Transfusion-Induced Immunomodulation Following Cancer Surgery: Fact or Fiction?

This is the authors' abstract. We don't add key points for this paper.

Exposure to a foreign antigen can induce both immune activation and immunosuppression. The immunomodulatory effect of allogeneic transfusions (blood products provided by a donor other than the patient) was first revealed by the positive association between graft survival and the number of allogeneic transfusions received by renal graft recipients (1,2), and this benefit persisted despite the use of the immunosuppressant cyclosporine (3). In patients undergoing surgery for cancer, such a post-transfusional immunomodulation might have detrimental effects by promoting infection, local tumor relapse, or metastasis. This commentary is a critical review of the literature conducted to determine if the transfusion of blood products during cancer surgery is itself capable of inducing immunosuppression and thereby increasing the risk of postoperative infection and/or tumor relapse and, if so, how this effect can be overcome. In addition, the biologic mechanisms that could contribute to a post-transfusional immunomodulation are reviewed and discussed. The first animal model studies (4–11) of the immunologic effects of transfusion yielded contradictory results depending on the study and/or tumor model; i.e., allogeneic transfusions stimulated (4–9,11) or reduced (6,10) tumor growth and/or metastasis formation. Moreover, some studies reported only transient enhancement of tumor growth (11) or no effect (7,9). These conflicting results may be due to the different experimental protocols used, i.e., type of tumor cells, doses, route of administration, and time of transfusion (before or after tumor cell inoculation). Blajchman and co-workers (12–14) performed extensive in vivo experimental (i.e., animal model) studies. Their major findings were as follows: Transfusion of allogeneic whole blood led to increases in tumor size and the number of metastases in C57BL mice inoculated with B16 melanoma cells and in DBA/2 mice inoculated with P815 mouse mastocytoma cells. This effect was not observed when the number of cells in the inoculum was doubled (12). Transfusion of allogeneic whole blood before inoculation of C57BL/6J mice with FSL-10 fibrosarcoma cells led to increases in the number of secondary metastases when compared with the results obtained with transfusion of syngeneic whole blood. This effect disappeared when the allogeneic blood was leukocyte depleted (depletion was accomplished by use of a second-generation leukocyte filter providing at least 2-log [99.5%] leukocyte depletion). In addition, injection of spleen cells from animals transfused with allogeneic blood products increased the number of metastases in the recipients relative to that in animals receiving spleen cells from animals transfused with syngeneic blood products. Identical results were obtained with the same experimental protocol but with the use of New Zealand white rabbits inoculated with VX-2 epithelial tumor cells (13). Transfusion of allogeneic whole blood after inoculation of C57BL/6J mice with FSL-10 fibrosarcoma cells led to increases in the number of secondary metastases when compared with the results obtained with transfusion of syngeneic whole blood. The effect disappeared when allogeneic blood was leukocyte depleted immediately after collection. In addition, the authors obtained the same effect with the combined leukocytes and platelets (i.e., buffy coat) of whole blood and found a similar but less marked effect with fresh plasma. Finally, when New Zealand white rabbits were inoculated with VX-2 epithelial tumor cells, the authors noted that transfusion of allogeneic blood that was not leukocyte depleted immediately after collection but after a week of storage gave results similar to those obtained with non-leukocyte-depleted allogeneic blood (14). Taken together, these results suggest that, at least in the context of specific experimental conditions, allogeneic transfusions promote tumor growth and/or metastasis formation. The effect is observed regardless of whether transfusions are given before or after the injection of tumor cells. In contrast, the detrimental effects of allogeneic transfusion are obtained when leukocytes are present in the transfused product and disappear if allogeneic blood is depleted of leukocytes prior to storage and transfusion. Landers et al. (15) reviewed 36 retrospective studies that analyzed the influence of perioperative transfusions (versus no transfusion) on relapses or survival of patients surgically treated for the following different types of solid tumors: hepatocellular carcinoma (16), breast cancer (17–26), head and neck cancer (27–34), lung cancer (35–41), soft-tissue sarcoma (42), osteosarcoma (43), gastric cancer (44–47), and prostate cancer (48–51). Twenty-one studies (16,18,21–24,27,29–32,35,36,40,42,43,45,46,48,49,51) demonstrated that transfusion was associated with a detrimental effect, and 15 studies (17,19,20,25,26,28,33,34,37–39,41,44,47,50) showed no effect. Landers et al. (15) concluded