Teachers have their job cut out recognizing pupils in their class who are doing better or worse than others and might need special attention. Separating out subsets of tumours that might benefit from more accurate diagnosis and tailored treatments is also difficult. Andreas Rosenwald et al. and Kerry Savage et al. have now identified a distinct subgroup of the non-Hodgkin's lymphoma diffuse large B-cell lymphoma (DLBCL), called primary mediastinal B-cell lymphoma (PMBL) using gene-expression profiling.

Diagnosis of PMBL using clinical and morphological features alone is imprecise, leading to an uncertain prognosis. Both research groups have devised a molecular signature to identify PMBLs and validated their predictor gene sets on samples from patients with DLBCL. Rosenwald et al., showed that the patients with molecularly identified PMBL had a better 5-year survival rate (64%) than total DLBCL patients (46%), so confirming previous clinical reports.

Both groups also showed that there are striking similarities between the molecular signature of PMBL and that of Hodgkin's lymphoma (HL) — the similarity in clinical presentation of these B-cell malignancies was already known. Savage et al. used a new gene-expression analysis technique, called gene-set enrichment, to confirm the shared molecular features of PMBL and HL. Rosenwald et al. found that 34% of the PMBL signature genes were also characteristically expressed in HL. Three genes that have previously been shown to be highly expressed in PBMLs — MAL, FIG1 and LFA3 — were present in both the PMBL predictor gene sets and in the HL profile.

So, can the individual genes in the molecular signatures tell us anything new about the aberrant signalling pathways in PMBL and HL? Rosenwald et al. found that the gene that best discriminated PMBL from other DLBCLs, and is also highly expressed in HL, was a gene close to JAK2 on chromosome 9, PDL2 . The PDL2 locus is amplified in about half of PMBL tumours and it encodes a regulator of T-cell activation. Rosenwald et al. suggest that expression of PDL2 might allow a malignant B cell to co-exist with T cells in the thymus, where this lymphoma arises. Both groups noted many other similarities, including low expression of B-cell receptor signalling molecules, and high expression of interleukin-13 signalling molecules (including JAK2 and STAT1), tumour necrosis factor family members and several targets of the NF-κB transcription factor (including TRAF1). Savage et al. went on to show that the NF-κB pathway was activated in almost all cases of PMBL, and suggest that this might be the mechanism by which PMBL and HL cells resist apoptosis. Despite these similarities, the two lymphoma types were clearly distinguishable by the expression of other genes — such as several mature B-cell genes that were only detectable in the PMBL cells.

Molecular diagnosis of PMBL could help guide management of patients with DLBCL and lead to development of targeted treatments.