- Email: [email protected]
Testing platelet components for bacterial contamination
Transfusion and Apheresis Science 45 (2011) 69–74
Contents lists available at ScienceDirect
Transfusion and Apheresis Science journal homepage: www.elsevier.com/locate/transci
Testing platelet components for bacterial contamination William G. Murphy a,b,⇑, Pauline Coakley b a b
School of Medicine and Medical Science, University College Dublin, Ireland National Blood Centre, Irish Blood Transfusion Service, James’s Street, Dublin 8, Ireland
a r t i c l e Article history:
i n f o
a b s t r a c t Bacteria in transfused platelets can cause serious morbidity and, rarely, death. Most contaminating bacteria enter the blood at the time of venepuncture. While many of these contaminants fail to grow in the platelet unit, storage of platelets at 20–24 °C facilitates growth of some organisms, and the cumulative risk of severe sepsis increases with the storage age of platelet components. Several methods have been developed or adapted to attempt to detect contaminating bacteria with high sensitivity and specificity, but the perfect test has yet to be found. Testing early in the platelet component’s shelf life, even using exquisitely sensitive culture-based tests, is compromised by major problems of sample error – there may be too few bacteria present at this stage to ensure that any practical sample volume contains even one of them. Culture techniques are too slow to be useful as a release test. On the other hand, available rapid tests are too insensitive to use early in the shelf life, and have yet to show convincingly that they are sensitive enough for testing close to the time of transfusion. Nevertheless testing for bacteria in platelet components represents a significant advance in blood transfusion safety, and prevents the transfusion of many hundreds of bacterially-contaminated platelet units each year. Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction The role that microorganisms play in our lives is widely known, though perhaps rarely appreciated. Our symbiosis with microscopic life on an evolutionary, individual and community level is truly extraordinary: think of the origin of mitochondria, or consider what would happen if biological waste and remains were not returned to the earth by microorganisms. So dependent are we on bacteria for life, for example, that colostrum must contain high levels of nitrite to compensate for the lack of salivary bacteria at birth to reduce dietary nitrate for the neonate [1]. The list is endless: the genome, the cell, the system, the organism and the society are all intimately intertwined with a microbiological ecosystem. The immanent beauty of this setup ⇑ Corresponding author at: National Blood Centre, Irish Blood Transfusion Service, James’s Street, Dublin 8, Ireland. E-mail address: [email protected] (W.G. Murphy). 1473-0502/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.transci.2011.06.005
however is seriously undermined by our attempts to live life on an individual level – viruses are not evolved to help us (even genes are not evolved to help us!), nor bacteria, nor the rest. Only we, peculiarly, may be evolved to support humanity, though we seem to be a bit hit and miss at that. Once the ecosystem is disturbed on any level, we are open to a breach of our beautiful symbiosis, and to the emergence of disease in our environment, that was not there before. The process of mutation at many levels in the complex organism of the planet constantly disturbs the equilibrium. The often lethal, seemingly malign, and probably limitless, emergence of bacterial resistance to antibiotics is one manifestation of this. The emergence into our culture of transfusion-adapted microorganisms is another. Once we began the practice of transfusing large volumes of blood from one human to another we changed again our relationship with the microbiological ubiquity. As we generate tactical responses, we open up other avenues either for pre-existing organisms to find a new niche
70
W.G. Murphy, P. Coakley / Transfusion and Apheresis Science 45 (2011) 69–74
to grow, or for new mutations to find a possible zone of survival success. Consider Yersinia enterocolitica, emerging as a lethal complication of blood transfusion in the 1970s and 1980s, seemingly from nowhere [2]. A common cause of food poisoning, and with a transient and often asymptomatic intravascular phase, it had not troubled the blood transfusion community beforehand, even though it did not noticeably change its epidemiology in the late 70s. The organism’s ability to survive the cold, to adhere to blood cells without being destroyed by them, and to resist complement destruction, allowed this passenger, which had probably previously contaminated hundreds, even thousands of blood collections without harm to the recipient, to emerge as a lethal contaminant when some change in blood processing or storage was introduced. This may have been the simple reduction of plasma content of red cell transfusions, a practice that became widespread from the end of the 1970s – almost all reactions were from plasma-reduced red cell concentrates [3]..
2. The natural history and ecology of bacterial contamination of blood components Many units of blood collected from donors are contaminated at the point of venesection. It is impossible to sterilise the deep skin, and at the very least, some bacteria are very likely to be transported from there to the collection bag. A whole-sample analysis for contaminating bacteria during collection would probably be needed to measure the incidence of contamination at this point: immediately upon exposure to the constitutive elements of the collected blood (which will continue to function for several hours at least), bacteria are killed, sequestered, or inhibited. However, holding blood at a temperature warm enough to clear the complement-sensitive and phagocytise the ingestible, may also increase the risk of allowing otherwise harmless spores to vegetate or other bacteria to proliferate, and emerge as pathogens in their turn. There is one more striking feature of the natural history of bacterial contamination of blood collections about which next to nothing is known, and which bears consideration: almost all of our knowledge relates to contaminating events that cause acute clinical reactions – we know little about the longer term effects of transfusion of bacteriallycontaminated blood components into immune-compromised hosts. Perhaps there are none – we are after all adapted to accommodate bacteria on our skin, in our lungs and guts, and in sizeable numbers in our blood from time to time – but it is difficult to imagine that patients who are troubled by their own commensals during periods of profound immune-suppression can universally accommodate 105 colony forming units of coagulase negative staphylococci down their central line with complete biological equanimity. This also applies to low levels of contamination in red cell units, and also in plasma – freezing does not sterilise, though in practice we often assume that it does. Platelets, and the patients who receive them, form their own microbiological niche, where benign, stable, skin commensals in one healthy donor can become a toxic,
sometimes lethal, invader in another. The constituents of this system are the organisms in or on the donor, the defences against invasion of these organisms deployed by the collector, the conditions in which the organisms that survive the collection process are maintained, and the state of the recipient. A purely technical approach to the problem would therefore mandate optimised removal of bacteria from the donor, optimised conditions of processing and storage to destroy or immobilise any surviving organisms, and validation of the effectiveness of these strategies on a global and individual level. In addition, any feasible strategy to boost the defence of the recipient would be identified and engaged. A more systematic approach might also set these in a broader context – there will probably always be bacteria either present or emerging that will escape destruction or detection, and changes to processing, storage or patient protocols may have unforeseeable consequences from trivial to the disastrous. In addition, large differences in transfusion rates of platelets between countries – where the rate in some countries is considerably more than in others with similar health care systems and outcomes, without any obvious reason – translate into differences in the rate of severe reactions per total population per year. The USA for example, will have nearly 50% more septic transfusion episodes per year than the UK, all things else being equal, since 50% more platelets per unit population per year are transfused there. Why are platelets so much more of a problem than red cells from the point of view of bacterial contamination? For red cells compared to platelets on a unit for unit basis, the problem is vanishingly small: one in several million red cell units causes bacterial reactions in patients (or serious, immediate ones at least) compared to one in several tens of thousands for platelets, close to two orders of magnitude of a difference. Yet the initial contamination rate of the red cell unit is the same. The striking difference is the conditions of storage – even brief exposure to cold immobilises almost all the organisms collected with the unit. There may be other, minor agents in generating the difference, but cold exposure, especially before much in the way of endotoxin or exotoxin can be generated, is certainly crucial. Bacterial reactions from red cells are almost always caused by organisms that grow reasonably well at low temperatures, or are associated with breaches of the cold chain. Platelets are denied this 99% effective defence – the comparison shows just how important cold storage probably is. (It is probably important to note that cold storage is an effective post collection strategy. The sterility of collection itself may be of minor importance.) Testing for bacteria in platelets is one way of trying to compensate for the lack of the effective bactericidal or bacteriostatic effects of cold.
