SEARCH

SEARCH BY CITATION

Over the last decade, nucleic acid amplification testing (NAT) has become a routine part of blood donor infectious screening in developed countries, as well as also being introduced in some developing countries [1–3]. NAT for human immunodeficiency virus (HIV) and hepatitis C virus (HCV) is commonly employed; the chief purpose for this testing is to detect potentially infectious units that are donated in the antibody-negative window period [4]. NAT for hepatitis B virus (HBV) is also used, but less commonly than HIV or HCV. HBV NAT has two potential applications: to detect window period units that are hepatitis B surface antigen (HBsAg)-negative and to detect occult HBV infection that usually occurs in donors who are anti-HBc (antibody to hepatitis B core)-positive [5–7]. This latter application is of most importance in countries with high prevalence of HBV infection, which precludes the use of the anti-HBc assay for blood donor screening. The complex issues involved in the decision to test for HBV DNA to detect window period and/or occult B infections are reviewed in detail in several recent publications [7–11].

Evolution of technology and testing

  1. Top of page
  2. Evolution of technology and testing
  3. Comparison of assays and assay systems
  4. Minipool vs. individual donation testing
  5. References

Initially, NAT was performed using either commercial assays or those developed in-house. Over the last several years, the use of commercial assays has increased. This can be attributed to a combination of factors: advances in automation, regulatory concerns and legal issues. The two major manufacturers are GenProbe, in conjunction with Novartis/Chiron, and Roche Molecular Systems.

Initially, NAT involved either separate tests for individual viruses (Roche assays or in-house testing) or a single screening test for multiple viruses in a multiplex format (Chiron assays). The principle of multiplex testing is to amplify and detect the nucleic acid of more than one virus in a single reaction tube using multiple pairs of primers and probes. A positive result obtained in multiplex testing must then be tested by discriminatory NAT assays for each individual virus in order to ascertain which viral nucleic acid is present in the donor sample.

Most countries began NAT testing in minipools of various sizes (16, 24, 48, 96, or 128 samples were common pool sizes in Europe and North America; in Japan, pool sizes were initially 500, but have subsequently deceased to 50 and more recently to 20) [1,12]. Due to the high concentrations of HIV and HCV RNA present during window period infection, this was and still is viewed as an acceptable method to improve blood safety [4]. Sample dilution via pooling would not be expected to significantly affect the yield of NAT screening, except perhaps in special circumstances where there is a high incidence of infection, such as has occurred in the HIV epidemic in South Africa [13,14]. Recently, there has been a major shift to smaller pool sizes (minipool of 6 samples) or individual donation (ID) testing. This is due to the development of better automation with newer assays as well as the desire to detect HBV DNA, which in many cases is present only in very low concentrations [5–11].

The assay first distributed by Chiron was the Procleix duplex assay, a multiplex assay that detects HIV-1 and HCV. The duplex assay remains in use in some countries, and is generally performed in a minipool format of 16 samples. More recently, the Ultrio assay, which detects not only HIV and HCV RNA but also HBV DNA, has been introduced in some countries [8,15]. Reasons for the introduction of this assay are a policy decision to include HBV DNA testing as part of routine blood donor screening and/or a decision to gain access to the automated instrumentation available with this newer assay. While Ultrio can be performed in minipools, current use is almost exclusively in the ID NAT format, with the exception of ongoing clinical trials in the USA that are evaluating minipools of 8 samples. Both Procleix and Ultrio use the method of transcription mediated amplification (TMA) to amplify viral nucleic acid via an isothermal reaction.

The initial Roche assays were the AmpliScreen assays in which HIV, HCV, and subsequently HBV were detected in separate reactions (i.e. a testing laboratory would use the same sample tube but would run one or more of the viral assays, depending on their policy for donor screening). In the USA, these assays are run in pools of 24 samples [16,17]. More recently, the Cobas s201 system has been introduced, which consists of a multiplex assay for all three viruses, usually run in pools of 6 [18]. In addition to detecting HIV-1, the manufacturer has designed the assay to be able to detect HIV-2. The Roche assays use the well-established polymerase chain reaction (PCR) method to amplify nucleic acids. The Cobas s201 system employs a kinetic PCR using the TaqMan methodology.

