Altered Neutrophil Maturation Patterns that Limit Identification of Myelodysplastic Syndromes

Transcription

1 Cytometry Part B (Clinical Cytometry) 82B: (2012) Original Article Altered Neutrophil Maturation Patterns that Limit Identification of Myelodysplastic Syndromes Sara A. Monaghan, 1 * Urvashi Surti, 1,2 Ketah Doty, 1 and Fiona E. Craig 1 1 Department of Pathology, School of Medicine, University of Pittsburgh, Pittsburgh, Pennsylvania 2 Pittsburgh Cytogenetics Laboratory, School of Medicine, University of Pittsburgh, Pittsburgh, Pennsylvania Background: Altered neutrophil maturation patterns have been reported useful for identification of myelodysplastic syndromes (MDS). Methods: Neutrophil maturation patterns based on CD11b, CD13, and CD16 were visually and numerically evaluated in 19 control, 23 MDS, 37 nondiagnostic for MDS (NDM) specimens, and 19 also processed 1 and 2 days subsequently. Results: In contrast to maturation patterns illustrated previously by others as normal, 84% of controls displayed diminished acquisition of CD16, imparting a contracted appearance. Such divergence from published normal patterns was usually mild-moderate, considered nonspecific, and associated with delayed processing: longer intervals between collection and processing (median 20.5 vs. 5.2 h), and following 1 and 2 days delay. Findings restricted to nonspecific contraction were found in 56% MDS and 78% NDM specimens. Evaluation for aberrant patterns was still performed with mild-moderate contraction present, but concern for over interpretation led to use of an equivocal-aberrant category. Nine cases had aberrant or equivocal-aberrant patterns (seven MDS, two NDM) with distinct visual alterations that differed from nonspecific contraction and had numerical evidence for a left shift: myeloblasts increased (67%) and least mature neutrophils (CD11b-/low, CD16-/low) increased (78%). Although evidence for a left shift was associated with MDS, it was also seen in NDM specimens with a synchronous left shift. Conclusions: Neutrophil maturation patterns that diverge from previously illustrated normal patterns, not specific for MDS, may be common in some settings. Laboratories seeking to implement FC evaluation for MDS must determine which findings have sufficient specificity for MDS within their own practice and patient population. VC 2012 International Clinical Cytometry Society Key terms: flow cytometry; myelodysplastic syndromes; bone marrow How to cite this article: Monaghan SA, Surti U, Doty K, Craig FE. Altered neutrophil maturation patterns that limit identification of myelodysplastic syndromes. Cytometry Part B 2012; 82B: Comprehensive flow cytometry (FC) evaluation of bone marrow specimens for multiple immunophenotypic features reveals abnormalities that help distinguish myelodysplastic syndromes (MDS) from normal hematopoiesis and from other causes of cytopenia(s) (1 6). Indeed, some studies have found that FC can assist with recognizing even difficult-to-diagnose low-grade MDS (7 12). Because of these encouraging findings, consensus statements from an international working conference acknowledge that immunophenotypic alterations detectable by FC may help support a high suspicion for MDS in cases where morphology and cytogenetic studies are not fully diagnostic (13). As one component of a comprehensive FC evaluation for MDS, many previous studies have included Additional Supporting Information may be found in the online version of this article. *Correspondence to: Sara A. Monaghan, M.D., Department of Pathology, The University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd., Dallas, TX , USA. sara.monaghan@utsouthwestern.edu Received 23 November 2011; Revision 1 February 2012; Accepted 7 February 2012 Published online 19 March 2012 in Wiley Online Library (wileyonlinelibrary.com). DOI: /cyto.b VC 2012 International Clinical Cytometry Society

2 218 MONAGHAN ET AL. assessment of neutrophil maturation. As neutrophilic precursors mature from promyelocytes to segmented neutrophils, they undergo characteristic changes in expression of CD11b, CD13, and CD16 (14,15). In contrast, MDS specimens may demonstrate altered expression of these antigens that does not conform to any stage of normal maturation. Initially, Bowen and Davis described decreased proportions of granulocytes expressing CD11b and CD16 in 64% of bone marrows involved by MDS (16). Subsequently, Stetler-Stevenson et al. demonstrated the utility of FC in the diagnosis of MDS and found abnormal patterns of granulocyte maturation in 70 and 78% of MDS patients, based on dotplots of CD11b vs. CD16 and CD13 vs. CD16, respectively, and did not detect abnormal patterns in healthy donors or other patients (17). Several studies have confirmed the utility of evaluating CD11b, CD13, and CD16 among maturing neutrophils for the identification of MDS, but many report a lower frequency of abnormalities in MDS than initially described and some found abnormalities in patients without MDS (9 11,14,18 21). In the largest study to date of comprehensive FC evaluation in MDS, Kern et al. reported abnormal patterns of CD11b vs. CD16 in only 28% and CD13 vs. CD16 in only 42.9% of 511 MDS patients (18). In addition, this study reported abnormalities in patients without MDS: CD11b vs. CD16 in 3.2% and of CD13 vs. CD16 in 9% of 277 patients. Altered patterns have also been reported by others occasionally in association with aplastic anemia, Hodgkin lymphoma, cytopenias secondary to various medical causes, therapy with granulocyte colony-stimulating factor and in marrow regeneration after transplantation (9,10,14,21). We describe our experience specifically with evaluating neutrophil maturation patterns based on plots of CD11b vs. CD16 and CD13 vs. CD16, which had been performed as part of a more comprehensive multiparameter FC assessment for MDS. Our findings are of note for a relatively high frequency of findings even among control specimens that diverged from those previously illustrated as normal by others (14 16). Such findings may pose a risk of being over interpreted as worrisome or abnormal. In addition, we illustrate some alterations in neutrophil maturation pattern that appear restricted to MDS. MATERIALS AND METHODS Specimen Selection and Review The study was