Pitfalls in the immunophenotyping of leukaemia and leukaemic lymphomas: survey of 9 years of quality control in The Netherlands. Dutch Cooperative Study Group on Immunophenotyping of Haematological Malignancies (SIHON).

During the last decade, biannual quality controls were performed in the Netherlands focusing on the immunophenotyping of leukaemic haematological malignancies. All results on 48 specimens obtained by 18-34 laboratories were analysed. The interlaboratory variability and percentages of discordant results from 30 markers were measured by assessing false positive or negative (cut-off 10%) results in comparison with median results of the group. The quality of the immunophenotypic diagnoses obtained from the interpretation of these markers in relation to clinical data was evaluated by scoring them as 'correct', 'minor fault', 'major fault', 'not based upon the markers used', and 'no diagnosis', CD3, CD8, CD19, CD61 and Sm lambda had the lowest percentage discordancy (sum of total negative and positive discordant values 5-7.5% of assays): CD13, CD15, cyCD22, CD33 and TdT scored worst with 14-20% cumulative discordancy. The analysis of each diagnosis yielded 78% acceptable immunophenotypic conclusions (correct 54% and minor fault 24%). It appeared that the major faults in immunophenotyping were caused by suboptimal antibody selection and erroneous interpretation of the results obtained, rather than by technical errors. Large differences per diagnostic category were observed, with the best scores for mature B-cell leukaemias, AMLs and common-ALL, and the poorest scores for T-cell malignancies which were correctly diagnosed in only 24-60% of specimens. Mature T-NHL and T-PLL were mistakenly diagnosed as T-ALL by 40% of the centres. Misinterpretation of TdT immunofluorescence or omitting this marker contributed significantly to these wrong diagnoses. A median of 4% of immunophenotypic diagnoses were not based on a correct panel of antibodies, but upon the morphology of the accompanying blood smear, and was often flawed by overinterpretation. In conclusion, both the technical performance of immunophenotyping of haematological malignancies in The Netherlands and the procedure by which a final diagnosis is obtained needs improvement, especially for T-cell malignancies.