Introduction

The majority of studies on perioperative blood transfusion in cancer patients have linked transfusion to adverse outcomes and decreased survival.1–7 Although a direct causal mechanism has not been elucidated, blood transfusion has been shown to induce anergy, T-suppressor cells, and clonal deletion.8 Underlying clinical variables surrounding the need for blood transfusion likely confound patient outcomes and interpretation of results. Our lab has shown that the acellular fraction of packed red cells (pRBCs) increases tumor cell proliferation and migration in vitro and augments pancreas cancer progression in vivo in an immunocompetent model of murine pancreatic adenocarcinoma.9–11 Interestingly, in vitro proliferation and migration increased with storage time, while augmented in vivo growth was evident with fresh blood, indicating pro-cancer mediators may be present in fresh (day 1) blood.9

Transfusion is a known risk factor for acute lung injury12 and a major independent risk factor for post-injury multiple organ failure following both massive and less severe trauma.13 During storage, plasma from pRBCs accumulates lipid mediators, which prime neutrophils, and have been implicated in transfusion-related acute lung injury.14–17 Transfusion of older blood is also associated with increased infections in severely injured patients.18 Additionally, receipt of blood products has been linked to more infectious complications and mortality in cardiac surgery patients.8,19 It has been hypothesized that the negative effects of blood transfusion may be mediated by “passenger leukocytes,” which release granules, anti-leukocyte and anti-HLA antibodies, apoptotic bodies, and other soluble mediators.20–22 However, clinical studies comparing leukocyte-reduced (LR) pRBCs versus non-leukoreduced (NLR) pRBCs have not demonstrated survival advantage with transfusion in patients with colorectal cancers.22–24 We hypothesize that stored packed red blood cells contain pro-cancer cytokines that augment tumor progression, and these soluble mediators increase with storage time and are attenuated with prestorage leukoreduction.

Materials and Methods

Blood

Following a protocol approved by the Colorado Multi-institutional Review Board, consent was obtained and ten healthy donors donated 450 mL of whole blood per American Association of Blood Banks criteria.25 Preparation of the acellular fraction of pRBCs has been described previously.9

Protein Expression in Stored pRBCs

Chemiarray™ Human Cytokine Antibody Array III Map (Chemicon International, Billerica, MA) analysis was used to investigate the expression of pro-cancer cytokines and growth factors in the acellular fraction of LR and NLR of day 1 (D.1) and day 42 (D.42) pRBCs. Aliquots of the acellular fraction of each tested unit of pRBCs were applied to the membranes according to the manufacturers’ instructions. Dulbecco’s minimal essential medium (DMEM, Invitrogen, Carlsbad, CA) was also analyzed as a negative control. Chemiluminescent images were acquired using a FlourChem™ imaging system (Alpha Innotech, Santa Clara, CA).

Enzyme-linked Immunosorbent Assay

ChemiArray™ analysis revealed increased expression of monocyte chemotactic protein-1 (MCP-1), regulated on activation, normal T cell expressed and secreted (RANTES), angiogenin, tumor necrosis factor-alpha (TNF-α), epidermal growth factor (EGF), and platelet-derived growth factor BB (PDGF-BB). To evaluate the concentrations of each of these soluble factors in stored pRBCs, separate aliquots of the acellular fraction of LR and NLR pRBCs from D.1, day 28 (D.28), and D.42 were quantified with enzyme-linked immunosorbent assay (ELISA) following the manufacturer’s instructions. ELISAs for MCP-1, RANTES, angiogenin, TNF-α, EGF, and PDGF-BB were purchased from R&D Systems (Minneapolis, MN).

Pan02 Cell Culture

The murine pancreatic adenocarcinoma line, Pan02, was obtained from the Developmental Therapeutics Program, NCI (Frederick, MD). Cells were maintained at 37°C in a mixture of 5% CO2 and 95% air in DMEM supplemented with 10% fetal bovine serum (FBS; Gemini Bio-Products, Woodland, CA) and 1% penicillin–streptomycin (Invitrogen).

