Comparison of automated versus manual neutrophil counts for the detection of cellular abnormalities in dogs receiving chemotherapy: 50 cases (May to June 2008)

Michelle C. Cora Department of Population Health and Pathobiology, College of Veterinary Medicine, North Carolina State University, Raleigh, NC 27606.

Search for other papers by Michelle C. Cora in
Current site
Google Scholar
PubMed
Close
 DVM, DACVP
,
Jennifer A. Neel Department of Population Health and Pathobiology, College of Veterinary Medicine, North Carolina State University, Raleigh, NC 27606.

Search for other papers by Jennifer A. Neel in
Current site
Google Scholar
PubMed
Close
 DVM, DACVP
,
Carol B. Grindem Department of Population Health and Pathobiology, College of Veterinary Medicine, North Carolina State University, Raleigh, NC 27606.

Search for other papers by Carol B. Grindem in
Current site
Google Scholar
PubMed
Close
 DVM, PhD, DACVP
,
Grace E. Kissling Biostatistics Branch, National Institute of Environmental Health Sciences, National Institutes of Health, 111 TW Alexander Dr, PO Box 12233, Research Triangle Park, NC 27709.

Search for other papers by Grace E. Kissling in
Current site
Google Scholar
PubMed
Close
 PhD
, and
Paul R. Hess Department of Clinical Sciences, College of Veterinary Medicine, North Carolina State University, Raleigh, NC 27606.

Search for other papers by Paul R. Hess in
Current site
Google Scholar
PubMed
Close
 DVM, PhD, DACVIM

Abstract

Objective—To determine the frequency of clinically relevant abnormalities missed by failure to perform a blood smear evaluation in a specific subset of dogs receiving chemotherapy and to compare automated and manual neutrophil counts in the same population

Design—Retrospective case series

Animals—50 dogs receiving chemotherapy with a total nucleated cell count > 4,000 nucleated cells/μL.

Procedures—50 blood smears were evaluated for abnormalities that have strong potential to change the medical plan for a patient: presence of blast cells, band neutrophils, nucleated RBCs, toxic change, hemoparasites, schistocytes, and spherocytes. Automated and manual neutrophil counts were compared.

Results—Blood smears from 10 (20%) patients had ≥ 1 abnormalities. Blast cells were identified on 4 (8%) blood smears, increased nucleated RBCs were identified on 5 (10%), and very mild toxic change was identified on 2 (4%). Correlation coefficient of the neutrophil counts was 0.96. Analysis revealed a slight bias between the automated and manual neutrophil counts (mean ± SD difference, −0.43 × 103/μL ± 1.10 × 103/μL)

Conclusions and Clinical Relevance—In this series of patients, neutrophil count correlation was very good. Clinically relevant abnormalities were found on 20% of the blood smears. An automated CBC appears to be accurate for neutrophil counts, but a microscopic examination of the corresponding blood smear is still recommended; further studies are needed to determine whether the detection or frequency of these abnormalities would differ dependent on chemotherapy protocol, neoplastic disease, and decision thresholds used by the oncologist in the ordering of a CBC without a blood smear evaluation.

Abstract

Objective—To determine the frequency of clinically relevant abnormalities missed by failure to perform a blood smear evaluation in a specific subset of dogs receiving chemotherapy and to compare automated and manual neutrophil counts in the same population

Design—Retrospective case series

Animals—50 dogs receiving chemotherapy with a total nucleated cell count > 4,000 nucleated cells/μL.

Procedures—50 blood smears were evaluated for abnormalities that have strong potential to change the medical plan for a patient: presence of blast cells, band neutrophils, nucleated RBCs, toxic change, hemoparasites, schistocytes, and spherocytes. Automated and manual neutrophil counts were compared.

Results—Blood smears from 10 (20%) patients had ≥ 1 abnormalities. Blast cells were identified on 4 (8%) blood smears, increased nucleated RBCs were identified on 5 (10%), and very mild toxic change was identified on 2 (4%). Correlation coefficient of the neutrophil counts was 0.96. Analysis revealed a slight bias between the automated and manual neutrophil counts (mean ± SD difference, −0.43 × 103/μL ± 1.10 × 103/μL)

Conclusions and Clinical Relevance—In this series of patients, neutrophil count correlation was very good. Clinically relevant abnormalities were found on 20% of the blood smears. An automated CBC appears to be accurate for neutrophil counts, but a microscopic examination of the corresponding blood smear is still recommended; further studies are needed to determine whether the detection or frequency of these abnormalities would differ dependent on chemotherapy protocol, neoplastic disease, and decision thresholds used by the oncologist in the ordering of a CBC without a blood smear evaluation.

