Reply to Kaestner et al.: Pioneering quantitative platforms for stored red blood cell assessment open the door for precision transfusion medicine

In a recent Perspective, we emphasized the lack of translatable tools to objectively assess the quality of stored red blood cells (RBC) for “real age” determination of stored units and rational donor-to-recipient selection prior to transfusion (1). We envisioned that lab-on-a-chip (LOC) technologies hold the potential for quantitative assessment of stored RBC quality using multidimensional quality metrics identified by -omics and machine learning. The integration of these technologies is expected to lead to precision transfusion medicine by addressing storage lesions and subsequent post-transfusion complications. We are pleased that our Perspective was received as visionary and compulsory by Kaestner et al. (2) whose efforts in the field warrant acknowledgment. Specifically, they developed a point-of-care device, Erysense, for the evaluation of flow properties of stored RBCs (3), which is further along in development than most LOC platforms we discussed. The microfluidic geometry in this device addresses clogging issues and avoids the need for complicated fabrication and flow instrumentation. The device quantifies impaired deformability of stored RBCs at various flow velocities and classifies morphological abnormalities through machine learning.

This device, together with other platforms to assess RBC deformability and morphology (4, 5), forms a strong foundation toward precision transfusion medicine. Still, despite the successes of these pioneering examples, several limitations and challenges lay ahead. First, storage lesions involve a list of multifaceted metabolic, biochemical, and morphological injuries (6). The accumulation of RBC subpopulations with irreversibly altered morphologies, while potentially a relevant proxy for in vivo performance (7), does not offer specific details on the molecular characteristics of the unit or inform unit-specific processing strategies. Thus, extensive preclinical and clinical validation is required to establish that impaired deformability sufficiently and universally reflects the real age of stored RBC units. Otherwise, a multidimensional quality index approach, which integrates the best quality predictors onto a single LOC platform, might be imperative. Second, the overall workflow of these platforms should be automated for successful translation to large-scale clinical adoption. The automation will ensure that all RBC units are sampled in a standardized high-throughput manner rather than via manual and error-prone labor. Also, the current assessment platforms use diluted samples (0.5% hematocrit for Erysense) while stored RBC units are packed (55 to 85% hematocrit). Dilution and washing should be performed on-chip to eliminate potential errors by manual handling. Eventually, the automation of operation, sampling, and processing is needed to eliminate the contamination risk, which is essential for safe transfusion. These limitations and challenges should be addressed in the next generations of these platforms.

Finally, we agree that scalable strategies for in vitro erythropoiesis are scientifically and clinically promising (8). Until these efforts evolve and find their way into the transfusion workflow and infrastructure, relatively easier steps can be taken. For example, more of the eligible population can be motivated to donate to address recurrent blood shortages; alternative storage practices can be considered to address blood wastage and storage lesions (9); and quantitative assessment of RBC quality can be performed to address the ethical dilemma and clinical consequences of donor-to-recipient selection.