M. Fournier et al., LABORATORY EVALUATION OF THE ABBOTT CELL DYN-3500 5-PART DIFFERENTIAL, American journal of clinical pathology, 105(3), 1996, pp. 286-292
The study reports the performance of the Abbott CD3500 automated hemat
ology analyzer for the enumeration and delineation of leukocyte popula
tions for both adult and pediatric samples, and the ability of this in
strument to detect the presence of abnormal cells, Samples from 542 in
dividual patients either attending medical practitioners or during hos
pitalization were examined and then subdivided for the purposes of thi
s study into 106 samples from newborn infants (<15 days), 145 samples
from older children (15 days to 14 years) with non-oncologic disorders
, 100 samples from normal adults, and 191 samples from oncology patien
ts (97 adults and 94 children). The leukocyte differentials provided b
y both the Abbott CD3500 and the Coulter STKS were compared with those
obtained from conventional morphology (two observers, total of 100 le
ukocytes). The sensitivities and specificities of the blast, immature
granulocyte (IG) and NRBC ''flags'' were also determined. For the non-
oncology adult (n = 100) and pediatric(n = 145) cohorts, automated dif
ferentials Here given in all samples with the CD3500, whereas the STKS
did not provide a differential analysis for 20 of the 145 (14%) pedia
tric samples, 11 of these were absent for no obvious reason, However,
for the evaluable cases, the performances of the CD3500 and the STKS w
ere broadly similar and generally correlated well with the manual refe
rence procedure. The results for the newborn samples were less consist
ent with wider 95% confidence intervals (CI) noted, For example, the C
D3500 (which reported a differential for all 106 samples studied) gave
CI values of +/-14.4% for neutrophils, +/-14.6% for lymphocytes and /-8.1% for monocytes. For comparison, the STKS (which did not provide
a differential in 15% of 79 samples analyzed; insufficient material be
ing available from the remaining 27 of 106 newborn samples) gave Ct va
lues of +/-21.9% for neutrophils, +/-23.5% for lymphocytes and +/-8.2%
for monocytes. For all samples, the sensitivity of the blast flag on
the CD3500 was 85% with a specificity of 91% (STKS: sensitivity, 75%;
specificity, 85%); the sensitivity of the CD3500 IC flag was 72% with
a specificity of 76% (STKS: sensitivity, 75%: specificity, 73%); and t
he sensitivity of the NRBC flag was 43% with a specificity of 94% (STK
S sensitivity, 37%; specificity, 88%). This study confirms competitive
performance levels for the CD3500 in the analysis of normal adult sam
ples and suggests positive performance advantages in the study of neon
atal, pediatric, and leukopenic samples.