Even if a regulatory feature is only active in one cell type, it's still relevant. Maybe it regulates a gene that's only active in one cell type, or a gene that is active in different pathways in different cell types. If you disrupt TF binding in that enhancer, you can still alter enhancer activity, still alter gene expression and still have a phenotypic effect.

Sid, This is the common problem of genetic variation based association study. Usually, we use PBMC as the DNA source to build the association between SNP, CNV with some specific phenotypes for which the disease source might be liver, lung or brain rather than PBMC itself. But we should clear that the genetic variance would bring light influence to our biological system in each tissue with long time (from your born), therefore, if we found some positive and significant association between SNPs and phenotype, we need do our best to interpret the mechanism behind the association.