As they sit in the fridge marinading in their own filthy excretions, human erythrocytes mutate into something grotesque, no longer resembling human blood and wholly incapable of carrying oxygen. It is unbelievable that anybody should want to infuse such material into their patients, and concerns have been raised about the safety of using particularly old cells as their near their expiry. In Australia, packed red blood cells can only be stored for a maximum of 42 days (which is slightly conservative - internationally, the maximum PRBC shelf life is 49 days). Surprisingly, international clinical trial data supports the safety of existing practice. CICM SAQs mention storage lesions often, but only Question 29.1 from the first paper of 2010 has actually asked anything directly about them.
During the golden age of physiology, when scientists wore hats and white lab coats, a series of articles was published which addressed the problems of packed red blood cell storage in exquisite detail.
Since 1956, some things have changed, but these papers remain classical. For the modern reader unburdened by reverence for the hoary fathers of medicine, there is an excellent article on this from 2010, which is unfortunately only available to the paying customers of ScienceDirect. A less excellent article is also available, and is free. Within it, there are nice clear graphs demonstrating the physiological changes which the RBCs undergo during their long refridgerated imprisonment. I have combined them into one uber-graph which turns this already complex subject into a confusing tangle of multicoloured lines.
The ageing of red blood cells is a complicated tyopic. It is possible to summarise it by dividing the process by affected variables, rather than by arbitrary time stages. Time stages would also work - some lesions occur earlier, and others later. As the above graph suggests, these changes begin their evolution immediately as the blood is sucked out by the phlebotomist; however clinically relevant storage lesions only appear after a few days of refrigeration.
The normal value for intracellular 2,3-DPG for an erythrocyte is arounf 4.5 mmol/L. After 24 hours of refrigeration, this value drops to about 3.0-3.5 mmol/L, and the entire 2,3-DPG supply is depleted after two weeks. This has dire consequences for the oxygen-carrying capacity of these cells. 2,3-DPG is an allosteric modulator of oxy-haemoglobin binding; its absence exhibits itself as a greatly increased oxygen affinity for any given partial pressure of O2 (a massive left shift of the dissociation curve).
The disadvantage of this is the utter functional uselessness of such transfused blood. Certainly, it might appear to carry oxygen, but it will not let go of it once it is in the tissues. This persists in vivo, well after the transfusion has completed - even after 8 hours 2,3-DPG levels are only 33% of their normal value.
Does this have a clinically relevant effect? perhaps not. Only one (rodent-themed) study has demonstrated a decrease in oxygenation when one uses older cells. The decrease in totally 2,3-DPG depleted cells was in the order of 15%. This finding could not be replicated, confirming that the oxygen carrying capacity of normoxic organisms is probably not affected. This makes a certain amount of sense. The top of the oxyhemoglobin concentration curve really does not move- only the middle and bottom. Given that there is an increased capacity for oxygen binding, one would expect that overall oxygen carrying capacity is not impaired. Rather, oxygen extraction is impaired. Thus, if one's patient had a circulating volume composed entirely of recently thawed PRBCs, their whole body oxygen extraction would be less than 25%, and their venous saturation would be high.
The levels of ATP within RBCs tend to drop slowly, and diminish to insanely low levels by the end of 5 weeks. Understandably, this has implications for pretty much every energy-dependent process which the cell may want to conduct. The absence of ATP promotes the dysfunction of ion pumps and other membrane proteins, which in turn results in all sorts of impaired cellular mechanisms, most notably the maintenance of RBC shape (which is a surprisingly active process).
As membrane proteins are damaged and lost, and as ion pumps begin to fail, so the starving erythrocytes begin to swell and to take on a more spherical or hemispherical appearance. Then, they shapeshift into monstrous spheroechinocytes, with numerous non-deformable protrusions of cellular membrane sticking out in all directions. . Several studies of red cell morphology following cold storage have concluded that this failure of normal shape decreases the deformability of erythrocytes, and thus decreases flow through the microvasculature.
The rheological handicap of stored spheroechonocytes may slow the capillary flow enough to generate a clinically significant deficit of systemic oxygen delivery, which might be felt most in those tissues which are hungriest. Brain and kidney come to mind. A study of microvascular perfusion which compared stored and fresh cells demonstrated that the stored erythrocytes reduced microvascular flow and functional capillary density by 63% and 54% respectively. The study concluded that the transfusion of stored cells resulted in "significantly malperfused and underoxygenated microvasculature".
This phenomenon is the consequence of transmembrane protein failure, and leads to an increased inflammatory response to the red cells. Such cells are also more osmotically fragile, and (with the loss of membrane elasticity) are even less deformable. They will not find it easy to slip through the microcirculation. Furthermore, such abnormal-looking cells will not survive the passage through the reticuloendothelial system. Haemolysis may ensue.
