Blood Transfusion-Induced Immunomodulation

Blood transfusion therapy is associated with many risks, including major or minor blood transfusion reaction, non-A non-B hepatitis, hepatitis B, and HIV infection. Blood transfusion may result in immunologic changes (immunomodulation) that are beneficial in some patients but harmful in others. After reports of increased renal allograft survival in patients receiving pretransplant transfusion ^[1], Gantt ^[2] questioned whether transfusion might diminish or retard the immune response. Clinical studies have shown a beneficial effect of blood transfusion on graft survival ^[3,4], but an adverse effect on cancer recurrence ^[5-10] and postoperative infection ^[10-14]. Approximately two thirds of 11,000 transfusions ^[15,16] given perioperatively are administered by anesthesiologists ^[17], and 26% are given inappropriately ^[18,19].

This review will focus on the immunomodulatory effects of transfusion therapy. It will begin with a brief review of the immune system and then discuss 1) the effect of transfusion therapy on modulating the immune system, and 2) transfusion-induced immunomodulatory effects on vascularized graft survival, cancer recurrence, and postoperative infections.

Immune System

Overview

The major function of the immune system is to recognize various pathogens and respond to them. Although the immune system evolved to inactivate pathogens, it is now known that it also protects the body against the development of tumors and the rejection of foreign tissue.

The immune system consists of an afferent (recognition) limb and an efferent (response) limb. The afferent limb is involved with alerting the host against the entry of foreign pathogens or foreign alloantigens. The afferent (response) limb may respond by cellular and/or humoral immunity and by secretion of lymphokines that modulate the immune response ^[20].

Human Lymphoid System

The lymphoid tissue is responsible for most of the body's immune response and contains about 10¹² lymphocytes, which produce approximately 10²⁰ antibody molecules ^[21]. The mature lymphocytes circulate to lymph nodes, spleen, or tonsils and remain in these secondary locations until they are activated by antigens and antigen-presenting cells ^[20].

T Lymphocytes

T lymphocytes (T cells) are so named because they are matured by the thymus. T cells become activated by antigens and by antigen-presenting cells (APCs) and are characterized by an antigen-binding molecule or T-cell receptor (TCR) on their surface. TCRs react with antigens that have been processed and presented by major histocompatibility complex (MHC) molecules on the surface of macrophages or APCs. The TCR and the CD3 chain form a complex that functions as the primary antigen recognition site on the cell surface Figure 1. In addition to the TCR/CD3 molecules, numerous accessory surface molecules are present Figure 2. They include the CD8 and CD4 molecules, which bind to MHC class I and class II molecules on the APCs Figure 1^[22].

Figure 1: Interaction of a peptide-major histocompatibility complex (MHC) complex on a presenting cell with a receptor on a T cell. The presenting cell can be a macrophage using a class II MHC or any tissue cell using a class I MHC. The T cells can be helper T lymphocytes carrying CD4 or cytotoxic T lymphocytes carrying CD8. The class I MHC binds CD8; class II binds CD4. The T-cell receptor not only has CD8 or CD4 molecules as components of its interaction system, it also has four different proteins that make up the CD3 complex; their role is uncertain. Also, bound to the CD4 or CD8 proteins is the lck protein tyrosine kinase, which can phosphorylate a CD3 component on tyrosine in response to an interaction of the T-cell receptor with MHC-peptide complex. The CD3 and lck proteins are apparently part of the response system used by T cells after encounters with an MHC-peptide complex antigen. (From Darnell J, Lodish H, and Baltimore D. Molecular cell biology. 2nd ed. Copyright (c) 1990 by Scientific American Books, Inc., New York, NY. Used with permission of W. H. Freeman and Company.)

Figure 2: Cell surface markers involved in T-cell activation and adhesion. Depicted are a number of T-cell surface molecules and their known ligands on the surface of antigen-presenting cells and/or the extracellular matrix. Many of these molecules are also known to deliver intracellular signals capable of interacting with the activation signal delivered through the T-cell receptor (TCR). HA, hyaluronic acid; FN, fibronectin; LN, laminin; Ag, antigen; PTP prime ase, protein tyrosine phosphatase; circles with dots represent phosphotyrosine residues; fyn (p59^fyn) and lck (p56^lck), protein tyrosine kinases. (Adapted with permission from Rothstein DM. The cellular immune response. In: Anderson KC, Ness PM, eds. Scientific basis of transfusion medicine--implications for clinical practice. Philadelphia: WB Saunders, 1994;124-46.)

