• Open Access

Hepatitis B virus testing by minipool nucleic acid testing: does it improve blood safety?


  • Susan L. Stramer,

    Corresponding author
    1. Scientific Support Office, American Red Cross, Biomedical Services, Gaithersburg, Maryland
    2. Holland Laboratory, American Red Cross, Biomedical Services, Rockville, Maryland
    • Address reprint requests to: Susan L. Stramer, PhD, Scientific Support Office, American Red Cross Biomedical Services, 9315 Gaither Road, Gaithersburg, MD 20877; e-mail: Susan.stramer@redcross.org.

    Search for more papers by this author
  • Edward P. Notari,

    1. Scientific Support Office, American Red Cross, Biomedical Services, Gaithersburg, Maryland
    2. Holland Laboratory, American Red Cross, Biomedical Services, Rockville, Maryland
    Search for more papers by this author
  • David E. Krysztof,

    1. Scientific Support Office, American Red Cross, Biomedical Services, Gaithersburg, Maryland
    2. Holland Laboratory, American Red Cross, Biomedical Services, Rockville, Maryland
    Search for more papers by this author
  • Roger Y. Dodd

    1. Scientific Support Office, American Red Cross, Biomedical Services, Gaithersburg, Maryland
    2. Holland Laboratory, American Red Cross, Biomedical Services, Rockville, Maryland
    Search for more papers by this author



Hepatitis B virus (HBV) DNA-positive yield since nucleic acid testing (NAT) implementation (minipools of 16 [MP16]) was reported for the first year. We have updated those figures, evaluated the current value of all HBV tests, calculated the HBV residual risk before and after the introduction of MP-NAT, and estimated residual risks with further improvements in HBV screening for US blood donations.

Study Design and Methods

All donations were screened by US-required serologic HBV tests and for HBV DNA by MP-NAT (Novartis/Gen-Probe). Further testing by individual-donation polymerase chain reaction (ID-PCR) confirmed various classes of MP-NAT–reactive or –nonreactive donations. The hepatitis B surface antigen (HBsAg)-yield method was used to calculate incidence and the incidence–window-period model used to define residual risk.


Of approximately 12.8 million donations screened during 2009 to 2011, a total of 1368 HBV confirmed positives including 941 by MP-NAT were observed (combined 4.32% positive predictive value) of which five were seronegative NAT-yield donations (1:2.6 million) and 25 HBsAg-yield (anti-HBc–nonreactive) donations from which an incidence of 1.62/100,000 person-years (vs. 3.43 during 2006-2008) and residual risk of 1:592,000 to 1:754,000 were calculated. With the addition of MP-NAT, and resulting 8.8-day window-period reduction, residual risks decreased to 1:765,000 to 1:1,006,000. Of the 1368 positives, 99.6% were detected by serology and 68.8% by MP-NAT; ID-PCR detected 427 more infected donors than MP-NAT.


HBV MP-NAT and decreases in HBV incidence (likely vaccine-related) in the United States have reduced residual risks to levels comparable to those of human immunodeficiency virus and hepatitis C virus and raise the question of the continued need for all three HBV markers for blood donation screening. Further reductions in residual risk will require the implementation of more sensitive HBV-NAT methods including ID-NAT.


American Red Cross


individual donation




minipools of 16


occult hepatitis B virus infection




signal-to-cutoff ratio.

Over many years, the estimated residual risk of transfusion-transmitted hepatitis B virus (HBV) infection in the United States has been markedly greater than that for hepatitis C virus (HCV) or human immunodeficiency virus (HIV). This has generally been attributed to the length of the hepatitis B surface antigen (HBsAg)-negative window period for HBV, as incidence rates did not differ greatly between the three viruses.[1, 2] HBV incidence rates would be expected to decrease in the donor population due to HBV universal vaccination policies in the United States and vaccinated donors presenting to donate. It has been speculated that the implementation of nucleic acid testing (NAT) for HBV DNA might reduce the window period and thus the residual risk, although early studies demonstrated that there may be little difference in sensitivity between minipool (MP) NAT and sensitive HBsAg tests.[3] More recent studies using US blood donors have shown HBV DNA yield by MP-NAT is low but comparable to that of HIV and HCV and includes detection of previously HBV-vaccinated donors with breakthrough HBV infection whose donations have unknown infectivity.[4, 5]

In June 2009, the American Red Cross (ARC) implemented automated triplex NAT for HIV, HCV, and HBV (Ultrio, Gen-Probe, Inc., San Diego, CA; and Novartis Vaccines and Diagnostics, Emeryville, CA) using MPs of 16 individual donation samples (MP16); Stramer and colleagues[5] reviewed the results of the initial year of testing in which the yield of MP-NAT (Ultrio, Gen-Probe, Inc. and Novartis Vaccines and Diagnostics) had little measurable impact on blood safety in detecting donations that were nonreactive by routine serologic markers of HBV (i.e., HBsAg and antibodies to hepatitis B core antigen [anti-HBc]). Other studies have shown that sensitive, individual-donation (ID)-HBV NAT detects significantly more DNA-positive, HBsAg-nonreactive donations than does MP-NAT, regardless of the anti-HBc status of the donor.[6, 7]

