An Evaluation of the Performance of Sysmex XE-2100 in Enumerating Nucleated Red Cells in Peripheral Blood

Archives of Pathology & Laboratory Medicine, Jul 2007 by Gulati, Gene, Behling, Eric, Kocher, William, Schwarting, Roland

* Context.-Automated methods of enumerating nucleated red blood cells (NRBCs) in blood are gaining acceptance in many laboratories.

Objective.-To evaluate the performance of Sysmex XE- 2100 in enumerating NRBCs in peripheral blood.

Design.-Automated relative number of NRBCs per 100 white blood cells (NRBC%) results for a total of 460 specimens run on the XE-2100 were compared with manual NRBC% results obtained by performing a 100-cell differential on aWright-stained smear prepared from each specimen. To assess within-day reproducibility, 64 specimens were rerun on the XE-2100, and a second 100-cell differential was performed on the original blood smear. Excel software was used for data analysis.

Results.-Regression analysis of automated NRBC% versus manual NRBC% yielded a correlation coefficient of 0.9712. No NRBCs were seen in the blood smear on 35 (15.1%) of 232 specimens with automated NRBC% of 0.1 to 1.9. The XE-2100 generated an NRBC% of 0.0 on 5 (6.8%) of 74 specimens, revealing 1 NRBC per 100 or more white blood cells by blood smear examination. The mean percent difference between duplicate automated results was 16.7 compared with 78.1 for the duplicate manual results. There were 9 instances in which the XE-2100 either did not detect the presence of more than 8 NRBCs per 100 white blood cells or generated an automated NRBC% of 18.1 or 18.8 when the smear revealed none. All of these were, however, flagged for smear review.

Conclusions.-Overall correlation between the automated and manual results was excellent. The automated method revealed better precision than the manual method. The number of specimens with false automated results was very small, and all were flagged for verification by a smear review.

(Arch Pathol Lab Med. 2007;131:1077-1083)

In humans, normoblasts or nucleated red blood cells (NRBCs), normally appear in the peripheral blood only during fetal life, and they may be seen at birth and in the first few days of postnatal life. The blood of a normal full-term newborn may reveal up to 10 NRBCs per 100 white blood cells (WBCs), and the blood of a fetus or a prematurely born baby usually contains a much greater number (up to 50 or more per 100 WBCs) depending on the gestational age. Normoblastemia is defined by the presence of a higher than normally expected number of NRBCs in the peripheral blood in this age group, or the presence of any NRBCs in the blood beyond the age of approximately 1 week after full-term normal birth. Normoblastemia is considered abnormal, and it is usually associated with either hypoxia or one or more clinical conditions, benign or malignant, that cause stress or damage to the bone marrow.

Until recently, in clinical laboratories NRBCs have been enumerated as part of the traditional 100-cell manual differential leukocyte count (DLC), and the findings have been reported as number of NRBCs per 100 WBCs, along with subsequent correction of the WBC count for their presence. Today, however, clinical laboratories equipped with modern hematology analyzers are embracing automated technology to detect, enumerate, and report NRBC counts. In some of these analyzers NRBC enumeration is performed as a routine part of the automated complete blood count (CBC) with DLC, whereas in others it is performed on specifically flagged specimens in a discrete mode of either CBC NRBC, or CBC with DLC NRBC. The results generated by the automated analyzers include both the relative number of NRBCs per 100 WBCs (NRBC%) and the absolute number of NRBCs per microliter of blood (NRBC#), along with the WBC count that has been automatically corrected for the presence of NRBCs. Clinicians and laboratory professionals are familiar with the clinical significance of the NRBC%, but the need and/or the benefit of additional reporting of the NRBC# remains unclear at this time due to lack of adequate pertinent studies. To the best of our knowledge many, if not all, laboratories equipped with such an automated analyzer continue to understandably report only the NRBC%.

To date, studies conducted to assess the performance of automated analyzers, specifically for NRBC counts, are limited in number, and individual studies have included a relatively small number of NRBC-containing specimens representing a wide range of NRBC values. The number of NRBC-containing specimens included in these studies ranged from 22 to 121, and the range of represented NRBC values was 0.1 to 900 per 100 WBCs.1-6 The published reports, although based on limited data, indicate a good overall correlation between the automated and manual NRBC results and reveal a comparatively better within- run precision for the automated method. None of the published studies, however, offers data adequate enough to answer without reservation a number of questions of interest and practical value to laboratory professionals in their daily practice. Such questions include the following: (1) Is the correlation between the automated and the manual method as good when the NRBC% values are in single digits as it is when the NRBC% values are in double or triple digits, or does the degree of correlation vary with the spans of NRBC values? (2) How much difference, if any, in the NRBC results generated by the two methods on a specimen can be expected and/or is acceptable, specifically in the low range of 0.1 to 10.0? (3) Is it possible, and if so, to what extent, that an automated analyzer may generate an NRBC% value when a blood smear examination does not reveal any NRBCs or vice versa? (4) What level of precision may be expected upon repeat NRBC analyses of a specimen by the automated method the same day (ie, within-day precision) compared with that obtained by performing repeat NRBC counts on the same smear by the manual method? (5) Are there instances when the automated analyzer may either miss flagging for the presence of a significant proportion of NRBCs or mis- flag for the presence of NRBCs, and if so, are there potential indicators to allow for appropriate resolution? Our current study was designed to answer all of these questions while assessing the performance of Sysmex XE-2100 (Sysmex America Inc, Mundelein, Ill), specifically in regard to NRBC enumeration, using a fairly large number of specimens representing a wide range of NRBC% values.