Introduction: Lipoprotein(a) concentrations are an important cardiovascular risk factor. A copy number variation (“KIV-2 repeat”) and largely unknown sequence variation in the LPA gene together determine Lp(a) variance. Depending on the number of repeats up to 70% of the LPA coding sequence is located in the KIV 2 repeat, which is not accessible to common sequencing and genotyping. 
Methods: We developed an ultra-deep sequencing approach and a bioinformatic variant analysis pipeline specifically tailored to the LPA KIV-2 repeat (https://github.com/genepi/lpa-pipeline). We deep sequenced the KIV-2 in 123 Central-European samples and additionally analyzed data of 2504 samples from 26 populations worldwide. The p.Arg20Ter mutation was followed-up by allele-specific qPCR in three German studies (KORAF3, KORAF4 and GCKD; n=11,390).
Results: We observed 14 different loss-of-function variants and >100 missense variants. Twenty two variants presented clear population specificity, arising to even high frequency in single populations. The “DraIII polymorphism” described earlier was identified as a G>A change at position 2106 of our amplicon. Loss-of-function variants were mostly rare except p.Arg20Ter, which presented moderate frequencies (1.1% 3.4%) in multiple populations. In three Caucasian studies (n_total=11,390) it showed a consistent Lp(a) reduction of 9.7-12.9 mg/dL Lp(a) (all p<10-10, 1.6-2% frequency) and was not captured by common GWAS hits. Interestingly, all p.Arg20Ter carriers possess also the splice site mutation rs41272114 (D’=0.96, R2=0.28).
Conclusion: The KIV-2 region contains a large amount of novel missense variation, partially population-specific patterns. The p.Arg20Ter mutation is moderately frequent in multiple populations and associated with low Lp(a). Perfect disequilibrium with rs41272114 suggests the existence of “double-null” Lp(a) alleles.

03 - A COMPREHENSIVE MAP OF THE VARIABILITY IN THE LIPOPROTEIN(A) KIV 2 REPEAT REGION AND FOLLOW-UP OF THE KIV-2 ARG20TER MUTATION IN 11,000 INDIVIDUALS