Transfusion of red blood cells after prolonged storage produces harmful effects that are mediated by iron and inflammation

Abstract

Although red blood cell (RBC) transfusions can be lifesaving, they are not without risk. In critically ill patients, RBC transfusions are associated with increased morbidity and mortality, which may increase with prolonged RBC storage before transfusion. The mechanisms responsible remain unknown. We hypothesized that acute clearance of a subset of damaged, stored RBCs delivers large amounts of iron to the monocyte/macrophage system, inducing inflammation. To test this in a well-controlled setting, we used a murine RBC storage and transfusion model to show that the transfusion of stored RBCs, or washed stored RBCs, increases plasma nontransferrin bound iron (NTBI), produces acute tissue iron deposition, and initiates inflammation. In contrast, the transfusion of fresh RBCs, or the infusion of stored RBC-derived supernatant, ghosts, or stroma-free lysate, does not produce these effects. Furthermore, the insult induced by transfusion of stored RBC synergizes with subclinical endotoxinemia producing clinically overt signs and symptoms. The increased plasma NTBI also enhances bacterial growth in vitro. Taken together, these results suggest that, in a mouse model, the cellular component of leukoreduced, stored RBC units contributes to the harmful effects of RBC transfusion that occur after prolonged storage. Nonetheless, these findings must be confirmed by prospective human studies.

Figure 1

Transfusion of stored RBCs. Transfusions of stored RBCs lead to increased RBC clearance, tissue iron delivery, and circulating NTBI levels, compared with transfusions of fresh RBCs, stored RBC-derived supernatant, or ghosts prepared from stored RBCs. All transfusion recipients were male C57BL/6 mice (8-12 weeks of age). The results are presented as mean (± SEM) except where specified. (A) Leukoreduced fresh FVB/NJ mouse RBCs (< 24-hour storage; n = 3; □) and stored RBCs (2-week storage; n = 5; ■) were transfused (400 μL at 17.0-17.5 g/dL of hemoglobin), and survival of transfused RBCs was calculated by dual-label flow cytometric tracking at 10 minutes, 30 minutes, 1 hour, 2 hours (only for stored RBCs), and 24 hours after transfusion. The results are from 1 representative experiment and are presented as mean (± SD). (B) A representative image of spleens obtained from mice 2 hours after transfusion with fresh RBCs or stored RBCs. (C) Mean spleen weight of mice transfused with fresh RBCs (n = 13) and stored RBCs (n = 13). (D) Aliquots (400 μL) of fresh RBCs (n = 13), stored RBCs (n = 13), washed stored RBCs (n = 13), stored RBC-derived supernatant (SN; n = 12), and ghosts prepared from stored RBCs (n = 8) were transfused. Total iron was measured in organs obtained at necropsy 2 hours after transfusion; the increases in iron are shown compared with levels measured in control, untransfused mice (n = 12). The results are combined from 3 separate experiments. (E) Mice were transfused as labeled (n = 5 per group) and plasma NTBI levels were measured 2 hours after transfusion. Note that the absence of an error bar indicates undetectable NTBI levels. The results are representative of 2 separate experiments; *P < .05; **P < .01; ***P < .001 compared with fresh RBC transfusions.

Figure 2

Macrophages are responsible for clearing transfused stored RBCs. All transfusion recipients and donors were syngeneic male C57BL/6 mice (8-12 weeks of age). (A) Mice were infused intraperitoneally with 2 mg of liposomal clodronate (n = 9) or control PBS-liposomes (n = 10) 48 hours before transfusion with stored RBCs. The 2-hour RBC survival was then measured. The 2-hour RBC survival (■) is indicated for each mouse and the horizontal bar indicates the mean. The results are representative of 2 separate experiments; ***P < .001 compared with treatment with PBS-liposomes. (B) Representative images of histologic sections of liver and spleen from mice treated with liposomal clodronate or control PBS-liposomes 48 hours before transfusion with stored RBCs, and stained with an anti–mouse F4/80 monoclonal antibody, as labeled. Note the absence of tissue macrophages in the liposomal clodronate–treated mice, as evidenced by the absence of brown staining cells. (C) Representative images of histologic sections from the liver of mice transfused with fresh or stored RBCs. Sections were stained with hematoxylin & eosin or with an anti–mouse F4/80 monoclonal antibody, as labeled. Arrows denote tissue macrophages that ingested RBCs. Brown staining is a result of F4/80 immunoreactivity of macrophages; the cytoplasmic staining is displaced to the periphery of the cells in mice transfused with stored RBCs because of the accumulation of ingested RBCs. Original magnification was ×400. Typical representative examples derived from 5 necropsies are shown.

Figure 3

Transfusion of stored RBCs induces dose-responsive proinflammatory cytokine responses. (A) Hemoglobinemia, as detected by absorbance, was observed in all mice (n = 8) transfused with stroma-free lysate derived from stored RBCs. Representative spectra of plasma (diluted 1:4 with PBS) obtained from mice 2 hours after transfusion with fresh RBCs (< 24-hour storage), stored RBCs (2-week storage), or stroma-free lysate derived from stored RBCs are shown. (B) Untransfused C57BL/6 mice (n = 13) or mice transfused with fresh RBCs (1u = 200 μL, n = 5; [ie, 1 human equivalent unit = 200 μL]; 2u = 400 μL, n = 17), stored RBCs (1u = 200 μL, n = 5; 2u = 400 μL, n = 17), washed stored RBCs (400 μL; n = 13), stored RBC-derived supernatant (SN, 400 μL, n = 12), ghosts prepared from stored RBCs (400 μL, n = 8), and stroma-free lysate derived from stored RBCs (400 μL, n = 8) were killed 2 hours after transfusion, and plasma cytokine levels were measured (as labeled); *P < .05; **P < .01; ***P < .001 compared with fresh RBCs.

