Analysis of canine haematopoietic neoplasms by using Sysmex the XT-2000iV

Background and objectives In dogs the differentiation between reactive and neoplastic proliferation of blood cells could be difficult and slide review by experienced hemathologist or expensive techniques such as flow cytometry are often required. The possibility to quickly identify neoplastic cells in blood and to predict the leukemic lineage could be useful to select the patients that need ancillary diagnostic test. In human medicine, automated methods to detect neoplastic cells and minimal residual disease in acute leukaemia have been developed. Also in veterinary medicine, the ability to identify leukaemic cells could be enhanced by using automated analyser such as Sysmex XT-2000iV (Sysmex, Kobe, Japan), which can analyze blood samples from several animal species. In addition, the analyser allows the operator to gate cell populations and to save these gates to analyse future samples or to re-gate specimens already stored in the database. This instrument would thus be used to preliminarily approach samples from dogs with hematopoietic neoplasms. In a previous study, we defined the gating procedures to design disease-specific gates able to differentiate chronic leukemia, acute leukemia, subleukemic conditions and reactive leukocytosis in dogs. The aim of this study is to collect information about the possible application of this instrumental approach in routine practice. Specifically, we would assess whether: 1) The analysis of data generated by Sysmex XT2000iV from eukemic vs non leukemic cases can differentiate “reactive” vs neoplastic conditions; 2) the analysis of data collected during the follow up can be used as a prognostic factor in luekemic and non leukemic cases. Material and methods Samples and diagnostic approach This study was done on 163 canine blood samples classified in the following groups based on haematology, bone marrow and lmph node cytology and immunophenotyping: - A: leukaemia (N=46), further sub-classified as acute myeloid leukemia (AML, N=4), acute lymphoid leukemia (ALL, N=12), stage V lymphoma (N=14), chronic lymphocytic leukemia (CLL, N=15), systemic mastocytosis (N=1) - Group B: neoplasia without blood involvement (N=41): lymphoma (N=35), mast cell tumour (N=5), hystiocytic sarcoma (N=1) on which cytology and immunophenotyping did not reveal the presence of neoplastic cells in peripheral blood. - Group C: non neoplastic reactive condition (N=33), with increased total WBC number - Group D: Healthy dogs (N=43): dogs without clinical or haematological changes. Analysis of Sysmex data and scattergrams the following parameters provided by the instrument were recorded and analyzed: - Total WBC counts in the WBC-BASO channel. - WBC flags such as the “positive” flag and/or the “grey area” WBC-DIFF differential. - High fluorescence intensity (HFI) events: events included in an “HFI gate” extending from the top of the normal canine profile (SFL level: 150) to the top of the scattergram. - Lysis resistant region (LRR) events: events included within a “LRR gate”, extending from above the few events classified as “basophils” to the upper part of the scattergram. - Scattergram profiles: The shape of the scattergrams was visually assessed by two independent observers, which classified each case, as normal (N), acute leukaemia/V stage lymphoma (AL), chronic lymphocytic leukaemia (CL) or reactive leukocytosis (RL). Data anlysis and interpretation Four different approaches were followed: Approach 1: diagnostic performance of each single parameter (WBC count, HFI or LRR events, presence of WBC flags and shape of the scattergram) Approach 2: definition of a diagnostic algorithm including all the parameters mentioned above. The cut-off values employed for HFI, LRR and WBC corresponded to those characterized by 95% Spec, as determined during the approach 1. Approach 3: generation of “leukemic flags at HFI>2.00, LRR>1.00 or WBC>37 x 103/l). Approach 4: grading system based, for of the single parameters mentioned above, on the severity of changes in respect to arbirtarily determined cut-offs. For each of the approaches described above the number of true positive (TP) true negative (TN), false positive (FP) false negative (FN) results was calculated and used to determine sensitivity (Sens) and specificity (Spec), which in turn were used to design a receiver operating characteristic (ROC) curve. Monitoring the follow up of leukemic cases Seven dogs (4 from group A and 3 from group B) were repeatedly sampled during the follow up an their results were examined with all the approaches mentioned above. Results Diagnostic utility of ysmex parameters All the pathologic groups had WBC counts higher than controls (Tab 1). The percentage of HFI and LRR events was significantly higher in leukemic dogs than in others. WBC flags were frequent both in dogs with reactive conditions and with leukemia. AL or CL profiles were found in 45 leukemic cases (88.2%) and rarely in dogs without leukemia. Group WBC x 103/l HFI events (%) LLR events (%) WBC flags A 114.9 ± 174.0*, †, ‡ (57.1) 5.4 ± 6.7 *,†, ‡(2.9) 1.8 ± 2.0*, †, ‡(1.0) 33*, †, ‡(71.8%) B 11.5 ± 5.7† (11.2) 1.0 ± 1.8 (0.3) 0.4 ± 0.5 (0.2) 4 (9.8%) C 29.8 ± 34.7‡ (19.5) 0.5 ± 0.8 (0.4) 0.4 ± 0.7 (0.2) 11‡ (33.3%) D 9.2 ± 3.2 (9.1) 0.2 ± 0.2 (0.2) 0.2 ± 0.1 (0.2) 1 (2.3%) Table 2: Mean ± SD (between parenthesis: median) recorded in each group * P<0.01 vs group B; † P<0.01 vs group C; ‡ P<0.01 vs. group D The ROC curves of WBC counts (AUC = 0.84), HFI (0.75) and LRR events (0.89) were statistically significant (P<0.001), indicating that all they can differentiate leukemic from non leukemic dogs. Nevertheless, at cut-off values chacterized by equally important Sens and Spec, both FN and FP were present (Tab 2). The FP rate decreases by increasing the cut-off but this would increase the FN rate, with heavy consequences, since FN dogs will escape the blood film review. Spec and Sens were moderate for WBC flags, but elevated for scattergram profiles, which however had a high interoperator variability. Cutoff Sens (%) Spec (%) FP FN WBC x 103/l 16.91 76.1 76.9 27 11 HFI events (%) 0.4 67.4 75.2 29 15 LRR events (%) 0.4 80.4 81.2 22 9 WBC flags yes 71.7 86.3 16 13 Scattergram profile yes 97.8 94.9 6 1 Algorithm Pos 100.0 77.8 26 0 “leukemic flag” Pos 97.8 94.9 16 0 Scoring system Pos 97.8 94.9 24 0 Table 2: Diagnostic performances recorded in this study All the “combined approaches” were characterized by no FN and by variable numbers of FP. This is an excellent result since it avoids to underdiagnose leulkmic cases. Most of th FP were reactive samples containing reactive lymphocytes or activated neutrophils. In routine practice, the diagonsis of leukemia in these cases would be ruled out by blood film analysis. Other FP dogs had neoplasms without blood involvement: this diagnosis was based on cytology and immuophenotyping but we cannot exclude that some of these dogs actually had circulating atypical cells that would have been detected by approaches (e.g. detection of abnormal DNA content) not frequently used in veterinary medicine. Diagnostic utility of ysmex parameters during the follow up All the 3 non-leukemic dogs remained non-leukemic during the follow up and the Sysmex analysis using “combined approaches” never provided FP results. With the exception of an atypical CLL case, leukemic dogs were correctly classified by the combined approach at first diagnosis: one of these dogs remained leukemic during the follow up and the presence of atypical cells was confirmed by the “combined approaches” mentioned above. The remaining two dogs responded to the treatments with a decrease of atypical cells in blood and Sysmex analysis was able to confirm this trend to normalization in term of decreased percentage of HFI or LRR events, although the “negativization” using the “combined approaches” occurred later than that recorded by flow cytometry. Again, on a practical point of view this “false positivity” is a minor problem since the presence of atypical cells would have been confirmed by blood film analysis or immunophenotyping. Conclusion Sysmex XT2000iV can be used to preliminarily screen blood from leukemic dogs or to monitor the follow up of dogs that respond to antineoplatic treatments. This can be done by looking at additional parameters such as the percentage of HFI or LRR events or at specific scattergram profiles. Morevoer, using a multiple data to design a “leukemic flag” is possible to identify the dogs that are potentially leukemic to be addresses for further examination, without the risk that leukemic dogs would escape a more detailed diagnostic approach.