Toward Translational Impact of Low-Glucose Strategies on Red Blood Cell Storage Optimization

Abstract

Transfusion of stored red blood cells (RBCs) to patients is a critical component of human healthcare. Following purification from whole blood, RBCs are stored in one of many media known as additive solutions for up to 42 days. However, during the storage period, the RBCs undergo adverse chemical and physical changes that are often collectively known as the RBC storage lesion. Storage of RBCs in additive solutions modified to contain physiological levels of glucose, as opposed to hyperglycemic levels currently used in most cases, reduces certain markers of the storage lesion, although intermittent doses of glucose are required to maintain normoglycemic conditions. Here, we describe an electrically actuated valving system to dispense small volumes of glucose into 100 mL PVC storage bags containing packed RBCs from human donors. The RBCs were stored in a conventional additive solution (AS-1) or a normoglycemic version of AS-1 (AS-1N) and common markers of stored RBC health were measured at multiple time points throughout storage. The automated feeding device delivered precise and predictable volumes of concentrated glucose to maintain physiological glucose levels for up to 37 days. Hemolysis, lactate accumulation, and pH values of RBCs stored in AS-1N were statistically equivalent to values measured in AS-1, while significant reductions in osmotic fragility and intracellular sorbitol levels were measured in AS-1N. The reduction of osmotic fragility and oxidative stress markers in a closed system may lead to improved transfusion outcomes for an important procedure affecting millions of people each year.

1. Introduction

Red blood cells (RBCs) are the most commonly transfused blood component,¹ and according to the National Blood Collection and Utilization Survey, nearly 30,000 units of RBCs are transfused daily in the United States alone.² Fortunately, the procedure for obtaining RBCs is rather simple, beginning with collection of whole blood from a donor into an anticoagulant solution (e.g., citrate-phosphate-dextrose or CPD) and subsequently separated by centrifugation, transferred into a bag containing an additive solution (e.g., AS-1) and stored at 2–6 °C for up to 42 days.

Throughout storage, the RBCs undergo irreversible metabolic and physiological damages collectively known as the “storage lesion”.³⁻⁵ These damages include lipid/protein oxidation, diminished release of adenosine 5′-triphosphate (ATP) from the RBC, membrane damage, reduced nitric oxide (NO) capacity, decreased 2,3-DPG levels, and other functional impairments.³ Though there is debate in the literature, the damage to the RBC during storage may contribute to clinical issues after transfusion.³ These can include insufficient nitric oxide bioavailability, iron overload, reduced tissue perfusion, inflammation, immunomodulation, and many others.³ Due to this, attempts have been made to prevent or reduce the storage lesion including pH adjustments,⁶ modifications to additive solutions,^7,8 cryogenic preservation,^9,10 hypoxic storage,¹¹ and the use of rejuvenating solutions to recover functional capabilities of RBCs.^12,13 To date, nearly all approaches to improve stored RBCs have not considered the significantly high level of glucose in whole blood collection and RBC storage solutions. Most anticoagulants and additive solutions contain glucose at concentrations at least 8 times greater than physiological levels and 5 times greater than levels often measured in the bloodstream of people with diabetes, a patient group whose RBCs often have many of the same properties as those of stored RBCs.^7,14

Recently, we reported that the hyperglycemic environment of traditional RBC storage conditions contributes to the same chemical and physical changes that comprise the storage lesion. Storage of RBCs in a normoglycemic (4–6 mM glucose) version of AS-1 (which we have called AS-1N) results in RBCs that maintain physiological size, are not rigid, are capable of increased ATP release, have reduced oxidative stress, and reduced osmotic fragility in comparison to AS-1.¹⁵⁻¹⁸

However, the storage of RBCs in AS-1N is not without shortcomings. For example, the maintenance of normoglycemic conditions required the addition of glucose to the stored RBCs every 2–4 days to prevent the RBCs from becoming hypoglycemic. Since the introduction of glucose into anticoagulants by Rous and Turner back in 1916, it has been well established that glucose is a necessary preservative for storing RBCs.¹⁹ Indeed, we have previously reported that hypoglycemic stored RBCs display significant levels of hemolysis.¹⁸ Therefore, our previous reports utilized a manual feeding method to maintain physiological glucose levels that involved opening the bag, adding concentrated glucose, and subsequently resealing the bag. This methodology cannot be reasonably translated to current practices due to breaches of sterility and labor requirements.

Here, we describe an automated glucose-feeding device intended to maintain normoglycemic blood storage in an autonomous manner while maintaining a closed system with the potential for translation to a clinical setting. Commercially available blood collection bags for smaller scale (>20 mL) storage of RBCs are employed in our model. This study aimed to improve the potential for normoglycemic RBC storage to be adopted in common practice through an automated glucose feeding device while assessing common storage lesion markers to highlight the benefit normoglycemic storage imparts on RBCs. Though not currently compatible with the current RBC storage infrastructure, the glycemic control system described here addresses significant limitations of the applicability of normoglycemic storage and is able to illustrate that autonomous and sterile glycemic control of stored RBC units is possible.

