Figure 3.
Correlation analysis and the robust prognostic effect of CD37 expression in patients with DLBCL. (A) A distribution plot showing that CD37− GCB-DLBCL (denoted by the yellow bar) more frequently had TP53 mutations (highlighted in red) or high levels of nuclear p50 (yellow), Myc (green), p-STAT3 (orange), and p65 (lighter red) expression compared with CD37+ GCB-DLBCL (denoted by the blue bar). (B) A distribution plot showing that CD37− ABC-DLBCL more frequently had high nuclear p50 (yellow) and survivin (pink) expression and BCL6 translocation (green), whereas CD37+ ABC-DLBCL more frequently had PI3K (blue) and CXCR4 (purple) overexpression. Each column in panels A and B represents 1 patient; cases without indicated abnormalities detected are shown in light blue (negative) or white (unknown). (C) In GCB-DLBCL, CD37 positivity predicted significantly improved survival, regardless of presence of TP53 mutations, p50high, Mychigh, p-STAT3high, GCET1high, and to a lesser extent, MYC translocations. Conversely, the adverse effect of CD37 negativity was independent of all these biomarkers. Particularly, CD37− patients without TP53 mutations and p50/Myc overexpression remained to have significantly worse survival than patients with CD37+ GCB-DLBCL. (D) In GCB-DLBCL, CD37 positivity predicted significantly better survival even when the patients had high IPI scores. (E) In ABC-DLBCL, the adverse prognostic effect of CD37 negativity was independent of p50, survivin, p63, PI3K, and CXCR4 expression and BCL6 translocations. In particular, CD37− patients without p50 and survivin overexpression remained to have significantly worse survival than patients with CD37+ ABC-DLBCL. (F) In ABC-DLBCL, CD37 and IPI had independent prognostic impact. The cutoffs for high/positive expression as indicated by p50+, Myc+, p-STAT3+, GCET1+, survivin+, p63+, PI3K+, and CXCR4+ in the figures were ≥20%, ≥70%, ≥50%, ≥50%, >25%, >5%, ≥70%, and ≥20%, respectively.