[Cutaneous lymphoproliferations: proposal for the use of diagnostic algorithms based on 2760 cases of cutaneous lymphoproliferations taken from the INCa networks (LYMPHOPATH and GFELC) over a two-year period]

Introduction: Taking as a base our retrospective study of 2760 cases of cutaneous lymphoproliferations from the LYMPHOPATH and GFELC networks, we analyzed the doubtful and discordant cases between non-expert and expert pathologists, and the interest of clinicopathological confrontation.

Material and methods: We defined the main diagnostic difficulties presented by cutaneous lymphoproliferations. We then designed and tested the algorithms on 20 random cases with 20 pathologists, in order to be used by any pathologist (not necessarily specialised in dermatopathology).

Results: The problematic differential diagnoses most frequently encountered are the following: MF or reactive dermatose; lymphoma without any other precision or reactive infiltrate; small B cell lymphoproliferation: lymphoma or reactive infiltrate; phenotyping of large B cell lymphoproliferation. We also analyzed less common problematic differential diagnoses, on the grounds that they are over- or under- diagnosed. Our test had a 72% success rate among the 20 randomly tested cases. The use of several algorithms for the same case is possible.

Discussion: Our study shows that an expert second-opinion is of interest in the area of cutaneous lymphoproliferations. A second opinion is useful for distinguishing a small B cell lymphoma from a HLR, and for defining a final diagnosis when the first pathologist doubts between lymphoma and reactive infiltrate. However, we demonstrate that for the problem MF or reactive dermatose, an initial clinicopathological confrontation produces more results than a second-opinion pathology review.

Conclusion: This is the first study of cutaneous lymphoproliferations that, without excluding reactionary infiltrates, concentrates on doubtful and discordant diagnoses between non expert and expert pathologists, and which has produced tested diagnostic algorithms.