To and fro on band count reporting and clinical utility
| To and fro on band count reporting and clinical utility |
|
|
|
November 2010 Published last March in CAP TODAY was an article, “For WBC differentials, report in absolute numbers,” by Joan E. Etzell, MD. Reader Eddie Ferguson, MT(ASCP), wrote soon after to say he “loved Dr. Etzell’s article,” noting that she made “valid points about the need to report the absolute numbers” and provided good case studies, and posing a few followup questions for Dr. Etzell. Here in CAP TODAY this month, for the benefit of all interested readers, are Ferguson’s questions and Dr. Etzell’s answers. Dr. Etzell, a member of the CAP Hematology and Clinical Microscopy Resource Committee, is director of the clinical hematology laboratory, University of California, San Francisco. Ferguson is senior hematology applications specialist, Beckman Coulter, Birmingham, Ala. Ferguson: I would like to know your thoughts on band neutrophils. The CAP has been on the leading edge for years in trying to persuade doctors and laboratories to include band neutrophils with segmented neutrophils. You made this point when you said in your article, “The functional ANC [absolute neutrophil count] includes the absolute number of segmented neutrophils and bands.” In your case studies, you have a separate category for immature granulocytes. What types of neutrophils are you including in this category? Metas? Myelos? Promyelos? CAP TODAY published in 1992 a Q&A that provided the reasons bands should not be reported separately.1 Dr. Etzell: The CAP TODAY Q&A on the significance of bands by Thomas F. Dutcher, MD, published in 1992, argues that bands should not be reported separately because: 1. “According to the generally accepted calculations of Rumke”2... “when a perfect technologist examines a perfect smear in a random search pattern of 100 cells and reports 10 percent bands, the patient-care physician can have 95 percent confidence that the true band count is a number somewhere between two percent and 21 percent. Conversely, a reported percentage between two percent and 21 percent could reflect a true band count of 10 percent. Assuming that a ‘normal’ band count is from zero to 10 percent, many false-positive counts occur with observed bands between approximately 11 percent and 20 percent, whereas many false negatives occur with observed counts from about five percent to 10 percent. The true band count cannot be assumed to be greater than 10 percent until the observed percentage is greater than approximately 25 percent.” 2. “Conflicting reports have been presented as to whether there might be clinical significance if an accurate band count could be obtained.” 3. “Another variable in the production of band counts is the identification of these cells by individual technologists. The greater the number of technologists performing differentials in a laboratory, the greater the variability expected.” The CAP Hematology and Clinical Microscopy Resource Committee continues to recommend reporting bands with segmented neutrophils to represent the total absolute neutrophil count. We also agree that the precision of the band count is poor due to both limited counting sample (usually a 100- to 200-cell manual differential) as well as individual technologist variation, even when a uniform definition of a band is used. In addition, the Clinical and Laboratory Standards Institute Reference Leukocyte (WBC) Differential Count (Proportional) and Evaluation of Instrumental Methods; Approved Standard—Second Edition (2007)3 recommends inclusion of bands with segmented neutrophils and says “the inclusion of ‘bands’ as a separate category of leukocyte is not supported by medical data.” The term “immature granulocytes” is generally used to denote cells more immature than a band neutrophil. In my laboratory, when immature granulocytes are reported as part of the leukocyte differential, this category includes metamyelocytes and myelocytes. We report promyelocytes and blasts in separate categories to emphasize the degree of left shift. Other laboratories may include promyelocytes in the immature granulocyte category, but blasts are generally reported separately since their numbers are important for diagnosis and classification of various neoplasms, including the acute leukemias, myelodysplastic syndromes, and myeloproliferative neoplasms. Ferguson: In 1994, CAP TODAY published an article by P. Joanne Cornbleet, MD, PhD, and Robert W. Novak, MD,4 on the results of the CAP’s experiment with using a “uniform” definition of bands. The conclusion: “. . . lack of consistent identification as well as the poor precision of band counts in a 100-cell differential may be significant factors in limiting the clinical utility of band enumeration. Individual laboratories may wish to review within-laboratory reproducibility of band counts in deciding whether combining segmented and band neutrophils is warranted.” Then we have the pediatricians and neonatologists, bless their hearts. A 2002 article in Pediatrics on this question cautions pediatricians/ neonatologists to know how their particular laboratory classifies segmented and band neutrophils.5 It looks as if the pediatricians /neonatologists are slowly backpedaling on the specific use of bands as a diagnostic criterion. This particular article references an “absolute band count.” Dr. Etzell: In addition to the publications you have mentioned, Dr. Cornbleet’s 