Vaccine-Induced HIV Seropositivity/Reactivity in Noninfected HIV Vaccine Recipients FREE

With an estimated 7500 incident human immunodeficiency virus (HIV) infections occurring each day worldwide, there is an urgent need to develop an effective prophylactic HIV vaccine.¹ Over the last 20 years many potential vaccine candidates have been developed and assessed in human clinical trials in more than 30 000 participants. These candidate vaccines have used a variety of approaches, including protein-,^2- 5 DNA-,^6- 7 HIV peptide–,^8- 10 and viral-vectored strategies.^11- 15 In addition, a variety of HIV targets or inserts have been studied, including group-specific core antigen gene (gag), polymerase gene (pol), negative factor (nef), and envelope gene (env) in a variety of combinations.¹⁶ Several of these candidate vaccines have advanced to large field trials.^{3- 4,17- 19}

HIV-1 vaccines have the potential of confounding interpretation of HIV tests because of the antibody induced by vaccination.^20- 22 Depending on the HIV-associated sequences used in the candidate vaccine, not only may the screening enzyme immunoassay (EIA) be reactive but the Western blot may also be difficult to interpret.²³ Participants in early-phase clinical trials, who are typically at low risk of HIV infection, may encounter difficulties with obtaining medical or disability/life insurance, donating blood or organs (which is based on a reactive EIA regardless of any confirmatory Western blot or RNA assay result), employment, and immigration owing to a false-positive HIV test result.

In volunteers who participate in later-stage efficacy trials, this becomes an even more complex issue, because these individuals are at higher risk for contracting HIV, and the candidate vaccines are generally known to be immunogenic.^{3- 4,17,19,24- 25} In the United States, the adoption of an “opt-out” HIV testing strategy recommended by the Centers for Disease Control and Prevention facilitates identification of individuals infected with HIV and thus allows earlier access to care.²⁶ However, this approach also has the potential to increase confusion around false-positive HIV testing resulting from vaccine-induced antibodies.

In this report we assess the occurrence of vaccine-induced seropositivity/reactivity (VISP; defined as reactive on 1 or more EIA tests and either Western blot–negative or Western blot–indeterminate/atypical positive [profile consistent with vaccine product] and HIV-1–negative by nucleic acid testing) associated with different vaccine delivery systems and HIV inserts studied by the National Institutes of Health–sponsored HIV Vaccine Trials Network (HVTN).

Data were combined from all phase 1 (n = 25) and 2a (n = 2) HIV-1 vaccine trials that had completed study follow-up visits (except for 1 phase 1 vaccine manufacturer–managed trial), conducted by HVTN clinical trial sites located in 9 countries (Botswana, Brazil, Haiti, Jamaica, Peru, South Africa, Thailand, and Trinidad and Tobago) and the United States between December 14, 2000, and January 15, 2010.

Clinical trial locations and participant demographics are shown in Table 1. Except for 2 phase 1 open-label trials, the trials were placebo-controlled, multicenter, double-blind, randomized trials. Participants were healthy, HIV-seronegative adults aged 18 to 60 years. At the end of each trial, volunteers were tested using the end-of-study (EOS) HIV testing algorithm described herein and in Figure 1, which established whether the participants had VISP. Each trial was approved by the institutional review board/ethics committees of the participating institutions, and participants provided written informed consent for study participation.

The phase 1 trials typically involved dose escalation. Except for 3 phase 1 trials that administered vaccines by the subcutaneous route and 1 phase 1 trial that compared different routes of administration for boosting, vaccines were administered intramuscularly. Because of the potential for participant unblinding, HIV testing was typically performed at 6-month intervals during the trials by the HVTN Laboratory Program using an algorithm able to distinguish true infection from VISP. Participants were informed about the risk of developing VISP and potential negative consequences thereof before signing the consent document. This information was also discussed with participants at study visits. As required by National Institutes of Health–sponsored trials, participants self-reported their ethnicity and race, using standardized categories (including “other”). Most studies were 12 to 18 months in length, and EOS testing typically occurred 6 to 12 months after the last vaccination.

