HLA-Bw4 identifies a population of HIV-infected patients with an increased capacity to control viral replication after structured treatment interruption


Correspondence: Dr Martin Stern, Immunotherapy Laboratory, Department of Biomedicine, University Hospital Basel, Hebelstrasse 20, 4031 Basel, Switzerland. Tel: +41 61 328 73 90; fax: +41 61 265 44 51; e-mail: sternm@uhbs.ch



After structured treatment interruption (STI) of treatment for HIV-1, a fraction of patients maintain suppressed viral loads. Prospective identification of such patients might improve HIV-1 treatment, if selected patients are offered STI.


We analysed the effect of previously identified genetic modulators of HIV-1 disease progression on patients’ ability to suppress viral replication after STI. Polymorphisms in the genes killer cell immunoglobulin-like receptor 3DLI (KIR3DL1)/KIR3DS1, human leucocyte antigen B (HLA-B) and HLA Complex P5 (HCP5), and a polymorphism affecting HLA-C surface expression were analysed in 130 Swiss HIV Cohort Study patients undergoing STI. Genotypes were correlated with viral load levels after STI.


We observed a statistically significant reduction in viral load after STI in carriers of HLA-B alleles containing either the Bw480Thr or the Bw480Ile epitope (mean adjusted effect on post-STI viral load: −0.82 log HIV-1 RNA copies/ml, P < 0.001; and −1.12 log copies/ml, P < 0.001, respectively). No significant effects were detected for the other polymorphisms analysed. The likelihood of being able to control HIV-1 replication using a prespecified cut-off (viral load increase < 1000 copies/ml) increased from 39% in Bw4-negative patients to 53% in patients carrying Bw4-80Thr, and to 65% in patients carrying Bw4-80Ile (P = 0.02).


These data establish a significant impact of HLA-Bw4 on the control of viral replication after STI.


Antiretroviral therapy (ART) enables long-term control of HIV-1 infection through suppression of viral replication in the majority of treated individuals. This leads to substantial immune reconstitution, significantly delays morbidity and mortality, and transforms HIV infection into a chronic disease [1]. However, ART is not curative and life-long pharmacological treatment is required, which can lead to numerous adverse effects. Depending on the drug combination applied, cardiovascular events, hepatotoxicity, neuropathy, renal dysfunction and lipodystrophy may occur [2]. Additional concerns regarding continuous ART are the induction of drug resistance, high costs, and treatment fatigue in patients. Structured treatment interruption (STI) strategies have therefore been explored in patients with viral replication suppressed under ART [3-6]. Overall, results were disappointing, with a significant proportion of patients showing rapid increases in viral load, declining CD4 T-cell counts, and an increased risk for disease progression [4, 7]. However, a subgroup of patients are able to suppress HIV replication for prolonged periods of time after STI [8]. A marker identifying such patients would be of great practical value and might renew interest in STI.

Several studies have identified genetic factors influencing the pretreatment set-point viral load and time to progression to AIDS in untreated patients: the first locus identified was human leucocyte antigen (HLA)-B [9]. The natural killer (NK) cell receptor pair killer cell immunoglobulin-like receptor 3DL1 (KIR3DL1)/KIR3DS1 followed [10]. More recently, whole-genome association studies provided the information that two single nucleotide polymorphisms, found in or close to the HLA complex, both correlate with HIV viral load in untreated individuals [11]. The first (rs9264942) is located 35 kbp upstream of HLA-C (HLA-C −35 C/T) and governs the level of surface expression of HLA-C [12]. The second (rs2395029) lies in the HLA complex P5 (HCP5) and is in strong linkage disequilibrium with HLA B*5701, the HLA-B allele associated with the strongest protection from disease progression. Interestingly, all of these polymorphisms are potentially associated with the function of NK cells, a subgroup of lymphocytes important in defence against viral infection: KIR3DL1 is an inhibitory receptor binding HLA-B antigens that carry the Bw4 epitope [13]. HLA-C is the ligand for the NK cell receptors KIR2DL1, KIR2DL2, KIR2DL3 and KIR2DS1 [14]. KIR–HLA interactions are important during NK cell development, as only NK cells carrying inhibitory KIR and their HLA ligands acquire full functional competence [15]. As antiviral effects of NK cells have been shown to operate most effectively in states of low viral load [16], we hypothesized that these polymorphisms may have a role in predicting which patients are able to maintain suppressed viral load after STI. We therefore studied the association of these polymorphisms with the evolution of viral load after STI in 130 Swiss HIV Cohort Study patients.

Patients and methods

Study subjects

Patients were recruited from Swiss HIV Cohort Study participants of the Strategies for Management of Anti-Retroviral Therapy (SMART), Swiss Spanish Treatment Interruption Trial (SSITT)/2nd SSITT, and STACCATO (A Trial of CD4 Guided Treatment Interruption, Compared to Continuous Treatment, for HIV Infection) trials [3-6]. Of note, patients enrolled in SSITT/2nd SSITT treated with granulocyte macrophage colony-stimulating factor were excluded from the present analysis. All patients gave written informed consent to their study treatment and to having their data analysed. One-hundred and thirty patients with viral load data available at set point, and before and after treatment interruption lasting at least 7 days, were included in the study. Mean pretreatment set-point viral loads were calculated, if more than one value was available within 6 months before initiation of antiretroviral treatment. If patients underwent more than one STI, only the first interruption was considered in the analysis. Demographic data for the patient population are summarized in Table 1.