performed in accordance with University of Pittsburgh Institutional Review Board guidelines. Bone marrow samples were identified during a 16- month period, which had been performed to evaluate cytopenias, but lacked a definitive pathologic diagnosis other than MDS, and had been submitted for both FC and morphologic assessment. Peripheral blood films, bone marrow aspirate smears, including Prussian blue iron stains, and biopsies were reviewed independently by 2 hematopathologists (SAM, FEC), along with available information from electronic medical records from University of Pittsburgh Medical Center. Dyspoietic forms, as delineated in the 2008 WHO Classification (22), were enumerated among 100 erythroid precursors, 100 maturing neutrophils, and 50 megakaryocytes if available (minimum 20). Forms likely to be more specific for MDS [i.e., erythroid multinucleation (>2 nuclei) or nuclear fragmentation, pseudo Pelger-Huët anomaly, agranular neutrophils and precursors, micromegakaryocytes, mature megakaryocytes with completely round nonlobated nuclei, megakaryocytes with widely separated nuclei] (22 27) were designated as severe dysplasia. Morphologic categories were assigned: no dyspoiesis, indeterminate (non-severe dyspoiesis affecting 10% of any cell line(s) and/or severe dysplasia limited to 5 9% of a cell line) or abnormal (circulating blasts <20% without recent G-CSF therapy, bone marrow blasts 5 19% or blasts <20% with Auer rods, or severe dysplasia affecting 10% of one or more cell lines). Cytogenetic studies included metaphase analysis and interphase fluorescence in situ hybridization (FISH) studies, performed using Abbott Molecular probes (Des Plaines, IL) to detect monosomy5/deletion5q, monosomy7/deletion7q, trisomy 8, and deletion 20q: LSI EGR1/LSI D5S23:D5S721 (5q31/5p15.2), LSI 7S486/CEP 7 (7q31/7 centromere), CEP 8 (8 centromere), and LSI D20S108 (20q12). Control specimens were collected during the same period and included 16 initial staging bone marrow evaluations negative for lymphoma and three specimens from patients that were evaluated for anemia or thrombocytopenia, which was attributed to iron deficiency based on a negative iron stain on a sufficient aspirate smear (Supporting Information Table 1). Specimens were excluded as controls if they had abnormal morphology or chromosomal abnormalities supportive for MDS. On the basis of morphologic, clinical, and cytogenetic findings (without knowledge of the FC findings), specimens were divided into those judged definitively diagnostic for MDS according to the 2008 WHO Classification (22) or Non-Diagnostic for MDS (NDM). Specimens diagnostic for MDS were identified from 21 patients, including 12 with refractory anemia with excess blasts (RAEB), 5 with refractory cytopenia with multilineage dysplasia (RCMD), one with refractory anemia with ring sideroblasts (RARS), one with MDS with isolated del(5q), one with refractory cytopenia with unilineage dysplasia (RCUD), and one with MDS, unclassifiable (Supporting Information Table 2). Two additional specimens also included in the MDS group were from patients who had prior bone marrow evaluations diagnostic for RAEB associated with a normal karyotype, who were receiving azacitidine at the time of this evaluation. Specimens from 37 patients were included in the NDM category (Supporting Information Table 3). Aspirate smears were not available for one specimen. Twenty-three of 36 (64%) NDM specimens lacked dyspoiesis and 10 (28%) had indeterminate morphology based on mild erythroid dyspoiesis (nuclear budding and/or megaloblastoid forms), mild megakaryocytic

3 ALTERED NEUTROPHIL MATURATION PATTERNS FOR IDENTIFICATION OF MDS 219 dyspoiesis (hypolobated forms, excluding completely round nonlobated nuclei), mild granulocyte dyspoiesis (megaloblastoid changes), and/or severe dysplasia in 5 9% of megakaryocytes. Morphology was abnormal in three specimens (8%) based on severe dysplasia in 11, 14, and 21% of the megakaryocytes, but no micromegakaryocytes were observed; these specimens were retained as NDM because this extent of dysplasia, together with the clinical information and normal cytogenetic findings, was judged insufficient for a definitive diagnosis of MDS. Thirty-six NDM specimens had sufficient metaphase cells for conventional cytogenetic analysis. A normal karyotype was identified in 34 specimens. Loss of chromosome Y was seen in 11 of 20 metaphases in one specimen. Another specimen had 11 metaphases with loss of chromosome Y, 2 with del(20)(q11.2;q13.3), and 8 with a normal karyotype. FISH studies for all 4 probes were performed for 36 specimens, but were limited to 5/5q and 7/7q in 1. FISH studies did not identify any additional abnormalities in NDM specimens and did not confirm the del(20q) found by metaphase analysis in one. Flow Cytometric Evaluation Heparinized bone marrow was filtered through a nylon mesh, suspended in phosphate-buffered saline (PBS) containing 1% sodium azide and 2% fetal calf serum, and stained on ice for 15 minutes. Red blood cells were lysed using an ammonium chloride solution. After washing with PBS/1% sodium azide, cells were fixed with formaldehyde (final concentration 1%). Cells were acquired on the day they were stained using a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA) with a total of 30,000 acquired events. As part of the routine clinical flow cytometric testing, the following four-color directly conjugated antibody combination was evaluated: CD16-FITC/CD13-PE/CD45-PerCP-Cy5.5/CD11b-APC (BD Biosciences, San Jose, CA). Stabilized peripheral blood (BD Multi-Check Control, BD Biosciences, San Jose, CA) was run daily to verify consistency of neutrophil staining for CD13 and CD16. Analysis was performed, using CellQuest software (BD Biosciences, San Jose, CA). An iterative, Boolean, strategy was used to isolate the following populations: lymphocytes, monocytes, granulocytes, myeloblasts and basophils (Fig. 1). Neutrophil maturation patterns were visually evaluated using dot plots of CD16 vs. CD13 and CD16 vs. CD11b (SAM, FEC). Results for control specimens were reviewed and any divergence from patterns identical to those previously illustrated by others as normal (14 16) was graded as either a mild-moderate or severe. Patterns seen among MDS and NDM