that the use of allogeneic blood products increases cancer recurrence after curative surgical resection. Unfortunately, the results of these retrospective studies must be interpreted with caution. Within a given study, various transfusion protocols and/or products (whole blood or red blood cells) were used. In addition, only 28 of these studies (16–20,23–28,30,32,33,35–40,42–44,47–51) attempted to control for some of the known confounding variables (e.g., age, tumor stage, preoperative hemoglobin levels, and tumor-specific prognostic factors). Of these 28 studies, 15 (16,18,23,24,27,30,32,35,36,40,42,43,48,49,51) showed that there was a statistically significant, independent effect of transfusion on tumor recurrence and/or survival. However, in these 15 studies, other important variables, such as the surgeon's ability (which strongly influences transfusion requirements) or the indications for transfusions, were not taken in account. Overall, although the need for transfusion has been found to be associated with poor prognosis in some studies, there is little evidence of a causal link between transfusion and poor prognosis. Three meta-analyses (15,52,53) of retrospective and nonrandomized prospective studies on the influence of transfusions during surgery for colorectal cancer have been published: Chung et al. (52), in an analysis of 17 studies, determined that there was a statistically significant increase in the relative risk (estimated at 1.69; 95% confidence interval [CI]41.31-2.19) of relapse or death because of cancer or any other cause among transfused patients relative to patients who were not transfused with blood product. Vamvakas and Moore (53), on the basis of 11 studies, determined that there was a statistically significant increase in the relative risk of relapse or death by cancer among transfused patients relative to patients who were not transfused (estimated at 1.37; 95% CI = 1.20-1.56). Landers et al. (15), on the basis of 29 studies, determined that there was a statistically significant increase in the relative risk of recurrence in transfused patients relative to patients who were not transfused (estimated at 1.33; 95% CI = 1.3–1.5). Unfortunately, these meta-analyses had several limitations. The criteria used to identify (15,53) or to select (15,52) the studies were not described. The quality and heterogeneity of the studies included in the meta-analysis were not assessed. In two of these meta-analyses (52,53), the relative risks were calculated on the basis of a composite end point (recurrence, death, or death by cancer, according to the study) that was collected at time points that varied from one study to another. In addition, in their analysis, the authors failed to take into account variability of follow-up periods. Lastly, confounding factors (e.g., age, cancer location, Dukes' stage, preoperative hemoglobin level, and duration of surgery) were frequently not taken into account, and the robustness of the results was not tested by means of a sensitivity analysis (especially one based on the quality of the studies). Despite the observation of statistically significant increases in relative risk associated with transfusion, these three metaanalyses (15,52,53) failed to prove definitively that transfusion, by itself, is associated with relapse, mainly because they were based on studies that did not randomly assign patients to treatment. In addition, these meta-analyses were unable to take into account several potential confounding factors. The most important study comparing allogeneic transfusion with autologous transfusion was conducted by Busch et al. (54). The aim of their study was to compare the effects of allogeneic and autologous transfusions on relapse-free survival at 4 years among patients with colorectal cancer. This randomized study involved 510 patients who underwent curative surgery at one of 14 participating centers during the period from August 1986 through November 1991. Among the 423 patients who actually underwent curative surgery, there was no significant difference (in an intent-to-treat analysis) in relapse-free survival between the allogeneic and autologous transfusion arms (66% versus 63%) at 4 years of follow-up. Similarly, there was no difference in relapse-free survival at 4 years in the following subgroup analyses: Three groups of patients who underwent curative surgery and actually received transfusions during the perioperative period (136 patients transfused with allogeneic products versus 102 transfused with autologous products versus 42 transfused with autologous and allogeneic products) Patients who underwent curative surgery and received only one or two units (allogeneic products only versus autologous products only) Nontransfused patients (autologous versus allogeneic, after randomization) In contrast, a statistically significant difference was found between transfused patients and nontransfused patients (4-year relapse-free survival rates 4 59% and 73%, respectively; twosided logrank test, P 4 .001). However, because these two groups were not determined by randomization, they were not