2. Results and Discussion

Previous in vitro work highlighting beneficial RBC function due to normoglycemic storage utilized a manual glucose feeding method that cannot be translated to clinical practice. To manually feed over 11 million RBC units that were collected in 2021 with glucose every day would not only be financially illogical but also logistically impossible.²⁰ Additionally, manual injection of a glucose-containing solution into stored RBC units would increase the risk of contamination and decrease patient safety. The motivation of this work was to overcome these limitations and demonstrate the potential of normoglycemic RBC storage for clinical translation through the maintenance of RBC glucose levels autonomously while maintaining sterility. Doing this required the design of a glucose-feeding device that could operate without intervention, maintain sterility, and not disrupt the current standard blood collection practices.

2.1. System Design, Modeling, and Validation

2.1.1. Device Design

Typical blood collection utilizes additive solutions (AS-1) that are hyperglycemic to store RBCs. For the remainder of the investigation, the hyperglycemic method will be termed AS-1, while normoglycemic storage will be termed AS-1N. This study implements a completely closed and automated glucose feeding device to dispense precise volumes of a concentrated glucose solution into the AS-1N stored RBCs. The design of the automated feeding system is shown in the computer-aided design (CAD) drawing in Figure 1. Similar to traditional intravenous (IV) systems, the automated glucose feeding design utilizes gravity to drive fluid flow into the stored RBC unit. The feeding solution is suspended approximately 90 cm above the stored RBC unit. The device delivers either 0.9% saline (for AS-1 stored RBCs) or 100 mM glucose in 0.9% saline (for AS-1N stored RBCs) by opening and closing the solenoid valve via a programmable Arduino Uno microcontroller, dispensing the feeding solution into the stored RBCs. The device delivered microliter volumes of concentrated glucose to normoglycemic stored RBCs to maintain physiological levels of glucose by controlling the frequency and duration of the valve opening. A detailed description and diagram of the device’s operation are described in the Supporting Information.

Figure 1.

CAD drawing of the automated glucose feeding device for normoglycemic stored RBCs. The pressure difference created between the 100 mM glucose bag and the stored RBC bag is driven by gravity since there is a 90 cm difference in height between the two. When the valve is closed, there is no fluid flow since the valve’s inability to expand exerts an equal pressure on the fluid. However, when it is opened, the pressure differential is realized due to the flexibility of the plasticized PVC bag. This drives fluid flow into the bag, therefore allowing feeding volume and frequency to be controlled by the opening and closing of the solenoid valve.

This closed system maintained complete sterility of the solutions throughout the duration of storage, once connected to the stored RBC unit. Additionally, its simplicity in using gravity-driven fluid flow avoids the use of complicated and expensive equipment and allows this system to be utilized with any commercially available blood storage bag, requiring no change to the current method of blood collection and storage.

2.1.2. Device Calibration and Characterization

The relationship between the valve opening period and the volume dispensed was characterized prior to use with stored RBCs using 8 different valve delay periods between 50 and 2000 ms. The volume dispensed from the valve was directly proportional to the valve delay time (Figure 2). The lower limit of the dispensing volume for the system was 176 ± 21 μL. Additionally, the theoretical dispensing volume calculated using pressure differentials in Bernoulli’s principle (described in the Supporting Information) is also shown and closely matches the experimental calibration curve resulting in a relative error in the slope of 0.67%. These results indicated that the valve operated as expected, delivering consistent microliter volumes that could be applied to the glycemic control of stored RBCs.

Figure 2.

Calibration curve depicting volume dispensed as a function of the solenoid valve delay period. The volume dispensed from the automated feeding system using various valve opening intervals (50–2000 ms) was determined by utilizing water (density = 1 g/mL) and a mass subtraction technique. The calibration curve is linear and can be used to determine the appropriate valve delay period that corresponds to the desired volume to be dispensed when feeding the concentrated glucose solution.

2.1.3. Glucose Utilization and Maintenance of Concentration

Once designed and calibrated, the glucose-feeding device was validated through its ability to maintain glucose concentrations in AS-1N RBC storage. The dispensing volume and frequency of the device remained constant throughout storage and were determined through an average glucose utilization rate over 44 days of storage (0.33 ±0.42 mM/day). The rate of glucose utilization by the RBCs stored in AS-1N is shown in Figure 3a. The average glucose utilization was 1.3 ± 0.3 mM/day on day 2, but it continually slowed throughout storage reaching 0.25 ± 0.07 mM/day by the end of storage. This drop in the glucose utilization rate can be explained by both the metabolism kinetics of cold storage, as well as the buildup of metabolic waste products. It is well established in the literature that low-temperature cold storage of RBCs slows the metabolic enzyme activity, resulting in reduced rates of glycolysis and therefore reduced glucose consumption.²¹ Additionally, the drop in metabolic activity is further exacerbated by the increase in metabolic waste products and oxidative stress in storage. The accumulation of lactic acid from glycolysis results in acidosis, while the buildup of reactive oxygen species (ROS) leads to the oxidation of proteins and enzymes.^21,22 Both processes result in a reduction of glycolysis and, therefore, reduced glucose consumption during storage, as shown in Figure 3a. By the end of storage on day 44, the RBCs operate at extremely low glucose consumption rates, reduced by more than 80% in comparison to day 2. Glucose metabolism of the RBCs may also be dependent on the extracellular glucose concentration. The main RBC glucose transporter, GLUT1, is a passive transporter, thus the glucose concentration relative to intracellular levels may impact the rate of transport, and therefore glucose utilization.²³ Though glucose utilization was not directly measured under hypoglycemic storage conditions, we have previously reported that the hypoglycemic storage of RBCs resulted in significant hemolysis, likely due to inadequate ATP generation.¹⁸

