Transfusion of red blood cells ≥35 days old: A narrative review of clinical outcomes

Expiration of red blood cells (RBCs) is based on laboratory assessments, historical precedent, good faith estimates from experts, and the logistical nuances of a country's blood bank system; it has never been tied to clinical outcomes. Measurements of pre-transfusion hemolysis and 24-h post-transfusion 51Cr-radiolabeled RBC recovery are often used as surrogates for efficacy as RBCs must remain intact and circulate to work.1 While efficacy benchmarks are accepted based on the observed survival of acutely transfused patients, including those massively transfused during trauma resuscitation and surgery, safety is less certain.2 Numerous studies have assessed clinical outcomes to determine the safety of long-stored RBCs. Among the largest studies, four international trials found no association between RBC age and increased mortality or multi-organ dysfunction score; however, “old” RBCs had a mean age of 22.0–28.3 days. Some patients received units near expiration, but none were randomized to a “≥35 days old” category.3-6 A study randomizing patients to receive RBCs in the final week of shelf life has never been approved and conducted. Alternative methods have estimated safety and effectiveness near expiration. Autotransfusion of 42-day-old RBCs suggests some quality markers are worsened (i.e., extravascular hemolysis, non-transferrin-bound iron,and in vivo recovery) while others are unaffected (i.e., intravascular hemolysis, oxygenation, microcirculation, and exercise duration).7-9 Most markers are not tied to FDA regulations and clinical relevance is not well established: autotransfusion of healthy volunteers is an imperfect surrogate for allogenic transfusion of ill patients. To conserve this limited resource and deliver a high standard of care, identification of clinical outcomes associated with RBCs ≥35 days old will help establish safety near expiration. In this narrative review, we summarize current information on patient outcomes after transfusion of units stored beyond 35 days, pathophysiological mechanisms of possible negative effects, and avenues for improvement. PubMed, Embase, and Cochrane Library were searched (BCR) for studies assessing clinical outcomes of RBC transfusion after ≥35 storage days. The results were filtered to exclude articles published before 1983, the year the first additive solution (AS-1) was licensed by the FDA.10 The search string was optimized to include all combinations and variations of “red blood cell,” “storage,” “duration,” and “≥35 days” or “near expiration.” Studies were excluded if they did not have a “≥35 days” treatment category, did not assess RBCs, or did not include a quantified measurement of clinical outcomes. Reviews and simulations were excluded. Due to a limited number of relevant results, studies were not excluded based on patient characteristics, sample size, or statistical protocol. The search identified 219 unique articles (Figure 1) of which 60 had relevant titles. Abstract and full-text review removed 53 articles based on exclusion criteria. Two additional studies were located with a manual search. Nine studies were selected for review: four secondary analyses of international randomized controlled trials (RCTs)3, 11-13 and five retrospective analyses of hospital transfusion databases (Table 1).14-18 TRIBE 2010–2015 Burn ≥18 years Proportion ≥35 days (303) INFORM 2012–2015 Inpatient ≥18 years At least 1 unit ≥35 days (4480) TRANSFUSE 2012–2016 ICU ≥18 years Exclusively >35 days (143) TRIPICU 2001–2005 PICU 3 days—14 years Single-center 2010–2014 Surgical—radical cystectomy ≥18 years Single-center 2002–2006 Cardiovascular acute care ≥18 years Single-center 2009–2015 Inpatient ≥5 years Multi-center 2008–2014 Inpatient ≥18 years Single-center 2009–2014 Surgical—GI resection ≥18 years In the current literature, the only assessments of RBC safety near expiration come from secondary or retrospective analysis of patients who happened to receive the oldest RBCs under the first-in-first-out inventory management system. The secondary analysis of RCT data is entirely multi-center and collectively represents 10 countries. In contrast, retrospective database studies primarily use single-center data and only represent two countries (Table 1). In this review, we delineate secondary versus retrospective as there is a notable difference in their findings. Eight studies evaluated all-cause mortality, five as in-hospital (two secondary analyses and three retrospective) and three as 90-day (one and two, respectively; Table 2). Six found no association between RBCs ≥35 days old and increased mortality.3, 11, 12, 14, 17, 18 Eikelboom and colleagues retrospectively reviewed cardiovascular acute care patients at a single center in Canada and found a modest mortality association following transfusion with at least