Secondary Logo

Journal Logo

Cardiovascular Anesthesiology

Fresh and Stored Red Blood Cell Transfusion Equivalently Induce Subclinical Pulmonary Gas Exchange Deficit in Normal Humans

Weiskopf, Richard B. MD*; Feiner, John MD*; Toy, Pearl MD*; Twiford, Jenifer BSN*; Shimabukuro, David MD*; Lieberman, Jeremy MD*; Looney, Mark R. MD; Lowell, Clifford A. MD, PhD; Gropper, Michael A. MD, PhD*

Author Information
doi: 10.1213/ANE.0b013e318241fcd5

Transfusion can cause lung injury. Severe injury, transfusion-related acute lung injury (TRALI), first described in 19831 with the acronym developed in 1985,2 remains a diagnosis of exclusion in patients experiencing new acute lung injury within 6 hours after having been transfused.3 The mortality of TRALI is thought to be approximately 5% to 20%,2,4 and the Food and Drug Administration has reported that TRALI was the leading cause of transfusion-related death each year from 2005 to 2009.5 Clinical experience and the NHLBI Working Group6 definitions for this severe form of pulmonary injury, TRALI, require radiologic evidence of bilateral pulmonary edema and hypoxemia for a positive diagnosis. We conjectured that it is possible that there might be lesser pulmonary dysfunction after transfusion that had not been noticed clinically and is not sufficiently severe to fit the definition of TRALI.

In a retrospective chart review of critically ill patients, without case cohort controls, Cornet at al.7 noted a decreased PaO2/fraction of inspired oxygen (FIO2) 24 hours after leukoreduced (LR) red blood cell (RBC) transfusion. Covin et al.8 reported a single case of acute transient pulmonary insufficiency after transfusion during surgery of an autologous unit of RBCs that when tested had significant lipid-priming activity. Whereas antibody mediation is an accepted etiology of TRALI, the latter case report and several laboratory experiments912 support the additional thesis that pulmonary injury after transfusion may also be caused by infusion of active biologic substances (cytokines, phospholipids) that activate neutrophils that have been primed by an existing abnormal physiologic state. However, there are no data from prospective randomized studies in humans that sought subtle changes in pulmonary function after transfusion, while at the same time comparing any such effects after transfusion of “fresh” RBCs with those that have been stored for substantial periods of time sufficient to accumulate the implicated biologically active compounds. Some retrospective analyses of databases have noted an association of transfusion of RBCs stored for a longer time with an increased incidence of adverse clinical outcomes than with transfusion of RBCs stored for lesser periods of time.13 However, other similar analyses,1416 including a recent larger, more carefully analyzed database of all transfused people in Denmark and Sweden, did not confirm those findings.17

Accordingly, we tested the following hypotheses: (1) that transfusion results in subclinical alterations of pulmonary function in normal humans who do not have an existing condition that causes neutrophil priming; (2) that transfused cytokines in blood components are associated with altered pulmonary function in normal humans; and (3) that transfused RBCs stored for a relatively long time compared with those stored for a lesser period produce subtle changes in pulmonary function.


To test the hypothesis that transfusion can produce relatively small, detectable changes of pulmonary function, and not only severe damage (with a low incidence), we assessed oxygenation and pulmonary dead space fraction. We assessed the latter because increased dead space fraction is an early feature of acute lung injury.18 These variables were assessed in healthy volunteers transfused with autologous stored (LR or not LR) or autologous fresh RBCs, so that results would not be affected by the presence of disease or concurrent medications.

Thirty-five healthy volunteers were studied with their signed written informed consent and with the approval of the IRB of the University of California, San Francisco (UCSF). Criteria for enrollment were that volunteers were required to be healthy as determined by history and physical examination, be nonsmokers, of age 18 to 35 years, have a body weight ≥110 lbs., a hematocrit ≥0.34, and be taking no medications other than for birth control.

After each volunteer gave their informed consent they were randomly allocated to 1 of 4 groups based on whether or not their donated blood was to be LR, and whether they would have the first study day as return of fresh or stored blood (see below). Randomization was accomplished using computerized code written for SAS version 9.1 (SAS Institute, Inc., Cary, NC) using blocks of 4 volunteers. The study coordinator accessed the randomization code for 1 volunteer at a time. The volunteers and the investigators involved in data collection and analysis of study end points were blinded to group allocation. The study coordinator and 1 physician not involved in data collection that day were not blinded for that day so that they could perform the required checks of the recipient and the units of RBCs to be transfused.

One unit of blood (450 mL) was withdrawn from each volunteer approximately 4 weeks before the study days, and a second unit approximately 3 weeks before. Blood was collected into Fenwal/Baxter bags with CPDA-1 (Fenwal Division, Baxter Healthcare Corp., Deerfield, IL) with or without (randomly allocated) an inline whole blood leukoreduction filter (Sepacell RS-2000; Asahi Kasei Kuraray Medical Co., Ltd.; Asahi Kasei Corp., Tokyo, Japan). The whole blood was separated by standard procedures in the UCSF blood bank and stored as packed RBCs (PRBCs) in the UCSF blood bank under standard conditions.

