Materials and Methods

Definition of reticulocyte types—Four types of reticulocytes were defined on the basis of the reticulum pattern revealed via supravital NMB staining (Figure 1). The reticulocyte classification categories were as follows: punctate-1 (< 10 dots of stained reticulum); punctate-2 (≥ 10 dots of stained reticulum); aggregate (any aggregates of stained reticulum, distributed diffusely within the cytoplasm); and ring-form (aggregates of stained reticulum that formed a ring—contiguous or discontiguous—around at least half of the reticulocyte nucleus).

View Full Size

Photomicrographs of each of the 4 avian reticulocyte types defined on the basis of the morphologic appearance after staining with NMB. A—Punctate-1 reticulocyte (< 10 dots of stained reticulum). B—Punctate-2 reticulocyte (≥ 10 dots of stained reticulum). C—Aggregate reticulocyte (any aggregates of stained reticulum, distributed diffusely within the cytoplasm). D—Ring-form reticulocyte (aggregates of stained reticulum that formed a ring—contiguous or discontiguous—around at least half of the reticulocyte nucleus). In this example, slightly discontiguous reticulum aggregates are present around the nucleus. Bar = 5 μm.

Citation: American Journal of Veterinary Research 69, 8; 10.2460/ajvr.69.8.1067

• Download Figure
• Download figure as PowerPoint slide

Comparison of reticulocyte-staining methods and evaluation of sample storage—A single 4-mL sample of whole blood was collected (immediately prior to euthanasia^a) from an injured red-tailed hawk that was brought to the veterinary medical hospital at the University of California, Davis; EDTA was used as the anticoagulant for the sample. The whole blood sample was diluted with NMB stain^b (1:1, 1:2, 1:3, and 1:4 [blood to stain]). Whole blood volumes of 10 to 20 μL were placed into plastic bullet tubes, and appropriate volumes of stain were added. The tubes were capped, mixed by repeated inversion, and allowed to incubate at room temperature (approx 22°C) for 5, 10, 30, or 60 minutes. Repeated inversion of the tubes was again performed before a direct smear of the contents of each was made. Sixteen smears (1 from each of the 4 diluted samples that underwent each incubation period) were evaluated by 1 observer (MPS). By use of the aforementioned reticulocyte classification scheme, the percentage of each reticulocyte type among 1,000 RBCs examined on each smear was determined.

A dilution of 1:3 (blood to stain) and a 10-minute incubation time were used to evaluate reticulocyte staining after 4, 12, 24, and 48 hours of storage of the EDTA whole blood sample at 4°C. Staining quality was subjectively evaluated for all smears, and the percentage of each reticulocyte type among 1,000 RBCs examined on each smear was determined. Degenerated erythrocytes, defined as cells with fragmented nuclear chromatin or deeply basophilic cytoplasm (or both), were counted separately when observed.

Comparison of polychromatophil and reticulocyte percentages—Samples of blood (anticoagulated with EDTA) from 31 birds that were examined at the veterinary medical teaching hospital were randomly selected from all avian blood samples submitted to the clinical pathology laboratory for routine CBCs. This group of birds included Psittaciformes (n = 16), Anseriformes (3), Falconiformes (7), and Strigiformes (5). Wright-Giemsa–stained smears^c were prepared from each blood sample as part of the CBC. An aliquot of whole blood (10 to 20 μL) was taken from each sample and added to NMB stain in a 1:3 dilution; samples were incubated at room temperature for 10 minutes, and a direct smear was prepared. Two observers (JLJ and MPS [observers 1 and 2]) each performed differential counts on 1,000 cells on the Wright-Giemsa– and the NMB-stained smears. Normochromic erythrocytes, polychromatophilic erythrocytes, and IECs (defined as round erythroid cells with basophilic cytoplasm) were counted in the Wright-Giemsa–stained smears (Figure 2). For comparison of different combinations of cells to reticulocytes, an adjusted polychromatophil percentage was calculated (ie, IECs were excluded from polychromatophil and total RBC counts) to express polychromatophils as a percentage of the combined number of polychromatophils and mature RBCs. A combined percentage of polychromatophils and IECs, which expressed all immature erythrocytes as a percentage of all erythroid cells, was also calculated. The 4 reticulocyte types (punctate-1, punctate-2, aggregate, and ring-form) were each counted in the NMB–stained smears and expressed as percentages. A combined percentage of aggregate and ringform reticulocytes was also calculated.

View Full Size

Photomicrograph of avian erythrocyte types defined on the basis of the morphologic appearance after staining with Wright-Giemsa stain. A—Normochromic erythrocyte. B—Polychromatophilic erythrocyte. C—Immature erythroid cell. Bar = 5 μm.