MHC I and MHC II antigens are cell surface glycoproteins that are responsible for alloreactivity and therefore rejection. The gene complex for human MHC is known as the human leukocyte antigen (HLA) complex. MHC genes that encode the targets for cytotoxic T lymphocytes' (CTLs') self-recognition are called class I MHC genes. These genes code for class I antigens at HLA A, B, and C loci Table 1. Several thousand molecules of each HLA A, B, and C antigen are expressed on every nucleated cell in the body.

Table 1: Class I and Class II Antigens: Structure and Function

The receptors on T cells recognize peptide fragments of antigens that are linear sequences of eight to 15 amino acids Figure 1. These peptides bind to MHC molecules on the cell's surface. The MHC molecules then present the peptides to a TCR. Class I MHC expresses peptides synthesized by the cell itself. These may be derived from normal self-proteins or from peptides from viral proteins made in a virusinfected cell Figure 3. Class II MHC binds exogenous protein which is endocytosed and broken down into peptide fragments.

Figure 3: Antigen processing and presentation. Endogenously synthesized or intracellular proteins (e.g., viral gene products) are degraded into peptides that are transported to the endoplasmic reticulum. These peptides bind to class I major histocompatibility complex (MHC) molecules and are transported to the surface of the antigen-presenting cell. CD8+ T cells recognize the foreign peptide bound to class I MHC by way of the T-cell receptor complex. Extrogenous antigen (e.g., bacterial) is endocytosed and broken down into peptide fragments in endosomes. Class II molecules are transported to the endosome in association with the invariant chain, bind the peptide, and are delivered to the surface of the antigen-presenting cell, where they are recognized by CD4+ cells. (Adapted with permission from Hanto DW, Mohanakumar T. Transplant immunology. In: Greenfield LJ, Mulholland MW, Oldham KT, Zelenock GB, eds. Surgery--scientific principles and practice. Philadelphia: JB Lippincott, 1993;461-500.)

T cells are divided into at least two subpopulations: CD-4 (helper) and CD-8 (killer) T-cells. The helper T lymphocytes, or T_H cells, utilize the CD-4 molecules to bind the class II MHC-peptide complexes that have been presented by macrophage APCs. The T_H cells recognize the antigen presented by the class II MHCs. The T_H cells then secrete lymphokines that stimulate other cells involved in the immune response ^[23].

Cytokines are nonantibody mediators of cellular immunity produced by activated lymphocytes ^[24,25]. As shown by Table 2, an individual lymphokine or cytokine may have multiple actions, with effect on immunomodulation, angiogenesis, hematopoiesis, septic shock, or antiproliferative activity against tumor cells ^[20]. Cytokines are known to have multiple effects on both T cells and B cells. Cytokines stimulate T_H cells through specific receptors on their surfaces. Activation of a T_H cell by an encounter with peptide-MHC complex induces formation of an IL-2 receptor, which responds to the IL-2 it secretes by activating autocrine growth ^[25].

Table 2: Partial List of Cytokines Secreted by T_H Cells or Macrophages in Response to Antigen^a

As B cells are maturing, they develop surface receptors for T-cell derived IL-2 and IL-4-6. These cytokines find receptors on activated B cells and stimulate growth of the B cells and their maturation to plasma cells. In summary, cytokines secreted by T_H cells promote antibody production from B cells and stimulate both T- and B-cell growth and maturation. T_H cell loss after HIV infection causes failure of the immune system in AIDS ^[26].

CTLs can be distinguished from T_H cells because they express the CD8 surface protein. CTLs recognize foreign antigens in the context of class I MHC and the CD8 molecule. Antibodies deal with intact foreign materials, whereas CTLs deal with cell-bound peptides. A classic CTL target is a virus-infected cell that displays fragments of viral glycoproteins on its surface bound to class I MHC. CTLs attach to the target cell and secrete proteases that form ion channels that depolarize the cell, thereby destroying its ionic and osmotic balance, resulting in cell death ^[20].

A third T cell, the suppressor T cell (T_S), suppresses B-cell activity and inhibits lymphocytotoxicity. The T_S cell appears to be associated with a CD8 surface protein that is different from the CD8 protein associated with CTL cells ^[20].

B Lymphocytes

B lymphocytes (B cells) originate in the bone marrow, mature into plasma cells, and are specialized for antibody production. During development, B cells pass through successive stages characterized by different patterns of cell surface marker expression and immunoglobulin (Ig) production. B lymphocytes express many different cell surface antigens Figure 4 as they pass from the pre-B cell to become activated B cells and then functional plasma cells ^[27]. These shortlived, noncirculating plasma cells are capable of producing 10 million antibody molecules an hour ^[28]. B-cell activation and response to antigen is directed by T_H cells and their secreted cytokines. The B cell can decompose a protein into peptides in association with MHC class II molecules and display them in a manner similar to the way an APC displays peptides in association with MHC class I or II molecules Figure 5.