Figure 3.

RBC glucose utilization rates and glucose tracking throughout storage. (a) Glucose concentration was monitored before and after feeding periods at 3 day intervals. The difference between glucose concentrations at each time interval was used to calculate the glucose utilization rates. The glucose utilization begins high at 1.3 mM/day, shortly after removal from the body, and begins to drop until plateauing to 0.25 mM/day toward the end of storage (n = 5). The average throughout the entirety of storage was 0.33 ± 0.41 mM/day (red dotted line). A trend line was also fit to the data (black solid line). (b) AS-1 (red squares) and AS-1N (green circles) stored RBCs were stored for up to 44 days, with glucose concentrations measured every 2–4 days. AS-1 stored RBCs exhibited glucose concentrations above saturation, steadily dropping through storage to 18.3 ± 4.42 mM. Though early time points saturated the glucometer, extrapolating the trend of the later time points (dotted line) back to time 0 estimated the original AS-1 glucose concentration at 37 mM. This trend line also resulted in an average glucose utilization of 0.428 ± 0.097 mM/day. The automated glucose feeding system was implemented to maintain glucose levels between 4 and 6 mM for AS-1N stored RBCs. Though some data points fell outside of this range, the average glucose concentration (solid line) was maintained within expected values for up to 37 days. (c) The glucose concentration of a singular set of stored blood products displayed the ability of the glucose-feeding device to maintain glucose levels within 4–6 mM completely autonomously for up to 37 days of storage. The glucose level at day 7 rose to 6.0 mM but did not exceed the normoglycemic range.

The glucose levels throughout the storage of 3 sets of RBC units stored in AS-1 or AS-1N are shown in Figure 3b. The glucose concentration dropped steadily throughout storage in AS-1 (AS-1 stored blood was fed with only 0.9% saline), reaching 18.3 ± 4.42 mM at the end of storage, still over 3 times greater than physiological conditions. Though AS-1 stored RBCs read above the measurement limits of the glucometer at early time points, a trend line fit to the later time points was extrapolated back to time 0, estimating the initial concentration to be approximately 37 mM, similar to the expected concentration (determined by a calculation taking into account the glucose initially in the storage solution) of 40 mM. Additionally, the slope of this trend line was calculated to be −0.428 ± 0.097 mM/day. This slope can be considered to represent the average glucose utilization of the hyperglycemic stored RBCs, considering that there was no addition of glucose to these cells. However, it should be noted that this trend line, and therefore the RBCs‘ average glucose utilization in hyperglycemic storage, does not represent all of the data points due to the saturation limits of the glucometer at 33 mM. The glucose data that read at saturation were not included in the fitting of the trend line. However, considering the pattern of decreasing glucose utilization throughout storage during normoglycemic storage as indicated in Figure 3a, it is likely that the hyperglycemic stored RBCs utilized more glucose at earlier time points, indicating that the 0.428 mM/day glucose utilization rate calculated here is likely an underestimate of the average utilization for AS-1 stored RBCs. As noted by the green circles and solid line in Figure 3b, AS-1N stored blood supplemented with glucose every 3 days was maintained within a physiologically relevant glucose concentration range (4–6 mM) for 37 days of storage. Though some samples periodically measured outside of this range, especially after 37 days in storage, the average (indicated by the solid line) was successful for up to 37 days. These deviations for particular RBC units can be explained by donor variation in glucose metabolism and time-dependent glucose utilization changes, as evidenced by Figure 3a. Not all donor RBCs metabolized glucose at the same rate, yet the automated feeding system dispensed the same concentrated glucose volume every 3 days based on the average glucose utilization rate determined from Figure 3a. This caused glucose levels to deviate from the normoglycemic range, especially at later time points when glucose utilization was extremely low, as shown in Figure 3a. The constant average dispensing volume was implemented to maintain simplicity in the initial design but is a clear limitation to be resolved in future iterations that can compensate for the changing glucose needs of the RBCs. These deviations from normoglycemia are also important to note when evaluating the biochemical measurement data, which may have impacted the results. Though the deviations from normoglycemia are a clear limitation, Figure 3c outlines that complete autonomous and sterile glucose control of stored RBCs is possible, supported by the maintenance of normoglycemic concentrations in a single AS-1N RBC unit for up to 37 days. The glucose tracking data in Figure 3c are representative of a single donor taken from the data in Figure 3b, showing that autonomous glycemic control between 4 and 6 mM is possible without any deviation for up to 37 days.