2002 article titled “Clinical utility of the band count” in Clinics in Laboratory Medicine is of note.6 It provides a comprehensive review of studies in which the band count was investigated for its clinical utility, in particular with regard to serious bacterial infection. In this article, Dr. Cornbleet reiterates the imprecision, lack of reproducibility among laboratories, and nonspecificity of the band count. Reference ranges for band counts vary widely in the literature, with upper limits ranging from five percent to 20 percent in adults, further complicating the clinical interpretation of a band count. Most physicians accept that there is little independent clinical utility of a band count in patients older than three months. In adults, multivariate analyses fail to identify bands as an independent predictor from neutrophil enumeration alone. Studies on the band count in infants/children from three months to 36 months of age show poor sensitivity and specificity as a screening test for occult bacteremia. The likelihood ratios in this patient population are suboptimal, and in many studies band counts are not as sensitive as the white blood cell or absolute neutrophil count in identifying occult infection in these children. The patient population for which clinicians have been most persistent in requiring a band count is infants under three months of age; they argue that these patients are less likely to develop typical symptoms and leukocytosis/neutrophilia in response to infection. Criteria used to stratify infants’ risk for infection use multiple clinical and laboratory parameters, often including a band count of greater than 1.5∞109/L or a band-to-total neutrophil ratio of greater than 0.2 (often referred to in the literature as an immature-to-total neutrophil [I/T] ratio) as evidence of infection. The absolute band count varies significantly over time in the neonatal period, and for these patients the I/T ratio is most often used. Unfortunately, the I/T ratio cutoff of 0.2 is somewhat arbitrary and overlaps the normal reference range in these patients, reducing the specificity of this parameter. Studies of absolute band count and I/T ratio in infants and neonates are inconsistent in their conclusions about the utility of these parameters for predicting infection. While some studies using multiple clinical and laboratory parameters, such as the Rochester criteria or Rodwell score, provide reasonable sensitivity for predicting infants with serious bacterial infections, these studies generally also use the overall WBC and neutrophil counts, and do not adequately delineate the contribution of the band count or I/T ratio as an independent predictor. Overall, the band count is imprecise, demonstrates suboptimal sensitivity and specificity, and has not been consistently demonstrated to provide information independent of the WBC and neutrophil count. The CAP Hematology and Clinical Microscopy Resource Committee does not support band counts for clinical decision-making because of the poor test performance; however, neonatologists and some pediatricians at our institutions still require testing for this purpose, largely because these patients may not develop obvious signs/symptoms and better, rapid tests may not be available. We do note, as you yourself referenced, publications in which pediatricians and neonatologists are recognizing the limitations of the band count. However, requests for band counts, especially in the neonatal and pediatric patient populations, are still common in the United States. A number of other markers are being investigated for their utility in neonatal sepsis, including acute-phase reactants (C-reactive protein, serum amyloid A, and procalcitonin), cytokines and chemokines (IL-6, IL-8, IL-10, IP-10, RANTES), and cell surface markers (CD64, CD11b).7 Of these, C-reactive protein is commonly used in neonates for diagnosing and monitoring sepsis and necrotizing enterocolitis. CRP has reasonable, but not optimal, sensitivity and specificity in neonates, but it is usually not elevated until six to eight hours after onset of symptoms. Other markers show promise, and perhaps they will be used clinically in this setting in the future. Ferguson: I have seen examples in which the WBC was 6.0–7.0 and the neutrophils were in the range of 60 to 70 percent and yet there was, by anyone’s definition, a significant percentage of bands but no ongoing infectious process. Could this be part of a human being’s fight-or-flight stress response? Dr. Etzell: It does appear that increased bands, and sometimes even a few metamyelocytes and myelocytes, can be seen in the circulation of patients undergoing disease-related stresses. As Dr. Cornbleet noted in her 2002 article, increased bands are not only seen with bacterial infections but also can be associated with viral infections and many other disorders, including various inflammatory processes (for example, collagen vascular disease, hypersensitivity reactions, gout), tissue damage/necrosis, neoplasia, seizures, intoxication, poisoning, metabolic abnormalities, hemolysis or acute hemorrhage, drugs, and myeloproliferative neoplasms. Conditions in infants that can result in increased bands in the absence of infection include meconium aspiration, asphyxia, seizures, periventricular hemorrhage, hypoglycemia, pneumothorax, and prolonged labor or oxytocin induction. References 1 Dutcher TF. Q&A, CAP TODAY, April 1992, 65–67. |
