Hindawi Cardiology Research and Practice Volume 2019, Article ID 7543917, 10 pages https://doi.org/10.1155/2019/7543917 Review Article Leukodepleted Packed Red Blood Cells Transfusion in Patients Undergoing Major Cardiovascular Surgical Procedure: Systematic Review and Meta-Analysis Daniel Simancas-Racines ,1 Ingrid Arevalo-Rodriguez,1,2 Gerard Urrutia,3 Diana Buitrago-Garcia,1 Solange Núñez-González ,1 Marı́a José Martı́nez-Zapata,3 Eva Madrid ,4 Xavier Bonfill,3 and Ricardo Hidalgo-Ottolenghi1 1 Cochrane Ecuador, Centro de Investigación en Salud Pública y Epidemiologı́a Clı́nica (CISPEC), Facultad de Ciencias de la Salud Eugenio Espejo, Universidad UTE, Quito 170129, Ecuador 2 Clinical Biostatistics Unit, Hospital Universitario Ramon y Cajal (IRYCIS), CIBER Epidemiology and Public Health (CIBERESP), Madrid 28034, Spain 3 CIBER Epidemiologı́a y Salud Pública (CIBERESP), Iberoamerican Cochrane Centre Biomedical Research Institute Sant Pau (IIB Sant Pau), Barcelona 08041, Spain 4 Centro Interdisciplinario de Estudios en Salud (CIESAL), Escuela de Medicina, Universidad de Valparaı́so, Cochrane Chile, Valparaı́so 2391415, Chile Correspondence should be addressed to Daniel Simancas-Racines; danielsimancas10@gmail.com Received 12 September 2018; Revised 21 December 2018; Accepted 23 January 2019; Published 25 February 2019 Academic Editor: Stephan von Haehling Copyright © 2019 Daniel Simancas-Racines et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Background. Leukocytes contained in the allogeneic packed red blood cell (PRBC) are the cause of certain adverse reactions associated with blood transfusion. Leukoreduction consists of eliminating leukocytes in all blood products below the established safety levels for any patient type. In this systematic review, we appraise the clinical effectiveness of allogeneic leukodepleted (LD) PRBC transfusion for preventing infections and death in patients undergoing major cardiovascular surgical procedures. Methods. We searched randomized controlled trials (RCT), enrolling patients undergoing a major cardiovascular surgical procedure and transfused with LD-PRBC. Data were extracted, and risk of bias was assessed according to Cochrane guidelines. In addition, trial sequential analysis (TSA) was used to assess the need of conducting additional trials. Quality of the evidence was assessed using the GRADE approach. Results. Seven studies met the eligibility criteria. Quality of the evidence was rated as moderate for both outcomes. The risk ratio for death from any cause comparing the LD-PRBC versus non-LD-PRBC group was 0.69 (CI 95% � 0.53 to 0.90; I2 � 0%). The risk ratio for infection in the same comparison groups was 0.77 (CI 95% � 0.66 to 0.91; I2 � 0%). TSA showed a conclusive result in this outcome. Conclusions. We found evidence that supports the routine use of leukodepletion in patients undergoing a major cardiovascular surgical procedure requiring PRBC transfusion to prevent death and infection. In the case of infection, the evidence should be considered sufficient and conclusive and hence indicated that further trials would not be required. 1. Introduction transfuse anemia patients who require infusion of red blood cell (RBC) to restore tissue oxygenation. However, PRBC Blood transfusion is an acute intervention implemented to transfusion is associated with an increasing risk of infectious solve life- and health-threatening conditions on a short-term and noninfectious adverse events, the most common among basis [1]. Packet red blood cell (PRBC), prepared by re- which are nonhemolytic febrile transfusion reactions, hu- moving plasma from whole blood, is typically used to man leukocyte antigen (HLA) alloimmunization and platelet 2 Cardiology Research and Practice refractoriness observed in multitransfused patients, and surgical patients. Therefore, the objective of this review was transmission of leukotropic viruses. One of the strategies to assess the effects of LD-PRBC in patients undergoing commonly used to prevent posttransfusion complications is major cardiovascular surgical procedure, who are more leukodepletion, that is, a process by which the white blood likely to suffer significant blood loss [20] and consequently cells (WBCs) are intentionally reduced by almost 99.99% to have a much higher probability to receive transfusions of PRBC. According to the current standards, PRBC residual blood products [18]. leukocytes require to be <5 × 106 cells per unit according to the FDA or <1 × 106 cells per unit according to the Council 2. Methods of Europe [2]. Several studies have focused on the advantages of leu- We conducted a systematic review of randomized clinical kodepleted PRBC for transfusion in cardiac surgery [3–8], trials (RCTs). The protocol was registered in PROSPERO, an colorectal surgery [4, 9–13], gastrointestinal surgery international prospective register of systematic review [4, 7, 14], and renal transplantation [15–17]. Cardiac surgery protocols (registration number: CRD42018103104). accounts for a large proportion of the blood transfusions administered each year. Transfusion rates have been re- 2.1. Inclusion and Exclusion Criteria. To be included in the ported from 7.8% to 92.8% for combination of coronary review, studies had to meet the following criteria: ran- artery bypass graft (CABG) surgery