The glycemic control of the automated feeding device overcomes a significant barrier to the application of normoglycemic RBC storage, specifically the required feeding of glucose to avoid hypoglycemia. Insufficient glucose or hypoglycemia prevents adequate ATP production for cellular functions, leading to hemolysis. Wang et al. showed that RBCs stored in normoglycemic conditions without the addition of exogenous glucose throughout storage exhibited hemolysis levels greater than 8%, 20 times greater than traditional storage, and a value that is unacceptable for clinical use.¹⁸ Normoglycemic storage therefore requires periodic glucose feeding to maintain acceptable lysis levels. The device described here enables glycemic control completely autonomously throughout storage, eliminating labor requirements and maintaining sterility.

2.2. Physical and Biochemical Measurements

2.2.1. Hematocrit Decreases While Hemolysis Increases

The hematocrit of RBCs (the volume of RBCs in solution relative to the total volume of the solution) stored in either AS-1 or AS-1N, along with the predicted model, is shown in Figure 4a. The initial hematocrit of both storage conditions was 59 ± 1% and decreased throughout storage to 50 ± 4% on day 44. The hematocrits of both AS-1 and AS-1N stored blood matched closely with the predicted values. Though the hematocrits of the stored RBC solutions decreased throughout storage, the average hematocrit of RBC units has been reported as 63.3 ± 9.3%, indicating that an RBC unit of 50% may still be sufficient in raising patient hemoglobin levels if transfused.²⁴ If not, however, utilizing a more concentrated glucose feeding solution of 400 mM would yield a hematocrit drop of only 2%, as indicated by the dashed line in Figure 4a. This predicted drop was calculated based on expected glucose volume additions to stored RBCs using a 400 mM glucose feeding solution and was not experimentally determined. The 100 mM glucose feeding solution was used in this design due to limitations on the minimum dispensing volume of the automated feeding system.

Figure 4.

Hematocrit and hemolysis of stored RBCs. (a) Hematocrit steadily dropped in both AS-1 and AS-1N stored RBCs as the automated feeding system dispensed either 0.9% saline or 100 mM glucose, respectively. The hematocrit is statistically equivalent at each time point between AS-1 and AS-1N stored RBCs (n = 7–10, error = standard deviation). Utilizing a more concentrated glucose feeding solution of 400 mm (dashed line) would lead to a hematocrit drop of less than 2%. (b) Hemolysis was statistically equal in both AS-1 and AS-1N stored RBCs at each time point. Additionally, hemolysis at the end of storage was statistically equivalent to the 1% threshold (dashed line) required for transfusions set by the FDA (n = 2–3, error = standard deviation).

Hemolysis levels for both AS-1 and AS-1N stored RBCs are shown in Figure 4b. The hemolysis levels at each time point were statistically equivalent between the storage conditions throughout the storage period. RBCs must be intact for them to function. Therefore, hemolysis is a frequently used metric of stored RBC health.²⁵ As RBCs age throughout storage, an increase in metabolic waste and oxidative stress along with a decrease in pH contribute to increased hemolysis.³ Hemolysis releases free hemoglobin into the extracellular space which is known to scavenge NO, a vasodilator that facilitates blood flow.^3,26 Donadee et al. showed that transfusion of the supernatant from stored RBCs with increased hemolysis and free hemoglobin led to increased vasoconstriction in comparison to stored RBCs with reduced hemolysis and free hemoglobin.²⁷ This phenomenon was attributed to the reduced bioavailability of NO after transfusion, a well-known issue in transfusion medicine. This is, in part, why the hemolysis threshold defining an acceptable RBC product for RBC transfusion in the United States is set to 1%.²⁸ In this study, AS-1N and AS-1 stored RBCs were below or statistically equivalent to 1% for the entirety of storage. Hypoglycemia is associated with increased hemolysis; therefore, it was crucial to ensure that normoglycemic storage and occasional transient hypoglycemic conditions, as shown in Figure 3b, did not cause increased cell lysis. Clearly, even during slight deviations from normoglycemia, there was no impact on the hemolysis level.

2.2.2. Osmotic Fragility and Sorbitol Levels Indicate Improved RBC Functionality

It is well established that membrane damages occur during storage, leading to the generation of echinocytes and spherocytes, increased cell fragility, and decreased deformability.³ These membrane damages, along with increased osmotic pressure from metabolic waste accumulation, are associated with increased osmotic fragility, or increased propensity for RBCs to lyse under hypotonic stress.^29,30 Additionally, it is also well known that increased oxidative stress during RBC storage can oxidize membrane proteins and lipids, leading to further increased RBC membrane damage.³ The extent of oxidative stress can be indirectly measured through intracellular sorbitol levels, an intermediate of the polyol pathway that exacerbates oxidative stress. Monitoring both osmotic fragility and sorbitol levels of stored RBCs provides useful information related to the progression of the storage lesion.