Each volunteer was studied twice, on 2 different days, separated by 1 week (Fig. 1). On each of the 2 study days, the volunteer rested in a semirecumbent position, with his or her head elevated at approximately 30° to 45°. The same elevation was maintained, as the angle of the lungs influences respiratory quotient (R) and dead space to tidal volume ratio (VD/VT).19 An IV cannula (16 or 18 gauge) was placed in each arm, and a 22-gauge cannula inserted in a radial artery after infiltration with a local anesthetic. After 10 to 15 minutes of quiet rest, the volunteer was fitted with a nose clip so that all breathing was through the mouth. On the first study day, the subject had a practice session for familiarization with the equipment and procedures. During test periods (15 minutes [95% confidence interval {CI}: 14–16 minutes] after hemodilution—blood withdrawal with maintenance of isovolemia—see below; and 66 minutes [95% CI: 65–67 minutes] after transfusion) subjects breathed room air through a rubber mouthpiece that was connected to a low-resistance valve (model 2630; Hans Rudolph, Inc., Kansas City, MO), into a Douglas bag. A low-resistance 3-way valve was interposed between the large-bore (43-mm diameter) tubing connecting the Rudolph valve and the Douglas bag, allowing the subject to exhale through the system without collection of expired gas until stability had been reached for at least several minutes. Stability was determined by constant monitoring of mixed expired partial pressure of CO2 (CO2 sensor P-61B and CO2 analyzer CD-3A; AEI Technologies, Pittsburgh, PA). These instruments were calibrated at the start of each study day, and several times during each study day. Electrical signals were recorded using a Macintosh computer with software written by one of us (JF) for LabVIEW (version 8.5 or 8.6; National Instruments Corporation, Austin, TX). After mixed expired CO2 stabilized, expired gas was collected for 3 minutes in a Douglas bag that had been previously completely evacuated by use of suction.

Figure 1
Figure 1:
Study procedures and mean times. Ninety-five percent confidence intervals of the times appear in the text. RBCs = red blood cells.

During each test period, in addition to determination of mixed expired partial pressures of O2 and CO2, blood was sampled from the arterial cannula for measurement of blood gases (ABL 500 or ABL 800; Radiometer America, Cleveland, OH), complete blood count, cytokine concentrations (see below), hemoglobin concentration, and oxyhemoglobin saturation (OSM 3 Hemoximeter; Radiometer America).

After initial measurements, 2 U (450 mL each) of blood were withdrawn through an IV cannula into a standard blood collection bag containing CPDA-1 (Fenwal/Baxter). These units were not LR on either study day and were processed into PRBCs by the UCSF blood bank in standard fashion. Albumin, human 5% (Grifols, Los Angeles, CA) was simultaneously infused through the IV cannula in the other arm, to maintain isovolemia (1.1 times the volume of blood removed20). A second set of measurements was made (see above; Fig. 1) and a sample of blood was taken 15 minutes (95% CI: 14–16 minutes) after completion of this procedure.

After a rest period of 60 minutes (95% CI: 57–63 minutes), 2 U of autologous RBCs (approximate volume 193 mL per unit) were infused for 28 minutes (95% CI: 26–30 minutes). On one study day, the infused erythrocytes were those stored for 3 to 4 weeks, and on the other study day, they were the erythrocytes withdrawn the day of the study (order of study days randomly allocated). Each volunteer was studied with both fresh and stored RBCs, on separate days. Stored erythrocytes were randomly allocated to be LR or not. On both study days, another set of measurements and collection of blood samples was performed 66 minutes (95% CI: 65–67 minutes) later.


We measured the following cytokine concentrations (minimal detectable concentration in picograms per milliliter) in volunteer blood samples each time pulmonary function was assessed: interleukin (IL)-1ra (0.49), IL-1β (0.23), IL-2 (0.13), IL-4 (0.03), IL-6 (0.2), IL-8 (0.16), IL-10 (0.17), granulocyte macrophage–colony stimulating factor (0.07), interferon-γ (0.16), tumor necrosis factor-α (0.55), and vascular endothelial growth factor (VEGF) (0.23) using a Bio-Rad cytokine kit (Bio-Rad Laboratories, Hercules, CA) and an ELISA-based assay (Luminex's xMAP® technology; Invitrogen Corp., Life Technologies, Carlsbad, CA). These cytokines were also measured in the RBCs at the time of infusion.

Primary End Point

The pulmonary gradient for oxygen—the difference between partial pressures of oxygen in the alveoli (PAO2) and arterial blood (PaO2)—(AaDO2) was assessed on each study day after isovolemic removal of 2 U of blood (before transfusion of PRBCs), and again after transfusion of the autologous PRBCs. The primary end point of the study was the difference of the effect of stored autologous RBCs with that of fresh autologous RBCs on the change of AaDO2 between these 2 times.

The PAO2 was calculated from the alveolar gas equation21:

R was assumed to be 0.86.22,23

Secondary End Point

The secondary end point was the difference in change of VD/VT from after withdrawal of 2 U of blood but before transfusion to after transfusion, between stored autologous and fresh autologous RBCs.

VD/VT was calculated as: (PaCO2 − PECO2)/PaCO2

where PECO2 is the partial pressure of CO2 in mixed expired gas.


We determined the number of volunteers to be studied by an a priori 2-tailed power analysis using preliminary data from healthy patients transfused during posterior spinal surgery. The analysis, assuming a correlation of 0.5 between changes in one condition compared with changes in the second condition, determined that studying 35 volunteers would give the study a power of 0.8 with an α of 0.05. We had planned to study 38 volunteers to allow for 3 volunteers to withdraw before completing the full study; however, that was not necessary because all completed the study.

