Here, we demonstrate empirically different strength of association between rare variants in ABCA1 with HDL-C using a random population-based sample in contrast to a phenotypic extreme-selected sample.