Migration Assay

Migration assays were performed using commercially available inhibitors of EGF, gefitinib (Iressa; AstraZeneca, Wilmington, DE) or PDGF, imatinib (Gleevec; Novartis, New York, NY). Pan02 cells were harvested into DMEM alone and loaded into the upper chamber of a Cultrex BME (Trevigen Inc., Gaithersburg, MD) Matrigel-coated 8-μm modified Boyden chamber insert (Becton Dickinson Labware, Franklin Lakes, NJ) with experimental media in the lower chamber, which contained the following reagents: DMEM alone (control), DMEM plus; D.1 LR and NLR plasma fraction of pRBCs with or without additional treatment with gefitinib or imatinib, and D.42 LR and NLR plasma of pRBCs with or without additional treatment with gefitinib or imatinib. Cells were kept at 37°C for 24 h. Membranes were then stained, excised, mounted on slides, and examined by using a Nikon inverted microscope (Eclipse E600; Nikon, Tokyo, Japan) at ×200 total magnification. Cell migration across the membrane was quantified in five to ten fields of view for each membrane, with two membranes for each treatment condition. For assays done with gefitinib, the inhibitor was added to the supernatant to create a concentration of 10 μg/ml and allowed to incubate for 1 h in the bottom chamber of the 24-well dish prior to cells being added to the insert. For assays done with imatinib, the inhibitor was added to the supernatant to create a concentration of 1 μg/ml and allowed to incubate for 1 h hour in the bottom chamber of the 24-well dish prior to cells being added to the insert. Data are presented as the mean number of cells that migrated across the membrane for each high-powered field (total magnification, ×200).

Proliferation Assay

Proliferation assays were also performed using imatinib. Cultured Pan02 cells were harvested using 1%Trypsin–EDTA (Invitrogen). The cells were placed in 96-well dishes (Falcon; BD Biosciences, San Jose, CA) in DMEM + 10% FBS for 6 h to allow cells to adhere. Media were then removed, and experimental media were added with various conditions as follows: DMEM alone (control), DMEM + 10% D.1 or D.42 from both LR and NLR plasma from pRBCs with or without additional treatment with imatinib. Cell proliferation was determined in triplicate at 24 h by using a commercially available MTS assay (CellTiter 96 Aqueous One Solution Cell Proliferation Assay; Promega Corporation, Madison, WI, Cat. No. G3580) according to the manufacturers’ guidelines, and absorbance was quantified using a FLUOstar OPTIMA plate reader (BMG LabTech, Offenburg, Germany). For assays done with imatinib, the inhibitor was added to the supernatant at a concentration of 1 μg/ml. Data are presented as a percentage of control (DMEM alone).

Statistical Analysis

Statistical analysis was by analysis of one-way variance using a post hoc Fisher’s exact test. Statistical significance was accepted as p values less than 0.05. All numerical data are presented as means±standard error. N ≥ 4 per group.

Results

Increased Expression of Pro-Cancer Mediators in Stored pRBCs

Screening ChemiArray Human Cytokine Antibody Array III Map analysis for growth and inflammatory mediators present in the plasma fraction of D.1 and D.42 pRBCs, LR, and NLR showed increased expression of MCP-1, RANTES, angiogenin, TNF-α, EGF, and PDGF-BB (Fig. 1).

Fig. 1
figure 1

Screening ChemiArray analysis from the day 1 (D.1) and day 42 (D.42), leukoreduced (LR) or nonleukoreduced (NLR), plasma fraction of stored pRBCs. Each dot in duplicate represents detection of a specific cytokine or growth factor. a Columns from ChemiArray analysis from D.1 and D.42, NLR and LR, blood showing increased expression of MCP-1 and TNF-α in duplicate. Solid black line separates two trials. The same experiments run with ChemiArray show elevated expression of EGF (b), angiogenin (Ang, c), RANTES (d), and PDGF-BB (e)

Full size image

Effects of Storage Time and Leukocyte Reduction on Pro-Cancer Mediators in pRBCs

MCP-1

Quantitative analysis of the concentration of MCP-1 in the plasma fraction of pRBCs is shown in Fig. 2a. MCP-1 levels increase with storage time in both LR and NLR blood. MCP-1 levels in D.1 LR blood, 86.3 ± 6.3 pg/ml, increase with storage time to 121.2 ± 6.1 pg/ml in D.42 LR blood (p = 0.007). Levels increase at a greater rate in NLR blood, 78.2 ± 7.3 pg/ml in D.1 NLR blood to 647.8 ± 220.7 pg/ml in D.42 NLR blood (p = 0.02). By D.42, there was a higher level of MCP-1 in NLR vs. LR blood, 647.8 ± 220.7 vs. 121.2 ± 6.1 pg/ml, respectively, (p = 0.05).