HBV residual risk estimates were last reported in 2009 by the ARC for blood donations in the United States as calculated by two different methods. The first was the traditional method where repeat donor incidence was estimated by seroconversion using a correction factor to adjust for the transient nature of HBsAg and then divided by the person-time of observation; incidence was then multiplied by the residual window period to determine risk. The second was a novel method where incidence in all donors was estimated from the number of HBsAg-confirmed-positive donations that were nonreactive by anti-HBc and thus termed HBsAg-yield donations; risk was then calculated by the incidence–window-period model. The advantage of the HBsAg-yield method is that incidence calculations for all donors are possible, and that for HBV, an incidence correction factor for the transient nature of HBsAg is not needed. Both methods produced similar results with residual risk for HBV before the implementation of MP-NAT estimated at 1 in 280,000 to 1 in 357,000 donations.[1] No such residual risk estimates have been performed since the implementation of MP-NAT in the United States.

A second year of testing by the ARC defined the positive rates for all three HBV markers. This has permitted further evaluation of the yield of seronegative, HBV DNA–positive donors detected by MP-NAT (Ultrio, MP16), as well as those donors who are HBsAg positive in the presence or absence of HBV DNA detected by MP-NAT. Using these determinations, we have updated HBV residual risk estimates based on HBsAg incidence measures and the observed yield of HBV MP-NAT, as well as projecting further reductions in residual risk by use of a newer version of Ultrio (Ultrio Plus) with improved HBV DNA detection and ID-NAT.[7]

Materials and Methods


During the period of July 1, 2009, through June 30, 2011, all donations to the ARC were tested by MP-NAT (Ultrio). The term MP-NAT will be used for such testing for the remainder of this article. In those cases where samples were tested individually by Ultrio, the term ID-NAT will be used. The procedures for triplex (HIV-1, HCV, and HBV) MP-NAT by Ultrio including reactive pool resolution and specific virus discrimination have been previously described.[4, 5] In summary, testing followed the previously published protocol for resolution of MP-NAT–reactive pools, HBV discrimination, and further confirmatory testing. In addition to MP-NAT, all donations were routinely tested for HBsAg and anti-HBc (PRISM ChLIA, Abbott Diagnostics, Abbott Park, IL). All donations testing repeat reactive for HBsAg were confirmed by PRISM neutralization as previously described. For this study, any sample that was repeatedly reactive by one or both of the two serologic tests and HBV DNA positive by discriminatory NAT (Ultrio dHBV) after MP-NAT reactivity was not further tested. Seronegative samples testing HBV ID-NAT–reactive (Ultrio), as well as HBsAg-reactive samples testing MP-NAT or anti-HBc nonreactive were further tested individually for HBV DNA by reverse transcription–polymerase chain reaction (RT-PCR; UltraQual 1000, National Genetics Institute, NGI, Los Angeles, CA). Lastly, all anti-HBc repeatedly reactive donations that were HBsAg and MP-NAT nonreactive were further tested individually for HBV DNA by RT-PCR (COBAS Ampliscreen Multiprep 1-mL sample extraction tested in triplicate, Roche Molecular Diagnostics, Pleasanton, CA). ID-NAT using PCR methods as outlined above will be termed ID-PCR. Viral load determinations were performed for all HBV ID-PCR–positive samples regardless of serologic status (SuperQuant, NGI). The 95% lower limit of detection for HBV DNA using Ultrio is 10.4 IU/mL (42-52 copies/mL),[4, 7-9] and that of the Ampliscreen multiprep assay when performed in triplicate, and that of UltraQual is 1 to 2 IU/mL (4-10 copies/mL).[4, 7, 10] Viral loads are expressed in copies/mL with the SuperQuant assay having a lower limit of quantitation of 100 copies/mL.[4] A factor of 4 to 5 copies/IU was used for all conversions from IU/mL to copies/mL.[4, 9] HBV DNA–yield donors were also tested for quantitative levels of hepatitis B surface antibody (anti-HBs; Monolisa, Bio-Rad Laboratories, Redmond, WA) as an indicator of HBV vaccine status.[4] To minimize NAT false positivity, all ID-PCR was performed from an independent sample (the retrieved frozen donation plasma component); all routine MP-NAT was performed using a dedicated sample contained in a plasma preparation tube (Becton Dickinson, Franklin Lakes, NJ) that was not used in other areas of the screening laboratory. All HBV DNA–reactive, seronegative donors identified through MP-NAT were retested by Ultrio dHBV in replicates of 10 using the retrieved frozen donation plasma component.