Table 2 lists the 25 different vaccine products tested in these trials, which were given alone or in combination as noted (clinicaltrials.gov identifiers and references corresponding to the HVTN protocols reported in Table 2 appear in the Box). The number of vaccinations ranged from 1 to 5. Products tested in the largest number of participants were the canarypox vector vaccine ALVAC HIV vCP1452 (Aventis Pasteur, Bridgewater, New Jersey), the 2 DNA Vaccine Research Center (VRC) VRC-HIVDNA009-00VP (4-plasmid) and VRC-HIVDNA016-00-VP (6-plasmid) vaccines, and the VRC recombinant adenovirus serotype 5 (Ad5) vector vaccine VRC-HIVADV014-00-VP.

HVTN 026: NCT00011037 (Cleghorn et al,¹⁵ 2007)
HVTN 039: NCT00027261 (Goepfert et al,¹² 2005)
HVTN 040: NCT00063778
HVTN 041: NCT00027365 (Goepfert et al,⁵ 2007)
HVTN 042: NCT00076063
HVTN 044: NCT00069030
HVTN 045: NCT00043511 (Mulligan et al,⁷ 2006)
HVTN 048: NCT00054860 (Gorse et al,¹⁰ 2008)
HVTN 049: NCT00073216
HVTN 052: NCT00071851
HVTN 054: NCT00119873
HVTN 055: NCT00083603
HVTN 056: NCT00076037 (Spearman et al,⁸ 2009)
HVTN 057: NCT00091416
HVTN 059: NCT00097838
HVTN 060: NCT00111605
HVTN 063: NCT00115960
HVTN 064: NCT00141024 (Jin et al,⁹ 2009)
HVTN 065: NCT00301184
HVTN 067: NCT00428337
HVTN 068: NCT00270218
HVTN 069: NCT00384787
HVTN 070: NCT00528489
HVTN 071: NCT00486408
HVTN 072: NCT00472719
HVTN 203: NCT00007332 (Russell et al,¹¹ 2007)
HVTN 204: NCT00125970

The EOS HIV testing algorithm consisted of 3 US Food and Drug Administration–licensed EIA tests administered equally in all trials (except for 182 participants in the earliest trials who had only an Abbott test [n = 11] or an Abbott and one other EIA kit run [n = 171]): (1) Abbott (Abbott Park, Illinois) HIVAB HIV 1/2 (rDNA); (2) Bio-Rad Genetic Systems (Hercules, California) HIV 1/2 Peptide, which was replaced in March 2006 with the next-generation Bio-Rad Genetic Systems HIV 1/2 Plus O EIA; and (3) bioMérieux (Marcy l’Etoile, France) Vironostika HIV-1 Plus O Microelisa System, which owing to product discontinuation was replaced in October 2007 with the Bio-Rad Genetic Systems rLAV kit.

Kits were used according to manufacturer's instructions. Participant specimens repeatedly testing reactive on any 1 or more of the 3 different EIA kits were further tested by Western blot. For early trials, a validated Western blot kit developed by the testing laboratory from commercially available antigens was used (California Department of Health Services Viral and Rickettsial Disease Laboratory, Richmond), which was replaced in June 2005 with the Bio-Rad Genetic Systems HIV-1 Western Blot kit. Blots were interpreted according to Centers for Disease Control and Prevention criteria and the manufacturer's product insert, requiring the presence of any 2 of the bands glycoprotein (gp) 41, gp120/160, or p24 for positivity.²⁷

Specimens testing positive or indeterminate had nucleic acid testing consisting of quantitative RNA polymerase chain reaction (PCR) performed to rule out true infection. During the time span of the study, the quantitative RNA PCR kits included Amplicor HIV-1 Monitor (Roche Diagnostics, Indianapolis, Indiana), Abbott Realtime HIV-1 (Abbott Molecular), and a validated real-time HIV-1 RNA assay developed at the Department of Laboratory Medicine, University of Washington, Seattle. Any participant with a positive RNA PCR result had a repeat sample drawn to confirm true infection.