with valve or other domized controlled trial conducted with patients of any age major surgical interventions [18]. undergoing a major cardiovascular surgical procedure (such Although blood transfusions are necessary in major as valve surgery, cardiac bypass, and aneurysm repair), cardiovascular surgery, several studies found that blood requiring allogeneic PRBC, with the aim of comparing LD- transfusions had also deleterious effects. Considering the PRBCs versus non-LD-PRBCs. Besides, studies had to report abovementioned reasons, this subgroup of the surgical results on death from any cause and infection from any cause population is of special interest for the analysis of the ef- (unspecified). We excluded studies with other designs or fectiveness and safety of transfusion practices. Several that included patients transfused with other blood com- studies have been showed that cardiac surgery is related with ponents as a principal intervention. tissue trauma, ischemia-reperfusion injury, and blood sur- face contact. These clinical settings induce systemic effects and release of inflammatory mediators, which are supposed 2.2. Search Strategy. We carried on sensitive electronic to play a role in the development of systemic inflammatory searches in the Cochrane Injuries Group Specialized Reg- response syndrome (SIRS), multiple organ dysfunction ister, the Cochrane Central Register of Controlled Trials syndrome (MODS), infections, and postoperative compli- (CENTRAL, the Cochrane Library), MEDLINE (OVID 1946 cations [19, 20]. Additionally, in cardiac surgery, PRBC are to present), EMBASE (Elsevier), LILACS, Clinical Trials frequently transfused and these transfusions have been register (http://www.clinicaltrials.gov), and the WHO In- found to be associated (dose-dependent) with an increased ternational Clinical Trials Registry Platform (http://apps. risk of postoperative infections and mortality after cardiac who.int/trialsearch/). We ran the most recent search on surgery [21–25]. It is not clear what the possible mechanisms June 10th, 2018 (see Supplementary material (available here) that clarify this association could be [26] although the for details). presence of allogeneic leukocytes in PRBC are hypothetical to play a fundamental role, probably by evolving into the 2.3. Screening, Data Extraction, and Assessment of Risk of Bias. inflammatory response after cardiac surgery. In support of Two review authors independently screened all titles and this reasoning, Bilgin et al. found higher concentrations of abstracts retrieved by the search against the selection criteria proinflammatory mediators (such as IL-6 and IL-10) during and obtained full texts when necessary. All decisions re- the postoperative period in cardiac valve surgery patients garding inclusion and exclusion were made by consensus. receiving allogeneic leukocyte-containing blood trans- Data extraction was performed in duplicate and risk of bias fusions compared with leukocyte-depleted blood trans- (RoB) assessment of the included studies, following the fusions [27]. This, in turn, would support the potential domain-based evaluation method described in the Cochrane benefits of the routine use of leukodepleted PRBC trans- Handbook for Systematic Reviews of Interventions [30]. fusion in the setting of cardiac surgery to reduce infectious As a support to establish our conclusions on the effects of complications [28]. leukodepleted PRBC, we developed a “Summary of Find- One previous review reported that there is no clear ings” table using the GRADE approach for assessing the evidence supporting the effectiveness of leukodepleted quality of evidence, according to the methods and recom- PRBC for preventing transfusion-related acute lung injury mendations described in the Cochrane Handbook for Sys- (TRALI) or reducing mortality and infectious or non- tematic Reviews of Interventions [30]. infectious complications in patients undergoing any type of surgery [29]. However, a considerable heterogeneity in the pooled estimation was found due to the inclusion of different 2.4. Statistical Analysis. We calculated the treatment pooled types of populations (oncology, trauma, and cardiac surgery effect for death from any cause and infection from any cause patients) which may have prevented the detection of ben- by means of the risk ratio (RR) with the corresponding 95% eficial effects in some particularly relevant subgroups of confidence intervals (CI), using the random-effects model Cardiology Research and Practice 3 approach for data pooling in the meta-analysis, which ac- 7999 records 4 additional records counts for statistical heterogeneity across studies and leads identified through identified through to a more conservative estimate of the effect. We estimated database searching other sources the statistical heterogeneity in the meta-analysis by using the I2 statistics [31]. All these analyses were carried out using RevMan 5.3 [32]. 