Data were tested for normal distribution using the Shapiro-Wilk test. Comparisons between groups and within groups (measurements made after removal of blood but before transfusion were compared with those made 66 minutes after transfusion) were performed using the Student paired t test. For data that were not normally distributed (cytokines), comparisons were made using the Mann-Whitney U and Wilcoxon signed rank tests. A nonparametric test, Spearman ρ, was used to test for correlation among cytokine concentrations. Examination of possible effects of order of experimental day (fresh or stored), AaDO2, PaO2, and VD/VT were compared for all volunteers using the Student paired t test. All tests were 2 sided. Statistical tests were performed using JMP (version 7.0; SAS Institute) or InStat (version 3.0b for Macintosh; GraphPad Software, Inc., La Jolla, CA). Statistical significance was accepted at P ≤ 0.05. Presented P values were not modified for analyses of multiple testing of cytokine concentrations and associations.

Data are presented as mean (95% CI) or median (interquartile range) unless otherwise indicated.


Thirty-five volunteers completed the study; 20 transfused with LR RBCs, and 15 transfused with non-LR RBCs. There were no statistical differences for gender, age, ethnicity, height, weight, or body mass index between those transfused with LR or non-LR RBCs (Table 1).

Table 1
Table 1:

“Fresh” RBCs were transfused 1.67 hours (95% CI: 1.53–1.80 hours), and “stored” RBCs were transfused 24.5 days (24.0–24.9 days) after collection.


Data from 3 volunteers were excluded from analysis (1 LR, 2 not LR) for AaDO2, PaO2, and VD/VT because of calibration errors of the blood gas instrument.

Primary Outcome Measure: AaDO2

There were no differences for absolute values or change of AaDO2 between LR and non-LR RBCs at any time point on either day (transfusion with fresh or 24.5-day stored RBCs; range of P values: 0.65–0.99). Therefore, the data from the 2 groups were pooled for comparison of transfusion with fresh or 24.5-day stored RBCs.

Transfusion of either fresh or stored RBCs increased AaDO2. The change in AaDO2 did not differ (P = 0.85) when volunteers were transfused with fresh (change: 2.8 mm Hg [0.8–4.8 mm Hg]; P = 0.007) or stored (change: 3.0 mm Hg [1.4–4.7 mm Hg]; P = 0.0006) RBCs (Fig. 2). The difference in AaDO2 from before to after RBC transfusion between fresh and older RBCs was 0.24 mm Hg (95% CI: −2.8 to 3.2 mm Hg). This value is 8% of the effect size, a value substantially less than that frequently required to establish “equivalence.”

Figure 2
Figure 2:
Alveolar to arterial difference for PO2 (AaDO2) from before (white bars) to after (horizontally striped bars) transfusion increased significantly after transfusion with red blood cells (RBCs) stored for 1.7 hours (*P = 0.007) and RBCs stored for 24.5 days (#P = 0.0006). The change of AaDO2 (filled bars) did not differ (P = 0.85) between the RBCs of different storage duration. Data are mean, 95% confidence intervals.

AaDO2 was greater (P = 0.021) after transfusion of stored (7.0 mm Hg [5.1–9.0 mm Hg]) than after fresh (4.8 mm Hg [2.4–7.1 mm Hg]) RBCs; however, this difference seems to have been influenced by a numerical, but not statistically significant, difference before transfusion (P = 0.12).

Arterial Blood Partial Pressure of Oxygen

There were no differences for PaO2 between LR and non-LR RBCs at any time point on either day (transfusion with fresh or 24.5-day stored RBCs; range of P values: 0.36–0.95). Therefore the data from the 2 groups were pooled for comparison of transfusion with fresh or 21-day stored RBCs.

PaO2 did not change with transfusion of fresh RBCs (change: −1.4 mm Hg [−3.8 to 1.0 mm Hg]; P = 0.25) but decreased by a small but statistically significant amount with transfusion of stored RBCs (2.0 mm Hg [−4.0 to −0.1 mm Hg]; P = 0.038; Fig. 3). However, the change of PaO2 did not differ between transfusion of fresh or stored RBCs (P = 0.62). Sixty-six minutes after transfusion, PaO2 did not quite reach a statistically significant difference (P = 0.063) between fresh (102.5 mm Hg [98.9–106.1 mm Hg]) and stored (100.3 mm Hg [97.9–102.9 mm Hg]) RBC groups.

Figure 3
Figure 3:
Change (filled bars) of PaO2 from before (white bars) to after (horizontally striped bars) transfusion did not differ (P = 0.62) between the red blood cells (RBCs) of different storage duration. Data are mean, 95% confidence intervals. *P = 0.038 compared with before transfusion of RBCs stored for 25 days.

Ratio of Respiratory Dead Space Volume to Tidal Volume

There were no differences for VD/VT between LR and non-LR RBCs at any time point on either day (transfusion with fresh or older stored RBCs; range of P values: 0.47–0.88).

VD/VT did not change with transfusion of either fresh (change: 0.00 [−0.02 to 0.02]; P = 0.84) or stored (0.01 [−0.01 to 0.03]; P = 0.47) RBCs (Fig. 4). There was no statistical difference between these tiny changes comparing transfusion of fresh RBCs with that of stored RBCs (P = 0.76).

Figure 4
Figure 4:
Change (filled bars) of ratio of respiratory dead space to tidal volume (VD/VT) from before (white bars) to after (horizontally striped bars) transfusion did not differ (P = 0.76) between the red blood cells (RBCs) of different storage duration. VD/VT did not change with transfusion of RBCs of either storage duration. Data are mean, 95% confidence intervals.