The destruction of RBCs both in the bag of cells and in the patient's bloodstream leads to the dispersion of free haemoglobin thoughout the circulation. This has certain consequences beyond the development of a mild hyperbilirubinaemia. Free hemglobin is actually an excellent nitric oxide scavenger. This reaction is treated briefly in the section on metabolism and clearance, inside the drug monograph on inhaled NO. To summarise, excess free haemoglobin results in the loss of nitric oxide, and thus in a reversal of normal mechanisms of NO-associated hypoxic vasodilation in poorly perfused organs. The result is vasoconstriction; essentially the opposite of what happens in vasoplegic shock. In the absence of NO-mediated bloodflow autoregulation, hypoxic tissues may be unable to attract a diversion of bloodflow, and shock will persist.
Though not strictly speaking a change of the red blood cells themselves, the waste products of their metabolism which accumulate in the bag are noxious enough to cause serious problems. For instance, one can readily identify a source of trouble in the massive amount of potassium and lactate which is sloshing wound in the extremely acidic environment of the 42-day-old blood bag. One might be tempted to conclude that hyperkalemia and lactic acidosis may result from a transfusion of too many units of old blood, but it does not seem to happen very often. A Mayo Clinic retrospective case series, spanning a term between 1988 and 2006, had unearthed only 16 cases of hyperkalemic cardiac arrest associated with PRBC transfusion. Compared to the total numbers of patients who had undergone blood trasnfusion at the Mayo Clinic over that time, this complication must be rare indeed - comparable to the risk of contracting HIV from the blood.
The process of slowly being pickled in your own wastes leads one to the intuitive conclusion that some of the changes will fill the environment with proinflammatory products of decay. The notion that packed red cell transfusion triggers an inflammatory response has been confirmed within the limitations of a murine model. It appears the inflammatory response requires membrane-encapsulated RBC haem, rather then the free haemoglobin which inevtiably forms. Furthermore, it seems that the transfusion of RBCs induces an acute phase response, and exacerbates inflammation induced by endotoxin. The free iron which is released after hameoglobin lysis ends up in the circulation, and this has the effect of sponsoring bacterial growth.
Of course, sceptics might point out that mice and mouse blood are not effective experimental surrogates for human blood transfusion research. This can be countered by the observations of investigators such as Guillermo A. Escobar's group, who in 2007 confirmed that stored PRBC transfusions seem to upregulate proinflammatory gene expression in the leukocytes of the transfusion recipient.The in-vivo inflammatory effects of transfusion was confirmed by McFaul et al, who in 2009 found that human packed red cell suspensions stored in standard solutions become increasingly proinflammatory as a function of storage time. The culprits seem to be lipid fragments which have been shed from disintegrating membranes of the hideously deformed stored red cells- we know this because the acellular products do not seem to have this effect. The immunoactivating effects of these lipid fragments have been recognised for some time - here is a 1997 study linking these "neutrophil-priming agents" to the pathogenesis of transfusion-associated acute lung injury (TRALI).
The effect of storage lesions on the composition of PRBCs is illustrated in this ABG, which was collected from a bag of refrigerated packed red blood cells.
The specific results are explored in greater detail elsewhere. In summary, there is extreme acidosis, hyperglycaemia, hyperkalemia, and hyponatremia. There is a massive right shift of the p50 which one might attribute to the acidosis; but when corrected for pH and temeperature the p50(st) is significantly left-shifted (17.39), suggesting that there is 2,3-DPG depletion. The sample is at 4º C and so alpha-stat interpretation probably also changes the numbers somewhat (i.e. the CO2 and pH will look different when warmed to body temperature, and the potassium will shift into the cells).
There has for some time existed a common blodbanking policy of using the oldest red blood cells on the shelf, in view of the fact that they were about to expire. In this fashion, stock was preserved.
Of course, this does not mean that all the PRBC units you ever get are at the borders of expiry. The mean age of transfused cells worldwide is about 16-21 days. A good article published by several luminaries of Australian ICU medicine discusses this practice and its consequences. The authors lament that "most clinical studies in this area have been observational in nature, retrospective in design, small in size, and subject to bias, leaving this issue unresolved for more than 20 years."
Critically ill patients may be more susceptible to these storage lesions than the average punter. For instance, lets take TRALI. There is a good reason to believe on a theroretical basis- that the neutrophil-activating properties of the abovementioned broken lipid fragments can lead to a "second hit" in an already injured lung, rendering it even leakier, and filling it with oedema fluid to the ongoing frustration of the intensivist. This is merely one argument - consider that ICU patients have many more transfusions than other patients, that ICU patients have higher circulating proinflammatory cytokine levels already, etc etc. Lastly, the microvascular underperfusion which results from the use of stored cells may be a major disadvantage to ICU patients who already have a uselessly defective microcirculation.