specimens that were similar to those seen in the control set were considered as either normal or nonspecific and were not considered aberrant. MDS and NDM specimens with patterns differing from those seen among controls were divided into three groups based on the degree to which the patterns implied dysregulated/asynchronous antigen expression: aberrant if the patterns were a blatant deviation from normal synchronous maturation and the patterns seen among controls; equivocal-aberrant if there were deviations that appeared to be asynchronous but they were more subtle and/or there was concern that nonspecific changes might be contributing to the patterns, and synchronous-shift if there were alterations that appeared consistent with a synchronous left shift to immaturity. Discordant assessments were reviewed to reach consensus. Using visual cluster analysis for the plot of CD16 vs. CD11b, granulocytes were divided into least mature stage I (CD11b-/low, CD16-/low), intermediate maturity stage II (CD11bþ, CD16-/low), and most mature stage III (CD11bþ, CD16þ). The regions were not fixed, but adjusted based on the best visual distinction of three clusters. The percentage of each stage of maturation was determined. Numerical results, including percentage of total granulocytes, percentages of the 3 stages of granulocytes and percentage of myeloblasts, were considered abnormal if they exceeded the control mean by 3SD. To test the effect of protracted delay in specimen processing, bone marrow samples obtained for lymphoma staging with no definitive evidence of involvement and adequate aspirate cellularity were processed at three time points: the same day as receipt and after the filtered sample had been stored at 4 C for 1 day and 2 days. The extent of visual divergence from presumed normal maturation patterns was independently (SAM, FEC) and blindly assessed for each time point, graded (mild-moderate or severe) and individually ranked from least to most severe. Numerical parameters evaluated included those as above and the geometric mean fluorescence intensities (MFI) for CD11b, CD13 and CD16 for total granulocytes. Numerical parameters were compared between the day 0 and day 1, and between day 2 and day 3. Statistical Methods Comparsions between unrelated groups for continuous variables were made assuming nonparametric distributions using the Mann Whitney U test. Comparisons were made using the Fisher s exact test for categorical variables. The Wilcoxon signed-rank test was used to compare the numerical parameters assessed in the group of samples evaluated at different time points. RESULTS Neutrophil Maturation Pattern for Control Specimens Three of 19 control specimens demonstrated patterns of CD11b, CD13 and CD16 expression that were identical to those previously illustrated by others as normal (14 16) (Table 1, Supporting Information Table 1). The presumed normal maturation pattern exhibited a sickle shape on the CD16 vs. CD13 dot plot characterized by an initial peak intensity for CD13 beginning above the third log decade with low CD16, followed by a decrement of CD13 more than 2 log decades and

4 220 MONAGHAN ET AL. FIG. 1. Logic used to isolate cell populations. Lymphocytes were defined as cells in the region with low to intermediate side scatter light properties (R1), bright CD45 (R3) and with a lack of CD13(R7). R3 and R7 were set to exclude B-cell precursors, which display less intense CD45. Monocytes included cells with low to intermediate side scatter light properties (R1), moderate to bright CD45 (R4), bright CD11b and CD13 (R8), and with a lack of the brightest CD16 seen among mature neutrophils (R9). Granulocytes were defined as cells with intermediate to high side scatter light properties (R2), intermediate CD45 expression (R5), positive for CD13 (R10), and with exclusion of cells previously defines as monocytes. Myeloblasts were defined as cells with low to intermediate side scatter light properties (R1), positive for CD45 (R6), and positive for CD13 (R10), but usually lacking CD11b (i.e., not R11). finally increasing CD13 that again peaked slightly above the third log decade along with an associated acquisition of CD16 that also extended beyond the third log decade (Fig. 2A). Increasing CD11b from low to ultimately approaching or extending just beyond the third decade, along with the acquisition of CD16 during the final stages of maturation, resulted in a pattern with three clusters or nodes on the CD16 vs. CD11b dotplot (Fig. 2B). The other 16 control specimens (84%) showed varying degrees of divergence from previously published normal and had a contracted appearance (Figs. 2C 2F). Contraction appeared largely due to diminished acquisition of CD16 among most mature neutrophils. The most severe contraction was seen in one control specimen with nearly complete loss of the distinction between the expected two peaks for high intensity CD13 on the CD16 vs. CD13 dot plot and, in this specimen, the contracted pattern was also readily apparent on the CD16 vs. CD11b dot plot (Figs. 2E and 2F). Contraction of a lesser degree compared with the most

5 ALTERED NEUTROPHIL MATURATION PATTERNS FOR IDENTIFICATION OF MDS 221 Table 1 Maturation Patterns Among Controls, MDS, and Specimens Nondiagnostic for MDS (NDM) Patterns Controls (n ¼ 19) MDS with subtypes specified (n ¼ 23) NDM (n ¼ 37) Granulocytes too sparse 0 1 (4%) RAEB (1) 1 (3%) Not classified (3%) Nonspecific (mild-moderate contraction) 15 (79%) 11 (48%) RAEB (4) 26 (70%) Receiving azacitidine for RAEB (1) RCMD (3) MDS-U (1) RARS (1) MDS with isolated del(5q) (1) Nonspecific (severe contraction) 1 (5%) 2 (9%) RAEB (2) 3 (8%) Similar to published normal * 3 (16%) 2 (9%) RCMD (1) 1 (3%) RCUD (1) Synchronous-shift (8%) Equivocal-aberrant 0 4 (17%) RAEB (3) 2 (5%) RCMD (1) Aberrant 0 3 (13%) RAEB (2) 0 Receiving azacitidine for RAEB (1) *Patterns published as normal from Refs severe case seen among controls was designated as mild-moderate. These cases appeared to have delayed acquisition of CD16 that lagged behind reacquistion of CD13 initially and this led to an altered, convex pattern for that portion of the CD16 vs. CD13 dot plot (Fig. 2C). The mild-moderate contracted pattern appeared subtle on the CD16 vs. CD11b dot plot (Fig. 2D). Neutrophil