The order of experimental day (fresh or stored) did not affect changes caused by transfusion for AaDO2 (P > 0.99), PaO2 (P = 0.67), or VD/VT (P = 0.77).


All cytokine concentrations were analyzed statistically using the Wilcoxon signed rank test, because nearly all were not distributed normally.

Transfused RBCs (In Vitro)

Concentrations of all measured cytokines were less in LR RBCs than fresh RBCs (all P < 0.0001). All cytokines in stored LR RBCs were less than in stored non-LR RBCs (all P < 0.0001 except granulocyte macrophage–colony stimulating factor, P = 0.008), except for IL-10 (P = 0.15).

In Vivo

Nearly all in vivo cytokine concentrations after blood withdrawal (“dilution”) did not differ between fresh and stored days or between those who were to receive LR or non-LR RBCs. Transfusion of LR RBCs on the fresh day did not change in vivo concentration of any cytokine in the LR group. In the non-LR group on the fresh day, transfusion of non-LR RBCs did produce a few statistically significant, but relatively modest changes in some cytokine concentrations. Transfusion of stored LR RBCs did not result in in vivo changes of most cytokines except for small statistically significant increases of IL-6 (median [interquartile range]: 0.3 pg/mL [0.1–1.2] to 1.5 pg/mL [0.4–3.5]; P = 0.018), IL-10 (3.4 pg/mL [0.6–5.3] to 5.3 pg/mL [3.7–8.8]; P = 0.016), and VEGF (8 pg/mL [5–11] to 12 pg/mL [8–15]; P = 0.014). Transfusion of stored non-LR RBCs resulted in significant, modest increases of IL-6 (0.2 pg/mL [0.1–0.9] to 2.8 pg/mL [1.2–3.2]; P = 0.040), IL-8 (1.7 pg/mL [1.0–2.4] to 3.1 pg/mL [1.8–4.1]; P = 0.020), and VEGF (11 pg/mL [5–16] to 18 pg/mL [13–31]; P = 0.0034) concentrations. In comparing LR with non-LR RBC transfusion, both on fresh and stored days, the only significant difference was a larger VEGF in vivo concentration after non-LR RBCs than after LR RBCs (P = 0.018; Fig. 5). Similarly, the only significant increase in cytokine concentration comparing the change from before to after transfusion between LR and non-LR was for VEGF for transfusion of stored RBCs (P = 0.0046).

Figure 5
Figure 5:
Transfusion of fresh (*P = 0.0046) or stored (leukoreduced [LR]: #P = 0.014; non-LR: †P = 0.0034) red blood cells (RBCs) increased in vivo vascular endothelial growth factor (VEGF) concentrations in recipients. VEGF increased (“change”) more after transfusion of non-LR RBCs than after LR RBCs (¥P = 0.0046). LR on fresh day (1.7 hours of storage) refers to the group that received LR RBCs on the stored day; all RBCs transfused after 1.7 hours of storage were non-LR. White bars are before transfusion; horizontally striped bars are after transfusion, and filled bars are change from before to after transfusion. Data are median, interquartile range (IQR).

With the exception of VEGF, in vivo concentrations of each cytokine correlated with in vivo concentrations of every other cytokine (all P < 0.05, most <0.0001; range of Spearman ρ: 0.18–0.80). The only cytokines with which VEGF correlated were IL-6 (P = 0.029; Spearman ρ = 0.18) and IL-8 (P < 0.0001; Spearman ρ = 0.52). VEGF was the only cytokine with in vivo concentrations that correlated with AaDO2 (P = 0.018; Spearman ρ = 0.21). The correlation was stronger when transfusion of only stored non-LR RBCs was considered (P = 0.0010; Spearman ρ = 0.38). Changes of in vivo concentration of all cytokines correlated with changes of all other cytokines (nearly all P < 0.0001; range of Spearman ρ: 0.33–0.87). No change in cytokine concentration correlated with change of AaDO2 (range of Spearman ρ: 0.004–0.17); however, correlation of change in VEGF with change in AaDO2 nearly reached significance (Spearman ρ = 0.21; P = 0.099).


The main findings of this study are: (1) autologous RBC transfusion produces a mild decrement in gas exchange in normal healthy adults; (2) the pulmonary effects do not differ between RBCs that are fresh or have been stored for >21 days; (3) the pulmonary impairment is not affected by leukoreduction of the transfused RBCs despite a larger concentration of cytokines in the non-LR stored PRBCs than in the LR stored PRBCs; (4) transfusion of autologous RBCs increased the in vivo concentrations of several cytokines; and (5) VEGF was the only cytokine tested with in vivo concentrations that correlated with AaDO2.

The data reported herein demonstrate that RBC transfusion results in a subtle decrement of pulmonary gas exchange for oxygen, supporting our hypothesis that lung injury related to RBC transfusion may produce a wider spectrum in humans than has been hitherto recognized, and not only overt TRALI. RBC transfusion may cause lung injury in animal models, but the limited data in humans supporting this concept emanate primarily from severe injury (TRALI) and retrospective analyses of databases. There have been 2 mechanisms postulated for such injury: (1) damage mediated by antibodies in donor plasma; and (2) damage induced by biologically active compounds in stored blood or blood components. Our experimental design allowed us to test the second possibility by transfusing autologous RBCs, thus eliminating the possibility of an antibody-mediated process. We demonstrated that transfusion of autologous RBCs causes a change in pulmonary gas exchange of oxygen, albeit small and in healthy volunteers, as well as an increase of concentrations of several in vivo cytokines after transfusion, thus implicating biologically active compounds as a likely etiology. Additional evidence points to the potential of neutrophil-platelet aggregates in producing organ damage.24,25 It is possible that the effects we observed might be accentuated by a patient's clinical condition before transfusion. Such patient risk factors for the development of TRALI have been identified recently in a large, prospective, observational study.26