comparable, meaning that the comparison is open to the same criticisms as the above-mentioned retrospective studies (16–51). This study (54) had some weaknesses. The effect on the patients of blood collection for autotransfusion was not taken into account in the autotransfusion arm of the study. Indeed, one study (55) has reported that donation of whole blood resulted in a decreased natural killer (NK) cell activity measured 5 days after donation with a persisting decrease 12 days later in cancer patients. In addition, the number of patients who were actually transfused was significantly higher in the autologous arm than in the allogeneic arm (74% versus 56%; two-sided chi-squared test, P<.001). Although the amount of blood lost during surgery was the same in the two arms, the higher number of transfused patients in the autologous arm was due to a statistically significant lower preoperative hemoglobin level in this arm (12.5 g/dL versus 14.1 g/dL in the allogeneic arm; two-sided Mann-Whitney test, P<.001), which probably resulted from blood collection for autotransfusion. The type of allogeneic blood product used was not described, and it was not stated whether it was the same in all centers and/or in all patients. Heiss et al. (56) conducted a clinical trial involving a smaller number of patients (120 patients) who underwent curative surgery for colorectal cancer at a single center during the period from November 1987 through March 1991. This randomized study compared the effects of autologous blood transfusions and leukocyte-poor (i.e., partial leukocyte depletion) allogeneic transfusions. Analysis of the 100 patients who actually underwent curative surgery showed no statistically significant difference in relapse-free survival (median follow-up time, 22 months) between treatment arms. Comparison of the subgroup of 38 patients in the autologous arm who received 0, 1, or 2 units of blood products with the corresponding subgroup of 31 patients in the allogeneic arm showed significantly longer relapsefree survival in the former (two-sided logrank test, P = .017). This analysis ignored the randomization, however, and the results must therefore be interpreted with caution. Indeed, patients who received 2 units or less of autologous blood may not have been comparable with those who received 2 units or less of allogeneic blood, even if the requirements were the same, in particular because of a lower preoperative hemoglobin value in the autologous transfusion group. Overall, results from these two trials (54,56), both of which studied the effects of autologous and allogeneic transfusions on patients undergoing surgery for colorectal cancers, indicated that there was not a statistically significant difference in relapse-free survival between the two arms. In view of the potential association between transfusionrelated immunosuppression and the amount of leukocytes found in the blood product (14), Houbiers et al. (57) performed a randomized study comparing the effects of leukocyte-poor red blood cell concentrates (mean value, 0.8 × 109 leukocytes/unit) with those of leukocyte-depleted red blood cell concentrates (mean value, 0.2 × 106 leukocytes/unit) in patients undergoing surgery for colorectal cancer at one of 16 centers during the period from June 1987 through December 1990. The end points were survival, relapse-free survival, and the risk of recurrence. Among the 1021 patients randomly assigned to one of the treatment arms, 150 were subsequently excluded for not fulfilling inclusion criteria. Of the remaining 871 patients, 697 actually underwent curative surgery. The results of this study indicated that there was no statistically significant difference between the patients receiving leukocyte-depleted red blood cell concentrates and those receiving leukocyte-poor red cell concentrates (analysis restricted to patients who underwent curative surgery) in survival (3-year rates: 73% versus 75%; two-sided logrank test, P 4 .97), disease-free survival (3-year rates: 64% versus 66%; two-sided logrank test, P 4 .96), or the risk of recurrence (3- year rates: 29% versus 27%; two-sided logrank test, P 4 .79). On the other hand, when transfused patients were compared with nontransfused patients (regardless of randomization), there was a statistically significant difference in survival (3-year rates: 81% versus 69%; two-sided logrank test, P=.001) and relapsefree survival (two-sided logrank test, P=.01) but not in the risk of recurrence (two-sided logrank test, P=.22). In a multivariate analysis that took into account the other prognostic factors, only the transfusion of more than 3 units of blood products was independently predictive of survival. Unfortunately, in this study (57), the indications for transfusion were not stated, and storage duration of the leukocyte-poor products was not specified (while the leukocyte-depleted products were transfused no later than 72 hours after collection). Patients were excluded after randomization, and the main analysis focused on 697 patients who underwent curative surgery, i.e., only 68% of the patients initially randomly assigned