Large Field Virtual Microscopy as a Tool for Quality Control and Blast Definitions in MDS and Leukemia. From the International Working Group on MDS.

All current classifications for myelodysplastic syndromes (MDS) require the identification of the blast population. From the FAB (1976) to the WHO (1999) new criteria such as the results of cytogenetic analysis have been introduced but the percentage of blasts remains a major factor in diagnosis, subclassification and prognosis. The definition of a blast cell is still based on those proposed in 1976 by the FAB group, but it is unclear that it has been applied in the same manner worldwide. Since the WHO has changed the definition for AML (minimum criterion is 20% blasts) and since RAEB has been divided into two groups (5 to 9 and 10 to 19% blasts respectively) it has become clear that the definition of blasts of granulocyte lineage should be more precise.

Definition: Experts in morphology from the IWG on MDS have proposed that all cells from undifferentiated blast (without granules) to but not including the promyelocyte would qualify as “blasts”. A promyelocyte starts with the appearance of a “clearly visible Golgi zone” independently of the number of granules. Mature granulocyte starts with disappearance of cytoplasmic basophilia and more mature nucleus.

Methodology: To verify agreement between experts on this new definition, a unique digital picture was produced from the bone marrow smear of an AML patient with FAB M2 by the association of 150 consecutive native pictures (600x800 pixels). The mosaic picture is 8340x2386 pixels (about 30 M in jpeg format or 16 M in pdf). All the 265 nucleated cells have been numbered and a drop down menu for choices was included. The picture was placed on the server of the MDS Foundation (http://www.mds-foundation.org/goasguenfollowup) and after evaluation the results were automatically sent to the MDS foundation center and registered. The five experts evaluated exactly the same cells (numbered from 1 to 265) and results were submitted for statistical analysis.

Results: for 176 cells (66.6%) there was complete concordance (5/5) and for 60 others (22.7%) concordance was 4/5. If we consider that concordance of at least 4/5 is acceptable, we obtained 89.4% good concordance. Moreover, 23 cells (8.7%) had a concordance of 3/5 while only 5 (1.89%) had a concordance of 2/5. We conclude that these definitions of blasts and promyelocytes are reproducible and may help to standardize the classification of AML and MDS. All experts would have produced the same diagnosis since standard deviations of the percentages of various cell types were very low.

Conclusion: the production of large field digital pictures may be very useful for education and quality control in morphology. In addition, use of such images may help to provide a new approach to difficult cells and may be useful in proposing new criteria for classification.