Figure 4: Cell surface antigen expression. The stages of expression of some B-cell antigens are indicated by a solid line. Cytoplasmic but not surface expression is indicated by a dashed line. (Adapted with permission from Ord DC, Tedder TF. The humoral immune response. In: Anderson KC, Ness PM, eds. Scientific basis of transfusion medicine--implications for clinical practice. Philadelphia: WB Saunders, 1994;124-46.)

Figure 5: The hapten-carrier complex activates B cells by first stimulating helper T (T_H) cells. The hapten-carrier complex is a protein to which several hapten molecules have been covalently coupled. The hapten portion binds to a B cell but cannot by itself initiate B-cell proliferation. The B cell internalizes and digests the hapten-carrier complex, and portions of the carrier peptide are displayed on the B-cell surface as a complex with class II major histocompatibility complex (MHC) protein. A macrophage cell also internalizes carrier protein and displays peptide fragments on its surface in association with class II MHC. Then T_H cells with appropriate receptors bind to the peptide-MHC complex displayed on the macrophage surface. Binding stimulates the T_H cells, which then proliferate, recognize the identical peptide-MHC complex on the B cell, and secrete factors that stimulate the B cell to grow. (From Darnell J, Lodish H, Baltimore D. Molecular cell biology. 2nd ed. Copyright (c) 1990 by Scientific American Books, Inc., New York, NY. Used with permission of W. H. Freeman and Company).

Antibody secretion by plasma cells requires a B lymphocyte to bind an antigen and then receive costimulation from T_H cells in order to differentiate and become an active plasma cell. IL-2, IL-4, and IL-5 stimulate B-cell maturation into plasma cells. After the B lymphocyte binds an antigen to its receptor, it differentiates and secretes antibody molecules ^[20,29]. IgA-producing plasma cells are located in the gut and bronchial mucosa; IgG-producing plasma cells are located in bone marrow, spleen, and lymph nodes; IgM-producing plasma cells are located in the spleen and lymph nodes; and IgE plasma cells are found in gut, mesenteric, and bronchus lymphoid tissue. Each antibody molecule consists of two classes of polypeptide chains, light and heavy.

Another type of activated B lymphocytes are memory B cells, which retain, for the life of the organism, a record of the antigens previously encountered so that a second encounter with the antigen can elicit a rapid, highly avid response ^[20].

Transfusion-Induced Immunomodulation

Mechanism

Exposure to allogenic tissue, including blood, can cause both allosensitization and immunosuppression ^[30,31]. The development of tolerance can be specific to the tissue donor; however, immune responses can be characterized in varying degrees by cross-reactivity or cross-specificity ^[32]. The best characterized clinical effect of transfusion-induced modulation is improved survival of renal allografts in previously transfused patients ^[3,4]. Transfusion-induced immunomodulation may be implicated as either a potential inhibitor of immunologic effector cells or as a stimulator of immunologic suppressor cells ^[30-34]. Some transfusion-induced down-regulated immune functions include decreased cytokine production, mitogen response, T_H cells, natural killer (NK) cells, and lymphocyte numbers; decreased function and cellmediated cytotoxicity; and increased T_s cell number and function ^[5]. NK cells are cytotoxic lymphocytes, their activity is enhanced by IL-2, and they are a first-line defense against virus-infected cells ^[35].

In order to describe the transfusion effect, we will first discuss nonspecific immune suppression and antigen-specific immune suppression after transfusion. The component(s) of transfused blood responsible for these immunomodulatory effects is (are) uncertain; however, there are data to suggest that white cells or plasma are important in modifying the host immune defenses ^[30,32]. Compared to packed cells prepared in the late 1970s, current packed red cells contain less than 10% of the plasma and 5% of the white cells ^[36]. The decreased quantity of plasma and white cells present in today's packed red blood cells may be responsible for the apparent decrease in the ability of packed cells to provide a modulating effect.

Nonspecific Immune Suppression after Blood Transfusions

The mechanism(s) by which transfusions produce nonspecific immunomodulation is (are) unknown. One hypothesis is that transfusions overload the reticuloendothelial system and that iron salts cause the suppressive changes ^[37,38]. Prostaglandin E₂ (PGE₂) production and IL-2 metabolism also play a role in down-regulating immune responses. PGE₂ production by monocytes is increased after transfusions ^[39,40]. PGE₂ is known to down-regulate macrophage class II antigen expression ^[41] and inhibit IL-2 production and target cell response to IL-2 ^[33].