3. Reducing bacterial risk Given therefore that bacteria in the collected platelet units are a fact of life, and that the use of cold as a sterilising agent is denied us, we are left with alternative removal techniques, and with attempting to test to sterility. There are effective removal techniques, though whether they
W.G. Murphy, P. Coakley / Transfusion and Apheresis Science 45 (2011) 69–74
provide any overall benefit in the presence of testing remains to be seen. There are also effective improvements that can be made to the collection techniques: no touch, improved skin disinfection [4,5] and initial sample diversion [6] may all reduce bacterial load in the initial collection, though whether these effects are additive is not clear – they do not make much material difference to the safety of red cell collections for example. Overnight warm hold, incubation with the donor leucocytes and perhaps plasma, at a biologically permissive temperature, and leucodepletion (perhaps only after prolonged warm hold) can reduce bacterial load [7,8] and is associated, for whatever reason, with a reduced incidence of platelet-associated bacterial sepsis. Shortening the shelf life also works. Testing works too, but it is not possible to test the platelet supply to sterility as things currently stand. The reasons are (1) that very low numbers of bacteria in a platelet unit near the start of its life may readily escape being captured in the sample [9], but still grow to lethality; and (2) that highly sensitive assays that detect 1–10 bacteria in the test sample are too slow, and efficient assays that give results quickly enough to be used immediately prior to transfusion are too insensitive, to get around this problem.
4. Clinical epidemiology of bacterial sepsis The rate of bacterial contamination of platelets will vary from place to place, as will the incidence of septic reactions from platelet transfusion. The skin cleaning technique, the use of diversion pouches, the length of storage and the method of platelet preparation will all influence the rate of contamination. The method of detection – the timing and volume of the sample and the testing method usedwill also affect the observed incidence. The observed rate of septic reactions is also subject to variables including the quality of the surveillance and reporting system, and the structure of the platelet supply. Accepting the limitations above, the contamination rate of platelet components, as measured with the BacT/ALERT system, is approximately in the range of 3 per 1000 to 1 in 5000, the lower incidence being associated with smaller samples cultured aerobically only, and the higher one with larger samples cultured both aerobically and anaerobically and with testing at the end of shelf life. In one series [6] rates of up to 0.95% were observed, where the sampling was done immediately after collection or manufacture using a 10 ml sample in a one-bottle test on BacT/ALERT. This may represent a true measure of the underlying contamination rate after venesection, before any auto-sterilisation takes effect in the collected or stored blood components. From the reports of large series of transfusions, the rate of septic reactions from bacterially contaminated platelets is in the order of 1:50,000–1:100,000 units of platelets supplied to hospitals. For death from bacterial contamination of platelets the rate is in the 1:250,000–1/500,000 range. The Serious Hazards of Transfusion group in the UK logged 40 episodes of platelet-related sepsis, with 11 deaths, in 10 years, comprising approximately 3000,000 platelet units shipped: i.e. approximately 1:75,000 septic reactions reported for platelets issued, and 1 in 273,000 deaths [10].
71
From the French Haemovigilance data a rate of 17 septic reactions per 1000,000 platelets issued has been reported [11] or 1 in 59,000. In the USA, following the introduction of a bacterial screening programme in the American Red Cross blood centres in 2004, the observed rate of sepsis was 1:59,000 over the first 26 months of the programme, falling slightly to 1/83,000 after the sample volume was increased to 8 mls from 4 mls in the BacT/ALERT test. Three fatal reactions were reported in the first phase of the programme, for a mortality rate of 1:498,711 [12]. Data on benefits from bacterial screening on studies of this scale are not available from the two large haemovigilance systems of the UK and France, neither of whom has introduced universal regular screening of platelet concentrates. 5. Testing strategies Platelets have a 5 day shelf life, or in some places, a 3 day shelf life [11], because of this bacterial problem. They are probably effective as haemostatic agents after 7 or 8 days of storage, and perhaps even longer, but the risk of bacterial disease increases with storage time [13], and limiting the storage time has been a universal response to this issue. A few groups have tried to get around this with testing protocols, but there are theoretical and practical obstacles. We have used a two stage testing protocol that has given us access to 7 day platelets on demand over several years with a theoretically high level of safety [14], and with reasonable logistical and economic demands, but the process requires sophisticated inventory management and a short supply chain to optimise value. To maximise efficiency – either safety of the product or cost of the process – a compromise must be made between time of sampling along the shelf life of the product, and the sensitivity of the test used. Near the start of the shelf life the bacterial numbers are low, and indeed contaminating bacteria may be in a quiescent phase – they may not be growing very much at all. Here, sample error rates will inevitably be high, even if the test used can detect very low numbers of bacteria in the test sample. Paradoxically, techniques that reduce the bacterial load in the initial collection – overnight hold and leucodepletion, for example – but do not sterilise the unit, will make testing even less effective at this point. Unlike the viruses of blood transfusion however, bacteria may increase their infectivity after they have been collected into a therapeutic preparation, and units without sufficient bacterial load to test positive at one point of their shelf life may be massively contaminated several days later – the platelet unit acting as an effective bacterial growth medium. We use a very large sample volume – at 2 7.5 mls around 5% of the total volume of the platelet preparation – sampled after a minimum of 12 h holding of apheresis platelets or after 12 h holding of the manufactured platelet pool. Nevertheless, testing again at outdate of previously sampled platelet units shows we miss around two thirds of contaminating bacteria in the initial sampling protocol [9]. While most of the bacteria that we miss are of dubious clinical significance, we have missed some instances of Bacillus cereus, and on one occasion, one split of a triple apheresis donation contaminated with Proteus mirabilis.