The osmotic fragility of RBCs stored under both AS-1 and AS-1N conditions is shown in Figure 5a. RBCs stored in AS-1N exhibited decreased osmotic fragility for up to 23 days of storage in comparison to AS-1. The hypotonic hemolysis percentage of RBCs stored in AS-1N began at 12.2 ± 6.9% on day 2 and rose throughout storage until reaching 74.7 ± 20.7% on day 30, statistically equivalent to the level of lysis in AS-1 stored RBCs. The osmotic fragility of the AS-1 stored RBCs began at 77.9 ± 9.3% lysis on day 2 and rose to 90.8 ± 0.3% lysis on day 44 but was statistically equivalent throughout storage. These results confirm and improve upon previous data that RBCs stored in normoglycemic conditions are more resistant to osmotic stress. This data, however, shows that this resistance persists for up to 3 weeks of storage, unlike our previous reports which indicated only a few days of benefit.¹⁷

Figure 5.

Osmotic fragility and intracellular sorbitol indicate that normoglycemic storage results in RBCs that are less prone to lysis and produce less markers of oxidative stress. (a) AS-1 stored RBCs exhibited significant hemolysis above 70% at each time point and were statistically equivalent throughout storage. AS-1N stored RBCs were more resistant to osmotic stress initially, with a cell lysis of 12.2% on day 2. AS-1N stored RBC lysis was statistically lower than AS-1 stored RBC lysis on days 1, 9, 16, and 23 (n = 3–4, error = standard deviation, *p < 0.05, **p < 0.01). (b) Intracellular sorbitol was measured enzymatically, showing that sorbitol levels were significantly lower in normoglycemic storage at each time point. Though rising throughout storage for both AS-1 and AS-1N stored RBCs, day 44 sorbitol levels using AS-1N remained below that of day 2 sorbitol levels using AS-1 (n = 4–5, error = standard deviation, *p < 0.05, **p < 0.01).

Additionally, sorbitol levels were monitored throughout storage as an indicator of the polyol pathway and indirectly oxidative stress. Intracellular RBC sorbitol levels for both AS-1 and AS-1N storage are shown in Figure 5b. Sorbitol levels for AS-1N storage initially measured 10.9 ± 7.1 nmol of gHemoglobin^–1 (gHb) and rose to 48.0 ± 31.9 nmol of gHb^–1 by day 44. The sorbitol levels for RBCs stored in AS-1 were higher than those for RBCs stored in AS-1N at each time point, beginning at 79.5 ± 32.4 nmol gHb^–1 on day 2 of storage and rising to 106 ± 43.4 nmol gHb^–1. Though not statistically significant, Figure 5b also shows that the sorbitol levels of AS-1N stored RBCs on day 44 were lower than the day 2 measurement of AS-1 stored RBCs. These results show that there is more sorbitol production, accumulation, and polyol activity in traditional storage relative to normoglycemic storage.

The polyol pathway first converts glucose to sorbitol, consuming a reduced nicotinamide adenine dinucleotide phosphate (NADPH) molecule, followed by the conversion of sorbitol to fructose, producing a reduced nicotinamide adenine dinucleotide (NADH) molecule.³¹ This pathway is markedly increased in diabetes patients and is reported as a significant contributor to oxidative stress.³¹ This is due to the increased consumption of reducing equivalents in the form of NADPH. NADPH is a vital molecule in the recycling of a key antioxidant, glutathione (GSH). With increased sorbitol accumulation, there will be an accompanied equivalent consumption and depletion of NADPH, leaving less available to recycle GSH and, therefore, more oxidative damage. These decreases in NADPH and GSH during RBC storage are, indeed, supported by the literature.^3,32

The excess sorbitol accumulation in AS-1 RBC storage and consequently increased glucose metabolism through the polyol pathway may be explained by the elevated glucose levels in hyperglycemic storage relative to those in normoglycemic storage. Indeed, RBCs from patients with diabetes have been reported to exhibit increased sorbitol levels that were directly proportional to their degree of hyperglycemia, suggesting that hyperglycemia renders excess glucose available for the polyol pathway, resulting in increased ROS levels.^31,33 This is further supported by the data in Figure 3, which illustrate that the average glucose utilization rate was higher under hyperglycemic storage in comparison to that under normoglycemic storage. The relative reduction of sorbitol production in normoglycemic storage, therefore, indicates that normoglycemic RBCs may exhibit less oxidative stress than RBCs stored under hyperglycemic conditions. This may also be able to explain the reduced osmotic fragility of AS-1N stored RBCs due to potentially increased oxidative damage in RBC storage causing accelerated membrane damage.

Additionally, the differences in osmotic fragility could also be explained by the buildup of metabolic waste products. The accumulation of waste, such as lactate, imparts osmotic stress onto the RBC, causing a greater influx of water when subjected to hypotonic 0.45% saline.³⁰ This is most likely why the normoglycemic RBCs deteriorate and increase in their osmotic fragility between days 9–30, resulting from a buildup of intracellular lactate that generates greater osmotic stress for the cell. Along with lactate, sorbitol and fructose (both polyol pathway byproducts) can contribute to osmotic stress as they are both membrane impermeable.³¹ The significant reduction in sorbitol accumulation and therefore reduced osmotic stress within the RBC during normoglycemic storage could also explain the reduced osmotic fragility.