The discriminatory assays used in the Procleix systems (duplex or Ultrio) can be run on the same testing platform as the primary screening assay. The assays are almost identical to the multiplex screening assay, with the exception that only one viral-specific detection probe is included in each discriminatory assay. The sensitivity of the discriminatory HIV and HCV assays has been shown to be equivalent to the screening multiplex assay [19]. In the Roche system, reactivity on the multiplex assay is currently being discriminated by use of the three viral-specific AmpliScreen assays, which require a different assay platform. In contrast to the Procleix systems where the screening and discriminatory assays have equivalent sensitivities, the Roche s201 multiplex assay has slightly greater sensitivity than the Roche AmpliScreen assays for individual viruses, if both test systems are performed in the ID format. In practice, however, the use of ID AmpliScreen tests to discriminate reactive multiplex test results (originally obtained from resolution of a six member minipool) is likely to be a reasonable approach in that the viral concentration is six times higher in IDs than in the screening pool, thereby compensating for the slightly lower AmpliScreen assay sensitivity. Currently, there are no large-scale published studies to substantiate the reliability of this proposed discriminatory algorithm. If all AmpliScreen assays are negative despite a reactive multiplex test, concern about adequate AmpliScreen sensitivity can be further addressed by running multiple replicates to enhance the assay's sensitivity. If the results of all three AmpliScreen assays remain negative, it may also be necessary to run an additional in-house or research NAT to exclude window period HIV-2 infection in jurisdictions where there is a sufficiently high incidence of HIV-2 infection.

Initially, each of the commercial assays had several manual steps. The newer assays are considerably more automated. When run in ID format, the Ultrio assay can be run using a single automated instrument (Tigris), which performs sample extraction, amplification and detection [18]. The Cobas s201 system is a modular system that uses three pieces of equipment: the Hamilton Microlab STAR pipettor, the COBAS Ampliprep for sample extraction and the COBAS TaqMan (CTM) Analyser for sample amplification and detection [18].

Comparison of assays and assay systems

  1. Top of page
  2. Evolution of technology and testing
  3. Comparison of assays and assay systems
  4. Minipool vs. individual donation testing
  5. References

Two major studies have recently been performed that directly compared the Ultrio and Cobas s201 assays in the formats in which they are likely to be used: ID testing for Ultrio vs. minipool testing of 6 samples using the Cobas s201 [18,20]. Both studies concluded that either of these test systems would be acceptable for screening blood donations for HIV, HCV, and HBV nucleic acids. More detail is available from the one published study in which Australian investigators tested approximately 10 000 samples obtained from Hong Kong blood donors [18]. This study showed similar analytic sensitivity, that is, the detection of a given number of international units (IU) of viral nucleic acid per ml of plasma at a 95% limit of detection (LOD), for HIV and HBV when both assays were conducted in the ID format but an enhanced sensitivity of the Ultrio assay for HCV. However, since the Cobas s201 assay will routinely be conducted in pools of 6, the authors noted that the analytic sensitivity of the Ultrio assays for all three analyses will be superior when used in actual blood donor screening. Their conclusion is subject to the caveat that they did not evaluate genotype variation, since only one standard was assayed for HCV, one for HBV, and two for HIV. The authors found no difference in the two assay systems with regard to clinical sensitivity; a total of four HBV DNA-positive donations were detected (all were occult HBV infection with low viral loads and positive anti-HBc tests), two by each assay. In an additional assessment, the Ultrio assay when used in pools of 4 or 8 was less sensitive in detecting HBV yield cases than when used in ID format. A similar result has been found in other studies [15,20], leading to the conclusion that HBV DNA screening may require ID testing with Ultrio, if the aim is maximal detection of low level viraemic units. As expected with this relatively small study of ~10 000 donor samples, the Australian study did not detect any HIV or HCV yield cases; therefore, the clinical sensitivity of the two assays with regard to these analysts could not be determined. Theoretically, the Ultrio assay in the ID format has been calculated to reduce the window period for HIV and HCV several days more than would the Cobas s201 assay in pools of 6 [20]; however, since most countries have a low incidence of HIV or HCV seronegative infections in blood donors, the differential yield of using Ultrio in place of Cobas s201 would be expected to be very small.