Dogs receiving chemotherapeutic drugs frequently develop transient cytopenias, most notably neutropenia and thrombocytopenia, because cells with the shortest circulating life span are most susceptible to the myelosuppressive effects of chemotherapy.1 Neutropenic patients have an increased risk of fever and sepsis, and patients with thrombocytopenia are at risk of hemorrhaging. As such, treatment modifications are often made when these cytopenias are detected, including the administration of prophylactic antimicrobials or granulocyte colony-stimulating factor or delayed administration or dose modification of chemotherapeutic drugs. To screen for these cytopenias, it is preferred that a patient's CBC be assessed the same day as scheduled chemotherapy.1 Ideally, microscopic examination of a stained blood smear is part of a CBC (ie, full CBC) and is most important in very sick patients or those with abnormal concentrations of leukocytes, erythrocytes, or platelets.2 Examination of the blood smear allows for confirmation of the automated cell counts and screening for potentially clinically relevant abnormalities, such as band neutrophils, toxic changes, hemoparasites, or nucleated RBCs. Preparation and examination of the blood smear does require extra test time, increasing the turnaround time for delivery of CBC results to the oncologist.

At the North Carolina State University Veterinary Teaching Hospital, the current acceptable standard of practice for stable oncology patients receiving chemotherapy is the option for the clinician to request an automated TNCC and differential without a blood smear evaluation when the TNCC is > 4,000 nucleated cells/μL (chemotherapy CBC) with the understanding that this test is to be used only in patients with recent baseline CBC results and no clinical signs of disease upon examination. This allows for expedited CBC results for the purpose of screening the patient for cytopenias on the same day as scheduled chemotherapy and alleviates some of the client cost accrued over the course of their animal's cancer treatment. Decreasing CBC turnaround time can be very helpful because there is limited time for the ordering and processing of the chemotherapeutic agents from the on-site pharmacy as well as for the administration of these drugs. With manual differential cell count as the gold standard, the 4,000 nucleated cells/μL TNCC cutoff value was established internally within the North Carolina State University Veterinary Teaching Hospital clinical pathology laboratory on the basis of a large population of chemotherapy patients. It was determined that in patients with an automated TNCC > 4,000 nucleated cells/μL, the automated neutrophil counts correlated well with manual neutrophil counts (unpublished data) and that these patients were uncommonly neutropenic (< 2,000 neutrophils/μL). Given these data and the well-established acceptability of use of an automated hematology analyzera for analysis of routine canine CBCs,3–5 the chemotherapy CBC was deemed an acceptable test in these particular patients. As part of the chemotherapy CBC, stained blood smears are made and saved in the event that a blood smear evaluation is needed or requested by the clinical pathologist or oncologist.

Although the differential cell count generated by an automated hematology analyzer is acceptable at the North Carolina State University Veterinary Teaching Hospital for prechemotherapy screening of cell counts, the potential exists for missing clinically relevant abnormalities by not performing a blood smear evaluation. The purpose of the study reported here was to determine the frequency of clinically relevant abnormalities missed by lack of performing a blood smear evaluation in dogs receiving chemotherapy with a TNCC > 4,000 nucleated cells/μL. In addition, because the blood smears were already being examined for abnormalities, the automated neutrophil counts were compared with the manual neutrophil counts (considered the gold standard) to determine whether the automated neutrophil counts were acceptable in this subset of dogs that are inherently medically compromised by their disease and chemotherapeutic treatments.