Maturation Pattern for MDS and NDM Specimens Granulocytes were too sparse to confidently assess maturation in one MDS and one NDM specimen (Table 1, Supporting Information Tables 2 and 3). Severe contraction, similar to that seen in one control, was present in two MDS and three NDM specimens and precluded further characterization of the maturation patterns. Patterns similar to published normal neutrophil maturation patterns were seen in 2 of 20 MDS specimens (RCMD and RCVD) and 1 of 33 NDM specimens, respectively. Mild-moderate contraction appeared to be the only feature present in 11 (48%) and 26 (70%) of the MDS and NDM specimens, respectively. Further assessment for possible aberrant features was still performed when mild-moderate contraction was suspected. Aberrant granulocyte maturation was present in 3 of 23 (13%) MDS specimens [RAEB2 (n ¼ 2) and 1 patient receiving azacitidine for RAEB2]. Aberrant maturation patterns were not detected in any of the 37 NDM specimens. Aberrant granulocyte maturation patterns included clear-cut asynchronous antigen expression as well as an obvious accumulation of cells suggesting a shift to immaturity and, sometimes, lack of continuity in the maturation pattern between granulocyte subsets (Figs. 3A 3D). Two of the three aberrant patterns were detected in specimens with limited morphologic evaluations (one with hemodilute aspirate smears and one that lacked aspirate smears). The former was diagnosed as RAEB2 based on detection of occasional blasts with Auer rods on hemodilute smears and a low blast percentage as estimated by immunohistochemistry on the biopsy. The latter had a prior bone marrow evaluation diagnostic for RAEB2 associated with a normal karyotype and was on azacitidine at the time of this FC evaluation. The potential contribution of azacitidine therapy on the granulocyte maturation patterns could not be systematically assessed since only two other patients were receiving azacitidine during this study (one MDS and one NDM specimen); however, the maturation patterns for these two patients displayed only mild-moderate contraction. Equivocal-aberrant patterns were present in 4 (17%) MDS specimens [RAEB2 (n ¼ 2), RAEB1 and RCMD] and 2 (5%) NDM specimens. Cases with an equivocalaberrant pattern demonstrated some of the changes seen in aberrant cases, including evidence for asynchrony and a shift to immaturity, but the changes were not as prominent and nonspecific contraction as a possible confounding factor was a concern (Figs. 3E and 3F). Notably, the 2 patients with NDM specimens and findings designated as equivocal-aberrant had complex medical conditions (e.g. renal failure, recent sepsis, history of liver transplant) and were on medications (e.g. filgastrim, cyclosporine, prednisone) that could conceivably contribute to cytopenias and/or altered maturation patterns (Supporting Information Table 3). Three NDM and no MDS specimens displayed a pattern classified as synchronous-shift, consistent with synchronous left shift (Figs. 3G and 3H). In addition, one NDM specimen displayed unique alterations not seen in other specimens during the study or during the experience of the authors practice; therefore, the pattern for this specimen was not placed in a group and the possibility was considered that the findings were spurious. Effect of Delayed Specimen Processing on Neutrophil Maturation To evaluate whether delayed specimen processing might have contributed to the nonspecific contracted

6 222 MONAGHAN ET AL. FIG. 2. Neutrophil Maturation Patterns in Control Specimens. (A) Pattern similar to that previously illustrated as normal by others, showing progression from promyelocytes to segmented neutrophils, resulting in a concave sickle shape on CD16 vs. CD13 dot plot; the pattern begins with a bright CD13 peak that diminishes more than 2 logs followed by a gradual increase of CD13 to its original intensity in conjunction with a synchronous increase of CD16 of more than 2 logs in intensity. (B) The same specimen as in (A) displaying the presumed normal pattern on CD16 vs. CD11b dot plot with 3 nodes comprised of dense concentrations of cells (node 1: low CD16, low CD11b; node 2: low CD16, bright CD11b; node 3: bright CD16, bright CD11b); the pattern begins with node 1, followed by node 2, and terminates with node 3. (C) A control specimen showing mild-moderate contraction on CD16 vs. CD13 dot plot with slightly diminished ultimate CD16 intensity and a convex shape due to deviation from the expected synchronous reacquisition of CD13 with increasing CD16. (D) The same specimen as in (C) displaying diminished ultimate CD16 intensity on CD16 vs. CD11b dot plot, but the contracted pattern is more subtle compared to CD16 vs. CD13 dot plot. (E) The single control specimen showing severe contraction on CD16 vs. CD13 dot plot with marked reduction in acquisition of CD16, which obscures the 2 expected CD13 peaks. (F) The same specimen as in (E) displaying markedly diminished distinction between node 2 and node 3 on CD16 vs. CD11b dot plot. effect on the maturation patterns, the time period between specimen collection and processing (i.e., staining and acquisition of events) was reviewed. For the control specimens, accurate times were available for eight specimens (median: 20.8 h; range: 5 30 h), including two with patterns similar to previously published normal patterns (5 and 5.2 h) and six with mild-moderate contraction (median: 23 h; range: h). Given the small number of control specimens with accurate time periods, the effect of delayed specimen processing was further analyzed using all specimens (i.e., controls, MDS and NDM specimens). The time intervals were FIG. 3. Neutrophil Maturation Patterns in MDS and NDM Specimens. (A, B) Patterns were classified as aberrant in a RAEB2 specimen due to large accumulation of CD16 low, CD13 bright positive cells with a lag in the expected down regulation of CD13 on the CD16 vs. CD13 dot plot, which appears asynchronous with the acquisition of CD11b on the CD16 vs. CD11b dot plot. (C, D) Patterns classified as aberrant in a different RAEB2 specimen based on a large accumulation of CD16 negative cells and also a lack of continuity in the pattern between CD16 low, CD13 low cells and the CD16 bright positive, CD13 bright positive