Group O Human Immunodeficiency Virus Type 1 Infection That Escaped Detection in Two Immmunoassays

CASE REPORT

Screening for human immunodeficiency virus (HIV) antibodies was performed on a 42-year-old Cameroonian woman during her first pregnancy, 3 months after her arrival in France, in July 1994. At that time, serum was reactive in two second-generation enzyme-linked sorbent immunoassays (ELISAs), a competitive assay (Welcozyme HIV recombinant; Murex Diagnostics, Dartford, United Kingdom) and an indirect assay (Axsym HIV-1/2; Abbott Diagnostics, North Chicago, Ill.), respectively (Table 1). Seropositivity was confirmed by Western blotting (New LAV Blot 1; Bio-Rad, Marnes-la-Coquette, France) showing strong reactivity to the Gag, Pol, and Env proteins except to gp120. Her CD4⁺ T-cell count was 317/mm³, and p24 antigenemia was undetectable (Murex HIV antigen monoclonal antibody). The patient was asymptomatic. Antiretroviral treatment was initiated in 1995 because her CD4 cell count was low, close to 300/mm³. She was treated first with zidovudine and then with a zidovudine-zalcitabine combination, which allowed an increase of 100 CD4⁺ T cells/mm³. However, she developed pancreatitis and myositis in November 1997, which necessitated discontinuation of the antiretroviral regimen.

As a moderate decrease of her CD4 cell count was observed, a highly active antiretroviral treatment was instituted in 1998 (zidovudine, lamivudine, and indinavir), after which the CD4 cell count increased to 700/mm³. Her plasma virus load was undetectable in all the samples tested using either HIV RNA 3.0 (HIV-1 bDNA) (Bayer Diagnostics, Berkeley, CA) or Roche Amplicor HIV-1 Monitor v1.5 (HIV-1 Monitor) (Roche Molecular Systems, Branchburg, N.J.). Highly active antiretroviral treatment was discontinued in September 2001 because of treatment intolerance and a CD4 cell count close to 700/mm3.

In 2003, after delivery of her second child, a complete analysis of virological markers (detection of HIV antibodies, lymphocyte culture, and plasma HIV RNA quantification) was performed on samples from the mother and infant. Surprisingly, ELISAs for HIV antibodies gave negative results for both the mother and the newborn using a third-generation (HIV-1/2 Access; Bio-Rad, France) and a fourth-generation (Vidas Duo; Biomérieux, Marcy l'Étoile, France) assay. Previously collected sera that were retested using these assays were also found to be negative for HIV-1 antibodies (Table 1). Lymphocyte cultures were negative both for the mother's and newborn's samples, as was HIV RNA detection for the infant. The maternal viral load at delivery was 5,568 copies/ml using the LCx HIV RNA assay (Abbott Diagnostics) but was undetectable by both the HIV-bDNA and Monitor assays. The discordant data obtained in both serological assays and molecular assays strongly suggested infection by an unusual HIV-1 variant.

A peptide-based serotyping assay was therefore performed with two serum samples collected from the patient. This assay is an ELISA based on synthetic V3 loop and gp41 peptides to discriminate between HIV-1 group M, HIV-1 group O, and HIV-2 infections (1, 8). The serum samples were strongly reactive to the V3 peptide of HIV-1 group O, suggesting infection by a group O variant (Table 2). However, and quite unusually in our experience, antibodies to the gp41 immunodominant epitope (IDE) were not detected. A group-specific reverse transcriptase PCR able to discriminate between HIV-1 groups M and O was performed using primers located in the transmembrane region of the env gene and covering the IDE (9). A DNA fragment was obtained only with group O primers. The amplified segment was sequenced (392 nucleotides), and identification of the strain was done using the neighbor-joining method. The sequence was compared with 50 reference sequences corresponding to the nine subtypes and major circulating recombinant forms (CRF01-AE and CRF02-AG) of HIV-1 group M and to different strains of HIV-1 group O, available from the HIV sequence database (http://hiv-web.lanl.gov ). Distances were calculated with the Kimura two-parameter method, as implemented in the MEGA program. Bootstrap analysis with 100 simulations was used to test the reliability of branching. The strain clearly belonged to group O (Fig. 1).

• Open in new tab
• Download powerpoint

FIG. 1.

Phylogenetic analysis of gp41 of the case strain. Values at nodes indicate the percentages of bootstraps in which the cluster to the right was found. The tree was rooted using the SIVCPZ CAM5 sequence as an outgroup.

Failure of the sera from this patient to bind to the group O IDE consensus peptide might be attributed to an unusual dipeptide motif (TR) located within the five-amino-acid loop of the IDE of the infecting variant (Table 2). Indeed, this TR motif was never found in any of the 64 available gp41 sequences of HIV-1 group O. Based on this observation, we investigated whether this divergent sequence was responsible for the lack of reactivity of the patient's sera to the group O IDE consensus peptide by preparing a synthetic peptide overlapping the IDE of the case's strain (see the sequence in Table 2). Serum samples from the case were tested in parallel by ELISA for this peptide, consensus peptide of IDE group O, and consensus peptide of IDE group M (1 μg/ml) based on a procedure described elsewhere (1). As shown in Table 3, the two serum samples from this patient bound the homologous IDE peptide, albeit with a relatively weak signal.

Sequential sera collected between 1994 and 2004 and preserved at −20°C were tested, when available in sufficient amount, with various immunoassays (Table 1). The results were positive with most second- and third-generation assays, except HIV-1/2 Access, and negative with two of the four fourth-generation tested assays (Genscreen plus; Bio-Rad and Vidas Duo; bioMérieux).

Over the past 20 years, considerable progress has been made in the abilities of immunoassays to detect antibodies to groups M, N, and O of HIV-1 as well as antibodies to HIV-2 (5). Nevertheless, failure to detect HIV antibodies due to HIV genetic diversity has been regularly reported (3). This case is remarkable because one third-generation and one fourth-generation ELISA each failed to detect HIV infection, whereas most second- and third-generation assays did detect HIV infection. Most of the third-generation assays include a recombinant p24 protein, which is an antigen conserved between HIV groups. Third- and fourth-generation assays use peptides or recombinant proteins, including the gp41 immunodominant epitope from HIV-1 groups M and O and HIV-2 (2). The unique IDE sequence of the group O variant described herein might explain the failure by several immunoassays. This hypothesis is supported by the fact that antibodies from the patient bound its homologous sequence, although weakly, whereas they did not bind a group O consensus sequence. Since the Access assay that includes a p24 antigen was negative, our results indicate that even addition of this conserved antigen does not guarantee 100% sensitivity in detecting HIV variants.

Most HIV-1 RNA quantification assays detect only HIV-1 group M, except the LCx assay, which detects group O; so discordant results between different assays can suggest infection by a variant HIV virus.

Although the prevalence of HIV-1 group O is low even in areas of endemicity, such as Cameroon (6), this case report highlights the impact of its high genetic diversity (4, 7) on efficacy of diagnostic tools and illustrates how the genetic diversity of HIVs in general challenges the diagnostics field.