Materials and Methods

Patient selection—Starting the first day of May 2008, canine chemotherapy CBCs (ie, an automated TNCC and differential without a blood smear evaluation performed when the TNCC is > 4,000 nucleated cells/μL; performed the morning on the day of a patient's scheduled chemotherapeutic treatment) were retrospectively chosen in consecutive order by searching the laboratory records of the North Carolina State University Veterinary Teaching Hospital until 50 were identified, followed by a fixed size sequential sample selection, which is a typical design for recruiting patients into a clinical study.6 The use of samples from 50 animals permitted detection of abnormality rates of ≥ 3% at a 0.05 level of significance with 80% power. The last or fiftieth patient was identified on June 10, 2008.

When a chemotherapy CBC was performed on a patient more than once before the 50 total samples were identified, only the first chemotherapy CBC on that patient was used for the study. Blood smears that were made and stained (but not evaluated) on the day of the chemotherapy CBC blood sample collection were retrieved from the archives. Blood samples were obtained as part of a regularly scheduled hematologic analysis related to the patients' chemotherapy treatments.

Sample analysis—The TNCC and neutrophil counts from the 50 identified chemotherapy CBCs were generated by an automated hematology analyzera the day of the blood sample collection. A 100-WBC differential count was performed on the corresponding retrieved blood smears within the monolayer with a 50× objective. Nucleated erythrocytes observed per 100 WBCs were recorded. If present, blast cells and band neutrophils were counted as part of the 100-WBC differential count. A manual neutrophil count was calculated by multiplying the neutrophil percentage from the differential count by the automated TNCC.

Under 10× objective magnification, the feathered edge, monolayer, and body of the smear were scanned for abnormalities (eg, microfilaria). Erythrocyte and leukocyte morphology were examined under 100× objective magnification. Platelet concentration and clumps were noted. The following were considered clinically relevant abnormalities: presence of blast cells or band neutrophils (any number), toxic change graded as previously described,7 > 4 nucleated RBCs/100 WBCs, hemoparasites, spherocytes (> 4/100× objective field), and schistocytes (> 4/100× objective field). Blast cells were defined as large cells with fine stippled chromatin, moderate to deeply basophilic cytoplasm, and at least 1 visible nucleolus. Cell lineage (eg, lymphoid or myeloid) was recorded when cell morphology was supportive of such. Additional diagnostic tests (cytochemical stains or cluster of differentiation markers) were not done. Spherocytes and schistocytes were chosen as clinically relevant morphological changes for this study because these particular morphologies have been found to strongly correlate with specific severe diseases (eg, hemangiosarcoma and immune-mediated hemolytic anemia)8 and therefore are more likely to result in a potential change of the medical plan (eg, further diagnostic testing or change in treatment). The 50 blood smear differential counts and evaluations were performed by 1 author (MCC), a clinical pathology resident, under the advisement of 2 board-certified clinical pathologists (JAN and CBG).

Statistical analysis—A statistical software packageb was used for calculations. Normality of data was assessed via the Shapiro-Wilk method. For statistical analysis, neither automated nor manual neutrophil counts were corrected for ≤ 4 nucleated RBCs/100-WBC count. Correlations between the automated neutrophil counts and manual neutrophil counts (used as the gold standard) were performed with the Spearman method. The r values were categorized as excellent (≥ 0.93), good (0.92 to 0.80), fair (0.79 to 0.60), or poor (< 0.59).9,10 Comparability of results was assessed by comparison of the mean values via the Wilcoxon matched-pair test. Bland-Altman plots were used to assess agreement between the 2 methods.11

Results

Animals—Dogs were receiving ≥ 1 of the following: vincristine, cyclophosphamide, prednisone, vinblastine, l-asparaginase, mechlorethamine, carboplatin, lomustine, doxorubicin, and mitoxantrone. Dogs had ≥ 1 of the following: lymphoma (n = 26), mast cell tumor (6), osteosarcoma (5), hemangiosarcoma (4), oral malignant melanoma (2), soft tissue sarcoma (2), mammary carcinoma (2), apocrine gland adenocarcinoma (2), transitional cell carcinoma (1), thyroid carcinoma (1), gastrinoma (1), multiple myeloma (1), and histiocytic sarcoma (1); 4 dogs had 2 neoplasms.