to treatment. The products compared were leukocyte-poor and leukocytedepleted red blood cell concentrates. The number of patients required was calculated to show a difference of in the end is that a decrease in leukocyte the in the leukocyte-poor products) have a difference from observed between the two arms, such a difference might have been found if products had been used. This is by the in with leukocyte-poor products relative to products even leukocyte was In addition, animal model studies have that a (i.e., in the leukocyte is to the effects by transfusion of blood products. Overall, these studies to the transfusion, by itself, induce immunosuppression to tumor recurrence in and, if so, is there any of this The link between perioperative transfusion and the of postoperative has been in three retrospective studies and in one nonrandomized study of these studies compared patients transfused with autologous products with patients transfused with allogeneic products and, in two studies with a of nontransfused patients. of these studies showed a statistically significantly increased of postoperative in the allogeneic as compared with patients transfused with autologous product or not However, despite multivariate analysis, all of these studies were not to take into account all known and potential confounding factors. In addition, the of patients in were The results to prove that transfusion is an independent risk for postoperative of patients who not have cancer but who perioperative transfusions the of between autologous transfusion and allogeneic transfusion, given the immunologic and risks associated with the On the other hand, the use of autologous transfusion in cancer patients has not been because it to be that this not increase the risk of tumor cell a the only randomized studies in which the of postoperative was the main end point have involved cancer the same of are as with tumor recurrence Heiss et al. a with the same patients who in the randomized study on tumor recurrence The end point was the of postoperative The analysis showed a statistically significant difference (two-sided chi-squared test, P = in the of postoperative infection as the of × or × and or a between the autologous arm in of patients) and the allogeneic arm in 17 of The difference persisted after for other risk factors, i.e., tumor and preoperative of the difference was no longer statistically significant when only infection was taken into account infection was by treatment more than a week and the need for or by of the to with of and In addition, Heiss et al. that the effect of blood collection for autotransfusion a confounding when the of allogeneic transfusion in the of postoperative In their study, Busch et al. (54) on the basis of an to the of such among patients who received allogeneic versus autologous transfusions and found no statistically significant difference between the treatment arms and et al. conducted a randomized study comparing transfusion of whole blood with and leukocyte and received red blood cells and in undergoing curative surgery for colorectal cancer. The end point was the of postoperative or in of patients who received whole blood, in one of patients who received and whole blood (two-sided chi-squared test, and in two of nontransfused patients (two-sided chi-squared test, The number of was not statistically different among the three groups of patients, two of patients, and of patients, Unfortunately, in this study, the analysis was not performed on an the indications for transfusion were not stated, patients who did not curative surgery were excluded after been randomly assigned to a treatment and the treatment groups were Houbiers et al. (57) found no statistically significant difference between patients randomly assigned to products and those randomly assigned to leukocytedepleted products versus two-sided chi-squared test, did however, a statistically significant difference between transfused and nontransfused patients versus two-sided chi-squared test, but this analysis ignored the and, as the results must therefore be interpreted with caution. et al. a randomized study comparing transfusion of leukocyte-poor red blood cell concentrates (mean value, × 109 leukocytes/unit) patients randomly assigned and and leukocyte-depleted red blood cell concentrates (mean value, 5 × 106 leukocytes/unit) patients randomly assigned and The end point was the of postoperative The of postoperative and the of were higher in the patients transfused with red blood cell concentrates than in patients transfused with leukocyte-depleted red blood cell concentrates versus test, P 4 and versus chi-squared test, In the randomized studies involved patients with only one type of cancer and yielded contradictory not evidence to the that perioperative transfusion by itself immunosuppression to increase the of postoperative and, of to determine whether it is to this potential effect. The immune cells and/or factors for the immunomodulatory effects of transfusion have not been cells major are the cells most frequently the and a in cell to and have been after transfusion In