Introduction:
Familial hypercholesterolemia (FH) is a genetic disorder influenced by at least three key genes related to low-density lipoproteins (LDL): LDLR, APOB, and PCSK9. Furthermore, genome-wide associations studies suggest a common variant, polygenic contribution to the disease and associated biomarkers, such as LDL-C concentrations in the blood. However, the genetic architecture of FH is understudied in non-European populations leading to inequities in the diagnostic yield in such patients.

Methods:
Leveraging the Mexican FH registry (www.fhmexico.org.mx), established in 2017, we analyzed 302 samples from 169 families using whole exome sequencing (WES; Twist Exome 2.0) combined with probes capturing 1.6 million genome-wide tagging SNPs (Twist diversity panel) with 2x150bp libraries on the Illumina NextSeq 2000 platform. For a subset of patients, we also performed 2x150bp whole genome sequencing (WGS) with Illumina, and long-read whole genome sequencing (lr-WGS) using the Oxford Nanopore PromethION platform. This enabled us to capture both common and rare genetic variants associated with FH and to perform validation of observed CNVs. We calculated a polygenic risk score (PRS) for LDL-C (PGS003035) and calibrated them using an ancestry-aware method that samples individuals with a similar genetic admixture distribution from our database of patients and reference genomes. A cohort of individuals with comparable ancestry and genetic composition was used as a population control for the study.

Results:
We identified pathogenic or likely pathogenic (P/LP) mutations in 125 FH patients from 40 families, with the majority attributed to LDLR. Notably, ~ 5% of the diagnoses were driven by copy number variants, including deletions in APOB and LDLR and a whole gene duplication of PCSK9. Considering an initial clinical diagnosis of FH, our molecular diagnostic yield is 67.8% for cases with more than one family member affected, and 34.2% for sporadic cases and suggest that such an approach has significant clinical utility.  Moreover, expanding the range of analysis to other genes in the LDL-cholesterol pathway, we identified one family with a multi-exon deletion of APOE. We also observed patients harboring P/LP variants in non-canonical FH genes including APOE, CREB3L3, and PLIN1, suggesting a possible phenotypic expansion for these genes in this disease. When compared with a control population from Mexico and stratified by carrier status of rare P/LP variants, we show that the distribution of the ancestry-adjusted PRS for LDL-C is shifted towards higher risk, with the effect being stronger for non-carriers. The proportion of patients above the 95th percentile, conferring a risk similar to being a heterozygous carrier of rare P/LP variants, was 8.1% for carriers, 11.7% for non-carriers, and 5.6% for the control population (as expected). By predicting LDL-year measurements from the risk scores, we show that individuals on the tail of the distribution obtain similar cumulative LDL levels as heterozygous FH patients.

Conclusion:
Overall, our analysis highlights the contributions of common and rare variants to FH in a Mexican cohort, providing a valuable model for understanding the role of genetic variation in disease manifestation. It also underscores the importance of incorporating diverse populations to reduce disparities in clinical diagnostics.