Figure 3. The ELL2 haplotype that predisposes for MM is identical to the ELL2 haplotype that influences IgA and IgG levels in healthy subjects.

In addition to the sentinel SNP rs56219066, the association between 5q15 identified in the MM and MM+MGUS meta-analyses was captured by numerous markers in strong linkage disequilibrium located in a ∼40-kb haplotype block in ELL2. To verify that the ELL2 haplotype associating with MM and MM+MGUS is identical to the ELL2 haplotype associating with Ig levels, we tested for association between each available SNP in the ELL2 region and IgA, IgG and IgM levels using the Icelandic Ig data set. We overlaid the log-linear regression P values for association with Ig levels with the meta-analysis P values obtained for the same SNPs for association with MM and MM+MGUS: (a) log-linear regression P values for association with IgA (red; left), IgG (red; middle) and IgM (red; right) overlaid on logistic regression P values for association with MM (blue); (b) corresponding results for MM+MGUS. The x axes indicate chromosomal positions. The y axes indicate −log₁₀ P values. Sizes of markers reflect degree of association with MM or MM+MGUS. As shown, all SNPs in the ∼40-kb haplotype block associating with MM or MM+MGUS associate with IgA and, to a lesser extent, with IgG. We did not observe any association with IgM. Taken together, ELL2 SNPs associating with MM and MM+MGUS associate with IgA and IgG and vice versa, further supporting that the ELL2 haplotype that predisposes for MM also influences Ig levels.