Citation: American Journal of Veterinary Research 69, 8; 10.2460/ajvr.69.8.1067

• Download Figure
• Download figure as PowerPoint slide

Comparisons were made between adjusted polychromatophil percentage and percentages of each of the 4 reticulocyte types and also the combined percentage of aggregate and ring-form reticulocytes. Comparisons were also made between the combined percentage of polychromatophils and IECs and percentages of the 4 reticulocyte types and also the combined percentage of aggregate and ring-form reticulocytes. To evaluate the potential for observer bias, the differential counts on 1,000 cells were repeated on all NMB-stained smears by both observers at the conclusion of the study.

Generation of a reference interval for polychromatophils in healthy psittacines—Clinic records from January 2003 to January 2007 were searched electronically for psittacines that underwent a routine wellness examination. A bird was considered healthy if no clinical abnormalities or other evidence of disease was detected via physical examination and if results of a CBC and plasma biochemical analyses were within published reference intervals for healthy psittacines.¹ Birds were excluded if they did not meet all of those criteria or if they were < 6 months old. Differential counts on 1,000 RBCs were performed on archived Wright-Giemsa– stained smears of blood from the birds that met the criteria for inclusion. Percentages of polychromatophils and IECs were recorded for each smear. Absolute polychromatophil counts were calculated by multiplying the polychromatophil percentage by the total RBC count (manual leukocyte counts were subtracted from the total nucleated cell count obtained from the hematology analyzer^d to obtain the total RBC count).

Statistical analysis—Descriptive statistics were generated for data from the comparison of staining methods.^e Interobserver and intraobserver variation in the percentage of ring-form reticulocytes was assessed via repeated-measures ANOVA.^f Polychromatophil and reticulocyte percentages were compared by use of a Spearman rank correlation procedure.^e Deming regression was used to compare polychromatophil and ringform reticulocyte percentages recorded by both observers.^e The distributions of both the percentage and absolute polychromatophil data were evaluated by use of the D'Agostino-Pearson test.^e A value of P < 0.05 was considered significant.

Results

Comparison of reticulocyte-staining methods and evaluation of storage—Staining quality of the 4 smears made from blood samples that were diluted 1:1 with NMB stain and incubated at room temperature for 5, 10, 30, or 60 minutes was unacceptable for differential counting because the reticulum was often faintly stained, thereby precluding differentiation of various reticulocyte types. Staining quality of the remaining 12 smears was assessed as acceptable for differential counting. The optimal staining protocol selected for use in subsequent analyses was a dilution of 1:3 (blood to NMB stain) and an incubation time of 10 minutes.

Mean ± SD percentages and CV values were calculated for each of the 4 reticulocyte types in the single sample of avian blood collected from a red-tailed hawk. Mean percentages of the reticulocytes were as follows: punctate-1 reticulocytes, 26.11 ± 3.20%; punctate-2 reticulocytes, 19.97 ± 4.36%; aggregate reticulocytes, 47.73 ± 4.40%; and ring-form reticulocytes, 6.19 ± 0.35%. The CV values for the 4 reticulocyte types were as follows: punctate-1 reticulocytes, 12.25%; punctate-2 reticulocytes, 21.82%; aggregate reticulocytes, 9.22%; and ring-form reticulocytes, 5.61%. The CVs were subjectively assessed as acceptable for reticulocyte percentages for blood samples stored at 4°C for 4, 12, and 24 hours prior to staining, with CV values of 9.10%, 13.44%, 6.87%, and 18.16% for ring-form, aggregate, punctate-2, and punctate-1 reticulocyte types, respectively. Degenerated erythrocytes comprised 0.6% of erythrocytes in the sample stored for 24 hours. Many of the erythrocytes in the sample stored for 48 hours prior to staining had glassy, deeply basophilic cytoplasm, which precluded reticulocyte evaluation; the sample quality was deemed unacceptable for reticulocyte counting.

Comparison of polychromatophil and reticulocyte percentages—On the basis of data collected from examination of Wright-Giemsa– and NMB-stained blood smears from 31 birds by the 2 observers, inter- and intraobserver variations were first calculated for ring-form reticulocyte types. Calculated F (test statistic for ratio of mean square numbers) values were low, and P values were > 0.05 for all comparisons (for interobserver variation, F = 0.072 and P = 0.790; for intraobserver variation for observer 1, F = 0.404 and P = 0.530; and for intraobserver variation for observer 2, F = 0.083 and P = 0.775). Because these results indicated a lack of significant interobserver or intraobserver variation, mean differential percentages for the 2 observers were calculated and used for further comparisons.