Stored RBCs have a relatively small amount of plasma stored and administered with them. Accordingly, the number of cases and incidence of severe lung injury and death attributed to RBC transfusion has been relatively low compared with those attributed to transfusion of plasma or platelets.5 Thus, the second etiology would seem more likely for many cases in which RBC transfusion may be implicated as causing lung injury. Data from laboratory experiments are not unanimous as to whether RBCs alone without a “first hit” and primed neutrophils can cause lung injury.10,2730 Our data demonstrate that RBC transfusion in healthy humans, without a first hit, and without a possible antibody-antigen mechanism can alter pulmonary function. Thus, the etiology of the changes observed in our volunteers must have been mediated by a mechanism not involving antibodies.

A potential mechanism for RBC-induced pulmonary injury is that of alteration of the pulmonary microvascular endothelium. Supernatant fluid from human RBCs stored for 3 days to 6 weeks increases the permeability of human lung microvascular endothelial cell monolayers.31 VEGF has a similar action,32 and increases pulmonary vascular permeability33 increasing pulmonary edema in mice.34,35 In our experiments, VEGF was the only cytokine in vivo concentrations that correlated with AaDO2. VEGF-induced increased permeability of pulmonary endothelial cells could explain our finding of increased AaDO2 with transfusion and may be a contributing factor to TRALI and transfusion-associated circulatory overload as well. However, our finding of an association of in vivo concentrations of VEGF with AaDO2 is not a determination of a cause-and-effect relationship.

It has been more difficult to implicate RBC transfusion as a cause of human clinical pulmonary dysfunction. Retrospective database analyses have been inconsistent in attempts to find associations between RBC transfusion and morbidities, including pulmonary dysfunction or infection,7,13,14,3641 prompting the initiation of several prospective randomized trials, including a large multicenter National Institutes of Health–sponsored effort (“RECESS”).42

Our results show that fresh RBCs and those stored for ≥21 days produced statistically equivalent subtle deficits in pulmonary gas exchange for oxygen. While proving our hypothesis that acute lung dysfunction after transfusion can have a subtle as well as severe component, we also simultaneously demonstrated that transfusion of stored RBCs did not produce more impairment than did transfusion of fresh RBCs in the volunteers that we studied. If cytokines are involved, this finding may suggest that the in vivo cytokine response could be of greater importance than the cytokine concentrations in the supernatant fluid of stored RBCs.

The lack of differences of PaO2 after transfusion would make clinical detection of such a subtle dysfunction difficult, because AaDO2 is not calculated frequently. The changes in gas exchange of oxygen that we observed after transfusion of 2 U of RBCs are sufficiently small that they would be difficult to observe in a clinical setting. This may help explain why a minor form of the disorder may not have been clinically recognized previously. Additionally, it is possible that transfusion of a larger volume of RBCs might produce a greater impairment of gas exchange that would be sufficient for clinical detection. Even if AaDO2 were to be calculated in patients, the absolute values of AaDO2 that we observed after transfusion, although representing statistically significant increases, were small and were within the normal range.43 Unless others examine changes in AaDO2 (as did we), rather than absolute values, the absolute values of AaDO2 would not be regarded as abnormal. Furthermore, AaDO2 is least in the supine position because of the lesser distance from the bottom to the top of the lung, and thus there is lesser variation of V˙A/Q˙.19,23 Because most patients would likely be supine if AaDO2 were to be assessed, detection of an AaDO2 abnormality would be even less likely. Additionally, patients breathing supplemental oxygen, but without their tracheas intubated, would have an inconstant FIO2, making accurate estimation of AaDO2 impossible. Lastly, our study design maximized the ability to detect changes by using repeated-measures analysis; this design and robust analysis would not be possible in a clinical setting. However, it is possible that these changes after transfusion might be more substantial after transfusions of larger volume.

The modest increases seen in IL-6, IL-8, and VEGF in the recipients of stored LR and non-LR RBC units may reflect a mild proinflammatory effect of these units, because these cytokines are mainly associated with acute inflammatory reactions.44 VEGF has a direct effect on vascular permeability through its ability to induce changes in endothelial cell adhesion and tight junction formation.32,45 VEGF increases pulmonary vascular permeability33 and overexpression of VEGF in mice produces pulmonary edema.34,35 Mild changes in vascular permeability would likely underlie the differences seen in AaDO2 in some of the transfusion recipients.

We found that VD/VT did not change with transfusion of either fresh or stored RBCs. Although increased VD/VT is a feature of early acute respiratory distress syndrome18 and is associated with increased mortality in those with acute respiratory distress syndrome, VD/VT is more sensitive to changes in pulmonary perfusion than ventilation.46 VD/VT is less in the supine than in the standing position19 owing to nonperfused, ventilated alveoli at upper lung regions when the lungs are longitudinal. Thus, it is not surprising that the changes in AaDO2 were not accompanied by changes in VD/VT, and implies that the changes we observed were not related to pulmonary perfusion.46


Several issues of our study deserve discussion. We did not find a difference between fresh and stored RBCs for the small, but statistically significant decrements in the primary outcome variable, pulmonary gas exchange for oxygen as assessed by AaDO2 or for PaO2. AaDO2 is a more reliable measure, because it is true gas exchange and accounts for other variables (R and PaCO2) that influence PaO2. Our study was powered to detect changes in AaDO2, but not PaO2. Similarly, it should be noted that our study design did not allow for repeated-measures analysis for comparison of LR and non-LR RBC transfusion and was not powered for this variable, and thus that comparison was less robust to permit detection of any difference(s) should it exist.

