Implementation of the Ogata flow cytometric scoring system in routine diagnostics of myelodysplastic syndrome

Abstract Background and Aims Compiling evidence has emerged for the relevance of flow cytometric assessment as a valuable part of the diagnostic work‐up of myelodysplastic syndrome (MDS). This study aimed at evaluating the implementation of a simple flow cytometric scoring system (FCSS), the Ogata score, in a routine diagnostic laboratory. Methods A total of 35 patient samples with a clinical suspicion of MDS were retrospectively assessed using the FCSS. The accuracy of the FCSS was evaluated on the basis of the final diagnoses of the patients. Results The final diagnoses included 17 MDS, 4 other myeloid cancers, and 14 reactive changes. Thirty‐two of 35 (91%) were correctly scored by the FCSS. All 3 incorrect scores were from samples classified as “other myeloid cancers.” Of the initial pathological evaluation of the bone marrows, 20% were inconclusive or incorrect. All inconclusive samples were correctly scored using the FCSS. Conclusion The FCSS evaluated here has high accuracy and low complexity. Cases with inconclusive pathological evaluation will especially potentially benefit from adding the Ogata score to the diagnostic work‐up. The system will be feasible to implement in most flow cytometry laboratories without the need for supplemental antibody panels. It should be emphasized that the FCSS, in our hands, provided poor discrimination between MDS and other myeloid clonal diseases.


| INTRODUCTION
Myelodysplastic syndromes (MDSs) cover a wide range of abnormalities, ranging from overt findings such as prominently skewed morphology, ring sideroblasts, and blast excess with or without cytogenetic aberrancies to subtle changes in unilinear MDS patients or patients presenting in very early stages of the disease. 1 In patients with subtle changes, the diagnosis can be very challenging. In recent years, compiling evidence has emerged for the relevance of flow cytometric assessment as a valuable part of the diagnostic work-up of MDS.  [2][3][4][5]. Pattern recognition provides high specificity and sensitivity but has multiple inherent disadvantages. 6 Often, it requires highly experienced staff and an extensive panel of antibodies. In addition, if an automated analysis is not applied, it suffers from high interinterpreter variation. The high number of parameters included in the evaluation prolongs the analysis and thus lowers the cost-benefits for a routine diagnostic setting. Some of the methods entail setting up local reference profiles, which also makes it less straightforward to implement (eg, De et al 7 ). Other approaches focus on unambiguous criteria, such as expression/no expression of aberrant markers or ratios between cell types. [8][9][10] This greatly facilitates the implementation in routine diagnostics but at the expense of lower predictive values compared with pattern recognition methods (eg, Satoh et al 8 vs Chung et al 3 ). For this aim, Ogata and his colleagues were pioneers when they suggested 4 cardinal parameters for a flow cytometric score for the diagnosis of MDS with little interexaminer variability. 8 The 4 cardinal parameters are (1) the percentage of CD34+ myeloid progenitor cells, (2) the frequency of B-cell precursors within the CD34+ compartment, (3) CD45 expression on myeloid progenitors relative to that on lymphocytes, and (4) the side-scatter (SSC) properties of neutrophils in comparison with lymphocytes. In 2009, the validation study of the first version of the 4-point scoring system was published. 11 In 2012, a multicenter validation study was published evaluating the same system, although reaching slightly different cut-off values for the respective scores. 12 In this study, we investigated the accuracy and the feasibility of the Ogata score in our routine diagnostic flow cytometric facility.
We suspected the section of samples reaching our lab to be slightly different from those in previous studies. In our setting, the flow cytometry facility is separated from the pathology department, and thus, the choice of panel is based only on the hematologist's sparse requisition notes. This results in a crude pool of samples for which a blast screening is requested. The flow cytometric report will often be completed before the pathologist evaluates the morphology and relates the findings to the patient's clinical history. Thus, the scoring system needs to provide reliable information for all samples submitted to blast screening. We placed special emphasis on identifying necessary changes in the present laboratory practice. The informative strategy was to exclude other causes such as reactive conditions, nutritional deficiencies, or drug side effects, as well as described by others. 13 Patients were selected for diagnostic bone marrow sampling when other tests fail to set a clear diagnosis. Typical indication would be unsolved consistent anemia below 10 g/L, thrombocytopenia below 100 billion/L ,and/or neutropenia below 1.8 billion/L. Eighty-three eligible patient samples were identified.