4022 records after duplicates removed We used trial sequential analyses (TSA) to estimate the required information size for death from any cause and infection from any cause in order to reduce the risk of random errors in our conclusions and calculating the re- 3948 records excluded based on quired information size for a meta-analysis. This analysis the title or abstract makes available an adjusted statistical threshold for benefits, 3 studies identified harms, or futility before the required information size was but with insufficient reached [33, 34]. By using this method, we aimed at con- information to include— trolling the risk of type I and type II errors due to sparse data listed in “studies and repetitive testing of accumulating data [33, 35–37]. awaiting 3977 records screened assessment” 3. Results 3.1. Literature Search Results. We initially identified 7,999 22 full-text articles records from the search strategies updated until June 2018 [23] excluded, with reasons: and four more from other sources. After removing duplicates, (i) Did not meet study 4,022 were manually screened, and 3,993 records were ex- population (N = 13) (ii) Did not meet study cluded for title and abstract. We reviewed the full text of 29 design (N = 7) studies, 22 of which were excluded. Finally, only seven RCTs 29 full-text articles (iii) Did not meet study with 3,154 participants were included into the qualitative and assessed for eligibility intervention (N = 2) quantitative analysis of this report [3, 5–8, 38, 39] (Figure 1). 7 studies included in the 3.2. Characteristics of the Included Studies. Two of the seven qualitative synthesis included studies only had abstract available [38, 39]. Three studies were carried out in the Netherlands (60%) [3, 6, 7]. All studies included adult patients, with mean ages greater 7 studies included in the than 60 years. Leukodepletion process was described only in quantitative synthesis (meta-analysis) three of the seven studies, using three different criteria (1.2 ± 1.4 × 106, 5 × 106, or 0.15 ± 0.02 × 106, leukocytes per Figure 1: Flow diagram of the literature search and study selection. unit) [3, 5, 6]. Sample size for the transfused patients ranged from 38 to 304 (mean 189) for the leukodepleted group and 31 to 303 and up to twelve months [38]. Two studies did not report the (mean 207) for the comparator group. In van Hilten 2004 follow-up time for the death from any cause outcome [7, 39]. study, we included only patients undergoing aneurysm re- The overall death from any cause at the last follow-up was pair, excluding gastrointestinal oncology surgery. From the 5.96% (79 events). The pooled RR for the comparison of LD- van de Watering 1998 study, we included the stored-filtered PRBC versus non-LD-PRBC was 0.69 (CI 95% � 0.53 to 0.90; (SF) group within the leukodepleted group and the packed I2 � 0%), thus showing a statistically significant reduction in cells (PC) group within the comparator group. Table 1 describes the main characteristics of included studies. Re- the risk of death from any cause with LD-PRBS (31% relative garding risk of bias, most studies were assessed as “unclear reduction) (Figure 2). risk” regarding selection bias (random sequence generation We conducted TSA analysis to determine the reliability and allocation concealment) due to lack of details in the of one of the outcomes of this systematic review: death from study report. Only one study was assessed as having “low any cause (Figure 3). TSA of LD-PRBC compared with risk” of bias for blinding of participants, personnel, and control non-LD-PRBC indicated that the optimal in- outcome assessors [6]. We considered the missing outcome formation size needed to reliably detect a plausible effect was data shown in van Hilten 2004 as having a “high risk” of bias, 5,187 patients. However, 2,771 (53.4%) patients had so far due to fact that losses are likely to be related to the main been collected. The cumulative z-curve of all trials crossed outcomes [7]. In addition, three studies were considered as the traditional boundary but did not cross the trial sequential at “unclear risk” of other bias [3, 7, 8, 38, 39]. monitoring boundary. The TSA α-spending adjusted 95% CI overlapped with no effect (RR 0.49 and RR 1.02, re- spectively); thus, the TSA yielded an inconclusive result 3.3. Target Death from Any Cause. In the included studies, about the true effect of LD-PRBC in preventing death from death was assessed at 30 days [8], 60 days [3], 90 days [5, 6], any cause. Therefore, for death from any cause outcome, 4 Table 1: Characteristics of the included studies. Age mean Leukodepleted group Comparator group ID Country by group Male (%) Transfused Leukodepleted Transfused Cointerventions (LD, C)∗ Type of surgery Filter Comparator patients,n definition patients, n Cardiac valve surgery Bilgin et al. with or without 0.15 ± 0.02 × 106 Cellselect- Buffy coat depleted Netherlands 65.3, 66.6 53 to 57 216 216 Platelets 2004 [6] coronary artery bypass per unit Optima packed cells graft Coronary artery bypass Boshkov et al. USA Unclear Unclear 304 graft and/or cardiac Unclear Unclear 258 Standard RBC No 2006 [38] valve replacement Open-heart surgery, Bracey et al. coronary artery bypass USA Unclear Unclear 170 Unclear Unclear 187 Standard RBC Unclear 2002 [39] graft, and valve replacement Connery et al. Coronary artery bypass USA 62.9, 66 71 to 74.2 38 Unclear Unclear 31 Standard RBC Platelets 2005 [8] graft van de Coronary artery bypass 1.2 ± 1.4 × 106 Cellselect- Buffy coat depleted Watering et al. Netherlands 62.9, 64.4 72.2 to 73.7 287 graft and/or cardiac 294 No per unit optima packed cells 1998 [3] valve surgery Acute aneurysm surgery van