Ten comparisons were assessed by use of Spearman rank correlation coefficient (Table 1). Correlation between adjusted polychromatophil percentage and the percentage of ring-form reticulocytes was strongest (ρ = 0.978; P < 0.001). Correlation between the combined percentage of polychromatophils and IECs and the percentage of ring-form reticulocytes was only slightly less strong (ρ = 0.976; P < 0.001). These 2 comparisons were plotted, and regression line equations were obtained (Figure 3). Correlation coefficients were substantially lower for the other 8 comparisons. The Deming regression model was used to compare data generated separately by each observer for polychromatophil percentage with data generated for ring-form reticulocyte percentage, and CVs were calculated for both measurement techniques. The CV for polychromatophil quantitation was 19.44%, whereas the CV for ring-form reticulocyte quantitation was 8.43%. The 95% confidence intervals were −0.017 to 0.003 for regression line intercept and 0.910 to 1.160 for regression line slope.

View Full Size

Scatterplot of adjusted percentage of polychromatophils (solid line) and combined percentage of polychromatophils and IECs (dashed line) versus percentage of ring-form reticulocytes determined in samples of blood collected from 31 birds. Datum points for polychromatophils only (diamonds) and for combined percentage of polychromatophils and IECs are shown. Direct smears were prepared from samples that were stained with NMB stain in a 1:3 dilution (blood to stain) and incubated at room temperature (approx 22°C) for 10 minutes. Wright-Giemsa–stained smears were also available from routine CBCs performed for each bird. Two observers each performed differential counts on 1,000 cells on the Wright-Giemsa– and the NMBstained smears. Regression line equations are displayed adjacent to their respective regression line. Notice that regression lines deviate from the line of identity (dotted line) at the higher end of the data range.

Citation: American Journal of Veterinary Research 69, 8; 10.2460/ajvr.69.8.1067

• Download Figure
• Download figure as PowerPoint slide

Generation of a reference interval for polychromatophils in healthy psittacines—On review of clinic records, 44 birds were defined as healthy adult psittacines according to the defined criteria. Differential counts on 1,000 RBCs were performed on archived smears of blood from these birds. Data for both the percentage and absolute number of polychromatophils were normally distributed, and no outliers were identified; thus, the reference interval for each was defined as the mean ± 2 SD value. The reference interval for percentage polychromatophils was 0.60% to 8.00% (mean, 3.66%). The reference interval for absolute polychromatophil count was 5,668 to 227,643 cells/μL (mean, 131,204 cells/μL).

Discussion

In the present study, a method of avian reticulocyte counting that was accurately reproduced by different observers was investigated. The 4 categories of reticulocyte morphologic types were representative of classification schemes used by previous investigators.^4,6 In our study, these criteria were more clearly defined, and in routinely stained blood smears, reticulocyte classifications were correlated with percentage of polychromatophils. For the 2 observers involved in the study, intraobserver CV for ring-form reticulocytes was low (5.61%), which suggested that counting this specific type of reticulocyte may be associated with a greater degree of reproducibility than that associated with counting any of the other 3 reticulocyte types. The results of our comparisons further indicated that the ring-form reticulocyte percentage correlated strongly with the polychromatophil percentage. Counting of ring-form reticulocytes is therefore recommended as the best method for avian reticulocyte quantification, given both the strong correlation between ring-form reticulocytes and polychromatophils and the higher degree of precision in counting ring forms versus other types of reticulocytes.

A separate category of IECs was created to include erythroid cells with a morphologic appearance that was more immature than that of polychromatophils; these cells are considered analogous to nucleated RBCs in mammalian blood. On the basis of findings in supravitally stained avian bone marrow,⁷ the staining pattern of IECs should be similar to that of ring-form reticulocytes, and we considered the potential impact of inclusion or exclusion of these cells in our study via comparisons with an adjusted polychromatophil percentage (with exclusion of IECs) and with the combined percentage of polychromatophils and IECs. Both polychromatophil categories quantitatively correlated strongly with ring-form reticulocytes, with only a slight decrease in correlation strength when IECs were included. Immature erythroid cells comprised 0.0% to 5.0% of all RBCs in the samples included in our study, which was a wide range of values. The practical implication of this finding is that inclusion or exclusion of IECs in a polychromatophil count appears to have a minimal, and likely clinically unimportant, effect on the percentage of ringform reticulocytes.