To summarize the risk of developing VISP for these diverse HIV-1 vaccine trials, vaccine regimens were grouped into general strategy (product) categories based on vaccine insert (DNA, peptide, or protein), viral vector (poxvirus, adenovirus 5/35, or alphavirus), or canarypox prime with peptide or gp120 boost. These general categories were further subdivided when VISP rates within a category substantially varied by product type. Participants who did not complete their vaccine regimen were included in the analysis based on the products they received. Since the specific HIV-1 gene inserts effect seropositivity/reactivity, we also present data by HIV-1 gene insert groupings.

For all categories, the VISP rate and exact 95% binomial confidence intervals (CIs) were calculated. The VISP rate was calculated as the number of participants testing reactive on 1 of the 3 different EIA kits tested divided by the number of participants tested. Because the HIV 1/2 (rDNA) kit nearly always detected VISP, we also included in the rate calculations those participants who had been tested with this kit but who were missing data from one or both of the other kits (n = 182). Additionally, positivity rates for the different EIA kits are provided based on all participants with a result for the kit. Sex, age (below vs above the median), and race (white vs black) differences were evaluated with Fisher exact tests. Statistical analyses were performed using SAS version 9.1 (SAS Institute, Cary, North Carolina). P values less than .05 were considered statistically significant. No adjustment for multiple testing was performed.

A total of 3014 participants were enrolled in the trials, of whom 2300 (76.3%) were vaccinees and 714 (23.7%) were recipients of placebo. Of the 2300 vaccinees, 2176 vaccinees (94.6%) had available EOS testing data (11 [0.5%] had EOS testing pending, and 92 [4.0%] did not have EOS testing performed, the majority owing to loss to follow-up; 16 [0.7%] were excluded from the analysis because they had become HIV-1 infected; and 5 [0.2%] were excluded for vaccine administration issues [eg, wrong product administered, product viability uncertain]) (Figure 2).

There were 2176 non–HIV-infected vaccine recipients with HIV test results available from the EOS testing algorithm at the time of data analysis. Although the algorithm specified that 3 different EIA kits be run, 182 participants (8.4%) from the earliest trials were missing kit results other than HIV 1/2 (rDNA) results from 1 (n = 171) or 2 (n = 11) kits.

The majority of participants (82%) in the analysis were from the United States (Table 1). Among US participants, 57% were men, 70% were white, and the median age was 29 years. African participants had a slightly younger median age of 25 years, and 49% were men. Participants from South America and the Caribbean were drawn from ethnic/racial groups representative of their respective countries, had a median age of 27 years, and 57% were men.

VISP occurred in 908 of the 2176 participants (41.7%; 95% CI, 39.6%-43.8%) (Table 3). Rates of VISP varied greatly by type of HIV vaccine administered. For viral-vectored vaccines, alphavirus replicon products containing the gag antigen alone were found to have the lowest VISP rate (1.0%; 95% CI, 0.0%-5.2%). In contrast, Ad5 products containing env antigens given alone or as a boost to a DNA prime were found to have the highest VISP rate (92.7%; 95% CI, 89.8%-95.0%). Within the DNA-only product category, VISP occurred in 18.9% (95% CI, 13.2%-25.7%) of participants receiving a VRC DNA product, whereas it occurred in only 1% (95% CI, 0.3%-2.6%) of participants receiving other DNA vaccines. For poxvirus products given alone or as a boost to a DNA or poxvirus prime, 53.4% (95% CI, 49.2%-57.7%) of recipients had VISP. This was similar to the 48.6% (95% CI, 41.9%-55.4%) occurrence for canarypox given in combination with a protein or peptide product. All 70 recipients of a gp140 vaccine had VISP, whereas recipients of other protein and peptide products given alone or as a DNA boost had a VISP rate of 0.5% (95% CI, 0.0%-2.6%).

The HIV 1/2 (rDNA) kit was selected for inclusion in the EOS testing algorithm because it is a commonly used diagnostic kit in the United States and has been noted to be particularly sensitive to detecting vaccine-induced HIV antibodies. Only 17 of 908 participants (1.9%) with VISP did not test reactive using the HIV 1/2 (rDNA) kit (Table 3). Within product categories, VISP occurred more frequently with the HIV 1/2 (rDNA) kit than with the other kits (except for the alphavirus replicon category, wherein a single participant was found to have VISP detected with only the HIV 1/2 Peptide kit). The gp140 product was unique in that 94.3% of recipients were reactive with all 3 kits. The other kits varied in their rates of reactivity, especially for viral-vectored vaccines (Table 3). Although kit data are summarized for all participants with a kit result, data are similar when limiting comparisons to participants tested with the same 3 kits.