one unit of RBCs stored ≥35 days compared with 7–14 days. They noted a “continuous and graded” increase in mortality with increasing RBC age, not a large jump in the final week.15 This suggests any negative impacts from accumulating storage lesions are incremental as the expiration date does not coincide with a sharp inflection on the mortality curve. MODS z = 1.24 (p = .213) Mortality = all-cause, in-hospital MODS = multi-organ dysfunction score Duration of ventilation was discounted due to high clinical variability (r2 = .007) Wound healing time z = .03 (p = .976) 90-day Readmit 21.4% (vs. 26.0%) (p = .80) Mortality = all-cause, 90-day Morbidity = composite Mortality = all-cause, in-hospital Morbidity = composite Increased mortality found in the ICU subset only. Mortality = all-cause, in-hospital Top cell: whole blood filtered RBCs Bottom cell: red cell filtered RBCs Mortality = all-cause, 90-day Morbidity = perioperative Goel and colleagues retrospectively assessed inpatients ≥5 years old at a single center in the United States. Using RBC age definitions of “exclusively ≥35 days” versus “exclusively ≤21 days,” the study found increased mortality in the ICU patient subset but not in the overall patient population.16 This suggests only the most critical patients may be susceptible to the negative effects of long-stored RBCs. An association between mortality and patient age, number of units transfused, and Medicare Severity-Diagnosis Related Group was also noted, recognizing any negative outcomes are multi-factorial and not strictly due to RBC age.16 The range of studied morbidity outcomes was quite broad (Table 2). Variables included composite morbidity, infection, multi-organ dysfunction score, wound healing time, duration of mechanical ventilation, length of stay, and readmission. Composite morbidity was defined as infectious, renal, thrombotic, respiratory, or ischemic events.14, 16, 18 Measured independently, infection included Clostridium difficile, septicemia, sepsis, surgical-site infection, pyelonephritis, drug-resistant infection, upper/lower respiratory tract infection, pneumonia, catheter-associated urinary tract infection, meningitis, ventriculoperitoneal shunt infection, and mediastinitis.13, 14, 18 All five studies measuring morbidity found increases; three were retrospective studies,14, 16, 18 and two were secondary studies.11, 13 Upon secondary analysis of RCT data, Flatman and colleagues found an increased infection rate in pediatric ICU patients, further supporting the suggestion that critically ill patients may be more susceptible to negative effects from transfusion of the oldest RBCs.13 Cartotto and colleagues found an increased duration of mechanical ventilation in adult burn patients; however, the authors discounted this finding due to high clinical variability.11 Meanwhile, the retrospective studies found increased morbidity and infection in radical cystectomy patients,14 increased morbidity and length of stay in general inpatients ≥5 years old,16 and increased morbidity, infection, and length of stay in GI resection patients.18 As expected given the overlapping factors, increased composite morbidity coincided with increased infection and length of stay.14, 16, 18 Infection was the outcome most commonly associated with transfusion of RBCs ≥35 days old. Three studies found increased infection, but neither retrospective study found a concurrent increase in mortality.14, 18 The secondary study did not measure mortality.13 Meanwhile, the retrospective study that found increased mortality in the ICU subset did not stratify the multi-variate analysis by sub-category of composite morbidity, but infection was noted to be the most common event (14.4%, p = .007).16 Infection has a broad spectrum of severity and may often be relatively mild. The findings suggest most infectious complications can be adequately managed to avoid mortality in most patients. Storage-lesion-induced infection appears to be the most plausible explanation for negative effects following transfusion of older RBCs.19 A popular hypothesis is that hemolysis releases free iron which fuels infectious bacteria. Humans have evolved systems, including transferrin, ferritin, hemopexin, and haptoglobin to rapidly sequester free-iron, heme, and free-hemoglobin, thereby quarantining it for host metabolic processes and denying it to invading pathogens.20 In evaluating the impact of RBC transfusion on non-transferrin-bound iron (NTBI) levels, one autotransfusion study found no change in NTBI with 35-day-old RBCs transfused to volunteers with E. coli lipopolysaccharide-induced endotoxemia.21 A second study found increased NTBI after autotransfusion of healthy volunteers with RBCs stored 42 days but not 35 days.8 This suggests that beyond 35 days of storage, RBC hemolytic susceptibility may release free-iron beyond the body's sequestration capacity. That