In addition to percentages of each reticulocyte type, a combined percentage of aggregate and ringform reticulocytes was also compared with the percentage of polychromatophils. This fifth category was added because ring-form reticulocytes arguably represent a subgroup of aggregate reticulocytes, such that a category of what might be considered all aggregate reticulocytes could show the strongest correlation with polychromatophils in avian species, as is true in domestic cats.^11,12 However, strong correlation between the percentage of polychromatophils and the combined percentage of aggregate and ring-form reticulocytes was not detected in our study.

Lack of uniformity in counting and reporting of avian reticulocytes has been identified as a likely cause of variation in published reticulocyte counts in chickens.¹⁰ The data from the present study supported this theory. In the 1950s and 1960s, reported values for reticulocytes in clinically normal humans also varied markedly, and several authors attributed this to a lack of uniform criteria for identification of reticulocytes.¹³ Standardization of a uniform reticulocyte-counting technique should reduce variation in proposed reference intervals in avian species, as has been the case for mammalian species.

Supravital staining of avian bone marrow has revealed the presence of a perinuclear ring of reticulum in cells in the early stages of erythroid cell development.⁷ Avian erythrocytes do not extrude their nucleus during maturation as do mammalian erythrocytes, but a maturation stage similar to that of a late metarubricyte was believed to be indicated by dissolution of the perinuclear ring of reticulum into a diffuse distribution of granular reticulum throughout the cytoplasm.⁷ The results of our study suggest that this transformation from a perinuclear to a diffuse distribution of reticulum is more likely to be analogous to the maturation of a polychromatophil to a mature erythrocyte because the percentage of ring-form reticulocytes (ie, those with perinuclear granulation) quantitatively correlated much more strongly with the percentage of polychromatophils than did the percentage of aggregate reticulocytes (ie, those with diffusely distributed granulation) or the combined percentage of aggregate and ring-form reticulocytes.

In the present study, the optimal avian reticulocyte staining protocol involved a 1:3 dilution of whole blood to NMB stain with an incubation time of 10 minutes. Ratios of whole blood to supravital stain ranging from 1:1 to 1:3 have been used previously for staining of blood from both avian and mammalian species.^2,8,14 Both the type and concentration of the supravital stain used must be taken into account when comparing staining methods. The 1:3 dilution was selected as optimal in our study because cytoplasmic staining was better defined at this dilution than at lower dilutions.

Refrigerated storage of whole blood for as long as 24 hours minimally affected reticulocyte staining and counting in our study, although there was a low percentage (0.6%) of erythrocytes that were degenerated in appearance and could not be included in the differential count. After 48 hours of refrigerated storage, many cells were degenerated and deeply stained, and reticulocyte counting could not be performed. Therefore, our recommendation is to store avian whole blood in a refrigerator for no more than 24 hours prior to reticulocyte staining and counting.

Hematology reference intervals for healthy birds can be difficult to establish, in part because of population sample size and sample volume limitations, and in part because of the wide phylogenetic diversity of avian patients. Because of the strong correlation between percentages of polychromatophils and ring-form reticulocytes detected in the present study, which included blood samples from birds in 4 different orders, we proposed that a polychromatophil reference interval could be used as a substitute for a reticulocyte (ring-form) reference interval. The ability to assess erythrocyte regeneration from examination of routinely stained blood smears would save time and avoid the need for collection of an additional sample volume for a reticulocyte count. The polychromatophil percentage reference interval generated in our study (0.60% to 8.00%) was consistent with published estimates^1,15 of polychromasia in healthy birds. An absolute polychromatophil reference interval was also generated, but no published estimates of this interval could be found for comparison. Because of the proportionate bias between percentages of polychromatophils and ring-form reticulocytes, the polychromatophil reference interval should be used only as an estimate of, rather than as a replacement for, a reference interval for ring-form reticulocytes. In addition, interobserver CV in the present study was considerably higher for counts of polychromatophils than it was for counts of ring-form reticulocytes (19.44% and 8.43%, respectively), which indicated that there was a greater degree of random error, and therefore greater imprecision, associated with counting polychromatophils than counting ring-form reticulocytes. Observer detection of polychromatophilia is based on color alone and is therefore likely to be a more difficult and subjective assessment than detection of reticulocytes.¹⁶ Decreased interobserver agreement in blood smear evaluation has been correlated with color vision deficiencies, and a pattern-based system rather than a color-based system is clearly advantageous to a practitioner with color vision impairment.¹⁷ For these reasons, reticulocyte counting is the preferred laboratory method for evaluation of regenerative erythrocyte responses in birds, and reference intervals for ring-form reticulocytes should be developed by laboratories in which avian reticulocyte counts are performed.