Identification of stage V in canine lymphoma: comparison of three methods.

Lymphoma is the most frequent hemopoietic neoplasm of dogs. B‐lineage high grade lymphomas represent the majority of cases (1,2) and DLBCL is the most frequent form (3). Clinical staging is a key point in the diagnostic process determining the degree of extension of the tumor and playing an important role for prognosis and therapy (4). The evaluation of bone marrow and peripheral blood infiltration is mandatory for a complete staging allowing the identification of stage V disease according to the WHO system (5). Aim of the study was to compare three different methods for detecting peripheral blood infiltration in patients with DLBCL 100 cases of DLBCL were included. Peripheral blood was submitted to the following analyses: CBC with ADVIA 120 (Siemens), MGG‐stained smear evaluation, flow cytometric immunophenotyping. Percentages of LUC and BASO (ADVIA 120), blasts out of a 200 cell differential count (MGG‐stained smears), CD21+ large cells (flow cytometric immunophenotyping) were recorded. Differential counting on MGG‐smears were performed both by an experienced and an unskilled haematologist; atypical lymphocytes and quantity of blasts at the thin edge of the smear were also considered, respectively. For each method specific cut‐off values to define stage V were calculated on 15 healthy dogs. Method comparison analyses (Passing Bablock regression and Bland‐Altman plots) were carried out and accuracy in defining stage V disease was calculated for each method. Flow cytometry was used as reference method in both cases. All the methods except blast count by unskilled hematologist reported significant constant and/or proportional error in quantifying neoplastic circulating cells. All methods presented very wide limits of agreement. All methods reported a low concordance with flow cytometry in detecting stage V. LUC and BASO presented 100% specificity but a very low sensitivity. The differential count by the unskilled haematologist reported the best sensitivity. Overall evaluation, considering likelihood ratios and diagnostic odd, indicated the differential count by the skilled operator as the best method; ROC curve reported 1% as the best cut‐off value with sensitivity and specificity of 63‐68% and 93‐96%, respectively. None of the methods considered resulted to be an adequate substitute for flow cytometry in quantifying blood infiltration. Thus, flow cytometry remains the first choice to define stage V in lymphoma patients. LUC and/or BASO percentages from ADVIA 120 are reliable in case of positive result. A blast number >1% in a 200‐cell differential count has moderate sensitivity and high specificity. Anyway, the methods evaluated cannot be used interchangeably for the quantification of blood infiltration