Fig. 2
figure 2

Quantitative determination of the effect of leukoreduction (LR) and storage time on protein concentrations identified by ChemiArray analysis were performed using enzyme-linked immunosorbent assay. a MCP-1 levels increase with storage time from day 1 (D.1) to day 42 (D.42) in LR (p < 0.01) and markedly increase in nonleukoreduced (NLR) blood (p = 0.02). There is also higher levels of MCP-1 in D.42 NLR blood compared to D.42 LR blood (*p = 0.05). b RANTES levels decrease in LR blood after D.1 (p = 0.01). Levels trend toward increasing levels of RANTES in NLR blood (p = 0.06). Compared to LR blood, NLR blood has higher RANTES levels at D.28 (*p < 0.01), and D.42 (*p < 0.001). c Angiogenin levels in NLR blood have a trend toward decreasing levels with storage time (p = 0.08). NLR has higher angiogenin levels compared to LR blood, where LR decreases angiogenin levels to undetectable (all *p < 0.001). d TNF-α levels are not different with either storage time or LR. e EGF levels increase with storage time in NLR blood from D.1 to D.42 (p = 0.001). No change of EGF levels with storage time was observed in LR blood. Compared to LR blood, NLR blood has higher EGF levels at D.28 (*p < 0.01) and D.42 (*p < 0.001). f PDGF-BB levels decrease with storage time in LR blood from D.1 to D.42 (p = 0.01). Level increases with storage time in NLR blood (p < 0.01). Compared to LR blood, NLR blood has higher PDGF-BB levels at D.1 (*p = 0.02), D.28 (*p < 0.01), and D.42 (*p < 0.001)

Full size image

RANTES

RANTES levels in the plasma fraction of pRBCs are shown in Fig. 2b. In LR blood, RANTES levels decrease with storage time. D.1 levels, 13.8 ± 1.8 pg/ml, are elevated compared to both 4.7 ± 2.2 pg/ml at D.28 (p = 0.01) and 3.0 ± 1.9 pg/ml at D.42 (p = 0.003). In NLR blood, there is a trend toward increasing RANTES levels with storage time, 12.0 ± 1.6 pg/ml at D.1 compared to 15.8 ± 0.7 pg/ml at D.42 (p = 0.06). NLR blood RANTES levels are increased compared to LR blood at both D.28, 14.8 ± 1.2 vs. 4.7 ± 2.2 pg/ml, respectively, (p = 0.004), and D.42, 15.8 ± 0.7 vs. 3.0 ± 1.9 pg/ml, respectively, (p < 0.001).

Angiogenin

Angiogenin levels in the plasma fraction of pRBCs are shown in Fig. 2c. Collectively, angiogenin levels were increased in NLR blood, 44.2 ± 3.7 pg/ml, vs. LR blood, 0 pg/ml (undetectable, p < 0.001), and elevated in NLR blood compared to LR blood at each time point, 52.6 ± 4.4 vs. 0 pg/ml at D.1 (p < 0.001), 42.7 ± 7.1 vs. 0 pg/ml at D.28 (p < 0.001), and 37.2 ± 6.5 vs. 0 pg/ml at D.42 (p < 0.001). There was no difference observed in angiogenin levels with storage time in NLR blood, with a trend of decreasing concentration, 52.6 ± 4.4 pg/ml at D.1 vs. 37.2 ± 6.5 pg/ml at D.42 (p = 0.08).

TNF-α

TNF-α levels in the plasma fraction of pRBCs are shown in Fig. 2d. No differences were observed with storage for LR blood, 1.1 ± 0.5 pg/ml at D.1 vs. 2.0 ± 0.9 pg/ml at D.42, and for NLR blood, 1.0 ± 0.4 pg/ml at D.1 vs. 2.0 ± 0.7 pg/ml at D.42 (both p = NS). TNF-α concentrations showed no difference between LR and NLR blood at any time point.