cells. Both the CD16 vs. CD13 and the CD16 vs. CD11b dot plots also reveal decreased intensity of CD16 on early stage neutrophils compared with other specimens illustrated in the figure. (E, F) Patterns designated as equivocal-aberrant in an NDM specimen because there is evidence for an accumulation of CD16 low, CD13 bright positive cells on the CD16 vs. CD13 dot plot. The lack of acquisition of CD16 is also visualized on the CD16 vs. CD11b dot plot. The patterns are not considered sufficient for designation as unequivocal aberrant because of concern for possible nonspecific contraction effect. (G, H) Patterns designated as synchronous-shift in an NDM specimen because the findings represent what is expected for synchronous maturation up to the myelocyte/metamyelocyte stages with an abrupt lack of evidence for further maturation on both CD16 vs. CD13 and CD16 vs. CD11b dot plots. available for 21 of 23 MDS specimens (median: 20 h; range: h) and 33 of 37 NDM specimens (median: 19.5 h; range: h) and there was no significant difference between these two groups or either group compared with control specimens. The time interval

7 ALTERED NEUTROPHIL MATURATION PATTERNS FOR IDENTIFICATION OF MDS 223 between specimen collection and processing was significantly different for specimens with a patterns similar to previously published normal patterns (median: 5.2 h; range: 4 22 h) versus mild-moderate contraction (median: 20.5 h; range: 4 30 h) (P ¼ 0.003) and for presumed normal patterns versus severe contraction (median: 20 h; range: h) (P ¼ 0.03), but the difference was not significant between mild-moderate and severe contraction. To further evaluate the effect of protracted delay in specimen processing on the neutrophil maturation pattern, a subset of specimens were processed and evaluated at three sequential time points (day 0, day 1 and day 2). Nineteen specimens were evaluated, including 10 control specimens and 9 additional specimens excluded from the control category for the following reasons: biopsy suboptimal (n ¼ 2), low-level involvement by lymphoma could not be completely excluded (n ¼ 5) and unexplained anemia in a patient undergoing lymphoma staging (n ¼ 2). Nonspecific contraction of the neutrophil maturation pattern was seen for many of these specimens and, as described above for controls, was graded as mild-moderate or severe. An overall change in the grade of contraction for a specimen over time was rare: no change for 17 of 19 specimens, an increase from normal to mild-moderate for one specimen between two time points, and a decrease from severe to mild-moderate for one specimen between two time points. However, more subtle changes over time were identified most easily on the CD16 vs. CD13 dotplot and became apparent upon ranking the specimens in order of severity of contraction (Fig. 4): increased contraction rank for 10 of 19 specimens between day 0 and day 1, and 17 of 19 specimens between day 0 and day 2, and decreased contraction rank for only one specimen between day 0 and day 1, and one specimen between day 0 and day 2. The association between contraction and protracted delay in processing was also supported by a significant decrease in CD16 MFI for the total granulocyte population on day 2 compared with day 1 (P ¼ ) and an increase in the proportion of stage II neutrophils (CD11bþ, CD16-/low) on day 1 compared with day 0 (P ¼ ). No significant change in CD11b and CD13 MFI was identified with delayed specimen processing. Numerical Assessment of Neutrophil Maturation Numerical assessment was performed to confirm the visual observation that specimens with aberrant and equivocal-aberrant alterations in neutrophil maturation pattern demonstrated an apparent accumulation of less mature neutrophil precursors and to further evaluate the specificity of these findings for the identification of MDS. Subset analysis of control specimens generated the following reference values (mean 3SD): myeloblasts 2.0% (0 4.1%), total granulocytes 74.4% ( %), stage I neutrophils 13.4%, (0 27%), stage II neutrophils 40.2% ( %), stage III neutrophils FIG. 4. Increasing Rank of Severity of Nonspecific Contraction with Delayed Specimen Processing. This specimen was processed and evaluated at 3 sequential time points (day 0, day 1 and day 2). On day 0, the CD16 vs. CD13 pattern resembles that previously illustrated as normal by others. On day 1 and day 2 there is subtly increasing mild-moderate contraction with diminished ultimate CD16 intensity and a convex shape due to an apparent lag in the reacquisition of CD13 relative to increasing CD % (3.5 89%). No individual control specimens exceeded reference values (Tables 2 and 3). Compared with controls, MDS specimens demonstrated significantly higher overall myeloblast percent, a lower proportion of total neutrophils, and higher proportion of stage I neutrophils (CD11b-/low, CD16-/low), but no significant change in the proportion of stage II or stage III neutrophils (Table 2). A significantly higher proportion of MDS cases compared to controls exceeded the reference values for these same population subsets (Table 3). There was no significant difference between NMD specimens and control specimens for overall myeloblast percent, total neutrophils,

8 224 MONAGHAN ET AL. Table 2 Numerical Assessment of Neutrophil Maturation: Overall Distribution of Cell Populations (%) According to Diagnostic Category Control specimens (n ¼ 19) MDS specimens (n ¼ 23) Specimens nondiagnostic for MDS (n ¼ 37) Population subsets Mean SD Mean SD p Mean SD p Myeloblasts NS Total Neutrophilic cells NS Stage I Neutrophils (Least mature) NS Stage II Neutrophils (Intermediate maturity) NS NS Stage III Neutrophils (Most mature) NS NS SD ¼ Standard deviation P values reflect comparison with results for Control cases (Mann Whitney U test). NS ¼ not statistically significant (P > 0.05). or neutrophil stages of maturation. Statistically significant percentages of NDM cases exceeding reference values for enumeration of population subsets were also not detected. The numerical data was also compared between groups based on neutrophil maturation pattern (Table 3). The two MDS specimens and one NDM specimen with the patterns similar to published normal patterns did not display any numerical alterations