HIV reinfection has limited impact on disease progression
1 November 2014. Related: Transmission.
Gareth Hardy, HIV i-Base
Superinfection with a second viral variant after initial HIV infection has been described in various cohorts. However, it is currently unclear whether superinfection is associated with a clinically significant impact on disease progression or management.
In this study, Keshet Ronen from the Fred Hutchinson Cancer Research Center, University of Seattle, examined a well-characterised, prospective cohort of HIV negative women sex workers in Mombasa, Kenya, for cases of superinfection. 
From a total of 146 women screened, Ronen and colleagues identified and specified the timing of 21 cases of superinfection. This is currently the largest superinfection cohort and previous analysis have reported an incidence of superinfection that was approximately half the rate of initial infection.
HIV infection in this cohort occurred with clades A, C and D. Superinfections occurred with both intrasubtype and intersubtype variations and were timed from 63 to 1895 days after initial infection. The researchers investigated frequent viral load and CD4 count measurements to determine the impact of superinfection on markers of disease progression, as well as time to clinical events, in this longitudinal analysis. Of the 146 women screened for superinfection, 144 were selected based on estimation of initial infection being timed to within a one-year window. Timing of superinfection was determined as the midpoint between the last singly infected and first dual infected time points.
In order to exclude changes in viral load that occurred prior to establishment of the viral set point, viral load data was only included from after the first six months of initial infection. Viral load [log10 transformed] and CD4 count [square root (sqrt) transformed] data were analysed using R. Linear mixed effect (LME) models to determine intercepts and rates of change. Cox proportional hazards regression was used to determine time to disease progression (CD4 T cell count <200 cell/mm3, initiation of ART or death).
Within the 144 women included in the screening analysis, there were 21 cases of superinfection. Of these, 133 women were included in the longitudinal data analysis as they had one or more viral load and CD4 T cell count measurements between 6 months post-initial infection and initiation of ART [median 10 measurements each]. Within this slightly smaller data set, there were 18 cases of superinfection. Comparing results of women who remained singly infected with women who ultimately acquired superinfection, the researchers found that there was a 0.009 log10 viral load/mL/month faster rate of viral load increase (p=0.0008), but no significant faster decline in CD4 T cell count, at 0.047 sqrt cells/mm3/month (p=0.06). Presence of genital ulcers at HIV acquisition, initial viral subtype and HLA-B alleles had no effect on the model parameters.
The researchers also sought to determine if there were differences in any of the measured parameters between women that remained singly infected and those that became superinfected, prior to the superinfection occurring. The association between superinfection and lower pre-superinfection viral load intercept was also short of statistical significance (-0.44 log10 copies/mL, p=0.06). When this data was adjusted for initial viral subtype, genital ulcers at initial infection and possession of HLAB*57, HLAB27 and HLAB*35 this association just became significant (-0.45 log10 copies/mL, p=0.05). No such relationship was found for CD4 T cell counts.
Within the superinfection cases, there was no significant association to higher viral load intercept following superinfection, compared with before (+0.21 log(10) copies/mL, p=0.09) with no relationship for CD4 T cell counts.
During study follow up, 91 of 144 individuals experienced a clinical event (defined as CD4 T cell count <200 cells/mm3, initiation of ART or death). There was no relationship between superinfection and incidence of clinical events (hazard ratio 1.07, 95%CI: 0.60 – 1.89; p = 0.76). The model predicted that there would be a 0.23 log(10) higher viral load/mL and that CD4 T cell counts would be 27 cells/mm3 lower after 5 years.
The researchers concluded that their data demonstrates a statistically significant acceleration in viral load increase and CD4 T cell count decline following superinfection, suggesting that the lower viral loads observed among individuals who subsequently acquired superinfection, compared with those who remained singly infected, may suggest that lower replication by the initial virus may predispose to superinfection.
However, and importantly, the differences in viral load were only just significant and are limited by small sample size. Lastly, the researchers suggest that the lack of an association between superinfection and clinical progression may result from the fact that the marginal differences in viral load and CD4 count were insufficient to cause differences in clinical outcome after 5 years.
This study is important for highlighting the difficulty in finding any clear signal for a clinically significant impact of superinfection.
The modeled prediction over five years of having a CD4 count that was 27 cells/mm3 lower or a viral load 0.23 log(10) copies/mL higher has no clinical significance, especially in the context of access to ART.
These results should provide doctors with an accurate reference for the likely safety for HIV positive people to not use condoms with other positive partners, in a situation where neither STIs nor pregnancy are a concern.
The concern about drug resistance is important if either partner is not on ART and the resistance profile is likely to be different. The transmission risk data from the PARTNER study and others, suggests that the caution about drug resistance may not be important in the context of undetectable viral load on ART.
- Ronen K et al. HIV-1 superinfection is associated with an accelerated viral load increase but has a limited impact on disease progression. AIDS. 2014 Aug 6. [Epub ahead of print].