IL-2 production by T_H lymphocytes is known to decrease after transfusion ^[42-44]. As discussed earlier, IL-2 is required for the activation and proliferation of B cells and for the generation of cytotoxic T cells; hence, a decrease in IL-2 production will result in a decrease in B-cell stimulation, antibody production, and impaired NK activity ^[45,46].

These nonspecific immunomodulatory effects after transfusion appear to be transient after one transfusion, but they become more pronounced and longlasting after multiple transfusions. Hemodialysis patients demonstrate increased nonspecific immune suppression up to 20 wk after transfusion ^[47].

Antigen-Specific Immunosuppression

In 1984, Terasaki ^[48] advanced a new theory that the "function of transfusion was to immunize patients and not to induce tolerance, or to serve as a donor selection mechanism." He proposed that transfusions did immunize recipients; however, if they were receiving azathropine (AZA), the AZA partially eliminated the clone of lymphocytes which were stimulated by the transfusion. When the patient subsequently underwent transplantation, the transplanted organ acted as a secondary stimulus and high dose immunosuppression at the time of transplantation killed or inactivated remaining clones of reactive cells. This clonal depletion theory suggests that the transfusion and immunosuppression result in removing or incapacitating the cells that would reject the graft.

Multiple donor-specific transfusions (DSTs) have also been thought to generate T_S cells ^[49-51]. T_S cells are thought to express a CD8 surface protein and are generated via-a multistep process that occurs over several days ^[52].

A third antigen-specific immunosuppressive mechanism after DST results from the generation of an antiidiotypic antibody. Jeme ^[53] proposed that TCRs or antibodies generated against the graft antigen in the primary immune response could act as "new" antigens. Antibodies formed against these "new" antigens are called antiidiotypic antibodies, and they compete for binding locations on the initial antibodies ^[54-56]. These antiidiotypic antibodies function as blocking antibodies to prevent the antibodies generated against the graft antigen in the primary immune response from combining with the graft antigen.

Blood Transfusion and Organ Transplantation

In the late 1960s, renal failure patients were often given transfusions for anemia. However, transfusion therapy became controversial when it was established that hyperacute rejection after kidney transplant was due to lymphocytotoxic antibodies arising from sensitization to HLA antigens. The most frequent causes of sensitization were identified as blood transfusion, pregnancy, and rejected grafts ^[23]. After the discovery of sensitization to HLA antigens, transfusions were avoided in patients on hemodialysis. In 1973, Opelz et al. ^[1] published the first study showing improved cadaveric graft survival rates in transfused recipients. Other investigators subsequently confirmed that transfused recipients had better graft survival rates ^[57-60]. Opelz and Terasaki ^[57] collected data on 870 transfused and 490 untransfused patients prior to transplantation and stratified them according to the type and number of blood components received (whole blood, packed red blood cells, or frozen blood). This study showed a 1 yr graft survival rate of 42% in patients who had not received prior transfusions and 71% in patients who had received more than 20 transfusions. At 4 yr, graft survival rates were 30% without and 65% with transfusion ^[57,58]. Beginning in 1978, many centers developed a policy of giving potential recipients three units of packed red blood cells at 15-day intervals ^[59,60]. Other prospective studies confirmed the earlier results with a 1-yr graft survival rate of 87% in transfused and 32% in nontransfused patients ^[60].

However, controversy remained over the type of blood product to be transfused, the number and timing of transfusions, and the sensitization rate. Opelz and Terasaki ^[57] reported that whole blood and packed cells, but not frozen blood, had a significant immunosuppressive effect. Others confirmed that packed cells, washed red blood cells, and other blood products were more effective than frozen red blood cells in improving graft survival ^[60,61]. Persijn et al. ^[60] showed that single unit transfusions, even if given years earlier, could prolong cadaveric kidney graft survival. Opelz et al. ^[58] found a corresponding increase in graft survival rates with increasing numbers of transfusions. Some studies show improvement in graft survival only in animals receiving whole blood transfusions at the time of transplantation ^[62].

Salvatierra et al. ^[49] reported the beneficial effect of DST in recipients of kidneys from living, related donors. These patients received three transfusions 2 wk apart from the prospective donor while continuously receiving AZA immunosuppression. As discussed earlier, DST in conjunction with AZA is thought to eliminate clones of reactive cells (clonal depletion theory).