Blood smear findings—Ten (20%) dogs had ≥ 1 clinically relevant abnormality. Blast cells were identified on 4 (8%) blood smears. Increased numbers of nucleated RBCs were detected on 5 (10%) blood smears. To illustrate the extent of change that would have occurred if knowledge of the nucleated RBCs was available at the time of the original reported automated CBC, a corrected WBC count was calculated ([TNCC × 100]/[100 + nucleated RBCs]) and compared with the automated count; a mean percentage change of −7.2% was found (Table 1). The presence of nucleated RBCs did not unmask a neutropenia (data not shown). Toxic change was identified on 2 (4%) blood smears and was graded as very mild in severity in both instances. Band neutrophils, schistocytes, spherocytes, and hemoparasites were not observed in any blood smear. No demonstrable platelet clumps were identified. All patients with thrombocytopenia diagnosed from the automated count also had low estimated platelet concentrations on the corresponding blood smears (data not shown).

Table 1—

Comparison of the absolute and percentage difference in the TNCC before and after correction for the number of nucleated RBCs/100 WBCs identified on blood smears from 5 dogs receiving chemotherapy with an automated TNCC > 4,000 nucleated cells/μL.

No. of nucleated RBCs/100 WBCsAutomated TNCC (× 103/μL)Corrected TNCC (× 103/μL)Difference (%)
56.245.944.8
55.054.85.0
65.214.925.6
926.5524.368.2
144.053.5512.3

Data obtained from 5 of 50 dogs receiving chemotherapy with an automated TNCC > 4,000 nucleated cells/μL.

Three blood smears had 1 blast cell identified/100 WBCs. These cells were similar in morphology. They were large, distinct round cells with small amounts of basophilic cytoplasm and a round or oval central nucleus with finely stippled chromatin and at least 1 nucleolus, consistent with immature hematopoietic cells. They did not appear erythroid in origin, but further characterization on the basis of morphology was not attempted because they did not possess convincing or definitive lymphoid or myeloid morphological features. An additional blood smear had 10 cells consistent with lymphoblasts. Specifically, these cells were round and medium to large (2 to 3 times the size of an erythrocyte) with small amounts of basophilic cytoplasm, a round to oblong to slightly indented central to eccentric nucleus with finely stippled chromatin, and 1 to 2 nuclear rings or nucleoli. Of the 4 slides, 2 were from patients being treated for lymphoma and 2 were from patients being treated for nonlymphoid neoplasms (osteosarcoma and high-grade soft tissue sarcoma with metastasis). The lymphoma patients were considered in strong partial or full remission on the basis of clinical findings. The slide with 10 blast cells/100 WBCs was from the patient considered in strong partial remission of stage IV-Va lymphoma and translated to a blast cell count of 990 blast cells/μL, with a WBC count of 9,850 WBCs/μL. A CBC with a blood smear evaluation 4 days prior showed no blast cells in circulation.

Neutrophil counts—There was excellent correlation (r = 0.96; 95% confidence interval, 0.94 to 0.98) over a wide range of neutrophil counts, including counts categorized as neutropenic (Figure 1). However, the median automated neutrophil count (5.17 × 103 neutrophils/μL) was significantly (P < 0.001) lower than the manual neutrophil count (5.52 × 103 neutrophils/μL). Bland-Altman analysis revealed a slight negative bias (mean ± SD difference) of −0.43 × 103/μL ± 1.10 × 103/μL between the automated neutrophil count and manual neutrophil count (Figure 2).

Figure 1—
Figure 1—

Scatterplot of neutrophil counts obtained by means of an automated hematology analyzer and manual method from 50 dogs receiving chemotherapy with an automated TNCC > 4,000 nucleated cells/μL.

Citation: Journal of the American Veterinary Medical Association 242, 11; 10.2460/javma.242.11.1539

Figure 2—
Figure 2—

Bland-Altman plot for neutrophil counts obtained by means of an automated hematology analyzer and manual method from 50 dogs receiving chemotherapy with an automated TNCC > 4,000 nucleated cells/μL. The dotted line is the bias (mean difference), and the dashed lines are 95% confidence intervals.