There is another possible interpretation of the etiology of our finding for changes in AaDO2. Inasmuch as the changes produced by fresh or stored RBCs did not differ, it is possible that they resulted from the augmentation of blood volume per se. The infusion of <300 mL RBCs (representing approximately <7% of the volunteers' blood volume) would have initially augmented blood volume by not more than this amount, but with time fluid shifts would have partially ameliorated toward normal both that increase and extracellular volume changes. Furthermore, all volunteers were healthy and relatively young, with compliant vessels and myocardium; the small augmentation would not likely have altered cardiac preload or afterload. We could have controlled partially for this variable by studying each volunteer a third time without withdrawing 2 U of blood, but with administration of an equivalent volume of colloid. However, pharmacokinetic differences between these colloids and RBCs would not provide an exactly equivalent control. Our assertion that the relatively small volume per se of RBCs transfused did not cause the changes we noted in gas exchange is supported by the stability of the volunteers' arterial blood pressure (4 mm Hg non–statistically significant difference) and was in keeping with blood pressure changes in similar volunteers undergoing isovolemic hemodilution, owing to changes in hemoglobin concentration and systemic vascular resistance20 (viscosity and total vessel cross-sectional area47).

Our calculations for AaDO2 included an assumption for the value of R, because we did not measure oxygen consumption. Regional lung R varies with body position and within the lung based on regional differences in V˙A/Q˙.19,23 Use of 0.86 is a reasonable estimate based on the body position of our volunteers, because in the supine position, any regional V˙/Q˙ variations are small and the semirecumbent position serves to lessen the differences of the erect position, and published values.22,23 Any difference between the value used and the true overall average value would have been small and have had a minor effect on calculated PAO2 and thus AaDO2. Furthermore, any difference would have been included in all calculations for AaDO2 and thus would not likely have affected the value of the primary outcome measure, the change in AaDO2.

The purpose of our study was to test whether stored RBCs could alter pulmonary function. We did not focus on mechanisms of a potential effect, because we did not know a priori whether there would be any changes with RBC transfusion. Assessments of cytokine concentrations and neutrophil priming could be a logical next step to further identify the mechanisms responsible for the changes we observed. Thus, our exploratory evaluations of potential mechanisms were limited, and further constrained by the covariation of most cytokines. The observed correlation of in vivo VEGF concentrations and AaDO2, and the knowledge of the action of VEGF on endothelial layers and pulmonary vascular permeability, are suggestive that VEGF may have a role in the small oxygenation changes we detected, but do not provide proof.

Although the magnitude of the observed changes was small, it is possible that these would be amplified with higher volume transfusions or in patients with underlying disease or with an inflammatory state with either primed neutrophils or increased pulmonary vascular permeability. This study is the first demonstration of decrement in gas exchange of oxygen after autologous RBC transfusion, and points to factors other than antibody-antigen reactions as an etiology.

We conclude that RBC transfusion causes subtle pulmonary dysfunction, as evidenced by impaired gas exchange for oxygen, supporting our hypothesis that lung impairment after transfusion includes a wide spectrum of physiologic derangements and may not require an existing state of altered physiology. These data do not support the hypothesis that transfusion of RBCs stored for >21 days is more injurious than that of fresh RBCs in healthy people. These findings may not apply to those at higher risk for pulmonary injury because of an underlying condition or disease. Similarly, we did not concomitantly transfuse plasma or platelets, and it is possible that these might accentuate the observed RBC-induced impairment. The data suggest that VEGF may have a role in the small oxygenation changes that we detected.


Name: Richard B. Weiskopf, MD.

Contribution: This author helped designed the study, analyze the data, write the manuscript, and review the manuscript.

Name: John Feiner, MD.

Contribution: This author helped design the study, acquire the data, analyze the data, write the manuscript, and review the manuscript.

Name: Pearl Toy, MD.

Contribution: This author helped design the study, write the manuscript, and review the manuscript.

Name: Jenifer Twiford, BSN.

Contribution: This author helped acquire the data, write the manuscript, and review the manuscript.

Name: David Shimabukuro, MD.

Contribution: This author helped acquire the data and review the manuscript.

Name: Jeremy Lieberman, MD.

Contribution: This author helped acquire the data and review the manuscript.

Name: Mark R. Looney, MD.

Contribution: This author helped acquire the data, write the manuscript, and review the manuscript.

Name: Clifford A. Lowell, MD, PhD.

Contribution: This author helped write the manuscript and review the manuscript.

Name: Michael A. Gropper, MD, PhD.

Contribution: This author helped design the study, analyze the data, write the manuscript, and review the manuscript.

This manuscript was handled by: Jerrold H. Levy, MD, FAHA.