Hilten Buffy coat depleted Netherlands 66, 71 Unclear 133 and elective aneurysm Unclear Unclear 128 No et al. 2004 [7] packed cells surgery Buffy coat depleted Ratio men/ Coronary artery bypass Wallis et al. 6 packed cells and red UK 61.7, 62.4 women: 2.6 to 176 graft and/or cardiac 5 × 10 per unit BPF4 333 No 2002 [5] blood cells concentrate 2.9 valve surgery with plasma reduction RBCs � red blood cells. ∗ LD: leukodepleted group; C: comparator group. Cardiology Research and Practice Cardiology Research and Practice 5 Leukoreduced PRBC Non-leukoreduced PRBC Risk ratio Risk ratio Risk of bias Study or subgroup Events Total Events Total Weight M-H, fixed, 95% CI M-H, fixed, 95% CI A B C D E F G Bilgin et al. 2004 20 216 30 216 26.5% 0.67 [0.39, 1.14] + + + + + + + Boshkov et al. 2006 15 304 25 258 23.9% 0.51 [0.27, 0.94] ? ? ? + ? ? + Bracey et al. 2002 10 170 14 187 11.8% 0.79 [0.36, 1.72] - ? ? ? + + ? Connery et al. 2005 1 38 1 31 1.0% 0.82 [0.05, 12.52] + ? ? ? + + ? van de Watering et al. 1998 1 287 1 294 0.9% 1.02 [0.06, 16.30] ? ? ? ? + + + van Hilten et al. 2004 31 133 33 128 29.8% 0.90 [0.59, 1.38] + + + ? - + + Wallis et al. 2002 1 176 10 333 6.1% 0.19 [0.02, 1.47] ? ? ? + + + + Total (95% CI) 1324 1447 100.0% 0.69 [0.53, 0.90] Total events 79 2 2 114 Heterogeneity: χ = 4.23, df = 6 (P = 0.65); I = 0% Test for overall effect: Z = 2.71 (P = 0.007) 0.01 0.1 1 10 100 Leukoreduced PRBC Non-leukoreduced PRBC Risk of bias (A) Random sequence generation (selection bias) (B) Allocation concealment (selection bias) (C) Blinding of participants and personnel (performance bias) (D) Blinding of outcome assessment (detection bias) (E) Incomplete outcome data (attrition bias) (F) Selective reporting (reporting bias) (G) Other bias Figure 2: Forest plot of included studies evaluating LD-PRBC versus non-LD-PRBC in patients undergoing a major cardiovascular surgical procedure: death from any cause outcome. Cumulative Death. Pc 7.87%; RRR 25%; alpha 5%; beta 20%; diversity 0% is a two-sided graph Z-score Death. Pc 7.87%; RRR 25%; alpha 5%; beta 20%; diversity 0% = 5187 8 7 leukoreduced PRBC 6 Favours 5 4 3 Z-curve 2 1 2771 –1 Number of patients non-leukoreduced PRBC –2 (linear scaled) –3 Favours –4 –5 –6 –7 –8 Figure 3: TSA calculated to reliably detect a 25% relative change in the incidence of death from any cause, assuming a control group event rate of 8.99% with a power of 80% at an alpha of 5%. Notes: DARIS: diversity adjusted required information size; Pc: event proportion in the control group; RRR: relative risk reduction in the intervention group; (a) type I error; (b) type II error; DIVERSITY: diversity (D-square). Dead: the required information size was 5,187 participants. The cumulative Z-score (blue line) did not cross the trial sequential monitoring boundaries for benefit (red lighter inward sloping line) after the seven trials. more RCTs are needed (Figure 3). According to GRADE the risk of infection with LD-PRBS (23% relative risk re- criteria, the quality of the evidence was moderate to low duction) (Figure 4). (Table 2). We conducted TSA to determine the reliability of one of the outcomes of this systematic review: infection from any cause. TSA of LD-PRBC compared with non-LD-PRBC 3.4. Target Infection. Regarding infection, five out of seven indicated that the optimal information size needed to re- included studies reported this outcome [3, 5–8, 38, 39]. liably detect a plausible effect was 1,315 patients. However, Incidence of infection after follow-up was 19.8% (494 the accumulate data of 1,852 participants constituted more events). The pooled RR for the comparison of the LD-PRBC than 100% of the optimal information size calculated. The versus non-LD-PRBC group was 0.77 (IC 95% � 0.66 a 0.91; cumulative z-curve of all trials crossed the traditional I2 � 0%), thus showing a statistically significant reduction in boundary as well as the trial sequential monitoring 6 Cardiology Research and Practice Table 2: Summary of findings: GRADE criteria. Leukodepleted packed red blood cells transfusion in patients undergoing a major cardiovascular surgical procedure Patient or population: patients undergoing a major cardiovascular surgical procedure transfused with allogeneic packed red blood cells (PRBC) Setting: hospital Intervention: leukodepleted (PRBC) Comparison: non-leukodepleted (PRBC) Anticipated absolute effects∗ (95% CI) Relative No. of Risk with non- Risk with Certainty of the Outcomes effect (95% participants Comments leukodepleted leukodepleted evidence (grade) CI) (studies) (PRBC) (PRBC) Death. Number of events of RR 0.69 the total number of 79 per 1.000 54 per 1.000 (42 (0.53 to 2771 (7 RCTs) ⊕⊕⊕◯ a,b TSA yielded an inconclusive transfused patients to 71) 0.90) Moderate result. reported Infection. Number of RR 0.77 events of the total number 259 per 1.000 200 per 1.000 (0.66 to 1852 (5 RCTs) ⊕⊕⊕◯ TSA yielded a of transfused patients (171 to 236) 0.91) Moderatea,b conclusive result. reported ∗ The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RR: risk ratio. The overall certainty in the evidence should be assessed for each important outcome using four or three categories (such as high, moderate, low, and/or very low) and definitions for each category that are consistent with the definitions used by the GRADE Working Group. aDowngraded