Zero tolerance

Blood transfusion will never be completely safe, although tremendous improvements have been made in the past 10 years, through improved screening technologies and hemovigilance. In recent years defining the acceptable level of transfusion risk, and the provision of resources to achieve the defined levels, has been a largely political process. The move from storing blood in glass bottles to plastic packs in the early 1950s greatly reduced the incidence of posttransfusion bacterial infection. For 4°C stored red cell (RBC) products, sepsis remains a rare event. The introduction of storage at 22°C for platelets (PLTs) in the early 1970s by the late Scott Murphy and colleagues,1 as forewarned at the time, led to an increased rate of posttransfusion bacterial infection, particularly when the shelf life of such products was briefly extended beyond 5 days.2 With improvements to the methods for detecting the major transfusion-transmitted virus infections of concern, bacteria now account for the majority of transfusion-transmitted infections. By use of sensitive tests for the presence of bacteria, approximately 400 per million PLTs tested are found to be contaminated although most of these are at a level which will cause little or no harm to patients.3 After much debate, bacterial screening of PLT products was introduced in the United States in March 20044 after earlier implementation in some European countries. (AABB Standard 5.1.5.1. states: “The blood bank or transfusion service shall have methods to limit and detect bacterial contamination in all platelet components. Standard 5.6.2 applies [skin disinfection].” All AABB accredited blood banks were required to meet Standard 5.1.5.1 by March 1, 2004.) With the agreement of the FDA, through the PASSPORT scheme, the option exists to reextend PLT shelf life to 7 days.5 This scheme, which depends on bacterial testing by both aerobic and anaerobic methods at 24 hours with an additional 24-hour quarantine and retesting at outdate (7 days), has the advantage of defining an acceptable risk, in this case “no worse than the current bacterial risk for 5-day stored platelets,” estimated by the FDA to be around 100 per million doses (1 in 10,000). These statements refer to the rate of test-reactive units rather than the subset of stored PLTs that cause actual clinical problems. As with any screening assay concerns relate to the sensitivity and specificity. Recent reports6-11(and C.V. Prowse, J. AuBuchon, and S. Wendel on behalf of the BEST Collaborative and ISBT Transfusion Transmitted Infection Subcommittee, bacterial testing practices survey 2004, unpublished) have shown that in combination with improved donor skin disinfection and diversion of the first few milliliters of donated blood,12 testing of PLTs with culture methods can prevent a significant number of bacterially infective PLTs from entering the blood supply (Table 1). Problems remain with the rate of false-positive tests, however, which leads to the discard of valuable product, and occasional test-negative units, which cause infection (false negatives). Evidence from the robust Canadian hemovigilance system records that diversion and testing has reduced the number of PLT-related posttransfusion infections by more than 70 percent,13 while surveillance in the United States suggests a 75 percent reduction in clinically evident bacteremia.6 A recent editorial in TRANSFUSION14 and a number of recent reviews12, 15, 16 have itemized the outcomes of such studies and the effects of the variable growth of bacterial species on test outcomes, together with their dependency on choice of assay method (including the choice of whether to undertake both aerobic and anaerobic assays if culturing), sampling time, sampled volume, the impact of retesting at later storage times, and the consequences for blood bank stocks, outdate rates, and age at issue.17, 18 For the purposes of this commentary, it is sufficient to emphasize that the more serious clinical sequelae of transfusing contaminated PLTs have been associated with species that are largely aerobic and multiply fast,15, 19 that European experience is that slow-growing contaminants may only become test-positive at a stage when up to 50 percent of units have already been transfused,9, 11 and that sepsis and death still occur, albeit at a much reduced rate, even when a robust bacterial testing system is implemented (Table 1). Retesting, or use of anaerobic as well as aerobic tests, has obvious financial implications. The ideal solution would be to have a sensitive and rapid point-of-issue bacterial detection test. Such tests are in development by a number of companies and beg the question of what sensitivity is required. It is generally accepted that initial bacterial levels in infected units are of the order of 1 to 10 per mL (300-3000 per adult dose) and need to increase around 10,000-fold to have a clinical impact. In March 2007, one company announced that they have applied to the FDA for licensure of an assay with a sensitivity of approximately 103 to 104 bacteria per mL and an assay turnaround time of 30 minutes20, 21 (license submitted March 15, 2007, to the FDA for 30-minute assay from Verax Pan Genera Detection based on the detectionof lipoteichoic acids [for Gram-positive bacteria] and lipopolysaccharides [for Gram-negative bacteria] with Pan Genera binding agents to directly bind to these targets). Reports of the impact of such a test in routine use are awaited with interest. In this issue of TRANSFUSION, Nussbaumer22 reports on an alternative approach, namely, bacterial inactivation. The method used, photoinactivation with a psoralen (amotosalen) and ultraviolet light, is licensed for clinical use in Europe with a shelf life of 5 days.23, 24 Bacterial spiking studies have shown this method can inactivate at least 3 log (99.9%; more normally at least 5 log, 99.999%) of a wide range of bacteria.25 Other companies are developing analogous inactivation technologies with similar bactericidal capacity.26, 27 For the likely levels of bacteria to be found in PLT products just after donation, laboratory testing suggests that any of these methods should minimize the risk of infection. In their study, which was