EGF

EGF levels in the plasma fraction of pRBCs are shown in Fig. 2e. In LR blood, there was no storage effect on levels, 216.4 ± 3.8 pg/ml at D.1 vs. 207.3 ± 7.2 pg/ml at D.42 (p = NS). In NLR blood, EGF levels increase with storage time. D.1 levels, 241.1 ± 13.1 pg/ml, increase by D.28 to 801.1 ± 130.3 pg/ml (p = 0.003), and further increase to 1,436.4 ± 238.6 pg/ml at D.42 (p = 0.001). NLR blood EGF levels are increased compared to LR blood at both D.28, 801.1 ± 130.3 vs. 216.8 ± 13.8 pg/ml, respectively (p = 0.002), and at D.42, 1,436.4 ± 238.6 vs. 207.3 ± 7.2 pg/ml, respectively (p < 0.001).

PDGF-BB

PDGF-BB levels in the plasma fraction of pRBCs are shown in Fig. 2f. In LR blood, PDGF-BB levels decrease with storage time, 7.3 ± 0.3 pg/ml at D.1 vs. 6.4 ± 0.1 pg/ml at D.42 (p = 0.01). In NLR blood, PDGF-BB levels increase with storage time. D.1 levels, 34.7 ± 9.7 pg/ml, increase by D.28 to 73.6 ± 2.3 pg/ml (p = 0.005), and further increase to 76.5 ± 1.7 pg/ml at D.42 (p = 0.003). NLR blood PDGF-BB levels are increased compared to LR blood at all time points, 34.7 ± 9.7 vs. 7.3 ± 0.3 pg/ml, respectively, at D.1 (p = 0.02), 73.6 ± 2.3 vs. 6.6 ± 0.2 pg/ml, respectively, at D.28 (p < 0.001), and 76.5 ± 1.7 vs. 6.4 ± 0.1 pg/ml, respectively, at D.42 (p < 0.001).

Migration of Pan02 Cells Treated with Gefitinib

Based on the differential expression of EGF in stored pRBCs, cellular migration of Pan02 cells was measured, and differences of media treated with the plasma fraction of pRBCs were compared when also treated with the EGF inhibitor gefitinib (Fig. 3). Exposure of Pan02 cells to D.1 LR, D.1 NLR, or D.42 LR blood had similar migration with or without treatment with gefitinib. However, migration of Pan02 cells treated with D.42 NLR blood was significantly reduced when treated with gefitinib, 245.9 ± 11.2 cells/high-powered field (hpf) in D.42 NLR blood vs. 164.6 ± 10.6 cells/hpf in D.42 NLR blood plus gefitinib (p < 0.001).

Fig. 3
figure 3

Pan02 cell migration when treated with the plasma fraction of stored pRBCs with or without gefitinib (G) was significantly decreased in cells exposed to “outdate,” D.42 NLR blood (*p < 0.001). Treatment with gefitinib did not affect Pan02 migration in cells exposed to “outdate” D.42 LR or any D.1 blood. HPF high-powered field, DMEM Dulbecco’s minimal essential medium

Full size image

Migration and Proliferation of Pan02 Cells Treated with Imatinib

Based on the differential expression of PDGF-BB in stored pRBCs, cellular migration of Pan02 cells exposed to the plasma fraction of pRBCs was measured to determine any effect of treatment with the PDGF inhibitor imatinib. In Pan02 cells treated with D.1 or D.42, LR or NLR blood, additional treatment with imatinib did not decrease cellular migration (Fig. 4a). In fact, treatment with imatinib showed increased migration in cells treated with D.1 LR (96.2 ± 6.4 in untreated vs. 153.5 ± 15.0 cells/hpf in treated, p = 0.002), D.1 NLR (148.8 ± 16.2 in untreated vs. 236.5 ± 17.5 cells/hpf in treated, p = 0.03), and D.42 LR (59.9 ± 8.6 in untreated vs. 138.2 ± 8.8 cells/hpf in treated, p < 0.001). These results led us to test the effects of imatinib treatment on Pan02 cell proliferation for cells exposed to the plasma fraction of pRBCs (Fig. 4b). Exposure of Pan02 cells treated with D.1 LR, D.1 NLR, and D.42 LR blood had similar proliferation with or without treatment with imatinib. However, proliferation of cells treated with D.42 NLR blood was significantly reduced when additionally treated with imatinib (181.1 ± 1.5% over control (DMEM alone) in untreated vs. 157.5 ± 2.1% over control in treated, p < 0.001).