It is important to note that osmotic fragility differences could also be due to inherent differences in the osmolarity of the additive solutions (AS-1 = 464 mOsmolar, AS-1N = 359 mOsmolar), or the osmolarity changes due to the addition of glucose/saline from periodic feeding. However, this is not likely, as previous reports evaluating osmotic fragility utilizing other additive solutions characterized by lower osmolarity, such as AS-3 (291 mOsmolar), and without periodic saline addition showed similar results to the data shown here with AS-1.¹⁷

Regardless of the mechanism, these data highlight a key relationship between sorbitol accumulation and osmotic fragility, indicating that normoglycemic storage can improve both, potentially leading to improved RBC health in storage. This relationship and its link to hyperglycemia are indeed supported by the literature. Jain et al. revealed that incubating RBCs in elevated glucose environments led to increased membrane lipid peroxidation and osmotic fragility, which was eliminated when the RBCs were treated with glucose metabolism inhibitors, antioxidants, ROS scavengers, or other oxidative pathway inhibitors.³⁴ In traditional hyperglycemic storage, more glucose is available to be metabolized in the polyol pathway, leading to increased levels of sorbitol and subsequently increased osmotic stress.

2.2.3. Lactate and pH Show No Significant Differences

Though excess available glucose in this study led to a detrimental oxidative effect (Figure 5), previous studies showed that traditional, hyperglycemic storage resulted in increased intracellular ATP and increased lactate production in comparison to normoglycemic storage without feeding.¹⁸ The increase in intracellular ATP and lactate production was slightly higher than that of AS-1N with feeding, suggesting increased glycolytic activity. A reappraisal of these results using automated and controlled feeding of the AS-1N solutions is shown in Figure 6. Here, lactic acid and pH measurements were used as surrogates for the glycolytic activity. As shown in Figure 6a, both AS-1 and AS-1N stored RBCs exhibited initial extracellular lactate concentrations of 4.2 ± 0.7 mM, which is slightly greater than the physiologically relevant 2 mM but similar to previous findings.^18,35 The lactate levels in both storage environments steadily increased throughout the storage period, reaching 39 ± 9.4 mM in AS-1 and 36 ± 2.6 mM in AS-1N. This trend of lactate accumulation is consistent with the literature.³ The lactate levels are statistically equivalent between storage conditions at each time point. The change in pH during storage is shown in Figure 6b. The pH in both conditions began at 7.7 ± 0.2 on day 2 and steadily dropped to 6.6 ± 0.1 on day 44. There is no statistically significant difference in the pH of the stored RBCs between storage conditions at any time point measured.

Figure 6.

Extracellular lactic acid (lactate) and pH show no differences between storage conditions. (a) In both AS-1 and AS-1N stored RBCs, the lactate in the extracellular space increased throughout storage. There is no difference in lactate accumulation between the two storage conditions (n = 5–6, error = standard deviation). (b) pH for RBCs stored in both AS-1 and AS-1N began at a physiologically relevant 7.7 and steadily dropped throughout storage as waste products from metabolism, including lactate accumulated. There was no statistical difference in pH measurements between RBCs stored in AS-1 or AS-1N at any time point (n = 4–6, error = standard deviation).

Though the lactate accumulation is consistent with the drop in pH, these data contradict our previous reports but may be explained by better glycemic control. The tight control of glucose at physiologically relevant levels shown in Figure 3b likely supplied enough glucose to saturate the glycolytic pathway. In our previous report, our manual glucose feeding regimen maintained a low-glucose environment; however, it was not monitored as frequently and may have led to longer hypoglycemic periods.¹⁸ The lactate and pH data in Figure 6 lead to the indirect conclusion that there is no significant difference in glycolysis rates between AS-1 and AS-1N storage, indicating that the rate of glycolysis is at its maximum under normoglycemic storage, and any additional glucose may simply be available for other detrimental pathways known to damage to RBCs (e.g., polyol pathway). Additionally, the statistically equivalent levels of lactate between the two storage conditions also eliminate the possibility of lactate accumulation as the culprit for increased osmotic fragility in AS-1. However, this buildup of lactate throughout storage supports the previous claim that the increase in the osmotic fragility of AS-1N stored RBCs between days 9–30 was most likely a result of increased intracellular lactate.