The number of blood donations identified as reactive versus nonreactive or positive versus negative by each test for a specific marker was determined for the targeted period. The number of donations was then compiled according to the testing analysis flow presented in Fig. 1. Donors having seronegative donation samples that were MP-NAT reactive and confirmed by HBV PCR, and whose retrieved plasma demonstrated repeat dHBV reactivity, were termed HBV DNA–yield donors; those who had HBsAg confirmed positivity in the absence of anti-HBc reactivity and were confirmed as DNA positive were termed HBsAg-yield donors. For the purposes of this study, a donor was defined as “HBV infected” on the basis of the confirmed presence of HBV DNA by the methods stated above and outlined in Table 1.

Figure 1.

HBV screening and confirmatory testing algorithm and results leading to 1368 infected ARC donors, July 1, 2009, to June 30, 2011. *Tested donations are listed as repeat reactive (RR), nonreactive (NR), or reactive (if repeat testing is not required for the standard screening algorithm). Each subsequent test in the presented algorithm eliminates any reactives in the prior step (see Materials and Methods). MP-NAT reactivity included discriminatory HBV reactivity. **See Results and Table 2. Briefly, the 1613 classified as not infected consisted of 1440 donors with unconfirmed (negative) neutralization results, 125 with neutralized-positive, low-level HBsAg reactivity (S/CO ≤ 2.0) including 124 ID-PCR negative (and one not tested); 50 of the 125 were recent HBV vaccinees and 75 had HBsAg and/or neutralization results that were not reproducible when retested. Finally, there were 48 HBsAg-reactive donors where neutralization and ID-PCR were not performed; all were MP-NAT nonreactive, 41 of whom had low-level HBsAg reactivity (S/CO ≤ 2.0) and were anti-HBc nonreactive (and thus considered false positive) and seven who were anti-HBc reactive. These seven ID-PCR not-tested donors plus the one ID-PCR, neutralization retest negative, anti-HBc nonreactive donor (above) had inadequate data to be considered true positive (infected). The 28,548 anti-HBc RR donors classified as not infected all were HBsAg nonreactive and MP-NAT nonreactive; in addition, 27,138 were ID-PCR negative and 1410 ID-PCR untested. All 171 MP-NAT reactives classified as not infected were ID-PCR negative, and nonreactive in 10-replicate testing from the retrieved frozen plasma unit (i.e., all 10 replicates were nonreactive). ***PPV = 1368/31,700 (total of all reactives) × 100 = 4.32%.

Table 1. Reactivity defining HBV-infected donors
+++Not done
++ or +
++Not done
Table 2. All testing results for HBV markers in 12.8 million ARC blood donations, July 1, 2009, to June 30, 2011*Thumbnail image of

Determination of incidence and residual risk

Incidence density for HBV was determined by using the previously described HBsAg-yield approach; the yield was defined as the number of samples that were confirmed as HBsAg positive, but anti-HBc nonreactive, as described previously.[1] Yield rate for HBV DNA was defined as the number of seronegative samples that were confirmed as HBV DNA positive by MP-NAT divided by the total number of donations. The incidence rate was determined by dividing the HBsAg-yield rate by the marker-positive window period (yield window) of 44 days (0.12 year), as defined previously.[1] Residual risk was determined by multiplying the calculated incidence rate per 100,000 person-years (pys) by the test-negative window period (in years), using the published figures of 38 or 30 days (0.104-0.082 years) for HBsAg depending on an infectious dose of 1 or 10 HBV DNA copies.[11] The impact of MP-NAT on residual risk calculated from HBsAg-based incidence was determined by subtracting the number of HBV DNA yield donations from the estimated number of donations from infected donors derived from the HBsAg-based risk calculation.

Approximations of 95% confidence intervals (CIs) were obtained through Monte Carlo simulation using computer software (Crystal Ball, Decisioneering, Denver, CO). A triangular distribution was assumed for a variable if no individual data were available for distribution fitting or if the distribution of the variable was uncertain. The upper and lower limits of the triangular distribution for the 44-day yield window were 61.2 and 27.1 days; those for the 38-day window period were 43.7 and 33.0 days, respectively, and those for the 30-day window period were 35.0 and 25.0 days, respectively.[1] Other 95% CIs were derived from normal distributions or, in the case of the NAT-yield window period, was extrapolated from the HBsAg-yield window and the ratio of NAT-yield to HBsAg-yield donors.[1]


