss of heterozygosity (LOH), applying explicit calling thresholds for deletions/duplications and for LOH segment length.
Core Finding: The wGRS Does Not Track Affected Status in This Family
The central—and counterintuitive—result is that the highest wGRS occurs in the unaffected father, followed by the affected offspring, with the unaffected mother having the lowest score. In other words, the aggregated common-variant burden across these established loci does not discriminate affected from unaffected within this pedigree, undermining the notion that familial aggregation in this case can be “explained” by simply carrying more known MS risk alleles. Adding to this, all individuals are homozygous for the wild-type allele at the HLA-DRB1*1501 tagging SNP, indicating that a major and frequently implicated HLA-associated risk signal is not captured here. The authors therefore conclude that, at least for this family, wGRS does not function as a reliable predictor of familial MS manifestation and should be interpreted cautiously when applied to individual families or to “MS-like” presentations.
Structural Signals Beyond GWAS Loci: Copy-Number Changes and a Notable LOH Segment
Array-based analyses identify several structural variants consistent with a recessive-model framing (two gains and two losses across specific chromosomal regions) and, most prominently, a sizeable LOH region on chromosome 15q21.1–q21.3 spanning ~5.6 Mb and encompassing 50 genes (31 listed in OMIM). While the reported copy-number changes do not overlap known GWAS MS loci, the LOH segment raises the possibility of unmasking recessive or regulatory effects in a region rich in biologically relevant candidates. The authors specifically flag MYEF2, implicated in suppressing transcription of the myelin basic protein gene, and SLC27A2, involved in long-chain fatty-acid metabolism, as conceptually interesting in an MS context—particularly given growing evidence that lipid handling and dietary fatty-acid composition can modulate neuroinflammatory outcomes. These observations are hypothesis-generating rather than definitive, but they provide a coherent rationale for why familial MS may sometimes reflect mechanisms not captured by common-variant polygenic scoring.
Implications: From Polygenic Scores to Integrated Genomics and Environment-Aware Models
This short communication usefully illustrates a boundary condition for polygenic risk scoring in MS: even when a wGRS correlates with population-level susceptibility, it may fail to explain clustering in a specific family, especially when major risk-tagging alleles are absent and when other genetic architectures (structural variation, recessive mechanisms, rare variants, regulatory effects) or shared environmental exposures predominate. The authors argue for deeper follow-up using next-generation sequencing and multi-omics (epigenomics, transcriptomics, proteomics) to identify pathways that could plausibly converge on demyelination and immune dysregulation in this pedigree. More broadly, the study underscores an emerging clinical-genomics principle: risk prediction for complex neuroinflammatory disease will likely require models that integrate rare and structural variants with context-sensitive environmental and developmental factors, rather than relying solely on the additive burden of known common polymorphisms.
Disclaimer: This blog post is based on the provided research article and is intended for informational purposes only. It is not intended to provide medical advice. Please consult with a healthcare professional for any health concerns.
References:
Akkad, D.A., Lee, DH., Bruch, K. et al. Multiple sclerosis in families: risk factors beyond known genetic polymorphisms. Neurogenetics 17, 131–135 (2016). https://doi.org/10.1007/s10048-016-0474-4