Though AS-1 was utilized as the standard for traditional hyperglycemic storage in this investigation, it is important to note that there are other additive solutions commonly used which contain lower glucose concentrations (AS-3, SAGM, and AS-7) albeit still hyperglycemic (>40 mM). However, our previous studies evaluated the performance of a panel of these other lower glucose additive solutions and found that osmotic fragility and sorbitol levels in these lower glucose hyperglycemic additive solutions still elicited similar results to AS-1.¹⁷

The system described and tested in this study enabled maintenance of normoglycemic conditions (4–6 mM) in stored RBCs for up to 37 days of storage, leading to reductions in osmotic fragility and sorbitol accumulation while maintaining pH values, lysis rates, and lactate production statistically equal to AS-1. Further studies utilizing larger stored RBC volumes with in vivo measurements are warranted to confirm the beneficial impact normoglycemic storage imparts to RBCs. This work not only confirms and improves upon previous results associated with normoglycemic RBC storage but also overcomes a significant barrier to the application of this blood storage technology. This study highlights that the maintenance of glucose levels in normoglycemic RBC storage can be successfully accomplished for up to 37 days completely autonomously while maintaining sterility. Though improvements to the design must be made, these results significantly increase the feasibility of normoglycemic storage in translation to clinical practice.

This work is not without its limitations. First, the deviations from the 4–6 mM glucose range in Figure 3b suggest that the consistency of the device must be improved. This can be accomplished in future device design iterations that can deliver lower volumes of concentrated glucose so more frequent and smaller glucose adjustments can be achieved. Additionally, implementing a feeding program that adjusts the feeding volume as the RBC glucose needs change according to the black solid trend line in Figure 3a would also lead to better glycemic control. However, it should be noted that the deviation from normoglycemia did not seem to impact the results significantly, considering the differences outlined in the osmotic fragility and sorbitol accumulation. Additionally, it is acknowledged that a gravity-driven glucose-feeding device with fixed geometry is incompatible with the current blood banking infrastructure. Future work has already begun to overcome this limitation with a peristaltic pumping system. This work simply serves the purpose of showcasing that the autonomous glycemic control of stored RBCs is possible. Even if implementing this system at a large scale is difficult due to its incompatibility with current storage geometries, its application may still be warranted for rare blood types or special transfusion circumstances that require high-quality RBCs. There is also no clinical support that normoglycemic RBC storage imparts any in vivo benefit. However, these data now make animal studies and clinical trials with normoglycemic storage realistically possible.

3. Methods

3.1. RBC Collection, Processing, and Storage

All blood collection procedures from informed and consented donors were approved by the Biomedical and Health Institutional Review Board of Michigan State University. Anticoagulants and additive solutions were prepared in-house. CPD was prepared from its components at the specified concentrations.³⁶ The additive solution, AS-1, was prepared in the same manner.³⁷ Components of the normoglycemic test solutions, CPD-N and AS-1N, are shown in Table 1. The solutions were autoclaved at 121 °C at 21 bar for 30 min to sterilize.

Table 1. Anticoagulant and Additive Solution Components and Concentrations^a.

component [mm]	CPD	CPD-N	AS-1	AS-1N
sodium citrate	89.4	89.4
monobasic sodium phosphate	16.1	16.1
citric acid	15.6	15.6
dextrose	129	5.5	111	5.5
sodium chloride			154	154
adenine			2	2
mannitol			41	41
pH	5.6	5.6	5.8	5.8

^aThe FDA-approved anticoagulant and storage solutions (CPD and AS-1, respectively) contain glucose concentrations 20-fold greater than physiological levels. The proposed counterparts, CPD-N and AS-1N, are similar in composition with the exception of a lower glucose content.

Prior to collection, 1 mL of the CPD or CPD-N anticoagulants was injected into nonsiliconized and untreated (no anticoagulant) 10 mL glass Vacutainer tubes (BD, Franklin Lakes, NJ). Next, approximately 7 mL of whole venous blood was collected into each tube from a consented donor. A total of 10 tubes were collected from each donor and split evenly into traditional processing and normoglycemic processing (5 hyperglycemic tubes and 5 normoglycemic tubes). The collected blood was allowed to sit for 30 min at room temperature, then centrifuged at 2000g for 10 min. The plasma and buffy coat layer were then removed by aspiration along with the top 2 mm layer of packed RBCs to minimize the retention of white blood cells in the stored product. No leukoreduction was performed due to the small blood volumes used in this work, and it is noted that this limitation may have an impact on the stored RBCs. The packed RBCs were then combined with AS-1 or AS-1N in a 50 mL conical tube using a 2:1 volume ratio. The RBCs were then injected into commercially available, sterile, 100 mL veterinary blood collection bags (ABRI catalog no. BG-DD 100 BAG) for a total stored RBC volume greater than 20 mL and connected to the automated feeding system at 2–6 °C. All nonsterile equipment/reagents were autoclaved or UV sterilized, while the valve was sterilized using 10% bleach with a sterile water flush. All RBC processing was conducted inside a biosafety cabinet. The stored RBC units were kept upright inside of beakers without mixing or agitation, including after the periodic addition of concentrated glucose. It is noted that without mixing after glucose addition, localized hyperglycemic environments may have been generated within the normoglycemic stored RBCs. In the pursuit of complete autonomy of the feeding system, manual mixing of the stored RBC bags was avoided when possible, such as after autonomous glucose addition. However, the RBC units were inverted and mixed thoroughly before sampling, including glucose measurements every 2–4 days, to ensure a homogeneous and representative sample. The stored RBCs were fed with either 0.9% saline or 100 mM glucose in 0.9% saline every 3 days. The feeding volume was consistent for each RBC unit and feeding period; however, this volume was determined based on the total volume of stored RBCs collected and varied from 250 to 312 μL.