During the 2-year study period, 12,772,651 allogeneic donations including all donation types (e.g., whole blood, pheresis) were tested. Of these, 12,740,951 (99.75%) were nonreactive in all tests, while a total of 1368 donors (0.011% or 1:9337) had one or more HBV-reactive screening results and confirmed as HBV DNA positive (see Materials and Methods). The overall testing scheme and results by HBV marker are summarized in Fig. 1 where reactivity to the three HBV screening markers is presented sequentially starting with HBsAg, where all HBsAg-reactive donors are included (2703); next anti-HBc–reactive donors that are HBsAg nonreactive are included (28,821). Finally, only those donations that are HBV DNA (dHBV) reactive derived from MP-NAT are included (176). The 1613 HBsAg-false-positive donors consisted of 1440 with unconfirmed (negative) neutralization results (1422 who were anti-HBc nonreactive and 18 who were anti-HBc repeat reactive but ID-PCR negative), 125 with neutralized-positive but low-level HBsAg reactivity (signal-to-cutoff ratio [S/CO] ≤ 2.0) of which 124 were ID-PCR negative (one not tested); 50 of the 125 were recent HBV vaccinees and 75 had HBsAg and/or neutralization results that were not reproducible when retested. Finally, there were 48 HBsAg-reactive donors where neutralization and ID-PCR were not performed; all were MP-NAT nonreactive, 41 of whom had low-level HBsAg reactivity (S/CO < 2.0) and were anti-HBc nonreactive (and thus considered false positive) and seven who were anti-HBc reactive. These seven ID-PCR not-tested donors plus the one ID-PCR, neutralization retest negative, anti-HBc–nonreactive donor (above) had inadequate data to be considered true positive (infected). The 28,548 anti-HBc–repeat-reactive donors classified as not infected all were HBsAg nonreactive and MP-NAT nonreactive; in addition, 27,138 were ID-PCR negative and 1410 ID-PCR untested. Of the 176 MP-NAT (Ultrio and dHBV)-reactive, seronegative donors, 171 (97%) were classified as false positive as the result of additional testing. All 171 were ID-PCR negative at index (NGI) and nonreactive in 10-replicate testing from the retrieved frozen plasma unit (i.e., all 10 replicates were nonreactive). When taken together, HBV screening using three markers for the 2-year period had 99.76% specificity and a positive predictive value of 4.32% for HBV infection, including a total of 31,700 reactive donations of which 1368 were classified as currently infected and 30,332 donors who were classified as not currently infected.

Figure 2 shows the overall distribution of reactivity by all HBV screening tests. Of those shown, the donors in overlapping areas of MP-NAT and serology include 936 (25 + 876 + 35) of the 1368 HBV-infected donors. Within each remaining cell, further testing was required to determine their HBV infection status. The results of final confirmatory testing including ID-PCR are shown in Fig. 3 where 171 seronegative, MP-NAT–false-positive donors, 25 anti-HBc– and HBsAg–false-positive donors, 1588 HBsAg-only–false-positive donors, and 28,548 donors testing only anti-HBc false positive were eliminated. The breakout of all HBV testing for the entire donor base plus all additional confirmatory testing is shown in Table 2.

Figure 2.

HBV reactivity from screening 12.8 million ARC donations resulting in 31,700 reactive donors, July 1, 2009, to June 30, 2011.

Figure 3.

HBV reactivity of 1368 infected ARC donations, July 1, 2009, to June 30, 2011. Defined as HBV MP- or ID-NAT reactive (see Materials and Methods and Results).

Of the 1368 HBV-infected donors, 1363 (99.6%) had serologic reactivity to one or both tests regardless of NAT status; in contrast, 936 (68.8%) had reactivity by MP-NAT regardless of serostatus; the remaining 427 donors were defined as HBV infected based on ID-PCR reactivity. Notably, there were only five MP-NAT–yield donors (1:2,554,530) and 25 HBsAg-yield donors (1:510,906), or a total of 30 donors with early acute infection (1:425,755). Of the NAT-yield samples, two were previously described during the 2009 to 2010 period having viral loads of 100 to 4100 copies/mL and both were anti-HBs nonreactive.[5] Similarly, the three additional HBV DNA–yield donations identified during the 2010 to 2011 period had viral loads ranging from 100 to 4400 copies/mL with two of three testing anti-HBs nonreactive while the third had borderline anti-HBs (6.86 mIU/mL). The 25 HBsAg-yield samples had viral loads ranging from 3300 to 380 million copies/mL.

When the overall data for the 2-year period are compared with those from the 2009 to 2010 period,[5] it is clear that there is no appreciable difference in terms of the rates of detection of HBV markers, whether singly or in combination (Table 3). It continues to be clear that ID-PCR detected a significant proportion of HBV-infected donors (941 detected by MP-NAT of the total 1368, 68.8%; Table 2). In fact, among the 273 donations that were classified as having chronic occult HBV infection (OBI), that is, anti-HBc reactive, HBsAg-nonreactive with detectable HBV DNA, only 35 (12.8%) were identified by MP-NAT (Tables 2 and 3). The remaining 238 (87.2%) anti-HBc–reactive donors who tested MP-NAT nonreactive and ID-PCR positive had viral loads of 100 to 800 copies/mL. In total, 0.91% of all anti-HBc–reactive donations contained HBV DNA.

