Trans-border LD?

Occasionally I see high R^2 values even between linkage disequilibrium blocks. The following plot was obtained from the German asthma family study; basically a haploview screenshot rescaled and inverted before using the Photoshop polar filter.

this image (c)
Clearly at 2′ 40 there is a marker that is associated also with marker on another block – there are even weak epistatic effects(p < 0.05). Does this indicate any disease selection?