1. Popovsky MA, Abel MD, Moore SB. Transfusion-related acute lung injury associated with passive transfer of antileukocyte antibodies. Am Rev Respir Dis 1983; 128: 185–9
2. Popovsky MA, Moore SB. Diagnostic and pathogenetic considerations in transfusion-related acute lung injury. Transfusion 1985; 25: 573–7
3. Toy P, Popovsky MA, Abraham E, Ambruso DR, Holness LG, Kopko PM, McFarland JG, Nathens AB, Silliman CC, Stroncek D. Transfusion-related acute lung injury: definition and review. Crit Care Med 2005; 33: 721–6
4. Keller-Stanislawski B, Reil A, Gunay S, Funk MB. Frequency and severity of transfusion-related acute lung injury: German haemovigilance data (2006–2007). Vox Sang 2010; 98: 70–7
5. FDA. Fatalities Reported to FDA Following Blood Collection and Transfusion: Annual Summary for Fiscal Year 2009. Silver Spring, MD: U.S. Food and Drug Administration, 2010;
6. Toy P, Popovsky MA, Abraham E, Ambruso DR, Holness L, Kopko P, McFarland J, Nathens A, Silliman CC, Stroncek D. National Heart Lung and Blood Institute Working Group on TRALI: transfusion-related acute lung injury—definition and review. Crit Care Med 2005; 33: 721–6
7. Cornet AD, Zwart E, Kingma SD, Groeneveld AB. Pulmonary effects of red blood cell transfusion in critically ill, non-bleeding patients. Transfus Med 2010; 20: 221–6
8. Covin RB, Ambruso DR, Engl KM, Kelher MR, Mehdizadehkashi Z, Boshkov LK, Masuno T, Moore EE, Kim FJ, Silliman CC. Hypotension and acute pulmonary insufficiency following transfusion of autologous red blood cells during surgery: a case report and review of the literature. Transfus Med 2004; 14: 375–83
9. Silliman CC, Bjornsen AJ, Wyman TH, Kelher M, Allard J, Bieber S, Voelkel NF. Plasma and lipids from stored platelets cause acute lung injury in an animal model. Transfusion 2003; 43: 633–40
10. Silliman C, Voelkel N, Allard JD, Eklzi D, Tuder R, Johnson J, Ambrusco D. Plasma and lipids from stored packed red blood cells cause acute lung injury in an animal model. J Clin Invest 1998; 101: 1458–67
11. Silliman C, Paterson A, Dickey W, Stroncek D, Popovsky M, Caldwell S, Ambruso D. The association of biologically active lipids with the development of transfusion-related acute lung injury: a retrospective study. Transfusion 1997; 37: 719–26
12. Silliman CC, Boshkov LK, Mehdizadehkashi Z, Elzi DJ, Dickey WO, Podlosky L, Clarke G, Ambruso DR. Transfusion-related acute lung injury: epidemiology and a prospective analysis of etiologic factors. Blood 2003; 101: 454–62
13. Koch CG, Li L, Sessler DI, Figueroa P, Hoeltge GA, Mihaljevic T, Blackstone EH. Duration of red-cell storage and complications after cardiac surgery. N Engl J Med 2008; 358: 1229–39
14. Gajic O, Rana R, Mendez JL, Rickman OB, Lymp JF, Hubmayr RD, Moore SB. Acute lung injury after blood transfusion in mechanically ventilated patients. Transfusion 2004; 44: 1468–74
15. Vamvakas EC, Carven JH. Length of storage of transfused red cells and postoperative morbidity in patients undergoing coronary artery bypass graft surgery. Transfusion 2000; 40: 101–9
16. van de Watering L, Lorinser J, Versteegh M, Westendord R, Brand A. Effects of storage time of red blood cell transfusions on the prognosis of coronary artery bypass graft patients. Transfusion 2006; 46: 1712–8
17. Edgren G, Kamper-Jorgensen M, Eloranta S, Rostgaard K, Custer B, Ullum H, Murphy EL, Busch MP, Reilly M, Melbye M, Hjalgrim H, Nyren O. Duration of red blood cell storage and survival of transfused patients (CME). Transfusion 2010; 50: 1185–95
18. Nuckton TJ, Alonso JA, Kallet RH, Daniel BM, Pittet JF, Eisner MD, Matthay MA. Pulmonary dead-space fraction as a risk factor for death in the acute respiratory distress syndrome. N Engl J Med 2002; 346: 1281–6
19. Riley RL, Permutt S, Said S, Godfrey M, Cheng TO, Howell JB, Shepard RH. Effect of posture on pulmonary dead space in man. J Appl Physiol 1959; 14: 339–44
20. Weiskopf RB, Viele MK, Feiner J, Kelley S, Lieberman J, Noorani M, Leung JM, Fisher DM, Murray WR, Toy P, Moore MA. Human cardiovascular and metabolic response to acute, severe isovolemic anemia. JAMA 1998; 279: 217–21
21. Fenn WO, Rahn H, Otis AB. A theoretical study of the composition of the alveolar air at altitude. Am J Physiol 1946; 146: 637–53
22. Marra M, Scalfi L, Contaldo F, Pasanisi F. Fasting respiratory quotient as a predictor of long-term weight changes in non-obese women. Ann Nutr Metab 2004; 48: 189–92
23. West JB. Regional differences in gas exchange in the lung of erect man. J Appl Physiol 1962; 17: 893–8
24. Looney MR, Matthay MA. Neutrophil sandwiches injure the microcirculation. Nat Med 2009; 15: 364–6