In summary, studies of the "transfusion effect" suggest that even a single transfusion may enhance renal graft survival. Optimal transfusion effect for most recipients of cadaver grafts is obtained with three to five transfusions of whole blood or packed red cells given before transplantation. The magnitude of the transfusion effect appears to be somewhat less than that for cyclosporin A. With the introduction of cyclosporin A, the beneficial effect of random transfusion has disappeared; however, the clinical importance of DST protocols has persisted in some situations ^[63-65]. The specific blood component(s) responsible for the transfusion effect has (have) not been identified; the duration of the effect is unknown and the mechanism is not established ^[66-69]. Although the transfusion issue is not completely resolved, most programs now avoid transfusion to minimize the risk of sensitization to HLA antigens.

Blood Transfusion and Cancer Recurrence

Friedman et al. ^[70] reported that patients undergoing cancer treatment consume 19% of transfused blood and are the largest subset of patients who receive transfusion therapy. Recent studies ^[71] show that between 35% and 65% of patients treated for cancer of the large bowel, breast, lung, kidney, stomach, or prostrate receive transfusions. The possibility of transfusion-induced immunosuppression in these patients has prompted a number of studies on the effect of transfusion on tumor recurrence after potentially curative surgery ^[72-75].

Significant differences in cancer recurrence rates have been reported in most clinical studies, and may be secondary to different tumor stages, degree of preoperative anemia, length of surgery, or blood transfusion. It has been suggested that transfusion is a marker for other factors that contribute to cancer recurrence. However, many studies utilizing multivariate analysis or meta-analysis of individual retrospective studies have found the relationship between cancer recurrence and blood transfusion to be unexplained by other variables. In the studies failing to show a statistically significant transfusion effect, most found a higher recurrence rate in transfused patients, and those failing to find significance did not find significance because of small study size ^[72]. Another consistent finding has been a higher rate of cancer recurrence and death when plasma and leukocyte-containing blood products are transfused (whole blood or fresh frozen plasma) compared to packed red cell transfusion ^[73]. Patients who receive transfusions of three units or less of packed red blood cells ^[74] have a reduced recurrence rate and increased length of survival after cancer surgery. Many ^[76-83], but not all ^[84-86], animal studies support the concept of transfusion-induced immunosuppression in the recipient and subsequent enhancement of tumor growth. While transfusion-induced tumor growth rate enhancement is inconsistently found in these studies, immune system down-regulation is a consistent finding ^{[79-85,87,88]}.

Colon Cancer Recurrence and Transfusion

A possible adverse effect of blood transfusion on colon cancer recurrence rate was first reported by Burrows and Tartter ^[89]. Other authors ^[90,91] also reported a detrimental effect of perioperative blood transfusion on the rate of colorectal cancer recurrence when comparing transfused and untransfused patients. Thirty-one studies ^{[73,74,89-117]} evaluating the relationship between transfusion and colon cancer recurrence were reviewed; two ^[74,111] were excluded because of incomplete data, and the relative odds ratio and 95% confidence intervals were determined in the remaining 29 studies Table 3, Table 4. The majority (90%) of the studies reviewed showed a significant increase in colorectal cancer recurrence in transfused patients with an overall relative risk of 1.33 and 95% confidence intervals of 1.3, 1.5 Figure 6.* The overall odds ratio found in this meta-analysis is similar to that reported in others ^[10,118,119]. We found no apparent explanation for the lack of a transfusion effect in the three ^[92,115,117] negative studies. Removal of these minority studies from this meta-analysis would clearly increase the strength of the resulting odds ratio. Because of inconsistent reporting of confounding variables (age, stage, preoperative hemoglobin, etc.), adjusting the relative risk with consideration of these variables was not possible.

Table 3: Colorectal Cancer Recurrence Rates With and Without Blood Transfusion, Meta-analysis

Table 4: continued from Table 3

Figure 6: Odds ratios (fill diamond) of colorectal cancer recurrence (5 yr; 95% confidence intervals) with transfusion. Solid line represents odds ratio of no increased risk (1.0). Broken line represents cumulative odds ratio of 1.33 with transfusion.

*Relative odds ratio of 1.33 equates to a 33% increased risk of cancer recurrence with transfusion.

Transfusion of plasma and leukocyte-containing blood components (whole blood, fresh frozen plasma, etc.) has been shown to increase colorectal cancer recurrence and death rates when compared to red cell only transfusion ^{[72,101,120,121]}. Nielsen et al. ^[122] reported reduced response to skin test antigens after colorectal cancer resection in patients given whole blood compared to those given red blood cells. Blumberg et al. ^[123] demonstrated that colorectal cancer recurrence was greatest after whole blood (any amount) transfusion. Four or more units of packed red blood cells gave the next highest recurrence rate. The lowest recurrence rate was found in patients given a transfusion of three units or less of packed red cells or in those given no transfusion.

