Hypersegmented Neutrophil Percentage Using Automated Digital Cell Morphology: A Simple Laboratory Parameter to Monitor Hydroxyurea Therapy in Sickle Cell Disease Patients

Fetal hemoglobin (HbF) induction with hydroxyurea (HU) remains the only specific pharmacological treatment for sickle cell disease (SCD) patients. Compliance is key to achieve high HbF levels, but assessing patient adherence to long-term treatment is difficult. Since patient interviews are often unreliable, most hematologists rely on mean corpuscular volume (MCV) and HbF levels to monitor compliance, but red blood cell indices change slowly. Hypersegmented neutrophils (HN) have been for long recognized in patients taking HU, and automated digital cell morphology platforms allow routine peripheral blood smears to be conveniently photographed and stored for analysis. We have designed a protocol to determine the percentage of circulating HN and have studied how this parameter compares with others commonly used in clinical practice during HU therapy, such as HbF, MCV, and reticulocyte count.

We collected blood samples from 38 patients with SCD in steady state, receiving an average dose of 20.5mg/kg/day of HU (range 10.2-33.3) at our Hematology outpatient clinic. Complete blood counts and HN percentage determinations were performed on a Sysmex XE-5000/Cellavision DM96 equipment. HbF levels were determined by HPLC (Bio-Rad). Statistical analysis was performed using GraphPad Prism 6.0 software. Standard protocol for image acquisition initially recorded 100 cells per smear and HN count was performed independently by two experienced laboratory personnel using the same image database, by counting cells with 5 or more nuclear segments. To increase precision due to the relatively low numbers of neutrophils in some samples, analysis was subsequently improved by acquiring 300 images per smear aiming to picture 100 neutrophils.

Mean hemoglobin level in our population was 8.6±1.3g/dL (mean±SD), MCV 103.8±14.2fL, reticulocyte count 222,950±129,090/uL, and HbF was 13.1±7.8% (range 2.1-30.9%). HN percentage as determined with 100 images per smear displayed only borderline correlation with HbF levels (P =0.094), but acquisition of 300 images per smear yielded over 90 neutrophil images in 35/38 samples. Average HN percentage was 14,5% (range 2,0-45,0%) and correlated positively with HbF (r=0.4172, P =0.009) and MCV (r=0.4301, P =0.007). As expected, HbF also correlated with MCV (r=0.5777, P =0.0001) and reticulocyte count (r=-0.489, P =0.003). Despite our limited number of patients, ROC curve analysis showed that HN percentage had an area under curve of 0.7241 to detect patients with HbF>20% (P =0.047). Patients with more than 6 segments per neutrophil were also more likely to have higher HbF. Average daily HU dose did not correlate with HbF or any of the parameters analyzed.

Lack of correlation between dose and HbF supports that medication adherence may indeed be suboptimal in the study population, but that may also be caused by individual differences in HU metabolism. While MCV still displayed the strongest correlation with HbF in our study, HN percentage performed similarly, with higher counts associated with higher HbF. Neutrophils have a much shorter half-life than erythrocytes, with an estimated half-life of hours rather than the 20 days calculated for sickle red blood cells. Therefore, digital cell morphology analysis enables clinical laboratories to determine HN counts that can change more quickly than HbF, MCV, or reticulocyte counts during HU therapy. Monitoring the number of HN may allow more timely assessment of compliance in patients starting HU or in those in need for HU in combination with sporadic blood transfusions, since neutrophil hypersegmentation should not be affected by changes in red blood cell mass. Further studies should investigate HN percentage as a potential surrogate marker of response to HU and of patient compliance.

Financial support: FAPESP