25. Keating FK, Butenas S, Fung MK, Schneider DJ. Platelet-white blood cell (WBC) interaction, WBC apoptosis, and procoagulant activity in stored red blood cells. Transfusion 2011; 51: 1086–95
26. Toy P, Gajic O, Bacchetti P, Looney MR, Gropper M, Hubmayr RD, Lowell CA, Norris PJ, Murphy EL, Weiskopf RB, Wilson G, Koenigsberg M, Lee L, Schuller R, Wu P, Grimes B, Gandhi MJ, Winters JL, Mair D, Hirschler N, Rosen RS, Matthay MA; for the TRALI Study Group. Transfusion related acute lung injury: incidence and risk factors. Blood 2011; [Epub ahead of print]
27. Silliman CC, Clay KL, Thurman GW, Johnson CA, Ambruso DR. Partial characterization of lipids that develop during the routine storage of blood and prime the neutrophil NADPH oxidase. J Lab Clin Med 1994; 124: 684–94
28. Kelher MR, Masuno T, Moore EE, Damle S, Meng X, Song Y, Liang X, Niedzinski J, Geier SS, Khan SY, Gamboni-Robertson F, Silliman CC. Plasma from stored packed red blood cells and MHC class I antibodies causes acute lung injury in a 2-event in vivo rat model. Blood 2009; 113: 2079–87
29. Vlaar AP, Hofstra JJ, Levi M, Kulik W, Nieuwland R, Tool AT, Schultz MJ, de Korte D, Juffermans NP. Supernatant of aged erythrocytes causes lung inflammation and coagulopathy in a “two-hit” in vivo syngeneic transfusion model. Anesthesiology 2010; 113: 92–103
30. Looney MR, Nguyen JX, Hu Y, Van Ziffle JA, Lowell CA, Matthay MA. Platelet depletion and aspirin treatment protect mice in a two-event model of transfusion-related acute lung injury. J Clin Invest 2009; 119: 3450–61
31. Rao RS, Howard CA, Teague TK. Pulmonary endothelial permeability is increased by fluid from packed red blood cell units but not by fluid from clinically-available washed units. J Trauma 2006; 60: 851–8
32. Lal BK, Varma S, Pappas PJ, Hobson RW II, Duran WN. VEGF increases permeability of the endothelial cell monolayer by activation of PKB/akt, endothelial nitric-oxide synthase, and MAP kinase pathways. Microvasc Res 2001; 62: 252–62
33. Mura M, Binnie M, Han B, Li C, Andrade CF, Shiozaki A, Zhang Y, Ferrara N, Hwang D, Waddell TK, Keshavjee S, Liu M. Functions of type II pneumocyte-derived vascular endothelial growth factor in alveolar structure, acute inflammation, and vascular permeability. Am J Pathol 2010; 176: 1725–34
34. Lee CG, Link H, Baluk P, Homer RJ, Chapoval S, Bhandari V, Kang MJ, Cohn L, Kim YK, McDonald DM, Elias JA. Vascular endothelial growth factor (VEGF) induces remodeling and enhances TH2-mediated sensitization and inflammation in the lung. Nat Med 2004; 10: 1095–103
35. Kaner RJ, Ladetto JV, Singh R, Fukuda N, Matthay MA, Crystal RG. Lung overexpression of the vascular endothelial growth factor gene induces pulmonary edema. Am J Respir Cell Mol Biol 2000; 22: 657–64
36. Frenzel T, Sibrowski W, Westphal M. Red-cell storage and complications of cardiac surgery. N Engl J Med 2008; 358: 2841
37. Hall SW. Red-cell storage and complications of cardiac surgery. N Engl J Med 2008; 358: 2841
38. Benjamin RJ, Dodd RY. Red-cell storage and complications of cardiac surgery. N Engl J Med 2008; 358: 2840–1
39. Leal-Noval SR, Jara-Lopez I, Garcia-Garmendia JL, Marin-Niebla A, Herruzo-Aviles A, Camacho-Larana P, Loscertales J. Influence of erythrocyte concentrate storage time on postsurgical morbidity in cardiac surgery patients. Anesthesiology 2003; 98: 815–22
40. Gong MN, Thompson BT, Williams P, Pothier L, Boyce PD, Christiani DC. Clinical predictors of and mortality in acute respiratory distress syndrome: potential role of red cell transfusion. Crit Care Med 2005; 33: 1191–8
41. Watkins TR, Rubenfeld GD, Martin TR, Nester TA, Caldwell E, Billgren J, Ruzinski J, Nathens AB. Effects of leukoreduced blood on acute lung injury after trauma: a randomized controlled trial. Crit Care Med 2008; 36: 1493–9
42. Steiner ME, Assmann SF, Levy JH, Marshall J, Pulkrabek S, Sloan SR, Triulzi D, Stowell CP. Addressing the question of the effect of RBC storage on clinical outcomes: the Red Cell Storage Duration Study (RECESS) (Section 7). Transfus Apher Sci 2010; 43: 107–16
43. Farhi LE, Rahn H. A theoretical analysis of the alveolar-arterial O2 difference with special reference to the distribution effect. J Appl Physiol 1955; 7: 699–703
44. Xiang M, Fan J. Pattern recognition receptor-dependent mechanisms of acute lung injury. Mol Med 2010; 16: 69–82
45. Bates DO. Vascular endothelial growth factors and vascular permeability. Cardiovasc Res 2010; 87: 262–71
46. Severinghaus JW, Stupfel M. Alveolar dead space as an index of distribution of blood flow in pulmonary capillaries. J Appl Physiol 1957; 10: 335–48
47. Murray JF, Escobar E, Rapaport E. Effects of blood viscosity on hemodynamic responses in acute normovolemic anemia. Am J Physiol 1969; 216: 638–42
© 2012 International Anesthesia Research Society