when compared with controls. All 9 specimens with aberrant or equivalent-aberrant maturations patterns (7 MDS and 2 NDM) demonstrated 1 or 2 numerical alterations to support a left shift [6 (67%) with increased myeloblast percent and 7 (78%) with increased stage 1 neutrophils]. Only occasional specimens with nonspecific contraction showed numerical evidence for a left shift (7 of 42, 17%), including 4 with increased myeloblast percent, 2 with increased stage 1 neutrophils and 1 with both alterations. In contrast, evidence for a left shift was found in 2 of 3 specimens (67%) with a synchronous-shift pattern, including 1 with increased stage 1 neutrophils and 1 with both increased stage II and decreased stage III neutrophils. Decreased total neutrophil percent were seen in 4 of 9 specimens (44%) with aberrant or equivalent-aberrant maturation patterns, 12 of 42 specimens (29%) with nonspecific contraction, 2 of 2 with granulocytes too sparse to assess pattern and none with synchronous-shift. Although comparisons should be regarded with caution due to low number of cases with aberrant or equivalent-aberrant maturations patterns, statistical analysis supported a significantly higher proportion of cases with increased myeloblasts, increased stage I neutrophils and decreased total neutrophils among those with aberrant or equivocal aberrant maturation patterns when compared with controls. The analysis also supported a significantly higher proportion of cases with increased myeloblasts and increased stage I neutrophils among those with aberrant or equivocal aberrant maturation patterns compared with those with non-specific contraction, but no difference for the proportion with decreased total neutrophils. DISCUSSION While applying FC methods to evaluate for MDS, we identified factors not previously emphasized that made interpretation of granulocyte maturation patterns difficult. Although some control, MDS and NDM specimens displayed patterns of CD11b, CD13, and CD16 similar to those illustrated by others as normal (14 16), many control specimens, as well as MDS and NDM specimens, showed a contracted appearance that was largely due to less intense CD16 among later neutrophil stages. This effect was usually a mild or moderate divergence from presumed normal and most easily appreciated on the CD16 vs. CD13 dot plot. However, the effect was also apparent on the CD16 vs. CD11b dot plot when more marked. This nonspecific contracted appearance was considered within the reference baseline for our laboratory and patterns with this effect were not considered abnormal or used to suggest a diagnosis of MDS. Other studies have not reported a high frequency of alterations in control and other non-mds specimens. The reason for the discrepancy is not clear. However, it is possible that similar patterns among non-mds cases have been observed by other laboratories and they may be at risk for over-interpreting a contracted appearance as abnormal and/or supportive for MDS if they use only the published illustrations of normal for a reference baseline. Rare cases in our study displayed a severely contracted appearance, including one control specimen; such a finding should preclude assessment of granulocyte maturation since it could completely obscure findings compatible with MDS. When contraction appears mild or moderate, evaluation of the granulocyte maturation pattern might still be possible, and was performed in this study. However, recognition of the potential contracted effect contributed to designation of some findings as equivocal-aberrant rather than aberrant because confident identification of subtle, but possibly more significant alterations was difficult. This conservative approach may account, at least in part, for the lower incidence of aberrant alterations in neutrophil maturation patterns seen in MDS in this study compared to that reported previously as asynchronous or abnormal by others (9,10,17,18). An association was observed between the nonspecific contracted pattern of granulocyte maturation seen in the current study and delays in specimen processing, as supported by significantly shorter time periods from specimen collection to processing for specimens with

9 ALTERED NEUTROPHIL MATURATION PATTERNS FOR IDENTIFICATION OF MDS 225 Table 3 Number and Proportion of Cases with Numerical Alterations a According to Diagnostic Category and Neutrophil Maturation Patterns Numerical alterations observed a Numerical alterations according to diagnostic category Numerical alterations according to neutrophil maturation pattern (MDS and NDM cases) Controls (n ¼ 19) MDS (n ¼ 23) NDM (n ¼ 37) Aberrant or equivocal-aberrant (n ¼ 9) Nonspecific contraction (n ¼ 42) Number: Aberrant (n ¼ 3) Number: Equivocalaberrant (n ¼ 6) Percent (aberrant and equivocalaberrant) Number: Mild-tomoderate (n ¼ 37) Number: Severe (n ¼ 5) Synchronous-shift (n ¼ 3) Percent (mild-moderate and severe) Number Percent Increased myeloblasts 0 10 (43%) 1 (3%) 2 4 6/9 (67%) b 4 1 5/42 (12%) 0 0% P ¼ NS NS Decreased Total 0 12 (52%) 6 (16%) 1 3 4/9 (44%) b /42 (29%) 0 0% Neutrophils P ¼ NS P ¼ Increased Stage I 0 7 (30%) 5 (13.5%) 3 4 7/9 (78%) b 2 1 3/42 (7%) 1 1/3 (33%) Neutrophils (Least Mature) Increased Stage II Neutrophils Decreased Stage III Neutrophils (Most Mature) P ¼ NS NS (3%) 0 0 0% 0 0 0% 1 1/3 (33%) NS NS (3%) 0 0 0% 0 0 0% 1 1/3 (33%) NS NS a Alterations defined as numerical results that exceed the reference values (control mean 3 SD). Statistical significance determined in comparison to Controls using the Fisher s exact test; P value included if <0.05; NS ¼ no statistically significant difference (P > 0.05). b Comparison of combined Aberrant and Equivocal-Aberrant cases, although limited by the low number of cases (n ¼ 9), with Controls using the Fisher s exact test demonstrates significantly more cases with increased myeloblasts (P ¼ ), decreased total neutrophils (P ¼ ), and increased stage I neutrophils (P <0.0001). Comparison of combined Aberrant and Equivocal-Aberrant cases with cases showing only nonspecific contraction using the Fisher s exact test demonstrates significantly more cases with increased myeloblasts (P ¼ ) and increased stage I neutrophils (P <0.0001), but no significant difference for decreased total neutrophils.