72
W.G. Murphy, P. Coakley / Transfusion and Apheresis Science 45 (2011) 69–74
In contrast, a test done close to the time of transfusion of the platelet unit, within a couple of doubling times of the fastest growing organism possible perhaps, has merely to detect a clinically significant bacterial load, and is not constrained by the niceties of Poisson or binomial distributions. It’s not completely obvious what constitutes a significant, or even more, an insignificant bacterial load. A platelet unit contains two or three hundred millilitres of fluid – a test with a limit of detection of 1000 organisms per ml may miss up to 105 bacteria in the transfusate. This is probably good enough to reduce the incidence of disease considerably, but does leave a large window of opportunity for bacteria to get through. 6. Techniques Culture-based techniques rely on some characteristic of the changes induced in a medium by growing bacteria – gas generation, acid production, to elicit a signal. They require (1) that bacteria grow in the culture conditions of the test and (2) that the culture is continued for long enough to allow the signal to be generated by the growth of bacteria. They can be very sensitive – these tests are capable of detecting as few as one colony forming unit (cfu) in the inoculated sample, and can reliably be expected to detect 10–100 cfu in the sample. The sensitivity of the test then relies predominantly on the size of the sample, and on the chances of getting bacteria into the test volume – i.e. the ratio of the sample size to the total volume of the platelet unit. Therefore the bigger the sample size the greater the chance of detecting them in the test, where the bacteria are few in number. Bacteria in platelets can be very few in number at the start of the shelf life – below 1 cfu per ml of platelet concentrate [9,15]. A huge problem with culture methods is that they do not give a once-for-all-time result. The test can give a positive result after several days of growth. This forces blood centres to issue the platelet unit while the culture is continuing in the laboratory, giving rise to the frequent complication of having to inform clinicians that their patient has received a unit of platelets that was contaminated with bacteria, at least at an earlier point in its shelf life. This is not trivial – it at times gives rise to changes in patient treatment that are of uncertain benefit or risk. This state of affairs arises usually for slow growing organisms, and generally does not appear to be associated with severe septic reactions, though systematic data are lacking. Culture methods are usually applied towards the start of the platelet shelf life, within a few hours to 2 days after collection of the donation, and the culture is left to run for between 1 or 2 days, up to the end of the shelf life of the platelet unit. The sample volume varies between 2 and 15 mls, and two of the three methods in general use allow a choice of aerobic culture only, or aerobic and anaerobic culture in two separate culture bottles. While obligate anaerobes are very rarely associated with serious reactions to platelet transfusions [16], the anaerobic medium accelerates the growth of some aerobes, and the second bottle allows for a bigger sample size, and from a purely scientific point of view, for limiting dilution
quantitation of the number of bacteria in the contaminated platelet unit [9]. Three culture methods have been in routine use. The BacT/ALERT system (BioMerieux, France) is by far the commonest to appear in reports, though it is very similar to the BACTEC (BD Diagnostic Systems, USA) and there is probably little to distinguish between them in effectiveness or in practice. They require that the culture bottle or bottles are inoculated with the test material; the cultures are then monitored regularly by machine for telltale changes in pH within the medium that can be observed in colour changes in a material visible from outside the bottle. The third method that has been introduced into routine practice is the Pall eBDS system (Pall Corporation, USA), which relies on the generation of carbon dioxide in the headspace of a closed culture system. The sample size is smaller than the other two, and the culture time is limited, but it has advantages in greater automation of readout and ease of sampling. Culture based methods are prone to false positives, where the machines are over sensitive to fluctuations in their environment that are not caused by pH changes in the culture bottles, or by problems in the system caused by the platelets in the inoculum, or by the medium the platelets are being stored in. There is also the problem where bacteria in the sample grow in the culture, but die off in the platelet bag that they came from initially. This is unresolvable at present, and leads to different definitions of what constitutes a true positive. Some blood centres report this situation, where bacteria are growing in the culture bottle but cannot be found on re-sampling of the initial platelet unit, as contamination during sampling or inoculation. However it may not be, and this will lead to underestimates (or overestimates if the opposite assumption is made) in the initial rate of contamination. Culture methods are also prone to false negatives, due mainly to the low number of bacteria in the platelet unit at the time of sampling. This is a serious limitation, addressed in part by delaying the time of sampling as long as possible and by increasing the sample size, but can never be completely overcome. Non-culture methods have the major advantage of speed of readout, which enables them to tell whether there are toxic levels of bacteria in the platelet unit at the time of testing. They therefore lend themselves to being applied at the point of release of the platelet unit from the hospital blood bank – i.e. within minutes of the time of transfusion. This allows for the maximum amount of time for any contaminating bacteria in the platelet unit to grow to the level of detection of the test system. However there is a very large trade off in practice between speed of readout and sensitivity of detection. The earlier tests – Gram stain, glucose consumption, pH measurement – were insensitive to seriously large levels of contamination in practice [17]. Newer methods, for example using microfluidics and antibody–antigen immobilisation to detect bacteria in a small sample, may have higher sensitivity, down to 105 bacteria in the platelet unit, and may have utility in hospital practice. Flow cytometry and PCR testing, which have higher
W.G. Murphy, P. Coakley / Transfusion and Apheresis Science 45 (2011) 69–74
sensitivity for bacteria in the test sample, currently can only test a small sample, which makes them too insensitive for early testing (compared to culture based methods which are equally capable of detecting bacteria in the test sample, but can process a much larger test volume), while they are probably too slow to be practicable for late testing.
7. Testing to improve inventory or to improve safety? Because platelet shelf life is limited by bacterial safety, it should be possible to prolong shelf life by ensuring that there are no contaminating bacteria growing in the platelet unit. This is a subtle difference in the purpose behind testing – to improve supply as well as to improve safety. It is hardly acceptable to keep safety at current levels while improving logistics. It is therefore necessary to show that safety of a tested unit is preserved during storage beyond day 5 of shelf life, not simply that storage to day 7 with testing is safer than storing to day 5 without testing. Thus we considered that it was not valid to assume that the new-found safety margin achieved by testing at day 1 or 2 could be compromised by storing for a further 2 days beyond day 5. By mathematical modelling alone it was obvious that day 1 or day 2 testing could not reliably improve safety beyond day 5 [14]; practical experience subsequently bore this out [9]. Even without lag phase growth retardation of contaminating bacteria, simple binomial distribution will prevent the reliable detection of bacteria capable of growing from one cfu per platelet unit to 105 cfu per platelet unit at day 7 by any reasonable sampling volume or interval prior to day 4 of storage. Lag phase growth characteristics will make this insurmountable problem even worse. Day 1 or 2 testing using any feasible sampling protocol or available assay lacks sufficient sensitivity to allow storage to be increased beyond day 5 without compromising the safety gains over the first 5 days. On the plus side, we showed that a day 4 sampling protocol was probably very effective in detecting any bacteria in the initial platelet unit [9], and have regulatory approval to extend platelet shelf life to day 7 on the basis of a second BacT/ALERT test at day 4.