Other studies have compared the analytic sensitivity of these newer assays to that of the older assays (Procleix duplex, AmpliScreen) that are still in use. In two studies, Ultrio analytic sensitivity was found to be comparable to Procleix for HIV and HCV [15,21]. With regard to the Roche reagents, the s201 system in pools of 6 has an enhanced analytic sensitivity for HIV, HCV, and HBV, when compared to AmpliScreen performed in pools of 24 [16,17].

The reported analytic sensitivity of various assays is given in Table 1.

Table 1.  Reported limits of detection of various NAT blood donor screening assays in IU/mla,b
AssayStudy authorHIVHCVHBV
50% LOD95% LOD50% LOD95% LOD50% LOD95% LOD
  • a

    The 50% and 95% limits of detection were determined by probit analysis of dilution series of WHO international standards or PeliCheck panels (Katsoulidou study [21]) whose values were converted from geq/ml to IU/ml. Numbers in parentheses represent the 95% confidence intervals of the point estimates when reported by the study authors.

  • b

    Older studies by Giachetti et al. [19] reporting on the ProCleix duplex, HIV discriminatory and HCV discriminatory assays and Lelie et al. [25] reporting on the ProCleix duplex and AmpliSceen HIV and HCV assays are excluded from this table, because results were reported in geq/ml rather than IU/ml.

  • c

    Two separate HBV standards were assayed; one from WHO and one from the Paul Ehrlich Institute.

  • LOD, limit of detection; NA, not applicable; NR, not reported.

UltrioKoppelman et al. [15]4·5 (3·2–6·3)24 (15–53)0·9 (0·7–1·0)4·6 (3·7–4·5)1·9 (1·5–2·6)11·0 (7·3–22)
     1·1 (0·7–1·5)c 8·5 (4·9–21) c
UltrioKatsoulidou et al. [21]5·1 (3·6–7·1)31·6 (19·4–68·9)0·9 (0·7–1·3)5·5 (3·5–11·3)2·6 (1·1–6·9)19·0 (7·1–337·1)
UltrioMargaritis et al. [18]6·1 (4·3–8·8)42·2 (24·8–99·3)0·9 (0·7–1·3)2·0 (1·4–7·4)1·8 (1·3–2·6)12·2 (7·3–29·2)
UltrioAssal et al. [20]5·337·70·98·81·912·3
Procleix dHIVKoppelman et al. [15]3·4 (2·4–4·7)26 (16–58)NANANANA
Cobas s201Margaritis et al. [18]6·9 (4·8–10·0)50·5 (29·6–118·2)1·4 (1·1–2·0)6·0 (3·9–12·5)1·3 (0·9–1·8) 8·4 (5·0–21·7)
Cobas s201Assal et al. [20]7·741·42·315·40·7 3·5
HBV AmpliScreenKleinman et al. [17]NANANANANR 5

Minipool vs. individual donation testing

  1. Top of page
  2. Evolution of technology and testing
  3. Comparison of assays and assay systems
  4. Minipool vs. individual donation testing
  5. References

Previously with minipool NAT screening, a donation was only classified as NAT reactive if it gave two independent reactive results: firstly as part of a reactive minipool and subsequently on pool resolution testing of the individual donation. Because of dilutional factors, the initially reactive (IR) minipool result would only be obtained if the viral nucleic acid concentration was far above the LOD of the ID assay. Any truly infected unit that gave a reactive result on minipool testing would also be expected to yield a reactive result on ID testing and on one of the discriminatory viral assays. Thus, if a reactive minipool yielded a reactive individual donation that was negative by discriminatory testing (termed a non-discriminatory reactive or NDR), it would be predicted to be a false-positive result. This prediction was substantiated in a large study of NDR donors identified by primary screening with the Procleix duplex assay in pools of 16 or 24 over a 3-year interval [22]. Additional NAT performed on the index sample and/or NAT and serology performed on follow-up samples in 462 NDR donors confirmed that all were false positives. The rate of such false-positive samples detected through minipool testing has been demonstrated to be very low; American Red Cross data show a false-positive rate of 1 in 40 000 for the Procleix assay in pools of 16, a rate much lower than that found in serologic screening assays [1]. This high assay specificity (i.e. low false-positive rate) is a consequence of the same sample needing to give two false-reactive results (one during minipool testing and one on ID resolution testing) in order to be classified as reactive [23]. These data indicate that the donors of such units do not need to be notified of their results nor prohibited from future donation. It may even be safe to transfuse such units provided that mathematical modelling demonstrates no increased risk from their use (see below). Despite these considerations, current regulatory policy in the USA is to discard the unit and indefinitely defer the donor [22].