10 226 MONAGHAN ET AL. patterns similar to previously published normal patterns of maturation compared with either those with mildmoderate contraction or those with severe contraction. In addition, evaluation of a subset of specimens that was processed on the same day as receipt in the laboratory and then again 1 and 2 days later demonstrated a significant decrease in total granulocyte CD16 fluorescence over time and an increasing rank in the severity of contraction judged visually. Admittedly, however, the increasing nonspecific contraction over time, during which the filtered cells were stored at 4 C, was visually subtle and an overall change from presumed normal to mild-moderate contraction was very rare. The subtlety of these changes noted in this study may explain why another group found enumeration of immature granulocytes based on CD11b and CD16 to be stable for up to 48 h for peripheral blood specimens when stored at 4 C (28). Granulocytes are biologically active cells and changes in antigen expression, including changes in CD11b, CD13, and CD16 expression, have been observed under different in vitro conditions as has been summarized (29). Since apoptosis results in shedding of CD16 from neutrophils (30 32), it is not surprising that prolonged time between collection and processing was associated with reduced CD16 expression. Anticoagulation with heparin leads to less rapid apoptosis compared with other anticoagulants (33,34). Thus, specimen processing of bone marrow samples within 24 hours of collection and the use of heparin anticoagulation has been recommended (2,5,29,35). The current study mostly conforms to these recommendations, but occasional specimens were processed beyond 24 h (up to 30 h in specimens with known time intervals). Other possible preanalytical specimen variables, such as temperature during specimen transport, differences in specimen processing and patient variables (e.g., levels of inflammatory cytokines, fever, and effect of medications), were not evaluated in the current study but may also contribute to alterations in antigen expression and/or neutrophil apoptosis (29,36 42). Given such potential variables, it is important for laboratories to standardize their own procedures for specimen collection, transportation and processing, run controls such as stabilized cells to ensure consistency of staining, and become familiar with the patterns of antigen expression that are seen in non-mds specimens in their own setting. Studies and recommendations being generated by the European LeukemiaNet have begun to illustrate how standardization of across multiple centers is an achievable goal for FC even for the evaluation of MDS (5,37). Further delineation of preanalytical variables that can alter granulocyte maturation patterns in an individual laboratory and in published studies will help avert the misinterpretation of alterations that do not signify MDS. Despite the confounding nonspecific contraction that affected the granulocyte maturation patterns of many specimens, aberrant granulocyte maturation patterns were identified in MDS specimens and were recognized based on clear-cut asynchronous antigen expression and, sometimes, a lack of continuity in the maturation pattern between granulocyte subsets. Although these aberrant maturation patterns were found with low frequency, and a definite conclusion cannot be made, it is possible that they have good specificity for MDS since they were not detected in any control or NDM specimens. The findings in this study also suggest that an aberrant maturation pattern can help support a diagnosis of MDS in cases with limited morphologic evaluation since definitive abnormalities were found in one specimen with hemodilute aspirate smears and one specimen that lacked aspirate smears. On the other hand, equivocal-aberrant patterns were not only present in four MDS cases but also in two NDM cases. Although it is possible that the latter two patients had emerging MDS, it seems more likely that an equivocal-aberrant pattern may not have sufficient specificity for MDS since both patients had complex medical conditions and were on medications that could contribute to cytopenias, dyspoiesis and, possibly, altered maturation patterns. The neutrophil maturation patterns in each category were carefully reviewed visually and numerically to characterize the features that help distinguish definitive aberrant maturation patterns from equivocal or nonspecific changes. Aberrant patterns not only revealed asynchronous antigen expression but also visual evidence to suggest an accumulation of immature granulocytic cells. Similar, but more subtle changes were seen among the specimens with patterns designated as equivocal-aberrant. To confirm these observations and to investigate the specificity of the accumulation of apparently immature granulocytes for MDS, the percentage of five population subsets was compared numerically: blasts, total neutrophils, and neutrophil stages I, II and III. This numerical analysis revealed that, compared with controls, MDS specimens had a statistically significant higher percentage of myeloblasts, lower percentage of total neutrophils, and higher percentage of stage I maturing neutrophils (least-mature subset). In contrast, NDM specimens did not differ significantly from controls. In addition, all specimens with visually identified aberrant or equivocal-aberrant patterns had 1 or 2 numerical alterations to support a shift to immaturity (increased blasts and/or increased stage I neutrophils), based on individual values exceeding the reference range (control mean þ/- 3SD). Although none of the numerical alterations were specific for an aberrant or equivocal-aberrant pattern, more specimens with these patterns had increased myeloblasts and increased stage I neutrophils compared with either controls or specimens with nonspecific contraction. Specifically, no significant association was identified between nonspecific contraction and neutrophilic left shift, with only occasional contracted patterns demonstrating an increased percent of stage I neutrophils and/or myeloblasts. Thus, numerical analysis supports the association between an apparent immunophenotypic shift to immaturity and MDS and between an apparent immunophenotypic shift to immaturity and