Busch et al. ^[124] evaluated recurrence of colorectal cancer in patients receiving autologous compared to allogeneic transfusions. They reported no difference in recurrence between groups; both groups had lower survival rates than patients receiving no transfusion. In a letter to the editor, Blumberg and Heal ^[125] pointed out that 14 of the 15 hospitals entering this study were supplied buffy coat depleted allogeneic blood. This implies that Busch et al. ^[124] in reality compared a white cell depleted blood transfusion group with an autologous group and found no significant differences in colorectal cancer recurrence rates. In contrast, a recent prospective study ^[126] demonstrated that allogeneic (leukocyte-poor packed red cells) but not autologous transfusion is a significant independent predictor of an increased colorectal cancer recurrence rate. No difference in survival rates between allogeneic and autologous transfusion groups has been reported for patients undergoing radical prostate surgery ^[127]. Of note is a report ^[128] of reduced NK cell function in humans after blood donation only.

With the exception of one study ^[129], blood transfusion has been demonstrated to be associated with a worse outcome after liver resection of metastatic colorectal cancer ^[130,131] and after resection of primary hepatocellular cancer ^[132]. In the largest study ^[130], the survival rate of metastatic colorectal cancer at 3 in untransfused patients was 60% compared to 40% in a group who received at least a one-unit transfusion of whole blood or packed cells.

Mastectomy for Cancer and Transfusion

Seven ^[133-139] of 10 publications ^[96,133-141] demonstrated a worse outcome in the transfused group. Hoe et al. ^[136] retrospectively studied 455 patients with Stage I breast cancer and reported a 5-yr survival rate of 53% for transfused patients vs 93% for untransfused patients. Herman and Kolodzieski ^[139] reported that patients with Stage I and II disease had 5-and 10-yr survival rates of 75% and 63%, respectively, in untransfused patients versus 66% and 49% in transfused patients (P = 0.005). Multivariate analysis demonstrated that blood transfusion, nodal involvement, histologic grade, and stage were significant risk factors. Tartter et al. ^[138] found a significant adverse transfusion effect on 5-yr survival rates (transfused, 51% vs untransfused, 65%). Three negative studies (no adverse transfusion effect) ^[96,140,141] were retrospective, multivariate studies with between 226 and 383 patients. There are no apparent explanations for the discrepancies between these studies.

Head and Neck Cancer Recurrence After Surgical Resection and Transfusion

Eight publications were reviewed; six demonstrated a significant adverse effect after blood transfusion ^[142-147] and two ^[148,149] showed no effect. In a meta-analysis of their own data and five other studies, Wooley et al. ^[145] reported a combined odds ratio for recurrence after transfusion to be 2.6, indicating a "clinically important adverse effect of transfusion of blood products on tumor recurrence in patients with advanced head and neck cancer." Likewise, Jackson and Rice ^[144] reported a 5-yr recurrence rate of laryngeal cancer of 14% (untransfused) vs 65% (transfused); the rate for oral/nasal cancer was 31% (untransfused) vs 71% (transfused). Johnson et al. ^[146] reported that the 3-yr recurrence-free survival rate was 73% (untransfused) vs 47% (three- or four-unit packed red cell transfusion), and the 2-yr recurrence-free survival rate was 40% with transfusion of five or more units of packed cells.

Transfusion and Lung Cancer Recurrence

Seven studies were reviewed ^[150-156]; the first five ^[150-154] reported a relative hazard ratio (transfused versus nontransfused) ranging between 1.24 ^[152] and 1.99 ^[150] for Stage I disease, an increased incidence of recurrence ^[152,153], or a decreased survival rate ^{[150,151,154]} after blood transfusion. Little et al. ^[155] reported the 5-yr disease-free survival rate with Stage I non-small-cell lung cancer to be 81% (untransfused) vs 53% (transfused). Multivariate analysis demonstrated that the only significant variable for 5-yr disease-free survival was transfusion. Pastorini et al. ^[156] reported a retrospective study of 283 patients with resected Stage Ia non-small cell lung cancer and found no difference in 8-yr survival rates (36% transfused vs 34% untransfused).

Soft-Tissue Sarcoma Recurrence and Transfusion

Rosenberg et al. ^[157] reported a prospective multivariate study after surgical resection of extremity soft tissue sarcoma and found the five year recurrence-free survival to be 70% (untransfused) versus 48% (transfused). The only variable other than transfusion associated with poorer outcome was primary tumor size. Chesi et al. ^[158] reported a similar reduced survival rate in a transfused group with osteosarcoma.