Almost all VISP was associated with vaccines that contain env, either alone or in combination with other HIV-1 antigens (Table 4). Only 8 participants with VISP received a vaccine not containing env: 5 received Ad5 (gag, pol, nef inserts) and tested reactive using the HIV 1/2 (rDNA) kit; 2 received gag DNA and tested reactive using the HIV 1/2 (rDNA) kit; and 1 received alphavirus replicon (gag insert) and tested reactive using the HIV 1/2 Peptide kit.

Within the vaccine subproduct categories, no sex or age differences in VISP were observed within geographical regions or for regions combined. One racial difference was observed for US participants receiving VRC Ad5 alone (93.2% white vs 50.0% black; P = .01; no participants in Africa received this product), although this is based on only 6 black individuals and was not observed for VRC DNA prime and Ad5 boost. This analysis was not adjusted for multiple testing.

For participants testing reactive with 1 or more EIA kits, all except 7 had a Western blot result (Table 5, Figure 2). Overall, among the 901 participants with a Western blot result, 92 (10.2%) had a positive result, 592 (65.7%) tested indeterminate, and 217 (24.1%) tested negative; however, the distribution of results varied by product. For DNA and fowlpox vaccines, 50.0% or more had negative Western blot results. The env-containing Ad5 vaccine given alone or as a DNA vaccine boost had the highest proportion of positive results (25.4% and 13.6%, respectively). Poxvirus regimens had higher percentages of indeterminate results (50.0%-97.3%) and positive results (<10%).

For participants with a positive Western blot result, all but 1 had a positive gp160 band (11 had missing data). For gag bands, all but 1 were positive for p24, and 44 (53.7%) were positive for p55 (30 received an Ad5 product, and 14 received a poxvirus-vectored vaccine; 10 had missing data). For pol bands, p51 was positive for only 6 participants (4 received an Ad5 vaccine and 2 received poxvirus; 10 had missing data), and no participant had a positive p31 band (11 had missing data). For those with an indeterminate Western blot result, p24 was positive for 91.1% who received a poxvirus-vectored vaccine regimen and for 17.8% who received other vaccines (3 had missing data). Data on other bands were not consistently available for participants with an indeterminate Western blot result who received a poxvirus-vectored vaccine. For other products, only the gp160 band had a substantial number of positive results, particularly for a gp140 vaccine, 83.9% of which were positive.

These data demonstrate that VISP is a common but highly variable outcome of trials of preventive HIV vaccines. The variability of the occurrence of VISP is dependent on the immunogenicity of the vaccine product, the HIV gene inserts, and the HIV testing kit used. This analysis did not specifically examine the vaccine dose, number of doses, administration schedule, and use of adjuvants as contributing factors to the development of VISP, because it would be difficult to assign relative importance to these factors in this cross-study analysis. By product category, rates ranged from 1% for an alphavirus replicon construct containing only a gag insert to 100% for a gp140 vaccine. VISP occurred most frequently with the HIV 1/2 (rDNA) kit (98% of those with VISP tested positive by this kit). An awareness of the potential occurrence of VISP allows for appropriate counseling and education to be provided to study participants.

Among participants with a reactive EIA result, Western blot distinguished only 24% as HIV-negative, with most (66%) having an indeterminate Western blot result with positive bands consistent with the genes included in the vaccine insert. A limitation of the HVTN data is that testing for VISP with multiple EIA kits is performed only at the end of the study, so it is not possible to distinguish effects of multiple inoculations with a given vaccine or the contribution of a single product when given as series of inoculations in combination with other products. Given the likely specificity for an insert (especially env-based inserts) to cross-react with a given detection system (EIA assay), it is difficult to predict the potential rate of VISP associated with future vaccines as delivery systems, inserts, and potential EIA detection assays are modified.