11 ALTERED NEUTROPHIL MATURATION PATTERNS FOR IDENTIFICATION OF MDS 227 visually identified aberrant or suspected aberrant granulocyte maturation patterns. Although the ability to detect an increased proportion of myeloblasts by FC in MDS is not a novel finding, further study may be warranted to assess the strength and utility of the association between an accumulation of other less mature neutrophil stages and MDS. However, the numerical changes associated with a shift to immaturity based on CD11b, CD13, and CD16 appear more likely to lack specificity for MDS when compared with clear-cut aberrant patterns identified visually. In particular, NDM specimens with visual patterns to support a synchronous left shift (synchronous-shift pattern) included one with increased stage I neutrophils and one with increased stage II (intermediate-stage) and decreased stage III neutrophils (most mature). This observation is in agreement with those of Stachurski et al., (9) who noted that subset changes between two stages of granulocyte maturation (mature and immature) were not specific for MDS and could be seen with altered specimen quality, eosinophils, hemodilution, genetic polymorphisms and bone marrow regeneration. More recently, Finn et al. (43) also recently reported difficulty in distinguishing some non-mds cases with a shift to granulocyte immaturity from RAEB and RCMD, despite their ability to separate most higher grade MDS from non-mds using information geometry and an overall representation of multidimensional FC data related to granulocyte maturation. Nonetheless, this latter study, along with another by Matarraz et al. (20), does illustrate that objective ways to assess the extent of deviation from normal, even for granulocyte maturation patterns, may be an attainable goal. Matarraz et al. found statistically significant differences between MDS and normal control samples when enumerating five population subsets similar (though not identical) to those in this study and they were able to discriminate MDS from other diseases when incorporating this information into a score based on the number and severity of multiple abnormalities. Furthermore, selectively utilizing additional antigens for assessment of granulocyte maturation patterns can help distinguish clearly deregulated patterns of antigen expression from synchronous altered proportions of granulocyte subsets. Starchurski et al. (9) found CD64/ CD10 and CD33/CD15 to be helpful in addition to CD11b, CD13, and CD16, but other groups have illustrated different combinations that may also be useful (2,5,14,21,44,45). In addition, a detailed atlas of normal patterns of antigen expression among hematopoietic populations, including maturing granulocytes, has recently been published and will guide further exploration of abnormalities that may be helpful to diagnose MDS (37). Clearly, software-driven analysis and methods more objective than visual pattern recognition have become a necessity not only for standardization but also to make the most of the complexity of FC information that is needed to evaluate for MDS. In summary, this study reinforces that occasional clear-cut aberrant granulocyte maturation patterns in MDS can be detected by FC. However, patterns that diverge from those illustrated by others as normal may be more frequent among non-mds specimens than previously reported. Some confounding variables may be more common for particular laboratory settings or patient populations and some will likely remain difficult to control despite better standardization. In particular, the attempt to validate an analysis for granulocyte maturation in specimens from patients with diseases or medications known or suspected to contribute to a left shift and nonspecific dyspoiesis can be problematic. Numerical analysis and more sophisticated methods of analysis are expected to yield more objectivity and reproducibility between laboratories, but these methods may also be confounded by effects nonspecific for MDS. It is well established that multiple parameters must be assessed to gain adequate sensitivity and specificity for the FC evaluation of MDS and this study only focused on a limited panel to assess granulocyte maturation. However, our study highlights the importance for each laboratory in its own setting to critically evaluate the utility of assessing individual alterations for the FC evaluation of MDS. ACKNOWLEDGMENT The authors thank John Kloke, PhD, Department of Medicine, University of Pittsburgh School of Medicine, for assistance with the statistical analysis for the delayed specimen processing. LITERATURE CITED 1. Elghetany MT. Surface marker abnormalities in myelodysplastic syndromes. Haematologica 1998;83: Loken MR, van de Loosdrecht A, Ogata K, Orfao A, Wells DA. Flow cytometry in myelodysplastic syndromes: Report from a working conference. Leuk Res 2008;32: Loken MR, Wells DA. The role of flow cytometry in myelodysplastic syndromes. J Natl Comp Canc Netw 2008;6: Ogata K. Diagnostic flow cytometry for low-grade myelodysplastic syndromes. Hematol Oncol 2008;26: van de Loosdrecht AA, Alhan C, Bene MC, Della Porta MG, Drager AM, Feuillard J, Font P, Germing U, Haase D, Homburg CH, et al. Standardization of flow cytometry in myelodysplastic syndromes: Report from the first European LeukemiaNet working conference on flow cytometry in myelodysplastic syndromes. Haematologica 2009;94: Della Porta MG, Lanza F, Del Vecchio L. Flow cytometry immunophenotyping for the evaluation of bone marrow dysplasia. Cytometry B Clin Cytom 2011;80B: Ogata K, Della Porta MG, Malcovati L, Picone C, Yokose N, Matsuda A, Yamashita T, Tamura H, Tsukada J, Dan K. Diagnostic utility of flow cytometry in low-grade myelodysplastic syndromes: A prospective validation study. Haematologica 2009;94: Ogata K, Kishikawa Y, Satoh C, Tamura H, Dan K, Hayashi A. Diagnostic application of flow cytometric characteristics of CD34þ cells in low-grade myelodysplastic syndromes. Blood 2006;108: Stachurski D, Smith BR, Pozdnyakova O, Andersen M, Xiao Z, Raza A, Woda BA, Wang SA. Flow cytometric analysis of myelomonocytic cells by a pattern recognition approach is sensitive and specific in diagnosing myelodysplastic syndrome and related marrow diseases: emphasis on a global evaluation and recognition of diagnostic pitfalls. Leuk Res 2008;32: Truong F, Smith BR, Stachurski D, Cerny J, Medeiros LJ, Woda BA, Wang SA. The utility of flow cytometric immunophenotyping in cytopenic patients with a non-diagnostic bone marrow: A prospective study. Leuk Res 2009;33: van de Loosdrecht AA, Westers TM, Westra AH, Drager AM, van der Velden VH, Ossenkoppele GJ. Identification of distinct prognostic