Table 3. Comparison of HBV-marker rates for ARC donations, July 1, 2009, to June 30, 2010, versus July 1, 2009, to June 30, 2011
 2009-2010 (n = 6,459,047 tested)2009-2011 (n = 12,772,651 tested)
  1. aRate per 100,000 donations.
  2. bAll entries are positive by MP-NAT and discriminatory (dHBV) and/or ID-PCR positive for HBV DNA; see Materials and Methods and Results.
  3. cRepeat reactive.
Confirmed infectedb69910.81,36810.7
MP-NAT positive (total)4777.49417.4
MP-NAT seronegative yield20.0350.04
HBsAg yield (MP-NAT positive, anti-HBc nonreactive)100.15250.20
HBsAg positive and anti-HBc RRc (MP-NAT nonreactive/ID-PCR positive)1001.541871.46
Anti-HBc only (HBsAg and MP-NAT nonreactive/ID-PCR positive)1211.92381.9
Anti-HBc only and MP-NAT positive210.33350.27

Of the 1368 HBV-infected donors, 876 (64%) had reactivity to all three HBV markers (Figs. 2 and 3). As described previously, not all HBsAg-positive and anti-HBc–reactive donors were Ultrio HBV-NAT reactive after MP-NAT; of the 1063 seropositive donors reactive to both HBsAg and anti-HBc (876 + 187, or 77.7% of the total 1368 HBV-infected donors), 187 (17.6%) were nonreactive by MP-NAT and only detected by ID-PCR with viral loads of 100 to 1900 copies/mL. In addition, there were two HBsAg-positive and anti-HBc–nonreactive donors who were also MP-NAT nonreactive but ID-PCR reactive. One of these two donors was previously described and had five copies HBV DNA/mL when initially tested, but was nonreactive when retested from the retrieved plasma unit by ID-PCR and 10-replicate discriminatory Ultrio HBV NAT; in addition, the donor was HBsAg nonreactive when retested, but close to the assay cutoff (retrieved plasma S/CO of 0.94 vs. index S/CO values of 1.73-1.77 for the initial and duplicate retests). The donor was interpreted as testing HBV PCR falsely positive.[5] A second HBsAg-positive donor was identified during 2010 to 2011 who also tested nonreactive by anti-HBc and MP-NAT but was ID-PCR positive with fewer than 100 copies/mL. Similarly, ID-PCR and 10-replicate discriminatory Ultrio HBV ID-NAT of the retrieved plasma unit failed to identify HBV DNA reactivity. However, low-level confirmed-positive HBsAg was observed in this donor when retested from the independent sample (S/CO values less than 2.0; same as the original index test results). Even so, this donor was also interpreted as HBsAg false positive. Likely, both of these low-level, HBsAg–neutralized-positive donors had some type of biologic false positivity directed to components in the PRISM assay.[1]

Table 4 outlines the estimated incidence rate for HBV infection within the donor population, with a point estimate of 1.62 per 100,000 pys (95% CI, 0.93-2.68) based on HBsAg yield. The incidence rate last published on the basis of data from November 2006 to July 2008 using the HBsAg-yield method was provided for comparison (3.43/100,000 pys; 95% CI: 2.18-5.35). Table 5 gives point estimates and 95% CIs of residual risk, based on the incidence estimates provided in Table 4 and window periods of 30 and 38 days for HBsAg, respectively, reflecting an infectious dose of 10 or 1 copy of DNA per 20 mL.[11] In the absence of HBV MP-NAT, we have estimated that the residual risk would be 1 in 592,000 to 1 in 754,000 donations depending on the selected window period, with respective 95% CIs of 1:1,031,000 to 1:350,000 and 1:1,351,000 to 1:452,000. In comparison, during the last published period from November 2006 to July 2008, before the implementation of HBV MP-NAT and using an HBsAg incidence of 3.43 per 100,000 pys calculated using the HBsAg-yield method, the estimated residual risk was 1 in 280,000 to 1 in 355,000 (overall 95% CI, 1:180,000-1:440,000).

Table 4. Calculation of incidence rates using the observed HBsAg-yield approach and ARC donations from June 30, 2009, to July 1, 2011, compared to an earlier period
StudyTotal number of donations (a)aYield (b)Yield rate (/100,000) (c = b/a)Yield window (years) (d)Incidence (/100,000 pys) estimate (c/d)Incidence (/100,000 pys) 95% CI
  1. aLetters in parentheses identify the numbers in columns for further calculations; e.g., yield (b) divided by the number of donations (a) equals yield rate (c).
HBsAg incidence for a previously published period (November 2006-July 2008)[1]10,156,237420.41444/365 = 0.123.432.18-5.35
HBsAg incidence for the observed period (July 1, 2009-June 30, 2011)12,772,651250.19644/365 = 0.121.620.93-2.68
Table 5. Calculation of residual risk using the observed HBsAg-yield approach and ARC donations from June 30, 2009, to July 1, 2011, compared to an earlier period
StudyIncidence (/100,000 pys) (a)aWindow periodRisk per 100,000 donations (c = a × b)Risk estimate (1/c × 105)95% CI
In daysIn years (b)Lower limitUpper limit
  1. aLetters in parentheses identify the numbers in columns for further calculations; e.g., incidence (a) multiplied by window period (b) equals risk (c).
HBsAg incidence for a previously published period (November 2006-July 2008)[1]3.43380.1040.3571:280,0001:440,0001:180,000
HBsAg incidence for the observed period (July 1 2009-June 30 2011)1.62380.1040.1691:592,0001:1,031,0001:350,000