This analysis significantly expands on previous studies of VISP, which were limited in scope to a relatively small number of candidate vaccines.^20,23,28 In contrast, our analysis includes data on 25 vaccine products given alone or in combination. The most extensive previous assessment of VISP focused on gp120, vaccinia constructs, and canarypox virus (ALVAC) HIV-1 vaccines with or without a protein boost. In the analysis by Ackers et al, none of 118 samples from studies of gp120 constructs were reactive with the HIV 1/2 (rDNA) or HIV-1 Plus O Microelisa System kits, and 13 (11%) were reactive using the rLAV EIA kit, which was not used for the HVTN protein/peptide trials.²⁸ The results with the HIV 1/2 (rDNA) and HIV-1 Plus O Microelisa System kits were similar to our results of 1 positive among 214 participants receiving a non-gp140 protein or peptide constructs. For trials of various generations of canarypox-vectored vaccines boosted with gp120, Ackers et al observed low proportions of VISP detected by the HIV-1 Plus O Microelisa System kit, which was confirmed in our analysis. For these regimens, they observed VISP proportions of 3% to 31%, depending on the trial, using the rLAV EIA kit (not used in our analysis for these regimens). They did not use the HIV 1/2 (rDNA) kit, whereas for this kit we observed that 48% of those receiving a canarypox vaccine given with a protein or peptide boost developed VISP.

The occurrence of VISP is dependent on the immune response to the vaccine product and the sensitivity of the kit to detect reactivity to the HIV antigens it contains. In these HVTN trials, VISP was rare for vaccines not containing an env insert. This is in contrast to a 41% rate observed among participants in some trials who received either an Ad5-vectored clade B HIV-1 monovalent gag or trivalent gag/pol/nef vaccine.²³ Although the difference may in part be attributable to the immunogenicity of different types of vaccines (HVTN non-env studies were primarily of alphavirus replicon or DNA vaccines), the difference also may be attributable to the EIA kits used. To increase the likelihood of detecting VISP, the HVTN EOS testing algorithm uses commercial HIV testing kits that each have a different set of HIV antigens. An example of the importance of the match (or mismatch) of vaccine insert to kit antigens is that in a study by Quirk et al, 2 of 432 recipients of non–env-containing Ad5 were reactive to env-only–containing rapid EIA kits.²³

Development of a rapid, accurate HIV test that is unlikely to cross-react with vaccine construct would be a useful approach for improved serologic tests. One promising approach uses sequences that are universally found in natural infections but if omitted from the vaccine insert would reliably differentiate vaccination from natural infection by serologic testing.²⁹ Until such an approach is available, we recommend the inclusion of an HIV RNA assay to differentiate vaccination from infection (Figure 1). Because participants with VISP may subsequently become infected with HIV, it is imperative that appropriate follow-up testing be conducted, including HIV RNA testing, to minimize potential misinterpretation of HIV test results.

HVTN trial data are limited in that testing for VISP occurs only at the end of the trial (typically 6-12 months following the final vaccination). Although participants are offered long-term HIV testing services through the HVTN if they have VISP, the HIV testing data from longer time points have not been systematically collected to date. Through requests for HIV testing services, the HVTN is aware that some participants continue to have VISP for many years (sometimes longer than 15 years) following their study participation. An observational study of VISP posttrial termination is currently in development in the HVTN.

Testing for VISP at the end of the study and providing participants with their VISP status is critically important to prevent social harms, incorrect HIV diagnosis, and inaccurate reporting to health agencies. A misinterpretation of VISP can be minimized by clinicians obtaining a complete patient history (eg, participation in an HIV vaccine trial) and interpretation of the Western blot and HIV RNA. However, given the added time and cost associated with obtaining this information, clinicians may overlook or not pursue this information. During the course of participating in an HIV vaccine study, the detection of VISP might influence a study participant's perception of the vaccine product received and may influence their behavior.³⁰ To date, social harms related to HIV testing outside the context of the study sites have been relatively rare. However, with the Centers for Disease Control and Prevention recommendations for routine HIV screening and “opt out” HIV testing practices, the education of participants and clinicians will be important to keep risk of social harms related to HIV testing low for HIV vaccine trial participants.