Gastric Cancer Surgery and Transfusion

A retrospective study ^[159] reported a statistically significant 5-yr survival difference of 80.8% (nontransfused) vs 57% (transfused); after proportional regression analysis this effect fell to marginal significance (relative risk ratio of 1.16). Two other reports ^[160,161] showed significantly lower 5-yr survival rates in patients receiving perioperative blood transfusion compared to those with no transfusion. A final report ^[162] of survival of 124-patients undergoing curative resection of gastric cancer showed an adverse transfusion effect (lower survival in the group given five or more units of packed red cells compared to one to four units). When Cox regression analysis was done on the data ^[162], it was shown that the transfused group was older, had a more advanced histopathologic stage, and had a greater incidence of total gastrectomy, causing the adverse effect of transfusion to fall out of significance.

Prostate Cancer Surgery and Transfusion

McClinton et al. ^[163] retrospectively studied 246 patients treated by transurethral prostate resection (TUR) using proportional hazards regression analysis and found only age at time of surgery and blood transfusion to be significant variables on survival rates. Davies et al. ^[164] found similar results in prostate cancer treated by TUR: a 17% (transfused) vs 66% (untransfused) 5-year survival rate. The third TUR study ^[165] demonstrated a modestly higher (not significant) prostate cancer-specific 5-yr survival rate in the untransfused group, but further data analysis found only metastatic stage to be a highly significant factor.

Ness et al. ^[127] studied 309 patients; 36 received no transfusion, while 179 received autologous only, and 94 allogenic transfusions. On multivariate analysis, lymph node and seminal vesicle involvement were the only significant risk factors, tumor recurrence and death due to tumor were higher (not significantly) in the allogenic transfused group. Heal et al.'s ^[166] retrospective and multivariate study of nonmetastatic prostate cancer treated by both prostatectomy and TUR demonstrated significantly lower survival and higher recurrence rates (relative risk of 2.82) in the transfused group.

Summary

The above data support the concept that allogeneic blood products increase cancer recurrence after a potentially curative surgical resection. A conservative trigger for blood product administration during cancer surgery seems justified by the data presented, especially since the adverse cardiovascular effects of hemodilution and low hematocrit have been shown to be less detrimental than previously thought ^[167].

Blood Transfusion and Postoperative Infection

The effects of allogeneic transfusion on bacterial infection in animal models are difficult to evaluate because of the known immunosuppressive effects of hemorrhage alone ^[168]. Hemorrhage-induced immunosuppression in mice resulted in a 100% mortality rate compared to a sepsis challenge mortality rate of 58% in a control group ^[169]. Hemorrhagic shock resulted in endotoxemia in 88% of rats ^[170], while clearance of an intravenous bacterial challenge was reduced if hemorrhagic hypovolemia preceded the challenge ^[171]. Waymack et al. ^[172,173] demonstrated a transfusion-induced macrophage inhibition and reduced cell-mediated immune function in burned rats, while those rats receiving a septic challenge after allogenic blood transfusion had a higher mortality rate compared to controls. Other ^[174-176], but not all ^[168,177], studies have demonstrated immunosuppression to bacterial challenge after allogeneic blood transfusion in animal models.

The extent of trauma has been related to postoperative immunosuppression with subsequent increased infection and mortality rates ^[178]. Step-by-step logistic regression analysis has been useful in defining the relationship of blood transfusion to infection while taking into consideration other frequently associated variables: age, shock, wound contamination, and other variables. This type of analysis has been used by Tartter ^[179] in groups of patients subjected to colorectal cancer surgery and has demonstrated a highly significant relationship between blood transfusion and postoperative infection.

Several studies report that allogeneic blood transfusion increases the incidence of postoperative infection. Included are patient groups involving penetrating abdominal injuries ^[12,180-182], elective colectomy for cancer ^{[179,183-186]}, coronary artery bypass ^[187-189], orthopedic surgery including total joint replacement ^[190,191], spinal fusion surgery ^[192], hysterectomy ^[193], and trauma ^{[180,194,195]}. These studies utilized stepwise logistic regression analysis of their data using multiple variables including age, sex, injury severity scores, shock, length of surgery, and admission hematocrit, and all concluded that homologous blood transfusion is a significant predictor of increased risk of postoperative infection. In several of these studies ^{[180,182,185,187,192,194,195]} the amount of blood transfused (a dose-response relationship) was shown to be a significant factor in the incidence of postoperative infection rate. Transfusion history has also been reported to be a significant risk factor for nosocomial infections distant from the primary injury site ^[12].