The current residual risk of 1 in 592,000 to 1 in 754,000 using HBsAg projects to 21.7 to 17.1 infectious (yield) donations that remain undetected based on 12,772,651 tested donations during the 2-year period. By subtracting the five observed MP-NAT–yield donations from these figures, the number of donations that remain undetected by MP-NAT may be projected; this calculates to 12.1 to 16.7 yield donations remaining after MP-NAT or a residual risk with MP-NAT of 1 per 1,006,000 to 1 per 765,000. Thus, the HBV residual risk declined due to both decreases in incidence and improvements afforded by the implementation of MP-NAT. An estimate of the window-period reduction attributable to MP-NAT was calculated from the ratio of the 25 HBsAg-yield (95% CI, 15.2-34.8)/five HBV DNA NAT–yield (95% CI, 0.62-9.38) donors identified in this study multiplied by the 44-day (95% CI, 27-61)[1] HBsAg-yield window, giving a point estimate of 8.8 days (95% CI, 3.0-14.1) as the HBV MP-NAT–yield window or the window-period reduction attributable to MP-NAT.


Over the 2-year period of July 2009 to June 2011, almost 13 million blood donations were collected and tested within the ARC system including tests for HBsAg, anti-HBc, and HBV DNA among other infectious disease markers. Testing for HBV DNA was performed by MP-NAT using an automated triplex test (Ultrio MP16). Data from the first year of testing have been published[5] and it is apparent that there has been no measurable difference in the results of testing over the second year as shown by the comparison of the previous and current results (Table 3). The conclusions drawn in the previous publication have not changed, with one possible exception, which is that we have been able to demonstrate a measurable impact of MP-NAT in combination with reduced incidence rates (likely vaccine driven) on the estimated residual risk for HBV even though the yield has been only 1 per 2.6 million donations with only one vaccine-breakthrough infection observed, again considerably lower than that reported during the validation study (1:410,540) with six vaccine-breakthrough infections observed.[4] The difference may be attributed to variability associated with rare events or to the fact that the ARC and non-ARC validation sites were selected based on locations we believed would have the highest rates of HBV infection (i.e., urban and the southern United States). We further examined the impact of variability in NAT-yield number and frequency of vaccine breakthrough by reviewing the next annual period (2011-2012); during that time eight HBV NAT–yield donors of 6.01 million ARC donations (1:751,250) were identified of which two were characterized as vaccine-breakthrough infections. Thus, variability attributed to low-frequency events, coupled with site selection, is the most likely explanation for the observed fluctuations year to year.

As previously reported, we continued to see no sample with confirmed detection of HBsAg in the absence of any other marker, further supporting the conclusion that use of HBV DNA MP-NAT might eliminate the need to test donations for HBsAg. However, HBV MP-NAT (Ultrio MP16) failed to detect a high number of HBV-seropositive donors including approximately 88% of OBI donors with viral loads of 100 to 800 copies/mL and nearly 20% of HBsAg-positive and anti-HBc–reactive donors with viral loads of 100 to 1900 copies/mL. These viral loads are below the sensitivity of HBV MP-NAT (Ultrio MP16), which has a 95% detection limit of 10.4 IU/mL × 16 (166 IU/mL) or nearly 700 copies/mL (estimated range, 664-830 copies/mL based on use of conversion factors[4, 9]) and may be below the detection levels of more sensitive HBV NAT methods used in the MP-NAT format having 95% detection limits of 2 to 5 IU/mL (8-25 copies/mL).[7-10] Thus, the combination of HBV MP-NAT plus anti-HBc in the absence of HBsAg results would clearly decrease the accuracy of donor messaging, cause confusion for donor counseling, and limit relevant information for subsequent referral of HBV-infected and likely false-positive donors into the medical system. In the absence of HBsAg screening, if such testing were to be deemed duplicative based on this and possibly future studies, at minimum, HBsAg testing would be needed for donor counseling. Each facility would have to weigh the cost–benefit of maintaining an HBsAg screening program as part of a total screening platform (e.g., PRISM) versus developing or maintaining diagnostic testing capabilities required for counseling of donors with anti-HBc and HBV DNA reactivity.