So what is the response to these concerns? The Australian Red Cross has a Service Policy Document describing the way they handle the issue of age in blood products. In it, they acknowledge the need to balance fresh blood availability with the management of their stock (if you give everybody only the freshest cells, the average stock age will increase, as the oldest cells get left on the shelves). Ultimately, the outcome will probably be some sort of balance where intelligent management of blood product requests will direct the young cells to those people most likely to be harmed by storage lesions.
ARIPI (Fergusson et al, 2012): Double-blind, randomized controlled trial in 377 premature low birth weight neonates. Primary outcome was a composite measure of major neonatal morbidities (eg. death, necrotising enterocolitis, intraventricular haemorrhage, and so forth). The infants were given a mean of 5 transfusions, each of 14 mL. Unfortunately, the blood was very fresh in both groups: the trial ended up comparing PRBCs of 5 days storage with PRBCs of 14.6 days storage (plus-minus 8.3 days). No difference in outcome was seen between the groups, which at this time was somewhat different to previously published data. The authors concluded that their trial focused on clinically important outcomes, whereas previous studies focused on physiological parameters and laboratory markers.
RECESS (Steiner et al, 2015): Double-blind, randomized controlled multi-centre American trial with 1098 cardiac surgical patients. Cells less than 10 days of age were compared to cells more than 21 days of age. The roulette wheel of transfusion resulted in the "old" group receiving cells which were on average 28 days old (+/- 6.7 days). The mean volume transfuded was around 3 units. All-cause mortality at 28 days was very similar between groups. The only difference which reached statistical significance was the hepatic MODS score (0.2 point difference), which was accounted for by the raised bilirubin in the "old" cell group. Because nobody would freak out over a slightly raised bilirubin, the authors commented that "a between-group difference of 1 point or less in the change in MODS is unlikely to be clinically significant or to warrant a major change in the practice of blood banking".
The ABLE trial (Lacroix et al, 2015): 2430 patients, across 64 centers in Canada and Europe. Packed cells 6 days old ended up being compared to cells after 22 days of storage (similar to the average age of transfused cells worldwide). Neither was there a mortality difference at 90 days, nor did the authors find any difference in their secondary outcomes (stay in ICU, duration of ventilation, requirement for dialysis, etc). However, all the PRBC units used in this trials were leucodepleted. In addition, the ABLE investigators transfused their patients at a lower threshold (roughly a Hb of 77), which means that their patients got fewer transfusions overall.
The limits of storage were not approached. As mentioned above, international standards for blood cell refrigeration permit cells as old as 49 days, whereas the ABLE cells were 22 days old on average, with a range of +/-8.4 days (which means some of the "old" cells were as old as 30.4 days, and other were as young as 15.6 days). It is impossible to say what would have happened if the authors intentionally asked for cells from the back of the shelf, nearing their 49-day expiry. However, the ARIPI investigators in 2012 mentioned that their decision to use standard issue cells was purely ethical in nature.The effects of using large volumes of elderly leucocyte-filled blood remain to be seen, but are predictably bad. If one tried hard enough, one would eventually find low-quality blood so old and vile that outcomes from transfusing it would be uniformly poor. This would represent a substantial departure from current practice, and the investigators raised etghical concerns regarding using cells with intentionally extreme storage lesions.
The Australian TRANSFUSE study enrolled 4994 patients, using a pragmatic approach (youngest cells available vs. standard protocol) to determine the survival benefit of using the freshest blood. The use of blood products in Australia is far from liberal (those few intensivists who still use a haemoglobin transfusion trigger will tend to transfuse to maintain a Hb above 70). Moreover, we have uniformly leucodepleted cells. The TRANSFUSE study was therefore never expected to replicate the Canadian experience. No mortality difference was found, nor any difference in the rates of persistent organ dysfunction, febrile nonhemolytic transfusion reactions, mechanical ventilation, renal replacement therapy, or ICU length of stay. Some might say that the difference between groups was insufficient to show the expected improvement as all of the blood used was relatively young (11.8±5.3 days vs. 22.4±7.5 days). However - and this is the most important point - the reason for this age distribution was that nobody ever uses 49-day-old blood, because it doesn't last that long on the shelf, ie. people end up transfusing it before it ages to that point. This was a pragmatic study. When you ring blood bank for an order of the house red, you can be sure that what you are getting is probably no older than 22 days, and that it will cause no major harm to your patient.