because one study has high risk of bias due to attrition bias; another study has other risk of bias at high risk of bias; three studies have unclear risk of bias in generation and allocation concealment of random sequence. bDowngraded due to high risk of bias; one study has high risk of bias due to attrition bias; another study has high risk of bias in a random sequence; five studies have unclear risk of bias in generation and allocation concealment of random sequence. Leukoreduced PRBC Non-leukoreduced PRBC Risk ratio Risk ratio Risk of bias Study or subgroup Events Total Events Total Weight M-H, fixed, 95% CI M-H, fixed, 95% CI A B C D E F G Bilgin et al. 2004 52 216 73 216 29.8% 0.71 [0.53, 0.96] + + + + + + + Connery et al. 2005 5 38 8 31 3.6% 0.51 [0.19, 1,40] + ? ? ? + + ? van de Watering et al. 1998 50 287 71 294 28.6% 0.72 [0.52, 1.00] ? ? ? ? + + + van Hilten et al. 2004 55 133 56 128 23.3% 0.95 [0.71, 1.25] + + + ? - + + Wallis et al. 2002 22 176 52 333 14.7% 0.80 [0.50, 1.27] ? ? ? + + + + Total (95% CI) 850 1002 100.0% 0.77 [0.66, 0.91] Total events 184 260 2 2 Heterogeneity: χ = 3.07, df = 4 (P = 0.65); I = 0% Test for overall effect: Z = 3.08 (P = 0.002) 0.01 0.1 1 10 100 Leukoreduced PRBC Non-leukoreduced PRBC Risk of bias (A) Random sequence generation (selection bias) (B) Allocation concealment (selection bias) (C) Blinding of participants and personnel (performance bias) (D) Blinding of outcome assessment (detection bias) (E) Incomplete outcome data (attrition bias) (F) Selective reporting (reporting bias) (G) Other bias Figure 4: Forest plot of included studies evaluating LD-PRBC versus non-LD-PRBC in patients undergoing a major cardiovascular surgical procedure: infection outcome. boundary. The TSA α-spending adjusted 95% CI, did not moderate according to the quality of the body of evidence overlap the zone of no effect (RR 1.0), and is compatible with available, but conclusive according to the TSA analysis. As for a potential benefit (RR 0.65 and RR 0.93, respectively); thus, the later outcome, the certainty in the result is also moderate the TSA yielded a conclusive result about the true effect of but not as conclusive as regarding the former result. LD-PRBC in preventing infection from any cause. Therefore, Several reviews and meta-analyses on postoperative for infection from any cause outcome, no more RCTs are infection and death related to the leukoreduction of blood needed (Figure 5). According to GRADE criteria, the quality products have been carried out previously [29]. However, as of the evidence was moderate (Table 2). far as we know, no systematic reviews or meta-analysis with TSA focused specifically in patients undergoing a major 4. Discussion cardiovascular surgery have yet been published. Neverthe- less, it is important to point out that numerous studies have In this systematic review, we showed that patients undergoing sought to demonstrate the benefit of leukocytes reduction a major cardiovascular surgery who were transfused with LD- from red blood cell concentrates in different scenarios. PRBCs might benefit from a decreased risk of infections and One previously published Cochrane systematic review death from any cause. The certainty for the first outcome is comparing LD-PRBC with non-LD-PRBC in all type of Cardiology Research and Practice 7 Cumulative Infection. Pc 25.9%; RRR 25%; alpha 5%; beta 20%; diversity 0% is a two-sided graph Z-score Infection. Pc 25.9%; RRR 25%; alpha 5%; beta 20%; diversity 0% = 1315 8 7 6 leukoreduced PRBC 5 Favours 4 3 Z-curve 2 1 1090 Number of –1 patients (linear scaled) non-leukoreduced PRBC –2 –3 Favours –4 –5 –6 –7 –8 Figure 5: TSA calculated to reliably detect a 25% relative change in the incidence of infection from any cause, assuming a control group event rate of 24.6% with a power of 80% at an alpha of 5%. Notes: DARIS: diversity adjusted required information size; Pc: event proportion in the control group; RRR: relative risk reduction in the intervention group; (a) type I error; (b) type II error; DIVERSITY: diversity (D- square). Infection: the required information size was 1,315 participants. The cumulative Z-score (blue line) crossed the trial sequential monitoring boundaries for benefit (red lighter inward sloping line) after the second trial (1,090 participants); thus, the risk of random error in the finding can be excluded. Therefore, it is not necessarily additional testing based on the assumed intervention effect of the RRR of 25%, an alpha of 5%, and a beta of 20% with respect to this result. surgical patients requiring transfusion, showed a non- cells as an intervention, as well as nonrandomized studies. significant decrease in the risk of infection (10 trials with Moreover, heterogeneity between studies was not taken 6,709 patients) and all-cause mortality (9 trials with 6,485 into account, and the risk of bias of included studies was patients). However, these results were limited by a signifi- not appropriately assessed. Furthermore, some relevant cant heterogeneity [29]. These findings contradict our results studies were not included, and patients undergoing major partially because the population included in that review was cardiovascular surgery were not evaluated independently. very heterogeneous which may have hidden the effect of the Finally, any of the