said, donor variation in RBC hemolysis at the end of storage is high.22 A second mechanism of possible clinical impact is the accelerated removal of RBCs from circulation. Surface senescence markers, including Band 3-IgG complexes and phosphatidylserine, tag RBCs for erythrophagocytosis while structural changes such as sphero-echinocytosis lead to mechanical removal in the spleen.23, 24 However, a simple loss of circulating RBCs may not be the whole story. A recent study of senescent RBC clearance in a mouse model discovered that erythrophagocytosis downregulates STAT1, a transcription factor for antiviral and antibacterial host response. This immunosuppressive effect was shown to be caused by the porphyrin moiety of heme, independent of free-iron.25 Transfusion of storage-degraded RBCs may therefore contribute to infection in two ways: by directly fueling pathogens with free-iron and by indirectly suppressing the host immune response following erythrophagocytosis of senescent RBCs. The current literature is exceptionally heterogenous. Among the reviewed studies, there were five definitions of “oldest” RBC age and none of the studies used the same definition of “fresh” RBC age (Table 1). Three studies assigned patients to the “old” category if they received at least one unit ≥35 days old. Depending on number of units transfused, this could underestimate the effect of RBC age as the “old” category may include patients who received predominantly “fresh” RBCs. The reviewed studies drew data from countries using AS-1/3/5 or SAGM additive solution with expiration dates of 35 or 42 days but did not stratify by storage solution. Furthermore, three out of four secondary analyses included a mix of countries with 35- and 42-day expirations.3, 11, 13 Countries with a 35-day expiration were therefore only represented in the control group while countries with a 42-day expiration contributed to both the control and treatment groups. None of the reviewed studies controlled for donor characteristics, which is gaining acceptance as a factor in storage quality.22 Analysis of donor age, sex, body mass index (BMI), hemoglobin, blood group, and time since previous donation found BMI >27 to be the only donor characteristic associated with increased hemolysis on day 42.26 However, this finding may not lead to worsened clinical outcomes. A study of 93,726 transfusion recipients found no association between donor age, BMI, hemoglobin, or smoking status and in-hospital mortality or length of stay.27 Future work should control for donor characteristics to draw more definitive conclusions about the impact of RBC age. Observational studies are beholden to the available sample size and may be underpowered resulting in decreased probability of identifying a true effect, increased probability of a false-positive result, or overestimation of the effect size.28 Indeed, the only secondary analysis to identify a worsened outcome had the smallest sample size and widest confidence interval of all reviewed studies.13 In light of the observed trend of increased infection among other reviewed studies, this finding may be accurate, but the effect size could be overestimated. Likewise, the retrospective study with the smallest sample size noted it was underpowered to detect several intended outcomes and did not find a treatment effect in said outcomes.14 Due to the relative rarity of transfusing blood ≥35 days old, studies with small sample sizes add value to the literature but should be considered accordingly. Observational studies also risk residual confounding.29 There is often wide variation in controlled covariates as demonstrated by the heterogeneity of confounder adjustment strategies in the reviewed studies (Table 3). Notably, the two retrospective studies that found increased infection but not mortality assessed abdominal surgery patients14, 18 who are known to have a higher rate of surgical site infection.30 Observational studies should be considered in the context of the controlled covariates. Age, sex, multiple trauma, cardiac surgery, other surgery, use of immunomodulatory drugs, Pediatric Risk of Mortality score, TRIPICU study group. [RBC volume was controlled by each patient only receiving 1 unit] The most notable trend was a correlation between data source and identification of worsened outcomes. 