73
The tests we have at present, though useful, are not the Holy Grail. 9. Where next – Is there any future for testing in the setting of pathogen reduction technologies? Three technologies for the elimination of microorganisms from platelet units exist – the Cerus (amotosalen/ UV light), Caridian (riboflavin/UV light) and Macopharma (UV light alone) technologies. All are effective against bacteria, and could be used for that purpose. Given the unavoidable failures from testing in the absence of a feasible total-unit testing device, these techniques are set fair to replace bacterial testing, should they prove to be practicable and affordable. This is all the more so, since eradication techniques such as these will be most efficacious when the testing techniques are most compromised, at the start of the platelet shelf life. Notwithstanding the firm findings of a 2007 consensus conference [18], penetration of pathogen reduction into the marketplace has been slow. Progress is being made however, and several countries have moved towards implementing pathogen reduction using one of the first two techniques. The Macopharma method, perhaps in all other respects the most attractive of the three, has an unexplained lack of utility against HIV. This would not compromise its bactericidal effect, but undermines any protection against window period virus infection. All three, however, have an Achilles heel in the fight against bacteria – they are not sporicidal, and require that the spores vegetate before the contamination can be eradicated. Whether in practice this remains a concern will take time to resolve. B. cereus is a serious problem: it appears in many clinical series of implicated organisms, and is probably responsible for approximately 10% of clinical disease from contaminated platelets. It can escape detection even with delayed testing, but may be more sensitive to pathogen reduction than to detection, since spores will be eliminated as long as they vegetate before the PR is executed in the manufacturing protocol. This is very likely to be at a point earlier in the process than the point where they become detectable in any test. 10. Conclusion
8. What is the perfect test? The ideal test is sensitive to very low levels of bacteria, has a low false positive rate, does not require the bag to be punctured for sampling, and is cheap, reliable and fast. Because the sensitivity, especially early in the shelf life, is almost entirely dependent on sample size, the ideal test will sample the whole bag contents without compromising the platelets, a far from simple task. It will sample them in situ, perhaps using a capture method within the bag that can be read from outside, and that does not leach reagents into the therapeutic product. There have been attempts to develop these, for example using incident and emitted light and detectable changes in the absorption or excitation characteristics on ligands between the bound and unbound states, and perhaps it is time to dust these ideas off again.
Bacterial contamination of platelets is a very serious issue, and is unavoidable. Reducing the incidence of contamination is important, though it will always be of limited efficacy. Detection of contamination is also important, but remains seriously compromised by the available techniques. Perhaps a new, light-based, whole-unit sampling technology will emerge to solve the problems of small samples and delayed growth, but for the moment pathogen reduction may be a quicker route to reducing residual risk from bacteria. Even if pathogen reduction is deployed, there will be residual concerns to be addressed – particularly the inevitable presence of spores in some units, and their potential for escaping elimination. Nevertheless testing for bacteria has significantly improved the safety of platelet transfusion, even if it has not been universally applicable for the prolongation of platelet shelf-life.
74
W.G. Murphy, P. Coakley / Transfusion and Apheresis Science 45 (2011) 69–74
References [1] Hord NG, Ghannam JS, Garg HK, Berens PD, Bryan NS. Nitrate and nitrite content of human, formula, bovine, and soy milks: implications for dietary nitrite and nitrate recommendations. Breastfeed Med; 2010 [Epub ahead of print]. [2] Jones BC, Saw MH, Hanson MF, Mackie MJ, Scott J, Murphy WG. Yersinia enterocolitica from blood transfusion. J Clin Pathol 1993;46:477–8. [3] Gibb AP, Poling N, Murphy WG. Failure to kill Yersinia Enterocolitica by plasma diluted to the concentration found in red cell units. J Clin Pathol 1996:49434–6. [4] Goldman M, Roy G, Frechette N, Decary F, Massicotte L, Delage G. Evaluation of donor skin disinfection methods. Transfusion 1997;37:309–12. [5] McDonald CP, Lowe P, Roy A, Robbins S, Hartley S, Harrison JF, et al. Evaluation of donor arm disinfection techniques. Vox Sang 2001;80:135–41. [6] de Korte D, Curvers J, de Kort WL, Hoekstra T, van der Poel CL, Beckers EA, et al. Effects of skin disinfection method, deviation bag, and bacterial screening on clinical safety of platelet transfusions in the Netherlands. Transfusion 2006;46:476–85. [7] Holden F, Foley M, Devin G, Kinsella A, Murphy WG. Coagulasenegative staphylococcal contamination of whole blood and its components: the effects of WBC reduction. Transfusion 2000 Dec;40(12):1508–13. [8] Högman CF. Aspects of platelet storage. Transfus Sci 1994;15:351–5. [9] Murphy WG, Foley M, Doherty C, Tierney G, Kinsella A, Salami A, et al. Screening platelet concentrates for bacterial contamination: low numbers of bacteria and slow growth in contaminated units
[10]
[11]
[12]
[13]
[14] [15]
[16]
[17]
[18]
mandate an alternative approach to product safety. Vox Sang 2008;95:13–9. Serious hazards of transfusion. Annual Report 2009. Accessed at; accessed February 2011. Pietersz RN, Engelfriet CP, Reesink HW, Wood EM, Winzar S, Keller AJ, et al. Detection of bacterial contamination of platelet concentrates. Vox Sang 2007;93:260–77. Eder AF, Kennedy JM, Dy BA, Notari EP, Skeate R, Bachowski G, et al. Limiting and detecting bacterial contamination of apheresis platelets: inlet-line diversion and increased culture volume improve component safety. Transfusion 2009;49:1554–63. Braine HG, Kickler TS, Charache P, Ness PM, Davis J, Reichart C, et al. Bacterial sepsis secondary to platelet transfusion: an adverse effect of extended storage at room temperature. Transfusion 1986;26:391–3. Murphy WG, Smyth J. Testing for bacteria in platelet concentrates: defining the parameters. Transfus Apheresis Sci 2001;24:247–9. Benjamin RJ, Wagner SJ. The residual risk of sepsis: modeling the effect of concentration on bacterial detection in two-bottle culture systems and an estimation of false-negative culture rates. Transfusion 2007;47:1381–9. McDonald CP, Hartley S, Orchard K, Hughes G, Brett MM, Hewitt PE, et al. Fatal Clostridium perfringens sepsis from a pooled platelet transfusion. Transfus Med 1998;8:19–22. Yomtovian R, Brecher ME. pH and glucose testing of single-donor apheresis platelets should be discontinued in favor of a more sensitive detection method. Transfusion 2005;45:646–8. Klein HG, Anderson D, Bernardi MJ, Cable R, Carey W, Hoch JS, et al. Pathogen inactivation: making decisions about new technologies. Report of a consensus conference. Transfusion 2007;47:2338–47.