aforementioned reviews did not perform intervention in specific subgroups of interest. In contrast, a trial sequential analysis, in order to control the risks of our study was specifically focused in patients undergoing type I and type II errors due to sparse data and repetitive major cardiovascular surgery. This surgery has been related testing of accumulating data in all of calculated meta- to a higher volume of PRBC transfused per patient compared analyses. to colorectal and gastrointestinal surgery, as well as the fact Applicability of this evidence to daily clinical practice is that the leukocytes are transfused to an already activated restricted for several reasons. Firstly, external validity may be inflammatory system caused by cardiopulmonary bypass limited to patients undergoing the same major cardiovas- [40]. Thus, it is reasonable to assume that the potential cular surgery procedures that have been included in this harms of using non-LD-PRBC are higher than in other review. Secondly, the identified studies did not adequately surgical scenarios. report several factors related to the transfusion of RBC In contrast, the findings of other systematic reviews are practices that need to be considered when interpreting the in accordance with our results. One systematic review results, such as the use of LD or non-LD platelets as a showed a 50% reduction in the risk of a postoperative cointervention, the timing of LD (pre-post-storage), and the infection [41], and another one analyzing “only patients type of the filter used, among other factors. Thirdly, the who received transfusion” found a statistically significant number of units transfused in major cardiovascular surgery reduction of 40% in postoperative infection risk, but a is massive in most of the cases compared with other surgical nonsignificant reduction on mortality [42]. However, these and nonsurgical transfusion clinical settings, and therefore, two reviews have several limitations. They included studies patients undergoing major cardiovascular surgery may that used other blood components apart from LD-PRBC suffer a posttransfusion complication is more likely. Finally, 8 Cardiology Research and Practice the studies reported different definitions for infections, and Acknowledgments the mortality was assessed in different time periods (30 days, 60 days, 90 days, and up to twelve months). This work was supported by Universidad UTE. Most developed countries currently recommend uni- versal LD-PRBC. However, high costs associated with this Supplementary Materials procedure, such as the direct costs of LD-PRBC and other associated costs (i.e., costs associated with maintaining dual Appendix 1: detailed search strategy. (Supplementary inventories of leukodepleted PRBC and non-leukodepleted Materials) PRBC), merit special attention [43]. As a main strength of this report, we applied Cochrane systematic review meth- References odology throughout all the process. However, despite our [1] A. G. Tsai, A. Hofmann, P. Cabrales, and M. Intaglietta, effort to include all published studies evaluating LD-PRBC “Perfusion vs. oxygen delivery in transfusion with “fresh” and for the prevention of infection and death from any cause in “old” red blood cells: the experimental evidence,” Transfusion patients undergoing major cardiovascular surgery, it is and Apheresis Science, vol. 43, no. 1, pp. 69–78, 2010. possible that not all studies were identified. The Bilgin 2004 [2] R. Sharma and N. Marwaha, “Leukoreduced blood compo- and Connery 2005 studies reported the use of platelets as a nents: advantages and strategies for its implementation in cointervention, which could intervene as a confounder in the developing countries,” Asian Journal of Transfusion Science, analyses. The study Kremke et al. concluded that platelet vol. 4, no. 1, pp. 3–8, 2010. transfusion of CABG is not associated with increased [3] L. M. G. van de Watering, J. Hermans, J. G. A. Houbiers et al., postoperative mortality [44]; on the contrary, the study “Beneficial effects of leukocyte depletion of transfused blood Mangano observed a strong relationship between peri- on postoperative complications in patients undergoing car- diac surgery: a randomized clinical trial,” Circulation, vol. 97, operative platelet transfusion and increased postoperative no. 6, pp. 562–568, 1998. mortality [45]. The effect of platelets on major cardiovascular [4] W. H. Dzik, J. K. Anderson, E. M. O’Neill, S. F. Assmann, surgery is not yet clear; however, we have decided to include L. A. Kalish, and C. P. Stowell, “A prospective, randomized the studies with the use of platelets due to their common clinical trial of universal WBC reduction,” Transfusion, practice, and we recommend analyzing this variable in future vol. 42, no. 9, pp. 1114–1122, 2002. studies. The small number of trials identified in our review [5] J. P. Wallis, C. E. Chapman, K. E. Orr, S. C. Clark, and raises concern about publication bias. However, we demon- J. R. Forty, “Effect of WBC reduction of transfused RBCs on strated by means of the TSA analysis that no additional RCTs postoperative infection rates in cardiac surgery,” Transfusion, need to be conducted in order to demonstrate the beneficial vol. 42, no. 9, pp. 1127–1134, 2002. effects in terms of preventing infection complications. [6] Y. M. Bilgin, L. M. G. van