This study again emphasizes that nearly 1% (0.91%, 273/29,909) of anti-HBc–reactive donors represent OBI by virtue of the detection of HBV DNA and that only approximately 12% (35/273) of these are detected by the Ultrio MP-NAT, as stated above. The frequency of transfusion-transmitted infection from these OBI cases is unknown, but such transmission does occur.[12-14] Further, it is hypothesized that such infectivity is related to the viral load in blood components from donors with OBI in that infectivity of OBI requires 10- to 1000-fold more virus than that observed during the early, acute window period.[14-16] In a comprehensive lookback study based on retained donation samples found positive for HBV DNA, Satake and colleagues[12] showed that only one of 33 components with low anti-HBc titers could be identified as infectious in recipients, whereas 11 of 22 anti-HBc–negative components proved to be infectious. None of the 16 identified anti-HBs–positive components showed serologic evidence of infectivity. However, a recent study from Europe suggests that the infectivity of OBI donations may be as high as 25% to 100% of transfused products. The number of DNA copies transfused was a major determinant of infectivity, as was the absence of anti-HBs in the product and the patient. All 15 anti-HBs–negative recipients of anti-HBs–negative OBI plasma units had evidence of HBV infection posttransfusion (as assessed by anti-HBc positivity and detected DNA), but none of three recipients of anti-HBs–positive plasma were infected. Viral loads in transmitted cases were 1049 (range, 117-3441) copies. The authors speculated that the lower rates of transmission from OBI donations in Japan likely reflected the fact that blood with higher viral titers were not transfused due to the exclusion of units with high-titer anti-HBc and those that are MP-NAT positive.[14] Thus, due to limitations in HBV DNA sensitivity, the ongoing use of MP-NAT does limit the identification of donors with OBI. However, as pointed out above, our study was not designed to measure the efficacy of ID-NAT, or lack thereof for MP-NAT, in detecting either OBI or early window-period HBV infections. Consequently, our data do not provide adequate information to support routine implementation of ID-NAT for blood donors. However, even with the use of ID-NAT for HBV, as routinely performed in South Africa, a breakthrough transmission from an HBV-infected, window-period donor was documented. Vermeulen and coworkers[17] reported that, among 2.9 million donations screened during the first 4 years of ID-NAT, 114 window-period donations were identified (1:25,627). A single transmission was recognized from a unit of red blood cells, with 99.7% HBV DNA sequence identity in the donor and recipient (Subgenotype A1); the estimated viral load was approximately 32 HBV copies based on 20 mL plasma. The observed risk was thus 1 per 2.9 million or 0.34 per million.

The increased amount of data during the 2-year period of this study permitted us to update the residual risk of HBV infection in the United States, projected from the ARC blood supply, based on incidence estimates determined by the HBsAg-yield method, which was selected for simplicity and because a prior study showed it to be equivalent to measures based on seroconversion.[1] This method gave a current incidence rate of 1.62 cases per 100,000 pys (95% CI, 0.93-2.68). This figure is approximately 47% of the estimate of 3.43 per 100,000 pys (95% CI, 2.18-5.35) that was developed for HBV-infected blood donors from 2006 to 2008, also using the HBsAg-yield method.[1] At that time, using the same methods, the point estimates of residual risk in the absence of MP-NAT were 1 per 280,000 to 1 per 355,000 (95% CI, 1:180,000-1:559,000); the range dependent on the selected window period of 30 or 38 days (related to an infectious dose of 10 or 1 copy/20 mL, respectively).[10] The corresponding figures for 2009 to 2011 were 1 per 592,000 to 1 per 754,000 (95% CI, 1:350,000-1:1,351,000; Table 5). Note that the 95% CIs for the incidence and residual risk figures for the two periods (2006-2008 vs. 2009-2011) do overlap to some extent, even when rates for 30- or 38-day window periods are considered separately. The decrease in incidence since 2006 to 2008 has been the major driver in the reduction of residual risk. While it is reasonable to consider whether this change related specifically to first-time or repeat donors, this does not appear to be the case. For the 2009 to 2011 period, we tested 2.23 million first-time and 10.54 million repeat donors. One of the DNA- and nine of the HBsAg-yield donations were from first-time donors (10 total) and four of the DNA- and 16 of the HBsAg-yield donations were from repeat donors (20 total). The overall frequency of yield donors was 2.4-fold higher among donations from first-time donors than among donations from repeat donors. A comparable ratio for the HBsAg-yield donations (2.41) was previously reported by Zou and coworkers.[1] This ratio likely reflects increased incidence among first-time donors along with the impact of multiple donations among repeat donors; clearly, the ratio has remained stable despite the overall reduction in incidence. A plausible explanation for the reduction in risk may be the impact of HBV vaccination policies in the United States as demonstrated in a prior study where overall 44% of 520 donors were anti-HBs positive with 65% anti-HBs positivity in the 16- to 29-year age group, which represents a significant portion of the donor population.[1]

The contribution of MP-NAT has also been demonstrated, reducing the risk to 1 per 765,000 to 1 per 1,006,000. The addition of MP-NAT has reduced the risk by 23% to 29%, although it must be recognized that these point estimates, although derived from nearly 13 million donations, are based on small numbers of acutely infected donors and thus would be expected to have a wide CI (for example, the 95% CI for 5 is 0.62-9.38 and for 25 is 15.2-34.8). We note that our point estimates of yield rates are based on population figures for donations and observed yield and thus do not strictly require CIs. However, such CIs are appropriate for window-period estimates and thus for risk estimates. They are also appropriate for comparison of data from different periods. The decreased risk afforded by MP-NAT does not exceed the 95% CI for the risk in the absence of MP-NAT (Ultrio MP16), so it cannot be clearly shown that the benefit of MP-NAT is significant even on the basis of 2 years' data. The greatest contribution to the observed decreased risk is attributable to a decrease in the incidence of new HBV infection among donors, continuing a trend that was previously reported in 2009.[1] The current estimates of HBV residual risk after the implementation of MP-NAT of 1 per 765,000 to 1 per 1,006,000 are now comparable to those published for HIV and HCV after the implementation of MP-NAT in the United States (1:1,467,000 for HIV and 1:1,149,000 for HCV).[2] These point estimates used the same methods for residual risk calculations and all assumed a log-linear increase in viral load over time. However, none of the HBV residual risk calculations include adjustments for incident donors who would not be detected due to the absence of HBsAg, such as from donors with vaccine-breakthrough infection.[4]