Figure 4

Transfusion of stored RBCs induces an acute phase response.(A) SAA1-luciferase reporter mice were transfused with 200 μL of either fresh RBCs (< 24-hour storage) or stored RBCs (2-week storage) and luciferase activity measured by noninvasive bioluminescence imaging at multiple times up to 24 hours after transfusion (n = 3 per group). Results are representative of 2 experiments. (B) Bioluminescence was quantified over the hepatosplenic region of SAA1-luciferase reporter mice transfused with fresh RBCs (n = 6; ) or stored RBCs (n = 6; ■); *P < .01. (C) Circulating SAA1 protein levels in SAA1-luciferase reporter mice 24 hours after transfusion with fresh RBCs or stored RBCs (n = 6 per group); *P < .01. Results are combined from 2 separate experiments.

Figure 5

Transfusion of stored RBCs synergizes with the inflammatory response to LPS. C57BL/6 mice were infused with a subclinical dose of LPS (E coli 0111:B4 strain; 30 μg per mouse by tail-vein injection) immediately followed by transfusion of 400 μL of fresh RBCs or stored RBCs. Mice were killed 24 hours after transfusion, and plasma cytokines were measured (n = 5 per group). Results are representative of 2 experiments. *P < .05; **P < .01; ***P < .001 compared with mice infused with LPS plus stored RBCs.

Figure 6

Plasma from mice transfused with stored RBCs enhances bacterial growth in vitro. (A) Plasma (100 μL) was obtained from mice 2 hours after transfusion with 400 μL of fresh RBCs (n = 15), stored RBCs (n = 24), stored RBC-derived supernatant (SN, n = 12), washed stored RBCs (n = 13), or ghosts prepared from stored RBCs (n = 8). Plasma was also obtained from control, untransfused mice (n = 14) or 24 hours after transfusion with stored RBCs (n = 8). Samples were incubated at 37°C with shaking with ∼ 1 × 10⁶ CFU of E coli, as labeled. Bacterial growth was monitored every 30 minutes by absorbance at 600 nm for up to 5 hours. Bacterial growth in plasma obtained from mice 2 hours after transfusion with stored RBCs or washed stored RBCs began diverging from all other groups at 2.5 hours of incubation in vitro, and AUC (in parentheses) for each group was significantly different as indicated. (B) Pooled plasma samples (100 μL) from mice 2 hours after transfusion with 400 μL of fresh RBCs or stored RBCs were supplemented with either ferric citrate (20μM), sodium citrate (20μM), bovine serum albumin (BSA; 80μM), or protoporphyrin IX (20μM), and then incubated at 37°C with shaking with ∼ 1 × 10⁶ CFU of E coli. Bacterial growth was monitored every 30 minutes by absorbance at 600 nm for up to 5 hours in replicates of 5 per group. AUC (in parentheses) for growth in plasma from mice transfused with fresh RBCs, supplemented with or without sodium citrate, BSA, or protoporphyrin IX, differed significantly from the other 3 groups. (C) Pooled plasma (n = 4) were incubated with the iron chelator, DFO (20 μM), or with the iron-chelated form FO (20 μM) and inoculated with E coli as shown for the previous experiment. The AUC (in parentheses) for growth in plasma with DFO significantly differed from all other groups. (D) Pooled plasma (n = 5) was incubated with the iron chelator, 2,2′-dipyridyl (400μM), with or without ferric citrate (133μM) and inoculated with E coli, as shown for the previous experiment. The AUC (in parentheses) for growth in plasma with 2,2′-dipyridyl significantly differed from all other groups; *P < .05. Results are representative of at least 2 experiments and are shown as mean (± SEM). Note that the absence of an error bar is indicative of highly reproducible replicates with pooled plasma.

Figure 7

DFO treatment decreases the proinflammatory response induced by transfusion of stored RBCs. (A) Mice were pretreated with a PBS vehicle control (n = 28) or with 3 mg of DFO, with (n = 15) or without (n = 31) the addition of equimolar ferric citrate, immediately before transfusion with stored RBCs (400 μL). Mice were killed 2 hours after transfusion, and plasma cytokine levels were measured; *P < .05; **P < .01; ***P < .001 compared with mice infused with PBS vehicle and transfused stored RBCs. (B) Bioluminescence was quantified for 24 hours after transfusion over the hepatosplenic region of SAA1-luciferase reporter mice transfused with 200 μL of fresh RBCs (n = 3; ), the PBS vehicle control and stored RBCs (n = 3; ■), or 3 mg of DFO and stored RBCs (n = 6; ); P = .095 at 4 and 6 hours after transfusion comparing vehicle-treated and DFO-treated mice. (C) Proposed mechanistic pathway (the “iron hypothesis”) explaining how transfusion of older stored RBCs may induce adverse effects in patients. Transfusion of stored, but not fresh, RBCs delivers an acute bolus of RBCs and RBC-derived iron to the monocyte/macrophage system resulting in oxidative stress and inflammatory cytokine secretion. Some of the macrophage-ingested iron is also released back into the circulation (ie, NTBI) where it can also cause oxidative damage and enhance bacterial proliferation. SIRS indicates systemic inflammatory response syndrome.