Contents lists available at ScienceDirect
Transfusion and Apheresis Science journal homepage: www.elsevier.com/locate/transci
Testing platelet components for bacterial contamination William G. Murphy a,b,⇑, Pauline Coakley b a b
School of Medicine and Medical Science, University College Dublin, Ireland National Blood Centre, Irish Blood Transfusion Service, James’s Street, Dublin 8, Ireland
a r t i c l e Article history:
i n f o
a b s t r a c t Bacteria in transfused platelets can cause serious morbidity and, rarely, death. Most contaminating bacteria enter the blood at the time of venepuncture. While many of these contaminants fail to grow in the platelet unit, storage of platelets at 20–24 °C facilitates growth of some organisms, and the cumulative risk of severe sepsis increases with the storage age of platelet components. Several methods have been developed or adapted to attempt to detect contaminating bacteria with high sensitivity and specificity, but the perfect test has yet to be found. Testing early in the platelet component’s shelf life, even using exquisitely sensitive culture-based tests, is compromised by major problems of sample error – there may be too few bacteria present at this stage to ensure that any practical sample volume contains even one of them. Culture techniques are too slow to be useful as a release test. On the other hand, available rapid tests are too insensitive to use early in the shelf life, and have yet to show convincingly that they are sensitive enough for testing close to the time of transfusion. Nevertheless testing for bacteria in platelet components represents a significant advance in blood transfusion safety, and prevents the transfusion of many hundreds of bacterially-contaminated platelet units each year. Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction The role that microorganisms play in our lives is widely known, though perhaps rarely appreciated. Our symbiosis with microscopic life on an evolutionary, individual and community level is truly extraordinary: think of the origin of mitochondria, or consider what would happen if biological waste and remains were not returned to the earth by microorganisms. So dependent are we on bacteria for life, for example, that colostrum must contain high levels of nitrite to compensate for the lack of salivary bacteria at birth to reduce dietary nitrate for the neonate [1]. The list is endless: the genome, the cell, the system, the organism and the society are all intimately intertwined with a microbiological ecosystem. The immanent beauty of this setup ⇑ Corresponding author at: National Blood Centre, Irish Blood Transfusion Service, James’s Street, Dublin 8, Ireland. E-mail address: [email protected] (W.G. Murphy). 1473-0502/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.transci.2011.06.005
however is seriously undermined by our attempts to live life on an individual level – viruses are not evolved to help us (even genes are not evolved to help us!), nor bacteria, nor the rest. Only we, peculiarly, may be evolved to support humanity, though we seem to be a bit hit and miss at that. Once the ecosystem is disturbed on any level, we are open to a breach of our beautiful symbiosis, and to the emergence of disease in our environment, that was not there before. The process of mutation at many levels in the complex organism of the planet constantly disturbs the equilibrium. The often lethal, seemingly malign, and probably limitless, emergence of bacterial resistance to antibiotics is one manifestation of this. The emergence into our culture of transfusion-adapted microorganisms is another. Once we began the practice of transfusing large volumes of blood from one human to another we changed again our relationship with the microbiological ubiquity. As we generate tactical responses, we open up other avenues either for pre-existing organisms to find a new niche
70
W.G. Murphy, P. Coakley / Transfusion and Apheresis Science 45 (2011) 69–74
to grow, or for new mutations to find a possible zone of survival success. Consider Yersinia enterocolitica, emerging as a lethal complication of blood transfusion in the 1970s and 1980s, seemingly from nowhere [2]. A common cause of food poisoning, and with a transient and often asymptomatic intravascular phase, it had not troubled the blood transfusion community beforehand, even though it did not noticeably change its epidemiology in the late 70s. The organism’s ability to survive the cold, to adhere to blood cells without being destroyed by them, and to resist complement destruction, allowed this passenger, which had probably previously contaminated hundreds, even thousands of blood collections without harm to the recipient, to emerge as a lethal contaminant when some change in blood processing or storage was introduced. This may have been the simple reduction of plasma content of red cell transfusions, a practice that became widespread from the end of the 1970s – almost all reactions were from plasma-reduced red cell concentrates [3]..