We have further estimated the potential impact of additional improvements in HBV detection by NAT; that is, the implementation of ID-NAT and/or the recently licensed Ultrio Plus assay (May 25, 2012, licensure date; Gen-Probe/Novartis);[18] these are summarized in Table 6. To estimate these reductions in residual risk, we used window-period estimates derived from the studies of Vermeulen and colleagues.[17] They calculated an infectious window period of 32.5 days to PRISM HBsAg detection (assuming a 50% infectious dose of 3.7 copies). The window period was further reduced to 15.3 days by Ultrio ID-NAT and 12.6 days with Ultrio Plus ID-NAT, a difference of 2.7 days for Ultrio versus Ultrio Plus (or approximately 1 HBV doubling time of 2.56 days).[3, 17] We calculated a point estimate of 8.8 days (25/5 = 44/8.8) for the MP-NAT–yield window before PRISM HBsAg detection. Thus, using these data in combination suggest that MP-NAT with Ultrio, MP-NAT with Ultrio Plus, ID-NAT with Ultrio, and ID-NAT with Ultrio Plus would reduce window periods by modeled point estimates of 8.8, 11.5, 17.2, and 19.9 days, respectively, relative to PRISM HBsAg alone. As pointed out above, these point estimates are subject to broad CIs. It should also be noted that all models appear to overestimate risk for a variety of reasons (e.g., underrecognition or reporting of transfusion transmissions, asymptomatic infections, recipient mortality) and thus do not necessarily correlate with observed transfusion transmission. In the case of Vermeulen and colleagues, the calculated risk of posttransfusion HBV infection with the use of ID-NAT (Ultrio) and window periods mentioned above was 73-fold greater than the observed rate (based on a single HBV transfusion transmission).[17] Regardless, Table 6 does show the relative contribution of improvements that may be afforded by implementing more sensitive HBV NAT methods in the United States (using current rates of incidence). As a point of illustration in this case, the ranges provided reflect uncertainty about the infectious dose of HBV, but those for MP-NAT with either Ultrio or Ultrio Plus overlap (1:765,000-1:1,208,000), as do those for ID-NAT Ultrio and Ultrio Plus (1:1,075,000-1:2,213,000). These data suggest that a more substantive reduction in residual risk comes from migration from MP16 to ID-NAT than from the addition of MP-NAT to serology. Other studies have shown that Ultrio Plus provides a significant improvement in HBV detection for acute infection even in the MP16 format and for late acute or chronic HBV infections, including OBI, in the ID-NAT format.[7]

Table 6. Projected point estimates of residual risk for differing test protocols and infectious window periods based on combining two different sets of models
ProtocolModeled window-period reduction (days)Resulting window periodProjected residual risk
  1. aThis study; using the incidence–window-period model as described by Zou and colleagues.[1]
  2. bWindow-period reduction from modeling by Vermeulen and colleagues using the seroconversion risk model of Weusten and colleagues;[17] the results of this modeling (2.7-day window-period difference between Ultrio and Ultrio Plus) have been applied to the data in this study.
MP16 (Ultrio)a8.829.2-21.21:765,000-1:1,006,000
MP16 (Ultrio Plus)11.5b26.5-18.51:843,000-1:1,208,000
ID-NAT (Ultrio)17.2b20.8-12.81:1,075,000-1:1,746,000
ID-NAT (Ultrio Plus)19.9b18.1-10.11:1,235,000-1:2,213,000

In summary, it is clear that the introduction of HBV NAT in the United States, even in the MP format, along with the HBV vaccination policy in the United States have made a measurable contribution to blood safety and decreased residual risk and will result in an examination of the continued need for all three HBV markers for blood donor screening in the United States and elsewhere. As of 2010, 27 countries have either implemented or plan to implement HBV NAT, of which 12 countries have mandated its use; most remain on MP-NAT, using protocols more sensitive than Ultrio MP16 used in the United States, but the trend is clearly for reduced MP size or ID-NAT.[19] Considerations of the HBV donor screening algorithm before any given change may include the transition to ID-NAT to maximize yield (for those countries that have retained MP-NAT) and/or pathogen reduction technologies, which together should allow multiple serologic tests to be deemed antiquated and a blood donation screening focus to be NAT based.


The authors thank Dr Shimian Zou for his advice, review, and support of the statistical methods. We also thank Dr Erin Moritz for statistical method support.

Conflict of Interest

No authors have a conflict of interest.