de Watering, L. Eijsman et al., “Double-blind, randomized controlled trial on the effect of leukocyte-depleted erythrocyte transfusions in cardiac valve 5. Conclusions surgery,” Circulation, vol. 109, no. 22, pp. 2755–2760, 2004. [7] J. A. van Hilten, L. M. G. van de Watering, J. H. van Bockel There is clear evidence for supporting the routine use of et al., “Effects of transfusion with red cells filtered to remove leukoreduction in patients undergoing a major cardiovas- leucocytes: randomised controlled trial in patients undergoing cular surgical procedure for preventing infection from any major surgery,” BMJ, vol. 328, no. 7451, p. 1281, 2004. cause. Based on TSA analysis, it is not necessary to conduct [8] C. P. Connery, I. K. Toumpoulis, C. E. Anagnostopoulos et al., more RCTs to assess the effects on infection complications “Does leukofiltration reduce pulmonary infections in CABG risk reduction. The quality of the evidence is moderate for patients? A prospective, randomized study with early results this outcome and therefore the certainty as well. As for death and mid-term survival,” Acta Cardiologica, vol. 60, no. 3, from any cause, a beneficial effect of LD-PRBC in patients pp. 285–293, 2005. undergoing a major cardiovascular surgical procedure was [9] L. S. Jensen, A. J. Andersen, P. M. Christiansen et al., “Postoperative infection and natural killer cell function fol- also observed, but more RCTs are needed to confirm our lowing blood transfusion in patients undergoing elective findings. More research could be justified specifically in colorectal surgery,” British Journal of Surgery, vol. 79, no. 6, those middle and low incomes countries in which LD-PRBC pp. 513–516, 1992. has not been implemented universally yet and/or where the [10] J. G. A. Houbiers, A. Brand, L. M. G. Van De Watering et al., costs of the procedure could be a barrier. “Randomised controlled trial comparing transfusion of leucocyte-depleted or buffy-coat-depleted blood in surgery for colorectal cancer,” The Lancet, vol. 344, no. 8922, pp. 573–578, Disclosure 1994. [11] L. S. Jensen, P. Kissmeyer-Nielsen, B. Wolff, and N. Qvist, Daniel Simancas-Racines is a Ph.D. candidate at the De- “Randomised comparison of leucocyte-depleted versus buffy- partment of Pediatrics, Gynecology and Obstetrics and Pre- coat-poor blood transfusion and complications after co- ventive Medicine, Universitat Autònoma de Barcelona, Spain. lorectal surgery,” The Lancet, vol. 348, no. 9031, pp. 841–845, 1996. Conflicts of Interest [12] I. Titlestad, L. S. Ebbesen, A. P. Ainsworth, S. T. Lillevang, N. Qvist, and J. Georgsen, “Leukocyte- depletion of blood The authors declare that there are no conflicts of interest components does not significantly reduce the risk of in- regarding the publication of this paper. fectious complications. Results of a double-blinded, Cardiology Research and Practice 9 randomized study,” International Journal of Colorectal Dis- inflammatory mediators and postoperative complications,” ease, vol. 16, no. 3, pp. 147–153, 2001. Critical Care Medicine, vol. 38, no. 2, pp. 546–552, 2010. [13] J. Skånberg, K. Lundholm, and E. Haglind, “Effects of blood [28] C. Boer, M. I. Meesters, M. Milojevic et al., “2017 EACTS/ transfusion with leucocyte depletion on length of hospital EACTA guidelines on patient blood management for adult stay, respiratory assistance and survival after curative surgery cardiac surgery,” Journal of Cardiothoracic and Vascular for colorectal cancer,” Acta Oncologica, vol. 46, no. 8, Anesthesia, vol. 32, no. 1, pp. 88–120, 2018. pp. 1123–1130, 2007. [29] D. Simancas-Racines, D. Osorio, A. Marti-Carvajal, and [14] P. I. Tartter, K. Mohandas, P. Azar, J. Endres, J. Kaplan, and I. Arevalo-Rodriguez, “Leukodepleted for the prevention of M. Spivack, “Randomized trial comparing packed red cell adverse reactions from allogeneic blood transfusion,” blood transfusion with and without leukocyte depletion for Cochrane Database of Systematic Reviews, vol. 12, no. 3, 2015. gastrointestinal surgery,” American Journal of Surgery, [30] J. P. T. Higgins and S. Green, Cochrane Handbook for Sys- vol. 176, no. 5, pp. 462–466, 1998. tematic Reviews of Interventions, The Cochrane Collaboration, [15] F. Sanfilippo, R. Bollinger, J. MacQueen, B. Brooks, and London, UK, 2011. J. Koepke, “A randomized study comparing leukocyte- [31] J. P. T. Higgins, S. G. Thompson, J. J. Deeks, and D. G. Altman, depleted versus packed red cell transfusions in prospective “Measuring inconsistency in meta-analyses,” BMJ, vol. 327, cadaver renal allograft recipients,” Transfusion, vol. 25, no. 2, no. 7414, pp. 557–560, 2003. pp. 116–119, 1985. [32] The Cochrane Collaboration and The Nordic Cochrane [16] G. Opelz, Y. Vanrenterghem, G. Kirste et al., “Prospective Centre, Review Manager (RevMan): Version 5.3, The evaluation of pretransplant blood transfusions in cadaver Cochrane Collaboration, The Nordic Cochrane Centre, kidney recipients,” Transplantation, vol. 63, no. 7, pp. 964– Copenhagen, Denmark, 2014. 