Citation: Journal of the American Veterinary Medical Association 242, 11; 10.2460/javma.242.11.1539

Discussion

In the present study, in which blood smears were evaluated for 50 dogs with a CBC TNCC of > 4,000 nucleated cells/μL receiving chemotherapy, clinically relevant abnormalities were found on 20% of the blood smears. These results highlight the potential to miss important information by not performing a blood smear evaluation as part of a CBC in patients with a TNCC > 4,000 nucleated cells/μL that are receiving chemotherapy. Blast cells were identified on 4 of the blood smears evaluated. Of these, 2 were in dogs with nonlymphoid neoplasms and were consistent with immature hematopoietic cells. In these animals, the presence of low numbers of these types of blast cells most likely represented bone marrow insult or mobilization of hematopoietic progenitor cells due to the chemotherapy treatments,12,13 rather than metastasis, and would not necessarily trigger a change in treatment or additional testing. Blast cells were identified on blood smears of 2 patients being treated for lymphoma. The presence of circulating blast cells in a dog with hematopoietic neoplasia, even in low numbers, would prompt reconsideration of remission status and, potentially, a change in the therapeutic plan or further diagnostic testing, such as the request for the blood smear to be reviewed by a board-certified clinical pathologist or additional tests to characterize or confirm the blast cell lineage.

It is not known whether the other clinically relevant findings identified in this study would have ultimately led to further diagnostic testing or a change in treatment; regardless, their presence does illustrate the potential abnormalities that may go undetected by not performing a blood smear evaluation. Toxic change of neutrophils is associated with conditions that include bacterial infections, septicemia, acute renal failure, immune-mediated hemolytic anemia, drug toxicoses, and neoplastic disorders.3,7 In the present study, 2 blood smears had toxic change categorized as very mild in severity, accompanied by leukocyte counts that were within reference intervals. In these clinically stable patients, this finding would most likely be interpreted as a result of the neoplastic disease or ongoing chemotherapy treatment. Similarly, the presence of nucleated RBCs in circulation may occur as a consequence of the myelotoxic effects of chemotherapy. With substantial metarubricytosis, leukocyte counts can be falsely elevated, potentially masking neutropenia. In this study, however, correction for the presence of elevated nucleated RBCs did not unmask a neutropenia.

Automated neutrophil counts had excellent correlation (r = 0.96) with the manual neutrophil counts. Although the median automated neutrophil count was significantly lower than the median manual neutrophil count, agreement was good with only a slight negative bias (−0.43 × 103/μL ± 1.10 × 103/μL). These results were not unexpected, given that the automated hematology analyzer used at the North Carolina State University Veterinary Teaching Hospital is considered highly accurate and well suited for blood cell analysis in dogs.4–6 Our data support the use of an automated CBC for the monitoring of neutrophil counts in these dogs. Adequate training on proper sample handling and operation of the hematology analyzer as well as knowledge of how to deal with flagged samples and interpret the generated CBC histograms is important in ensuring that accurate automated neutrophil counts are reported.

There are 2 major limitations to this study. One limitation relates to the fact the decision to request either a chemotherapy or full CBC for a given patient is made by the oncologist. Such choices likely influenced the results of this study. For example, full CBCs are routinely ordered when the erythron evaluation is expected to reveal clinically important abnormalities or provide information about disease status, such as in the case of dogs with leukemia or those with excised splenic hemangiosarcoma, where the appearance of schistocytes or increased nucleated RBCs could signal the development of gross metastases. The exclusion of these dogs from the chemotherapy CBC population would have lowered the percentage of observed clinically relevant abnormalities. On the other hand, because our samples were collected at an arbitrarily designated point, rather than at a particular treatment in a chemotherapy protocol, some of the dogs had previously undergone ≥ 1 erythron evaluation by full CBC. In these cases, chemotherapy CBCs may have been ordered despite the high likelihood of finding clinically relevant abnormalities, given that the treatment plan was not dependent on re-identification or quantification of such findings. Inclusion of such patients potentially would have led to an overestimation of the percentage of case animals with abnormalities. Overall, the net effect of these opposing factors was presumably minor, particularly considering that the same clinical criteria to decide between a chemotherapy and full CBC will likely be in general use.

The other limitation relates to the diversity of chemotherapeutic drugs and neoplastic disease in the study population. This study was a preliminary attempt to determine what, if any, clinically relevant hematologic abnormalities were being missed. It is clear from the results of this study that abnormalities may be missed, and further investigations with more defined study populations and criteria are needed.