As mentioned above, ID NAT is now being used more often due to its ability to detect low levels of HBV DNA and/or to allow the user access to automation. Screening with ID NAT has introduced additional complexity related to determining whether the IR NAT result (assuming that serology for all viruses is negative) represents detection of a new viral infection or a false-positive result. It has been demonstrated that false-reactive NAT screening results occur much more commonly with ID NAT as compared to minipool NAT screening; this is because a sample is classified as reactive after a single result rather than after two reactive results when initially screened in minipools. Several investigators have reported high rates (0·13 to 0·85%) of IR, non-discriminated (NDR) or IR, non-repeatable reactive (NRR) when ID NAT screening is performed [14,24].

An algorithm for determining whether an IR ID NAT result is a true or false positive is needed to avoid the loss of large numbers of donations and the deferral of large numbers of blood donors that could result from conservative regulatory policies applied to IR NAT results on ID screening. The simplest approach of using a single repeat screening NAT assay result or the results of discriminatory assays of equal or lesser sensitivity has an inherent difficulty, if the positive donation has a low viral load, as can occur in very early window period infection with any of the three viruses or in occult late HBV infection. As a consequence of stochastic sampling at low viral concentrations (i.e. the small volume sampled will sometimes not contain any viral nucleic acid), a donation at an assay's LOD will sometimes but not always be detected by the same assay or another assay with similar analytic sensitivity. This theoretical prediction has recently been experimentally verified in a study that used samples of HIV-, HCV-, and HBV-infected patients on antiviral therapy to simulate the low viral loads that are present in blood donors with window period infection [21]. In this study, approximately half of the HIV- and HCV-infected patients and 16% of the HBV-infected patients detected by the Ultrio assay were not detected by the respective discriminatory assays. Therefore, a negative test result on a repeat NAT screening assay or on a NAT discriminatory assay does not definitively rule out infection.

These considerations have lead to several proposed algorithms involving additional testing to evaluate and further classify ID NAT IR donations and donors. Currently, there is no standardized algorithm used internationally. Approaches used in different countries include using the results of a repeat NAT assay and a discriminatory test (Ultrio and the dHxV assays – equivalent to two NAT replicates), repeating the NAT assay in duplicate without running the discriminatory assays (equivalent to two NAT replicates), and repeating the NAT assay in duplicate combined with testing by the discriminatory assays (equivalent to three NAT replicates). Such testing may be performed on the same sample source or may use an alternative sample source (e.g. the plasma unit) from the same donation. Mathematical modelling involving viral incidence rates, assay sensitivity at the 50 or 95% LOD, and the length of the probable infectious window period can be used to calculate the likelihood that an NDR or NRR unit tested by a particular algorithm will be infectious [24]. Although it is possible that in some jurisdictions such calculations may allow for transfusion of the donation with the IR ID NAT result, it is more likely that such modelling will justify using a particular testing algorithm to establish that donors with such results do not need to be notified of their results and can continue to donate. The most extensive dataset to evaluate the use of such algorithms has been obtained from the ID NAT testing programme in South Africa, which includes follow-up NAT and serology results from ID NAT IR donors in order to more definitively classify the index donations [24]. Analysis of these data should provide important insights.

References

  1. Top of page
  2. Evolution of technology and testing
  3. Comparison of assays and assay systems
  4. Minipool vs. individual donation testing
  5. References