3.2. Automated Glucose Feeding Device Calibration

The automated feeding system was calibrated using water with a density of 1 g/mL (see the Supporting Information). Briefly, water was dispensed via the feeding device into tubes of known mass. Subtracting the mass of the tubes before dispensing from the mass of the tubes after dispensing enabled calculation of the volume dispensed for each valve opening period.

3.3. Glucose Measurements

Every 2–4 days, an aliquot of stored blood (<20 μL) was drawn via the sampling port on the bag. The glucose concentration of these samples was determined by using a blood glucose glucometer (AimStrip Plus) and blood glucose test strips (AimStrip). Glucose measurements were obtained before and after dispensing to determine the glucose utilization rate for a particular period by subtraction.

3.4. Hematocrit and Hemolysis Assay

The hematocrit of the removed aliquots of RBCs was determined with a hematocrit centrifuge (CritSpin M960-22, StatSpin, Westwood, MA) and a microcapillary reader (StatSpin). To measure hemolysis, a separate aliquot was removed and centrifuged at 500g for 10 min followed by collection of 70–90 μL of the supernatant into a separate sample tube. The remaining packed RBCs were then further centrifuged at 2000g for 15 min. Approximately 20 μL of the cell pellet was collected in a separate sample tube. Six hemoglobin (Sigma-Aldrich) standards were then prepared in the range of 0–1 mg/mL hemoglobin by diluting the stock in Drabkin’s reagent to create 1 mL standards. The hemoglobin absorbance bands were measured at 540 nm by using a spectrophotometer (Molecular Devices, SpectraMax M4). Using eq 1 below, the lysis percentage was determined for each sample collected at each time point where [HbSN] is the supernatant hemoglobin concentration (mg/mL), Hct % is the hematocrit percentage of the original sample, and [HbRBC] is the hemoglobin concentration of the packed RBC samples (mg/mL)

1

3.5. pH Measurement and Lactate Accumulation

The pH of RBC aliquot supernatants after centrifugation was measured using a micro pH electrode (OHAUS, STMICRO5). Lactate was measured using a fluorescent enzymatic assay with lactic acid dehydrogenase (LDH = 20 U/mL) in the presence of 5 mM NAD⁺ and 0.1 M TRIS. Supernatant samples were removed of residential LDH via 10 kDa MWCO spin filters, and 100 μL of the sample supernatant was then added to 100 μL of the enzyme assay mixture or blank assay mixture. LDH converts extracellular lactate to pyruvate and NADH, which is fluorescent and is proportional to the concentration of lactate. The fluorescence of the samples was measured on a spectrophotometer (Molecular Devices, SpectraMax M4) (340/460 nm) and compared to standards to determine lactate concentration. Any signal from NADH present in the supernatant was subtracted using the blank assay mixture sample.

3.6. Osmotic Fragility and Intracellular Sorbitol Assays

Exactly 25 μL of RBCs was removed from storage and added to 500 μL of each of 3 solutions: 0.9% saline, 0.45% saline, and 0.0% saline (pure water). The samples were inverted a few times and allowed to incubate at room temperature for 30 min. All samples were centrifuged at 1500g for 5 min. Afterward, the supernatants were collected and diluted in 18.2 MΩ·cm purified water at a ratio of 1:10 for analysis. Next, 200 μL of each sample was added to a clear 96-well plate in triplicate with absorbance read at 540 nm. The lysis percentage was determined using eq 2 below

2

RBC intracellular sorbitol was measured every 7 days of storage beginning on day 2 using a sorbitol dehydrogenase assay as reported previously.¹⁷ Sorbitol levels were used as indirect measurements of oxidative stress. All sorbitol measurements were reported with respect to hemoglobin levels in the RBC aliquots.

3.7. Statistical Analysis

All statistical testing was performed using two-way ANOVA followed by Tukey’s Honest Significant Difference testing at a 95% significance level. GraphPad Prism software was used for statistical analysis. Statistical analysis data are only shown for data that displayed statistical significance. Anywhere from 2 to 10 donors were used for assay measurements depending on availability and measurement difficulty.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsptsci.4c00018.

• Methods involved in device calibration, feeding system operation, device modeling, and back-end coding of the Arduino microcontroller (PDF)

Author Contributions

L.D.S. designed the device, performed the experiments, analyzed the data, and wrote the manuscript. L.S.S. helped in performing the research, validating the device, processing the blood, and making edits to the manuscript. S.B. contributed to collecting data and processing the blood. T.J.M. reviewed the data and the manuscript. D.M.S. supervised the research, offered guidance in experimental design, and reviewed/edited the manuscript.

The authors declare no competing financial interest.