2. The natural history and ecology of bacterial contamination of blood components Many units of blood collected from donors are contaminated at the point of venesection. It is impossible to sterilise the deep skin, and at the very least, some bacteria are very likely to be transported from there to the collection bag. A whole-sample analysis for contaminating bacteria during collection would probably be needed to measure the incidence of contamination at this point: immediately upon exposure to the constitutive elements of the collected blood (which will continue to function for several hours at least), bacteria are killed, sequestered, or inhibited. However, holding blood at a temperature warm enough to clear the complement-sensitive and phagocytise the ingestible, may also increase the risk of allowing otherwise harmless spores to vegetate or other bacteria to proliferate, and emerge as pathogens in their turn. There is one more striking feature of the natural history of bacterial contamination of blood collections about which next to nothing is known, and which bears consideration: almost all of our knowledge relates to contaminating events that cause acute clinical reactions – we know little about the longer term effects of transfusion of bacteriallycontaminated blood components into immune-compromised hosts. Perhaps there are none – we are after all adapted to accommodate bacteria on our skin, in our lungs and guts, and in sizeable numbers in our blood from time to time – but it is difficult to imagine that patients who are troubled by their own commensals during periods of profound immune-suppression can universally accommodate 105 colony forming units of coagulase negative staphylococci down their central line with complete biological equanimity. This also applies to low levels of contamination in red cell units, and also in plasma – freezing does not sterilise, though in practice we often assume that it does. Platelets, and the patients who receive them, form their own microbiological niche, where benign, stable, skin commensals in one healthy donor can become a toxic,
sometimes lethal, invader in another. The constituents of this system are the organisms in or on the donor, the defences against invasion of these organisms deployed by the collector, the conditions in which the organisms that survive the collection process are maintained, and the state of the recipient. A purely technical approach to the problem would therefore mandate optimised removal of bacteria from the donor, optimised conditions of processing and storage to destroy or immobilise any surviving organisms, and validation of the effectiveness of these strategies on a global and individual level. In addition, any feasible strategy to boost the defence of the recipient would be identified and engaged. A more systematic approach might also set these in a broader context – there will probably always be bacteria either present or emerging that will escape destruction or detection, and changes to processing, storage or patient protocols may have unforeseeable consequences from trivial to the disastrous. In addition, large differences in transfusion rates of platelets between countries – where the rate in some countries is considerably more than in others with similar health care systems and outcomes, without any obvious reason – translate into differences in the rate of severe reactions per total population per year. The USA for example, will have nearly 50% more septic transfusion episodes per year than the UK, all things else being equal, since 50% more platelets per unit population per year are transfused there. Why are platelets so much more of a problem than red cells from the point of view of bacterial contamination? For red cells compared to platelets on a unit for unit basis, the problem is vanishingly small: one in several million red cell units causes bacterial reactions in patients (or serious, immediate ones at least) compared to one in several tens of thousands for platelets, close to two orders of magnitude of a difference. Yet the initial contamination rate of the red cell unit is the same. The striking difference is the conditions of storage – even brief exposure to cold immobilises almost all the organisms collected with the unit. There may be other, minor agents in generating the difference, but cold exposure, especially before much in the way of endotoxin or exotoxin can be generated, is certainly crucial. Bacterial reactions from red cells are almost always caused by organisms that grow reasonably well at low temperatures, or are associated with breaches of the cold chain. Platelets are denied this 99% effective defence – the comparison shows just how important cold storage probably is. (It is probably important to note that cold storage is an effective post collection strategy. The sterility of collection itself may be of minor importance.) Testing for bacteria in platelets is one way of trying to compensate for the lack of the effective bactericidal or bacteriostatic effects of cold.
3. Reducing bacterial risk Given therefore that bacteria in the collected platelet units are a fact of life, and that the use of cold as a sterilising agent is denied us, we are left with alternative removal techniques, and with attempting to test to sterility. There are effective removal techniques, though whether they
W.G. Murphy, P. Coakley / Transfusion and Apheresis Science 45 (2011) 69–74
provide any overall benefit in the presence of testing remains to be seen. There are also effective improvements that can be made to the collection techniques: no touch, improved skin disinfection [4,5] and initial sample diversion [6] may all reduce bacterial load in the initial collection, though whether these effects are additive is not clear – they do not make much material difference to the safety of red cell collections for example. Overnight warm hold, incubation with the donor leucocytes and perhaps plasma, at a biologically permissive temperature, and leucodepletion (perhaps only after prolonged warm hold) can reduce bacterial load [7,8] and is associated, for whatever reason, with a reduced incidence of platelet-associated bacterial sepsis. Shortening the shelf life also works. Testing works too, but it is not possible to test the platelet supply to sterility as things currently stand. The reasons are (1) that very low numbers of bacteria in a platelet unit near the start of its life may readily escape being captured in the sample [9], but still grow to lethality; and (2) that highly sensitive assays that detect 1–10 bacteria in the test sample are too slow, and efficient assays that give results quickly enough to be used immediately prior to transfusion are too insensitive, to get around this problem.
4. Clinical epidemiology of bacterial sepsis The rate of bacterial contamination of platelets will vary from place to place, as will the incidence of septic reactions from platelet transfusion. The skin cleaning technique, the use of diversion pouches, the length of storage and the method of platelet preparation will all influence the rate of contamination. The method of detection – the timing and volume of the sample and the testing method usedwill also affect the observed incidence. The observed rate of septic reactions is also subject to variables including the quality of the surveillance and reporting system, and the structure of the platelet supply. Accepting the limitations above, the contamination rate of platelet components, as measured with the BacT/ALERT system, is approximately in the range of 3 per 1000 to 1 in 5000, the lower incidence being associated with smaller samples cultured aerobically only, and the higher one with larger samples cultured both aerobically and anaerobically and with testing at the end of shelf life. In one series [6] rates of up to 0.95% were observed, where the sampling was done immediately after collection or manufacture using a 10 ml sample in a one-bottle test on BacT/ALERT. This may represent a true measure of the underlying contamination rate after venesection, before any auto-sterilisation takes effect in the collected or stored blood components. From the reports of large series of transfusions, the rate of septic reactions from bacterially contaminated platelets is in the order of 1:50,000–1:100,000 units of platelets supplied to hospitals. For death from bacterial contamination of platelets the rate is in the 1:250,000–1/500,000 range. The Serious Hazards of Transfusion group in the UK logged 40 episodes of platelet-related sepsis, with 11 deaths, in 10 years, comprising approximately 3000,000 platelet units shipped: i.e. approximately 1:75,000 septic reactions reported for platelets issued, and 1 in 273,000 deaths [10].
71
From the French Haemovigilance data a rate of 17 septic reactions per 1000,000 platelets issued has been reported [11] or 1 in 59,000. In the USA, following the introduction of a bacterial screening programme in the American Red Cross blood centres in 2004, the observed rate of sepsis was 1:59,000 over the first 26 months of the programme, falling slightly to 1/83,000 after the sample volume was increased to 8 mls from 4 mls in the BacT/ALERT test. Three fatal reactions were reported in the first phase of the programme, for a mortality rate of 1:498,711 [12]. Data on benefits from bacterial screening on studies of this scale are not available from the two large haemovigilance systems of the UK and France, neither of whom has introduced universal regular screening of platelet concentrates. 5. Testing strategies Platelets have a 5 day shelf life, or in some places, a 3 day shelf life [11], because of this bacterial problem. They are probably effective as haemostatic agents after 7 or 8 days of storage, and perhaps even longer, but the risk of bacterial disease increases with storage time [13], and limiting the storage time has been a universal response to this issue. A few groups have tried to get around this with testing protocols, but there are theoretical and practical obstacles. We have used a two stage testing protocol that has given us access to 7 day platelets on demand over several years with a theoretically high level of safety [14], and with reasonable logistical and economic demands, but the process requires sophisticated inventory management and a short supply chain to optimise value. To maximise efficiency – either safety of the product or cost of the process – a compromise must be made between time of sampling along the shelf life of the product, and the sensitivity of the test used. Near the start of the shelf life the bacterial numbers are low, and indeed contaminating bacteria may be in a quiescent phase – they may not be growing very much at all. Here, sample error rates will inevitably be high, even if the test used can detect very low numbers of bacteria in the test sample. Paradoxically, techniques that reduce the bacterial load in the initial collection – overnight hold and leucodepletion, for example – but do not sterilise the unit, will make testing even less effective at this point. Unlike the viruses of blood transfusion however, bacteria may increase their infectivity after they have been collected into a therapeutic preparation, and units without sufficient bacterial load to test positive at one point of their shelf life may be massively contaminated several days later – the platelet unit acting as an effective bacterial growth medium. We use a very large sample volume – at 2 7.5 mls around 5% of the total volume of the platelet preparation – sampled after a minimum of 12 h holding of apheresis platelets or after 12 h holding of the manufactured platelet pool. Nevertheless, testing again at outdate of previously sampled platelet units shows we miss around two thirds of contaminating bacteria in the initial sampling protocol [9]. While most of the bacteria that we miss are of dubious clinical significance, we have missed some instances of Bacillus cereus, and on one occasion, one split of a triple apheresis donation contaminated with Proteus mirabilis.
