Cell of origin in practice: Hans against expression profiling, and what it changes

The split was found by microarray in 2000: some diffuse large B-cell lymphomas express the genes of a germinal-centre B cell and some express the genes of a B cell that has been activated, and the germinal-centre group lived significantly longer (Alizadeh 2000). It was confirmed on 240 patients, where BCL2 translocation and c-REL amplification were found only in the germinal-centre group (Rosenwald 2002).

What a report usually shows is not that test. Most laboratories run the Hans algorithm: three immunohistochemistry stains read in a fixed order, CD10, then BCL6, then MUM1. On 152 cases with a microarray comparison, Hans called 42% germinal-centre and 58% non-germinal-centre, with five-year overall survival of 76% against 34% (Hans 2004). The reference method now exists as a NanoString assay usable on paraffin, but it is not run everywhere, and the two methods disagree on a meaningful minority of cases. That is why a careful report says germinal-centre or non-germinal-centre rather than germinal-centre or activated: the immunohistochemistry cannot tell the activated type from the unclassified type.

What it changes today is the honest part. Cell of origin is prognostic and it decides eligibility for trials, and every randomised attempt to act on it in first line, by adding ibrutinib, bortezomib or lenalidomide to R-CHOP for the activated group, has so far failed to change survival in the whole population. Nobody is denied R-CHOP because of it, and nobody is given a different regimen because of it outside a trial. A patient told their lymphoma is the activated type should be told what that does and does not mean.