9 of 11 worsened outcomes came from four single-center studies,14-16, 18 three of which were from the same institution.14, 16, 18 In contrast, the multi-center studies collectively measured eight mortality and morbidity outcomes, yet found only a single worsened morbidity outcome.3, 11-13, 17 An unrelated systematic review of 12 meta-analyses found single-center studies overestimate treatment effects compared with multi-center studies. The authors concluded, “results from single-center trials should be cautiously used for decision making.”31 A common reason for not randomizing patients to receive RBCs ≥35 days is the challenge of inventory management. Under a first-in-first-out inventory system, relatively few units remain beyond 35 days.32 Randomizing patients to receive long-stored RBCs would require deliberately aging a portion of the inventory; it would be logistically burdensome to ensure units of the requisite type and age are available on demand. Units intentionally held to the end of shelf life are also more likely to out-date before the presentation of a matching recipient. A controlled trial that successfully navigates these logistical complexities would add a novel perspective to the literature. Another oft-cited reason is the hesitancy of ethics committees to approve such protocols due to perceived inequity of benefit compared with fresher RBCs. However, the original protocol for the RECESS trial, which passed multiple Institutional Review Boards, slated a treatment category to receive the oldest blood in the inventory. The protocol was restricted to RBCs ≥21 days following concerns from the NIH Office of Research Protections which demonstrates the lack of ethical consensus (JRH). Stored RBCs are licensed until day 42 and transfusion until expiration is within the standard-of-care. Randomizing patients to a ≥35 day treatment category should be possible and would help settle the question of safety at the end of shelf life. In the 1970's, recognition that three-week storage resulted in wastage of almost a third of collected units led to the development and licensure of additive solutions allowing storage up to 42 days. Over time, these additive solutions have been associated with greater than 99% utilization of collected blood.33 Globally, the most popular additive solutions are AS-3, SAGM, and its AS-1 and AS-5 variants; MAP is exclusive to Japan. Despite an approved 42-day expiration, the Netherlands, UK, Germany, and New Zealand limit SAGM to 35 days because they perceive an improved quality of fresher products and their inventory systems can handle the restriction without excessive outdating.34 Japan restricts MAP to 21 days because irradiation, used to reduce the risk of transfusion-associated graft-versus-host disease in their genetically homogenous population, is detrimental to storage quality.34 Restricting shelf life may not be feasible for countries with more dispersed populations, less developed national blood systems, or varying needs for large-scale/high throughput product availability. While most countries maintain the status quo of 35–42 day storage in legacy additive solutions, improved options are available. Phosphate-adenine-glucose-guanosine-saline-mannitol (PAGGS-M) improves storage quality through the maintenance of intracellular phosphate stores and GTP concentrations.35 Although PAGGS-M would not meet FDA regulations,36 it was licensed in Germany for 49 days to support the 1993 United Nations peacekeeping mission in Somalia and remains in active use today.35 Likewise, AS-7 was developed with the support of the US Army to improve longer-duration RBC storage in austere locations. The only new additive solution licensed by the FDA in over 30 years, AS-7 was shown to outperform AS-1 in measures of hemolysis and 24-h recovery. On day 42, AS-7 RBCs demonstrated 0.29% hemolysis and 88% mean 24-h recovery.37 PAGGS-M and AS-7 performance suggests current storage limits are not absolute. Narrative reviews are inherently biased. This review lacked a systematic protocol to compare outcomes and assess the validity of each study; conclusions should be considered accordingly. Furthermore, the literature is scant so exclusion criteria were minimized to capture as many studies as possible resulting in the inclusion of exceptionally heterogenous studies. The reviewed studies were almost unanimous in finding no increased mortality following transfusion of RBCs ≥35 days old, except for two retrospective, single-center studies.15, 16 All five studies assessing morbidity outcomes found increases, primarily infection, but a concurrent increase in mortality was only seen in the ICU subset of a single study.16 This suggests any complications associated with older RBCs may be survivable for most patients. The authors of this review have served in settings where resources were limited by geographic, social, economic, and/or political forces, including active war zones. Under these circumstances, maximizing shelf life is crucial. The literature suggests that currently available RBCs are generally safe for up to 42 days of storage, and there is no compelling evidence for reducing storage duration. Brian C. Riley, Lynn G. Stansbury, and Rida A. Hasan have no financial conflict of interest with the contents of this study. John R. Hess is the inventor of AS-7 and a paid consultant to Hemerus, LLC, the company licensing the AS-7 patents.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.