72
W.G. Murphy, P. Coakley / Transfusion and Apheresis Science 45 (2011) 69–74
In contrast, a test done close to the time of transfusion of the platelet unit, within a couple of doubling times of the fastest growing organism possible perhaps, has merely to detect a clinically significant bacterial load, and is not constrained by the niceties of Poisson or binomial distributions. It’s not completely obvious what constitutes a significant, or even more, an insignificant bacterial load. A platelet unit contains two or three hundred millilitres of fluid – a test with a limit of detection of 1000 organisms per ml may miss up to 105 bacteria in the transfusate. This is probably good enough to reduce the incidence of disease considerably, but does leave a large window of opportunity for bacteria to get through. 6. Techniques Culture-based techniques rely on some characteristic of the changes induced in a medium by growing bacteria – gas generation, acid production, to elicit a signal. They require (1) that bacteria grow in the culture conditions of the test and (2) that the culture is continued for long enough to allow the signal to be generated by the growth of bacteria. They can be very sensitive – these tests are capable of detecting as few as one colony forming unit (cfu) in the inoculated sample, and can reliably be expected to detect 10–100 cfu in the sample. The sensitivity of the test then relies predominantly on the size of the sample, and on the chances of getting bacteria into the test volume – i.e. the ratio of the sample size to the total volume of the platelet unit. Therefore the bigger the sample size the greater the chance of detecting them in the test, where the bacteria are few in number. Bacteria in platelets can be very few in number at the start of the shelf life – below 1 cfu per ml of platelet concentrate [9,15]. A huge problem with culture methods is that they do not give a once-for-all-time result. The test can give a positive result after several days of growth. This forces blood centres to issue the platelet unit while the culture is continuing in the laboratory, giving rise to the frequent complication of having to inform clinicians that their patient has received a unit of platelets that was contaminated with bacteria, at least at an earlier point in its shelf life. This is not trivial – it at times gives rise to changes in patient treatment that are of uncertain benefit or risk. This state of affairs arises usually for slow growing organisms, and generally does not appear to be associated with severe septic reactions, though systematic data are lacking. Culture methods are usually applied towards the start of the platelet shelf life, within a few hours to 2 days after collection of the donation, and the culture is left to run for between 1 or 2 days, up to the end of the shelf life of the platelet unit. The sample volume varies between 2 and 15 mls, and two of the three methods in general use allow a choice of aerobic culture only, or aerobic and anaerobic culture in two separate culture bottles. While obligate anaerobes are very rarely associated with serious reactions to platelet transfusions [16], the anaerobic medium accelerates the growth of some aerobes, and the second bottle allows for a bigger sample size, and from a purely scientific point of view, for limiting dilution
quantitation of the number of bacteria in the contaminated platelet unit [9]. Three culture methods have been in routine use. The BacT/ALERT system (BioMerieux, France) is by far the commonest to appear in reports, though it is very similar to the BACTEC (BD Diagnostic Systems, USA) and there is probably little to distinguish between them in effectiveness or in practice. They require that the culture bottle or bottles are inoculated with the test material; the cultures are then monitored regularly by machine for telltale changes in pH within the medium that can be observed in colour changes in a material visible from outside the bottle. The third method that has been introduced into routine practice is the Pall eBDS system (Pall Corporation, USA), which relies on the generation of carbon dioxide in the headspace of a closed culture system. The sample size is smaller than the other two, and the culture time is limited, but it has advantages in greater automation of readout and ease of sampling. Culture based methods are prone to false positives, where the machines are over sensitive to fluctuations in their environment that are not caused by pH changes in the culture bottles, or by problems in the system caused by the platelets in the inoculum, or by the medium the platelets are being stored in. There is also the problem where bacteria in the sample grow in the culture, but die off in the platelet bag that they came from initially. This is unresolvable at present, and leads to different definitions of what constitutes a true positive. Some blood centres report this situation, where bacteria are growing in the culture bottle but cannot be found on re-sampling of the initial platelet unit, as contamination during sampling or inoculation. However it may not be, and this will lead to underestimates (or overestimates if the opposite assumption is made) in the initial rate of contamination. Culture methods are also prone to false negatives, due mainly to the low number of bacteria in the platelet unit at the time of sampling. This is a serious limitation, addressed in part by delaying the time of sampling as long as possible and by increasing the sample size, but can never be completely overcome. Non-culture methods have the major advantage of speed of readout, which enables them to tell whether there are toxic levels of bacteria in the platelet unit at the time of testing. They therefore lend themselves to being applied at the point of release of the platelet unit from the hospital blood bank – i.e. within minutes of the time of transfusion. This allows for the maximum amount of time for any contaminating bacteria in the platelet unit to grow to the level of detection of the test system. However there is a very large trade off in practice between speed of readout and sensitivity of detection. The earlier tests – Gram stain, glucose consumption, pH measurement – were insensitive to seriously large levels of contamination in practice [17]. Newer methods, for example using microfluidics and antibody–antigen immobilisation to detect bacteria in a small sample, may have higher sensitivity, down to 105 bacteria in the platelet unit, and may have utility in hospital practice. Flow cytometry and PCR testing, which have higher
W.G. Murphy, P. Coakley / Transfusion and Apheresis Science 45 (2011) 69–74
sensitivity for bacteria in the test sample, currently can only test a small sample, which makes them too insensitive for early testing (compared to culture based methods which are equally capable of detecting bacteria in the test sample, but can process a much larger test volume), while they are probably too slow to be practicable for late testing.
