Diagnosis of haematological malignancies in the era of total laboratory automation: comparison of the Advia 2120 to immunophenotyping and morphology
The incidence of leukaemia in South Africa is 2.5 per 100 000 and has increased due to HIV. Accurate and timeous diagnosis of leukaemia directly impacts success of patient treatment and consequent survival. Usually the Full Blood count (FBC), white blood cell (WBC) differential count and review of the peripheral blood smear alerts the clinician to the possibility of leukaemia. However the number of qualified and skilled technologists in peripheral and central laboratories is on a continual decline making the performance of the critical function of peripheral blood review a challenge. The Advia 2120 haematology analyser performs a WBC and differential count using principles of flow cytometry and the cytograms generated can be used to classify haematological malignancies through the Peroxidase and nuclear density analysis (PANDA) classification system. The presence of myeloperoxidase (MPO) activity in 3% or more leukaemic blasts confirms acute myeloid leukaemia, and enzyme activity can be detected by immunophenotypic analysis or conventional cytochemistry . Research on the comparison of the Advia 2120 and manual morphologic assessment in the classification of leukaemias is limited in the South African setting, where leukaemia often coincides with infection. The aim of this study was to determine if the FBC, differential count and cytogram assessment by the Advia 2120 using the PANDA classification is as reliable as morphologic assessment in the initial classification of haematological malignancies from peripheral blood samples when using flow cytometry as the gold standard.. 150 cases of confirmed leukaemia were collected. The diagnosis obtained from either PANDA analysis and/or morphological assessment was compared to the diagnosis obtained by immunophenotypic analysis. Secondly, the MPO activity obtained by the Advia peroxidase cytogram was compared to the MPO obtained by conventional methods of immunophenotypic analysis and/or cytochemistry. Using the PANDA analysis system, only 48% (72/150) of cases overall were accurately classified. The inaccuracy was 9.3% (14/150) and 42.7% of cases could not be classified. The positive predictive value (PPV) was 88%. The most significant finding was all of the acute Page | iv promyelocytic leukaemia (APL) cases (8/8) had a distinct pattern and were accurately classified on cytogram analysis alone. Accurate sub-classification of other types of acute myeloid leukaemia using PANDA analysis alone was inconsistent. However, the accuracy in classifying leukaemia was improved when the Advia cytogram was used in conjunction with morphological analysis, as 90% (135/150) of cases were accurately classified. The sensitivity and specificity of the peroxidase cytogram in evaluating myeloperoxidase (MPO) activity was 85% and 88.6% respectively. The agreement between cytogram peroxidase activity and the reference methods was 89.1% and the Cohen’s kappa was 76.9%. To the best of our knowledge, there is no data comparing peroxidase activity on the cytogram to other methods. In conclusion, it was shown that the routine use of the Advia cytograms in conjunction with the morphology findings provides invaluable information in the initial screening of leukaemia. In cases with indistinct morphology, the cytograms have the potential to aid in a provisional classification. The peroxidase activity from the cytogram could be used as a surrogate marker for myeloperoxidase activity in leukaemia. Moreover, a tentative diagnosis of an APL is possible by simple analysis of the cytogram resulting in earlier diagnosis which could be life-saving.
A research report submitted to the Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, in partial fulfilment of the requirements for the degree of Master of Medicine in the branch of Haematology. Johannesburg, March 2015