Anergy to skin testing was shown to be related to transfusion exposure in a nontraumatic (gastrointestinal hemorrhage) clinical model ^[196]. The anergic patient group had a significantly higher incidence of sepsis compared to the nonanergic (and lower transfusion exposure) group. Reduced NK cell cytotoxicity and low helper/suppressor cell ratios have been reported after transfusion for hemophilia ^[197,198], hemolytic anemias ^[197,199], and leukocyte-containing whole blood transfusion during colorectal cancer surgery ^[183]. Similar lymphocyte dysfunction has also been reported in patients with end stage renal disease after blood transfusion ^[200].

The immunosuppressive effects of blood transfusion have been reported to be long-lasting. Women who had received transfusions as much as 4.5 yr earlier had evidence of immune suppression (reduced lymphocyte function) ^[201]; colon cancer patients who had received transfusions several years prior to admission demonstrated reduced NK cell cytotoxicity and helper/suppressor cell ratios compared to a similar group of nontransfused patients ^[202]. Up to 18 mo of immune down-regulation (low T-cell and total lymphocyte counts) were found after bowel surgery for nonmalignant disease in transfused versus nontransfused patients ^[203]. In some studies, leukocytes have been implicated in the immunomodulation ^[204], while plasma has been implicated in others ^[67]. Jensen et al. ^[183] reported, in a randomized trial of patients having colorectal surgery, that the group of patients receiving leukocyte-depleted whole blood had significantly fewer postoperative infections than patients who received whole blood (2% vs 23%). A recent study found no differences in postoperative infection rates after colorectal cancer surgery between a leukocyte-depleted red cell group and a packed red cell without buffy coat group; both groups had a higher infection rate than a nontransfused control group ^[205]. The transient effect of anesthetics on lymphocyte function [reduced blastogenic response to phytohemagglutinin, reduced lymphokine production ^[206], and reduced helper/suppressor lymphocyte ratios ^[207]] or the effects of routine surgical stress ^[206,208] on lymphocytes are not implicated in this transfusion-induced immunosuppression ^[206,208].

Infections other than bacterial infections have also been reported to be exacerbated by blood transfusion. Shorter intervals from infection to full-blown AIDS ^[209], poorer survival after AIDS diagnosis ^[210], and recent onset cytomegaloviral infection ^[211] all support transfusion-induced immunosuppression.

Five reports ^{[190,212-215]} failed to demonstrate an increased rate of postoperative infection after blood transfusion. It should be noted that two ^[213,215] did demonstrate an increased (insignificant) postoperative infection rate in transfused patients, while another ^[214] demonstrated a significant transfusion effect on univariate, but not multivariate, analysis. Fernandez et al. ^[190] reported an infection rate of 6.9% in those receiving allogeneic blood, 5% for recipients of autologous blood, and 4.9% for nontransfused patients (no significant differences), while an infection rate of 11.9% was recorded for those requiring both allogeneic and autologous blood. Analysis of the group of patients who received both allogeneic and autologous blood may provide insight into the discrepancy between the "negative" study by Fernandez et al. ^[190] and the positive (significant transfusion effect) study by Murphy et al. ^[191]. Since both studies were done in similar groups of orthopedic patients, those patients who have higher than normal intraoperative blood loss may have the important (and frequent) problem of the confounding variables of hypotension, hypovolemia, and anemia. This may explain why Fernandez et al. ^[190] found a higher postoperative infection rate only in those patients requiring both allogeneic and autologous blood, i.e., the group with the greatest blood loss. On multivariate logistic regression analysis of their data, Fernandez et al. demonstrated that allogeneic whole blood transfusion was, in fact, a significant risk factor for postoperative infection.

In conclusion, a large number of studies reporting an association of bacterial infection with perioperative blood transfusion have been reviewed. Alexander ^[216] summarized the available evidence implicating blood transfusion with infection as "circumstantial." Interpretation of clinical studies is made difficult by the continuing problem of uncontrolled variable confounding, as previously discussed. The large number of clinical studies cited strongly suggests that perioperative transfusion results in a higher incidence of infection and/or sepsis. Techniques, such as third generation leukocyte filters ^[217] for leukocyte removal, may be warranted for use in the infection-prone or immunocompromised patient in the present, while well designed prospective studies will hopefully define the role of blood and/or specific blood components in the risks of development of postoperative bacterial infection. A continued awareness of the risks of allogeneic blood and techniques to reduce patient exposure, including a lower transfusion trigger, intraoperative and postoperative readministration of shed blood, preoperative autologous donation with or without erythropoietin ^[218], and "bloodless surgery," are all important in the effort to avoid allogeneic transfusion. Spence et al. ^[219] have recently demonstrated the importance of the amount of surgical blood loss and postoperative mortality. Those patients who lost 500 mL or less had an overall mortality of 0%, while a group with intraoperative blood loss exceeding 500 mL had a significantly (P < 0.025) increased mortality rate.