addition, of cells of cells cells (i.e., cells capable of a or cells of and of (i.e., of cells) are mechanisms that have been in studies involving the effect of blood transfusion on studies in that the immunologic effects of transfusion are to the of these studies are to two studies by et al. from the same of patients. patients undergoing surgery for colorectal cancer were randomly assigned at the time of transfusion to whole blood patients) or leukocyte-depleted whole blood The control was a of nontransfused patients The transfusion of whole blood was associated with a in cell a in the and an increase in 2 and in comparison to the transfusion of leukocytedepleted whole blood or of transfusion. of these between the of patients receiving depleted whole blood and the of patients who were not the influence of transfusion on by their was in and, in a in cell can be into on the basis of and with as as immune have been and and induce the of in These and and are involved in most by the of and by the of cells are involved in the of and (i.e., Overall, an important in On the other hand, the of in cells. These cells mainly and an important in the immune by and by are involved in by an as as the and activation of by cells have an between and by cells and by cells by Overall, by several associated with a could be associated with a and, with effects in of or immune On the other hand, of a could be a critical for the of after and the effects of transfusion on the of stimulated spleen in a i.e., cells in of allogeneic whole blood was associated with as by an at and transient of in the an increased of measured on that was persisting on and a in the of at In study, et al. studied the effect in mice of fresh allogeneic transfusion versus the transfusion of allogeneic blood for 15 days at 4 to donor was reduced in both In addition, transfusion of whole blood was associated with the of The in the observed in these might not be to the Indeed, a in vivo of allogeneic leukocytes has been in mice and not in rabbits or In addition, several mouse are a or the of In cells to a of in However, and immunomodulation has been observed in a model (13). et al. reported in a that was increased in a of patients transfused with allogeneic whole blood when compared with that in patients transfused with autologous blood or with that in nontransfused patients. they reported a statistically significant test, P = increase in in the of patients transfused with autologous blood compared with the of nontransfused patients. effect as as the route of could be involved in such a Indeed, a of the antigen to induce a induce The mechanisms these effects of antigen are not of the cells are cells if the antigen is a of these cell types and the of the specific a On the other hand, when the antigen is antigen may be by cells that not such as are present in both in the blood and in the which is the first by Moreover, of antigen to increasing and or induce a of immunologic that cells The influence of leukocyte antigen between donor and blood transfusion on the type of immune versus observed after transfusion has been in several studies Analysis of cell after transfusion has that a specific can if donor and an least immune could to such a specific an with the activation of on the transfused allogeneic by an such as or between donor and and an with these cells cells that have taken allogeneic from the donor of in immunosuppression is by the observation that the of an after transfusion in the of a is associated with in the of cells a given mechanisms have been to these an between and allogeneic cell and/or of in both in the of the cells. However, more studies have been unable to a specific effect of on et al. have reported that blood transfusion, of an activation of the Similarly, study found that by did not the Lastly, a decrease in in patients has been after transfusion, and this effect was found to be most in recipients of blood is that this of activity was observed other nontransfused allogeneic cell and could therefore contribute to a immunosuppression. The of an between donor and could donor cells to in the and therefore induce a specific not only by such as those but by In allogeneic for one in the and induce a specific In a survival of allogeneic cell in the blood is by a study of et al. who reported a survival in receiving transfusion. the of no study the influence of between donor and on the has been factors can influence the immune in cancer patients. In to the tumor itself the means used to the cancer, i.e., surgery and can induce immunosuppression. and/or postoperative transfusion may be an capable of inducing or this immunosuppression. of several animal have in particular experimental conditions, a effect of allogeneic transfusion on tumor and/or metastasis formation. This effect to be mainly to leukocyte of the transfused blood products. randomized clinical studies have been unable at the present time to the potential effects of allogeneic transfusion on the immune to the tumor and to in cancer patients. In the of a immunomodulation after cancer surgery has not been studies are required to and to the