(P-BT-30) Validation of Hematocrit Measurement for HPC-A Cell Products: Comparative Analysis of Automated vs. Manual Techniques

Background/Case Studies: Hematocrit (HCT) is a key parameter in cellular therapy, particularly for evaluating the quality, yield, and safety of hematopoietic progenitor cell apheresis (HPC-A) products. Accurate HCT assessment supports both clinical decision-making and compliance with regulatory guidelines, which emphasize thorough characterization of cellular therapy products. While automated hematology analyzers are widely used in clinical laboratories, their use for low-HCT cellular therapy samples has not been fully validated. This study evaluates the agreement between automated and manual HCT measurement methods to assess their reliability and interchangeability in HPC-A product testing.

Study

Design/Methods:

This systematic validation study compared HCT measurements using automated (Sysmex XN-9100) and manual (StatSpin) methods across a range of sample types. HCT levels were assessed in both undiluted and diluted specimens, with white blood cell (WBC) counts serving as internal controls to verify dilution accuracy. The study was conducted in three phases:

1. Ten sets of Sysmex XN Check Control samples (Normal, Abnormal 1, Abnormal 2) were tested undiluted and diluted.


2. 22 peripheral blood (PB) samples were similarly analyzed.


3. 25 HPC-A samples were tested using both methods.

Statistical tools included correlation analysis, Bland-Altman plots, linear regression, and error metrics (MAE, RMSE).



Results/Findings: QC: Sysmex showed better precision in undiluted controls (MAE: 0.58 vs. 1.24; RMSE: 0.71 vs. 1.61), but StatSpin aligned more closely with expected values at low HCT ranges (R²: 0.88 vs. 0.82).

PB: Methods strongly correlated in undiluted samples (r = 0.968), but less so in diluted ones (r = 0.761). Bland-Altman analysis showed minimal bias (0.05) with narrow limits of agreement [-0.74, 0.84].

HPC-A: Moderate correlation (r = 0.56, p = 0.0023) and systematic bias (+2.19, StatSpin higher). WBC concentration was the main predictor of method discrepancy (R² = 0.845, p < 0.001), and WBC moderately correlated with HCT differences (r = 0.46). HPC-A HCT (StatSpin) values showed strong correlation with WBC, indicating WBC as a predictor of HCT levels (Figure 1).

Conclusions: Sysmex provides reliable HCT measurements across standard ranges but shows limitations in low HCT zones typical of HPC-A products. StatSpin remains the preferred method for critical low HCT measurements but exhibits systematic overestimation in HPC-A samples due to WBC interference. The strong predictive relationship between WBC concentration and method discrepancy enables development of correction models for improved clinical accuracy. Method-specific calibration should be considered when using these techniques interchangeably in cell therapy applications.