967, 1997. [33] J. Wetterslev, K. Thorlund, J. Brok, and C. Gluud, “Trial [17] C. Hiesse, M. Busson, C. Buisson et al., “Multicenter trial of sequential analysis may establish when firm evidence is one HLA-DR-matched or mismatched blood transfusion reached in cumulative meta-analysis,” Journal of Clinical prior to cadaveric renal transplantation,” Kidney In- Epidemiology, vol. 61, no. 1, pp. 64–75, 2008. ternational, vol. 60, no. 1, pp. 341–349, 2001. [34] J. Wetterslev, K. Thorlund, J. Brok, and C. Gluud, “Estimating [18] E. Bennett-Guerrero, Y. Zhao, S. M. O’Brien et al., “Variation required information size by quantifying diversity in random- in use of blood transfusion in coronary artery bypass graft effects model meta-analyses,” BMC Medical Research Meth- surgery,” JAMA, vol. 304, no. 14, pp. 1568–1575, 2010. odology, vol. 9, p. 86, 2009. [19] V. A. Ferraris, S. P. Ferraris, S. P. Saha et al., “Perioperative [35] J. Chandler, R. Churchill, J. Higgins, T. Lasserson, and blood transfusion and blood conservation in cardiac sur- D. Tovey, “Methodological standards for the conduct of new gery: the Society of Thoracic Surgeons and the Society of Cochrane Intervention Reviews,” December 2013, http:// Cardiovascular Anesthesiologists clinical practice guide- editorial-unit.cochrane.org/sites/editorial-unit.cochrane.org/files/ line,” Annals of Thoracic Surgery, vol. 83, no. 5, pp. S27– uploads/MECIR_conduct_standards%202.3%2002122013_0. S86, 2007. pdf2014. [20] M. L. Williams, X. He, J. S. Rankin, M. S. Slaughter, and [36] J. Brok, K. Thorlund, C. Gluud, and J. Wetterslev, “Trial J. S. Gammie, “Preoperative hematocrit is a powerful pre- sequential analysis reveals insufficient information size and dictor of adverse outcomes in coronary artery bypass graft potentially false positive results in many meta-analyses,” surgery: a report from the Society of Thoracic Surgeons Adult Journal of Clinical Epidemiology, vol. 61, no. 8, pp. 763–769, Cardiac Surgery Database,” Annals of Thoracic Surgery, 2008. vol. 96, no. 5, pp. 1628–1634, 2013. [37] J. Brok, K. Thorlund, J. Wetterslev, and C. Gluud, “Apparently [21] M. D. Cabana, C. S. Rand, N. R. Powe et al., “Why don’t conclusive meta-analyses may be inconclusive-trial sequential physicians fol-low clinical practice guidelines? A framework analysis adjustment of random error risk due to repetitive for improvement,” JAMA, vol. 282, no. 15, pp. 1458–1465, testing of accumulating data in apparently conclusive neo- 1999. natal meta-analyses,” International Journal of Epidemiology, [22] L. T. Goodnough and A. Shander, “Patient blood manage- vol. 38, no. 1, pp. 287–298, 2009. ment,” Anesthesiology, vol. 116, no. 6, pp. 1367–1376, 2012. [38] L. K. Boshkov, G. Chien, D. VanWinkle et al., “Prestorage [23] H. R. Benson, “An introduction to benchmarking in Leukodepleted of transfused packed red cells is associated healthcare,” Radiology Management, vol. 16, no. 4, pp. 35–39, with significant ongoing 1–12 month survival benefit cardiac 1994. surgery patients,” Blood, vol. 108, p. 578, 2006. [24] H. Gombotz, P. H. Rehak, A. Shander, and A. Hofmann, [39] A. W. Bracey, R. Radovancevic, N. A. Nussmeier et al., “Blood use in elective surgery: the Austrian benchmark “Leukocyte-reduced blood in open heart surgery patients: study,” Transfusion, vol. 47, no. 8, pp. 1468–1480, 2007. effects on outcome,” Transfusion, vol. 42, p. 5S, 2002. [25] A. B. A. Vonk, M. I. Meesters, W. B. van Dijk et al., “Ten-year [40] G. A. Ortolano, G. S. Aldea, K. Lilly et al., “A review of patterns in blood product utilization during cardiothoracic leukofiltration in cardiac surgery: the time course of reper- surgery with cardiopulmonary bypass in a tertiary hospital,” fusion injury may facilitate study design of anti-inflammatory Transfusion, vol. 54, no. 10, pp. 2608–2616, 2013. effects,” Perfusion, vol. 17, no. 2, pp. 53–62, 2002. [26] R Covin, M O’Brien, G Grunwald et al., “Factors affecting [41] N. Blumberg, H. Zhao, H. Wang, S. Messing, J. M. Heal, and transfusion of fresh frozen plasma, platelets, and red blood G. H. Lyman, “The intention-to-treat principle in clinical cells during elective coronary artery bypass graft surgery,” trials and meta-analyses of leukoreduced blood transfusions Archives of Pathology & Laboratory Medicine, vol. 127, in surgical patients,” Transfusion, vol. 47, no. 4, pp. 573–581, pp. 415–423, 2003. 2007. [27] Y. M. Bilgin, L. M. G. van de Watering, M. I. M. Versteegh, [42] D. Fergusson, M. P. Khanna, A. Tinmouth, and M. H. J. van Oers, and A. Brand, “Effects of allogeneic leu- P. C. HébertHébert, “Transfusion of leukoreduced red blood kocytes in blood transfusions during cardiac surgery on cells may decrease postoperative infections: two meta-analyses 10 Cardiology Research and Practice of randomized controlled trials,” Canadian Journal of Anesthesia/Journal Canadien d’anesthésie, vol. 51, no. 5, pp. 417–424, 2004. [43] A. E. Tsantes, E. Kyriakou, and G. K. Nikolopoulos, “Cost- effectiveness of leucoreduction for prevention of febrile non- haemolytic transfusion reactions,” Blood Transfusion, vol. 12, no. 2, pp. 232–237, 2014. [44] M. Kremke, M. K. Hansen, S. Christensen, M. Tang, J. J. Andreasen, and C.-J. Jakobsen, “The association between platelet transfusion and adverse outcomes after coronary artery bypass surgery,” European Journal of Cardio-Thoracic Surgery, vol. 48, no. 5, pp. e102–e109, 2015. [45] D. T. Mangano, “Multicenter study of perioperative ischemia research group: aspirin and mortality from coronary artery bypass surgery,” New England Journal of Medicine, vol. 347, no. 17, pp. 1309–1317, 2002.