sponsored by the company marketing the amotosalen approach, Nussbaumer and colleagues assessed the outgrowth of seven species of bacteria, inoculated at three different but low levels (1-1000 bacteria per unit or 0.003-3 bacteria per mL, designed to mimic the low levels of bacteria in fresh contaminated units), with an automated microbial detection culture method (BacT/ALERT, bioMérieux, Durham, NC) of split apheresis units with and without amotosalen treatment. The study claims that inactivation prevents growth of all the species assessed, whereas particularly for slow-growing species, culture methods may not detect contamination or only detect it at too late a time to prevent transfusion. There are a number of design concerns about this study, some of which are inherent in the choice of adding very low levels of bacteria such that large volumes would need to be assayed to detect the contaminant, even assuming that such low levels of bacteria are not inactivated by plasma and/or PLTs per se.28 The study did not involve any replicates, used amotosalen at a time 12 to 24 hours later than is usually undertaken, and interprets the results of automated microbial detection system BacT/ALERT testing in two different modes. For the test (inactivation) arm, a negative culture test is taken as evidence of inactivation, whereas in the control arm the assay is used as evidence of failure to detect a true contaminant, there being no independent assay employed to define or titrate contamination other than for the primary inocula. Given these limitations, a strict interpretation of the data would require a conclusion that only those study pairs in which growth was shown in the control arm but not in the test arm provide evidence of the benefits of the inactivation treatment. This result only occurs in 14 of the 21 (three levels of inoculation for seven species) study pairs. Although the experiment bears repetition with a more robust design, let us take a leap of faith and assume that the conclusions are correct. This assumption is not unreasonable, because it is known (see Table 1) that the testing approach is not completely effective, whereas the inactivation approach appears to deal with any likely level of contamination more than adequately.25 What would this mean for the provision of PLTs? The most relevant starting point is a Dutch assessment of the relative cost-effectiveness of testing and inactivation9 which concluded that testing, while not as effective, was much more cost-effective (by more than 5-fold; Fig. 1). It is worth repeating, however, the assumptions made by Janssen and colleagues9 in developing this model. First, based on past Dutch experience, they assume a test reactive rate of approximately 4000 per million PLT units (higher than most recent studies listed in Table 1) and equate this to subclinical infection. Rather than assess the rate of confirmed-reactive units, they base their incidence of clinically important transfusion transmitted sepsis on local hemovigilance data (1% of test-positive units), which they admit could be an underestimate due to underreporting and therefore adjust upward by a factor of 10. It is assumed, on good grounds cited within their study, that 20 percent of clinically septic cases would result in death. These are not unreasonable assumptions but may need adjustment in the light of recent data, which indicate lower incidence rates partly explained by the fact that the Dutch studies use aerobic and anaerobic screening (which may have a 5- to 10-fold higher yield). They thus assume that testing is 90 percent effective, while noting that over 50 percent of test-reactive units are transfused before the result is known. Again this may not be unreasonable if slow-growing species have less clinical impact, although as indicated above the recent North American data would suggest a figure closer to 75 percent than 90 percent effective. They make no allowance for the cost of rejecting initial test-reactive units nor of the potential impact of extending PLT shelf life beyond 5 days, for example, in reducing wastage rates. Basis cost-effectiveness model (based on Janssen et al.9). On the inactivation side, they allow for the reduced potency of treated PLT units23, 24 and the resultant increased dosing, but make no allowance for the limits on PLT volume and RBC content that the amotosalen process requires, which may lead to rejection of a few percent of PLTs produced. This loss may be ameliorated by procurement via apheresis. Appropriate allowance is made for the efficacy of pathogen inactivation for reducing the transfusion-related risk for human immunodeficiency virus, human T-lymphotropic virus, and hepatitis B and C but not for any long-term toxic effects arising from the pathogen reduction process. Their analysis also does not consider any efficacy against a putative emerging pathogen, which would be difficult to quantify. It is noteworthy that the French Transfusion Service recently introduced amotosalen treatment of PLTs on the Island of Reunion as a risk reduction measure in the face of the rapidly emerging strain of Chikungunya virus in the Indian Ocean area last year.29 In the face of emergent microbial infection such technology has great appeal. Certainly the availability of methylene blue photoinactivation methods was of great reassurance when, at the time of importing fresh-frozen plasma for children to address variant Creutzfeldt-Jakob disease concerns in the United Kingdom, we had notification of the emergence of West Nile Virus. If all the above is taken into account, the model of Janssen and colleagues may well be a reasonable estimate. The pending introduction of alternative methods of pathogen reduction for PLTs and the possibility of rapid point-of-issue bacterial tests, however, may lead to a reduction in the marginal costs for further improving safety. Finally, possible moves toward therapeutic rather than prophylactic PLT therapy, the lowering of PLT transfusion triggers, and the promise of safe small molecules capable of stimulating PLT production via the thrombopoietin receptor may make a somewhat more expensive “conventional” PLT product more attractive.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.