Solving the Puzzle of Hans' Algorithm (HA) Methodological Shortcomings: Canadian Harmonization of Hans' Algorithm for...

Hans' algorithm (HA) is the most frequently used surrogate biomarker scheme for subtyping Diffuse large B-cell lymphoma (DLBCL) by the cell-of-origin (COO) into GCB and non-GCB subtypes. The originally published positive and negative predictive value (PPV and NPV) against gene expression profiling (GEP) subtypes were less than perfect and were not fully reproducible in published literature. Furthermore, little is known about how the HA performs in clinical practice. The Canadian Association of Pathologists National Standard Committee for High Complexity Testing (CAP-ACP NSCHCT) initiated a Canada-wide project to assess the current diagnostic accuracy of the HA in clinical practice, harmonize the analytical phase of the IHC assays used for HA, and to optimize its overall diagnostic sensitivity and specificity against GEP subtypes. We divided a DLBCL cohort (n=96), where COO was defined by GEP (Lymph2CX, NanoString technologies) into training (TC, N=45) and validation cohort (VC, N=51) and tissue microarray (TMA)-TC and TMA-VC were constructed. Selected major Canadian laboratories applied their routine CD10, Bcl-6, and MUM1 IHC protocols and sent stained slides to the reference laboratory for review. The IHC protocols from laboratories were designated as "weak" (1/10), "moderate" (4/10), and "strong" (5/10) based on their overall analytical sensitivity. The results of the central review HA readouts were compared with GEP results. Furthermore, in TC, the original HA readout was modified to adjust the cutoff to overall IHC protocol sensitivity. The new readout criteria were also assessed in the VC set. The original HA readout showed good results against GEP for low sensitivity protocol only. For all other laboratories that had IHC protocols with moderate and high analytical sensitivity, a readout was adjusted to higher IHC protocol analytical sensitivity using a cutoff of >30% of >2+ staining intensity. This modification yielded significantly improved diagnostic accuracy against GEP even without any changes to the IHC protocols and was widely applicable to the range of analytical sensitivities of IHC protocols, which are currently in use. IHC biomarkers for HA can be highly accurate and harmonized across different laboratories for clinical application if the following criteria are met: (i) testing laboratories use standardized reference materials to set up and monitor analytical sensitivity, and (ii) the pathologist's readout and cutoff are adjusted to the overall IHC protocol analytical sensitivity.