On the basis of our results, neutrophil counts appear to be accurate from the automated hematology analyzer used at the North Carolina State University Veterinary Teaching Hospital in this subset of dogs receiving chemotherapy. Clinically relevant abnormalities were found in 20% of the blood smears from this population of dogs, including the presence of blast cells in lymphoma patients. As such, this study revealed that there is the potential to fail to identify potentially important hematologic abnormalities when a blood smear evaluation is not performed. Although an automated CBC can be used to check neutrophil counts, on the basis of the results of this study, a microscopic examination of the corresponding blood smear to check for clinically relevant abnormalities is still recommended; further studies are needed to determine whether the detection or frequency of these abnormalities may differ depending on chemotherapy protocol, neoplastic disease, and the decision thresholds used by the oncologist in the ordering of this test.

ABBREVIATION

TNCC

Total nucleated cell count

a.

Advia 120, Siemens Healthcare Diagnostics, Deerfield, Ill.

b.

Prism, version 5.0, GraphPad Software Inc, San Diego, Calif.

References

  • 1. Chun R, Garret LD, Vail DM. Cancer chemotherapy. In: Withrow SJ, Vail DM, eds. Withrow and MacEwen's small animal clinical oncology. 4th ed. Philadelphia: WB Saunders Co, 2007;163192.

    • Search Google Scholar
    • Export Citation
  • 2. Stockham SL, Scott MA. Fundamentals of veterinary clinical pathology. 2nd ed. Ames, Iowa: Blackwell Publishing Professional, 2008;53106.

    • Search Google Scholar
    • Export Citation
  • 3. Moritz A, Fickenscher Y, Meyer K, et al. Canine and feline hematology reference values for the ADVIA 120 hematology system. Vet Clin Pathol 2004; 33: 3238.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4. Welles EG, Hall AS, Carpenter DM. Canine complete blood counts: a comparison of four in-office instruments with the ADVIA 120 and manual differential counts. Vet Clin Pathol 2009; 38: 2029.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5. Becker M, Mortiz A, Giger U. Comparative clinical study of canine and feline total blood cell count results with seven in-clinic and two commercial laboratory hematology analyzers. Vet Clin Pathol 2008; 37: 373384.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 6. Hulley SB, Cummings SR, Browner WS, et al. Designing clinical research. 3rd ed. Philadelphia: Lippincott Williams & Wilkins, 2007;2736.

    • Search Google Scholar
    • Export Citation
  • 7. Aroch I, Klement E, Segev G. Clinical, biochemical, and hematological characteristics, disease prevalence, and prognosis of dogs presenting with neutrophil cytoplasmic toxicity. J Vet Intern Med 2005; 19: 6473.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8. Stockham SL, Scott MA. Fundamentals of veterinary clinical pathology. 2nd ed. Ames, Iowa: Blackwell Publishing Professional, 2008;107221.

    • Search Google Scholar
    • Export Citation
  • 9. Tvedten H, Weiss D. The complete blood count and bone marrow examination: general comments and selected techniques. In: Willard MD, Tvedten H, Turnwald GH, eds. Small animal clinical diagnosis by laboratory methods. 3rd ed. Philadelphia: WB Saunders Co, 1999;1130.

    • Search Google Scholar
    • Export Citation
  • 10. Lara-Garcia A, Hosoya K, Iazbik C, et al. Evaluation of a point-of-care hematology analyzer for use in dogs and cats receiving chemotherapeutic treatment. J Am Vet Med Assoc 2008; 232: 14881495.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307310.

    • Search Google Scholar
    • Export Citation
  • 12. Kessinger A, Sharp JG. The whys and hows of hematopoietic progenitor and stem cell mobilization. Bone Marrow Transplant 2003; 31: 319329.

  • 13. Lapidot T, Petit I. Current understanding of stem cell mobilization: the roles of chemokines, proteolytic enzymes, adhesion molecules, cytokines and stromal cells. Exp Hematol 2002; 30: 973981.

    • Crossref
    • Search Google Scholar
    • Export Citation
All Time Past Year Past 30 Days
Abstract Views 105 0 0
Full Text Views 873 798 12
PDF Downloads 180 103 5
Advertisement