| Patients and samples
Subsequent exclusion criteria were a) >AML (acute myeloid leukemia) diagnosis (regardless of the actual blast count in the flowcytometric analysis), (b) >2% monoclonal lymphocytes by flow cytometry, and (c) <400 CD34+ events in the collected sample. In total, 35 samples were included in the final assessment.

| Flow cytometric analysis
The heparinized samples were kept at ambient temperature and processed within 24 hours following aspiration. The majority of samples were processed between 12 and 24 hours post aspiration.
All samples had been processed using wash-stain-lyse procedure. A mastermix containing the following antibodies had been used: 7-μL A total score of 2 to 4 points indicates MDS. See Figure 1 for details about gating and score calculation.

| Pathology assessment
Examination of the bone marrow included an evaluation of bone marrow aspirates and trephine biopsies. Besides this, a peripheral blood film from the patient was examined in parallel. Blast excess was evaluated from a differential count on peripheral blood films and bone marrow  Table 1     It is worth noticing that all samples which in this study were inconclusive after initial pathology assessment received the correct diagnostic label in the FCSS. In this study, they constituted 20% of the samples.
These patients would have undergone additional examinations and experienced prolonged time to diagnosis. The implementation of a reliable scoring system will be of particular benefit for those patients.
The implementation of this scoring system entails some pitfalls. The first challenge to consider is the fact that the scoring system is designed as an evaluation on the bone marrow. However, bone marrow aspirate is subject to potential hemodilution. The degree of hemodilution is not easily determined, and thus, the sensitivity of the system is expected to vary. Several parameters differ from bone marrow to blood, especially in the myeloid compartment. The Ogata score evaluates the degree of granulation on the neutrophilic granulocytes. It was originally suggested to evaluate this parameter on fully mature cells, 11 but in the final version of the scoring system, this was abandoned because it did not improve the specificity. 12  events. These alternative exclusion criteria allowed 11 of the 29 samples originally dismissed, to be included in the study. The inclusion of samples with possible hemodilution will weaken the sensitivity of the scoring system because these samples will have misleadingly low CD34+ events, and thus, less will obtain a point for >2% myeloblasts.
We suggest a practical threshold of 1000 collected CD34+ cells in a patient sample to imply eligibility for the Ogata score evaluation.
This would allow quantification of a 5% B-cell progenitor population (50 events) with a CV of 14%. 15 An inadequate threshold in the flow cytometric data collection constitutes another closely related pitfall. The threshold is most often based on forward scatter (FSC) properties. An inadequately high threshold will create a biased loss of B-progenitor-related events because these have lower FSC than does the myeloblasts (see Figure 2).

| Limitations of this study
Our exclusion criterion of 400 CD34+ events has, in practice, excluded all patient samples with less than 1.3% CD34+ cells. Some MDS samples would potentially be in this group due to hemodilution, but it is assumed that the exclusion is biased toward more reactive cases. It is expected that the MDS diagnosis is increasingly difficult to determine with decreasing blast count. Hence, the expected specificity and sensitivity of the scoring system, if implemented with the recommended criteria, is likely to be somewhat poorer than indicated by this study. Furthermore, we cannot exclude the possibility of bias in the initial patient cohort. Not all patients suspected of MDS have their bone marrow assessed by flow cytometry, and the analysis is, for the most part, used merely as a screen for blast excess. The choice of flow cytometric analysis is dependent on the individual physician planning the diagnostic assessment. Thus, there is a risk of bias toward high-risk MDS in the study cohort. It is well established that the sensitivity of this and other scoring systems is highest for the high-risk groups. 18

| CONCLUSION
We find that the Ogata score system for flow cytometric assessment of bone marrows suspected for MDS is feasible to implement, even in small laboratories with no hematopathologists employed. Our findings support the previously reported high accuracy in the segregation of MDS and reactive conditions. However, in our cohort, the system could not distinguish between MDS and other myeloid neoplasms.
We suggest a practical threshold for data collection at minimum 1000 CD34+ events. This will limit the effect of hemodilution.

CONFLICTS OF INTEREST
The authors have no competing interests.