7. Testing to improve inventory or to improve safety? Because platelet shelf life is limited by bacterial safety, it should be possible to prolong shelf life by ensuring that there are no contaminating bacteria growing in the platelet unit. This is a subtle difference in the purpose behind testing – to improve supply as well as to improve safety. It is hardly acceptable to keep safety at current levels while improving logistics. It is therefore necessary to show that safety of a tested unit is preserved during storage beyond day 5 of shelf life, not simply that storage to day 7 with testing is safer than storing to day 5 without testing. Thus we considered that it was not valid to assume that the new-found safety margin achieved by testing at day 1 or 2 could be compromised by storing for a further 2 days beyond day 5. By mathematical modelling alone it was obvious that day 1 or day 2 testing could not reliably improve safety beyond day 5 [14]; practical experience subsequently bore this out [9]. Even without lag phase growth retardation of contaminating bacteria, simple binomial distribution will prevent the reliable detection of bacteria capable of growing from one cfu per platelet unit to 105 cfu per platelet unit at day 7 by any reasonable sampling volume or interval prior to day 4 of storage. Lag phase growth characteristics will make this insurmountable problem even worse. Day 1 or 2 testing using any feasible sampling protocol or available assay lacks sufficient sensitivity to allow storage to be increased beyond day 5 without compromising the safety gains over the first 5 days. On the plus side, we showed that a day 4 sampling protocol was probably very effective in detecting any bacteria in the initial platelet unit [9], and have regulatory approval to extend platelet shelf life to day 7 on the basis of a second BacT/ALERT test at day 4.
73
The tests we have at present, though useful, are not the Holy Grail. 9. Where next – Is there any future for testing in the setting of pathogen reduction technologies? Three technologies for the elimination of microorganisms from platelet units exist – the Cerus (amotosalen/ UV light), Caridian (riboflavin/UV light) and Macopharma (UV light alone) technologies. All are effective against bacteria, and could be used for that purpose. Given the unavoidable failures from testing in the absence of a feasible total-unit testing device, these techniques are set fair to replace bacterial testing, should they prove to be practicable and affordable. This is all the more so, since eradication techniques such as these will be most efficacious when the testing techniques are most compromised, at the start of the platelet shelf life. Notwithstanding the firm findings of a 2007 consensus conference [18], penetration of pathogen reduction into the marketplace has been slow. Progress is being made however, and several countries have moved towards implementing pathogen reduction using one of the first two techniques. The Macopharma method, perhaps in all other respects the most attractive of the three, has an unexplained lack of utility against HIV. This would not compromise its bactericidal effect, but undermines any protection against window period virus infection. All three, however, have an Achilles heel in the fight against bacteria – they are not sporicidal, and require that the spores vegetate before the contamination can be eradicated. Whether in practice this remains a concern will take time to resolve. B. cereus is a serious problem: it appears in many clinical series of implicated organisms, and is probably responsible for approximately 10% of clinical disease from contaminated platelets. It can escape detection even with delayed testing, but may be more sensitive to pathogen reduction than to detection, since spores will be eliminated as long as they vegetate before the PR is executed in the manufacturing protocol. This is very likely to be at a point earlier in the process than the point where they become detectable in any test. 10. Conclusion
8. What is the perfect test? The ideal test is sensitive to very low levels of bacteria, has a low false positive rate, does not require the bag to be punctured for sampling, and is cheap, reliable and fast. Because the sensitivity, especially early in the shelf life, is almost entirely dependent on sample size, the ideal test will sample the whole bag contents without compromising the platelets, a far from simple task. It will sample them in situ, perhaps using a capture method within the bag that can be read from outside, and that does not leach reagents into the therapeutic product. There have been attempts to develop these, for example using incident and emitted light and detectable changes in the absorption or excitation characteristics on ligands between the bound and unbound states, and perhaps it is time to dust these ideas off again.
Bacterial contamination of platelets is a very serious issue, and is unavoidable. Reducing the incidence of contamination is important, though it will always be of limited efficacy. Detection of contamination is also important, but remains seriously compromised by the available techniques. Perhaps a new, light-based, whole-unit sampling technology will emerge to solve the problems of small samples and delayed growth, but for the moment pathogen reduction may be a quicker route to reducing residual risk from bacteria. Even if pathogen reduction is deployed, there will be residual concerns to be addressed – particularly the inevitable presence of spores in some units, and their potential for escaping elimination. Nevertheless testing for bacteria has significantly improved the safety of platelet transfusion, even if it has not been universally applicable for the prolongation of platelet shelf-life.
74
W.G. Murphy, P. Coakley / Transfusion and Apheresis Science 45 (2011) 69–74
References [1] Hord NG, Ghannam JS, Garg HK, Berens PD, Bryan NS. Nitrate and nitrite content of human, formula, bovine, and soy milks: implications for dietary nitrite and nitrate recommendations. Breastfeed Med; 2010 [Epub ahead of print]. [2] Jones BC, Saw MH, Hanson MF, Mackie MJ, Scott J, Murphy WG. Yersinia enterocolitica from blood transfusion. J Clin Pathol 1993;46:477–8. [3] Gibb AP, Poling N, Murphy WG. Failure to kill Yersinia Enterocolitica by plasma diluted to the concentration found in red cell units. J Clin Pathol 1996:49434–6. [4] Goldman M, Roy G, Frechette N, Decary F, Massicotte L, Delage G. Evaluation of donor skin disinfection methods. Transfusion 1997;37:309–12. [5] McDonald CP, Lowe P, Roy A, Robbins S, Hartley S, Harrison JF, et al. Evaluation of donor arm disinfection techniques. Vox Sang 2001;80:135–41. [6] de Korte D, Curvers J, de Kort WL, Hoekstra T, van der Poel CL, Beckers EA, et al. Effects of skin disinfection method, deviation bag, and bacterial screening on clinical safety of platelet transfusions in the Netherlands. Transfusion 2006;46:476–85. [7] Holden F, Foley M, Devin G, Kinsella A, Murphy WG. Coagulasenegative staphylococcal contamination of whole blood and its components: the effects of WBC reduction. Transfusion 2000 Dec;40(12):1508–13. [8] Högman CF. Aspects of platelet storage. Transfus Sci 1994;15:351–5. [9] Murphy WG, Foley M, Doherty C, Tierney G, Kinsella A, Salami A, et al. Screening platelet concentrates for bacterial contamination: low numbers of bacteria and slow growth in contaminated units
[10]
[11]
[12]
[13]
[14] [15]
[16]
[17]
[18]
mandate an alternative approach to product safety. Vox Sang 2008;95:13–9. Serious hazards of transfusion. Annual Report 2009. Accessed at