Fig. 4
figure 4

a Pan02 cell migration in cells treated with the plasma fraction of stored pRBCs with or without imatinib (I). Treatment with imatinib increased cell migration in cells treated with D.1 LR (*p < 0.01) and NLR (*p = 0.03), as well as D.42 LR blood (*p < 0.001). b Pan02 cell proliferation in cells treated with the plasma fraction of stored pRBCs with or without imatinib reveals imatinib decreased cell proliferation in cells treated with D.42 NLR blood (p < 0.001). HPF high-powered field, DMEM Dulbecco’s minimal essential medium, percent over control (DMEM only)

Full size image

Discussion

Multiple factors were identified in pRBCs that have tumorigenic properties and may augment tumor growth and progression. Of the six factors observed in stored pRBCs, angiogenin, TNF-α, and RANTES do not accumulate with storage, but are elevated in fresh blood. These factors may contribute to the observed detrimental effect of tumor progression with the transfusion of fresh blood as previously demonstrated in an immunocompetent model of pancreatic adenocarcinoma.9 Additionally, all mediators, with the exception of TNF-α, are significantly reduced with leukocyte reduction, indicating a potential role in removing white blood cells in the treatment of cancer patients, although it remains undetermined in clinical studies whether LR blood improves outcomes compared to NLR blood.22

Tumor infiltration with macrophages has been shown to negatively correlate with outcomes in about 80% of solid tumors.26–28 Recently, neutrophils have also been implicated as important effector cells and are integral in the angiogenic switch.29 MCP-1, RANTES, TNF-α, and PDGF are all potent immune chemoattractants, especially for macrophages and neutrophils, which may lead to possible detrimental effects by providing these factors through transfusions of blood products. Additionally, upregulation of RANTES and MCP-1 has been shown in breast and cervical cancer, and assists in angiogenesis.30–32 TNF-α has also been shown to be a cancer-promoting cytokine in breast cancer, important in the endothelial-to-mesenchymal transition.32

Elevated levels of angiogenin, one of the most potent angiogenic factors, is associated with aggressive disease in many cancers including gastric, pancreatic, colorectal, urothelial, brain, melanoma, and non-Hodgkin lymphoma.33,34 Angiogenin is present in NLR blood and does not change with storage time. Leukocyte reduction eliminates angiogenin from stored pRBCs.

Overexpression of EGF and its receptor have been linked to poorer outcomes in lung, colon, breast, pancreas, bladder, and head and neck cancer.35,36 EGF is a strong mitogen, increasing proliferation and survival, and is an angiogenic factor.36 Our study reveals that EGF levels increase with storage time in NLR blood, and these increases in EGF are abrogated with leukocyte reduction. In vitro treatment with gefitinib of Pan02 cells that are exposed to D.42 NLR blood, which correlates to a blood product that has the highest levels of EGF, attenuated cellular migration.

PDGF-BB is involved in angiogenesis, lymphangiogenesis, acts as a mitogen, and is elevated in some forms of cancer.37 PDGF levels in blood increase with storage time in NLR blood, not in LR blood. In vitro treatment with a PDGF inhibitor of Pan02 cells exposed to D.42 NLR blood, which correlates to a blood product that has the highest levels of EGF, attenuated cellular proliferation but not migration.

Conclusion

These data demonstrate the presence of pro-cancer cytokines and growth factors in pRBCs, which are either present at the highest level in fresh blood or accumulate significantly during storage. Although the negative effects of blood transfusion are likely multi-factorial, the tumorigenic mediators identified in the acellular fraction of pRBCs may represent possible therapeutic targets to offset negative effects of transfusion on clinical outcomes. Further examination of pretreatment of blood products or patients with specific inhibitors such as gefitinib and imatinib is warranted. As many of these pro-cancer cytokines were reduced by prestorage leukocyte reduction, this practice is likely beneficial, although it has not been demonstrated in current trials examining colorectal cancers.22 Together, these data reiterate the importance of limiting the use of blood products in patients with underlying malignancies, and further strategies to avoid transfusion should be pursued.