Table 1. Patient population
  1. aSSITT (5,6), Swiss Spanish Intermittent Therapy Trial; SMART (4), Strategies for Management of Anti-Retroviral Therapy; STACCATO (3), A Trial of CD4 Guided Treatment Interruption, Compared to Continuous Treatment, for HIV Infection.
Patient age (years) [median (range)]41 (22–71)
Gender [n (%)] 
Male87 (67)
Female43 (33)
Source of infection [n (%)] 
Blood products1 (1)
Injecting drug user13 (10)
Heterosexual contact57 (44)
Men who have sex with men55 (42)
Other/unknown4 (3)
Ethnicity [n (%)] 
White110 (85)
Black14 (11)
Hispano-American2 (2)
Asian4 (3)
Number of treatment interruptions [median (range)]2 (1–5)
Trial [n (%)] 
SITTa 76 (58)
2nd SSITTa 18 (14)
SMARTa 13 (10)
STACCATOa 25 (19)


The KIR genotype was assessed using sequence-specific primer (SSP) polymerase chain reaction (PCR) [17]. Alleles of KIR3DL1 differing in cell surface expression were discriminated by intermediate resolution allele-specific PCR into those carrying alleles with high (*h; 3DL1*001, *002, *003, *006, *008, *009, *015 and *020), low (*l; 3DL1*005 and *007), or no surface expression (3DL1*004) using a combination of PCR SSP protocols previously described [18-20]. Patients were grouped into those expressing two high expression alleles (*h/*y) and those expressing at least one low-expressing allele (*l/*x) [21]. The KIR3DL1*004 allele, which is not expressed on the cell surface, was analysed separately.

KIR3DL1 ligands were typed using a real-time PCR method that discriminates between the two types of HLA-Bw4, Bw4-80Thr and Bw4-80Ile [22]. The HLA-C35 single nucleotide polymorphism (SNP) (rs9264942) was typed using a pre-designed custom assay using TaqMan chemistry (Applied Biosystems, Foster City, CA). The SNP in HCP5 (rs2395029) was typed by direct sequencing (forward primer 5′-3′ ACGATTCTCCTCACACTTACA; backward primer 5′-3′ TCTCTCCCAAAACCACACTC).

Statistical analysis

Viral load data were compared using nonparametric tests (the Mann–Whitney U-test and the Kruskal–Wallis test). Values are reported as medians and interquartile ranges (IQRs). Correlations between variables were assessed by calculating Spearman's rho. Generalized linear models were used to test the impact of the polymorphisms on the control of viral replication in multivariate fashion. All P-values reported are two-sided. To account for multiple testing, we considered associations of P ≤ 0.01 as significant.


HIV viral load at set point, on and off ART

The distribution of pretreatment set-point viral loads, as well as viral loads before and after STI, is shown in Figure 1. The median pretreatment viral load was 4.73 log HIV-1 RNA copies/ml (interquartile range 4.14–5.74 copies/ml). Immediately before treatment interruption, viral load was below the detection level in the majority (71%) of patients. Viral load increased after STI to a median of 3.06 log copies/ml (IQR 1.46–4.61 log copies/ml; Fig. 1a).

Figure 1.

HIV viral load in patients at the pretreatment set point, before and after structured treatment interruption (STI). Horizontal lines represent medians, the box represents the interquartile range, and whiskers represent the 2.5 and 97.5 percentiles (a). A significant correlation was detected between the post-STI HIV viral load and the interval since interruption of ART (P = 0.002) (b), and between the set-point viral load and the post-STI viral load (P < 0.001) (c).

The median interval between treatment interruption and viral load assessment off ART was 21 days (IQR 18–43 days). A significant correlation was found between this interval and the rise in HIV copy number (P = 0.002; Fig. 1b). In addition – and as previously demonstrated [6, 23] – the pretreatment set-point viral load correlated significantly with the post-STI viral load (P < 0.001). The duration of STI and viral load at pretreatment set point were therefore included in multivariable analyses.

Impact of genetic polymorphisms on control of viral load

Eighty-nine patients (68%) carried at least one HLA-B Bw4 allele. Bw4 alleles can be further separated into those carrying isoleucine or threonine at position 80 (Bw4-80Ile and Bw4-80Thr, respectively). Functionally, alleles with isoleucine act as strong ligands, whereas alleles carrying a threonine act as weak ligands of KIR3DL1 [24]. The former were detected in 52 patients (40%) and the latter in 37 patients (28%), whereas 41 patients carried no Bw4 alleles (32%). Patients not carrying a Bw4 allele showed a median post-STI viral load of 3.24 log copies/ml (IQR 2.21–4.29 log copies/ml), whereas the median post-STI viral load was 2.39 log copies/ml (IQR 0–3.62 log copies/ml) in Bw4-positive patients (P = 0.003; Fig. 2a). No difference was found between carriers of 80Thr and 80Ile subgroups of the Bw4 (median increase 2.40 and 2.39 log copies/ml, respectively; P = 0.66; Fig. 2b).

Figure 2.

Post-structured treatment interruption (STI) log HIV viral load is significantly lower in patients carrying the human leucocyte antigen (HLA)-Bw4 epitope (a). Within Bw4-positive patients, no significant differences were seen between patients carrying Bw4-80Thr and Bw4-80Ile (b), between patients carrying low and high surface expression alleles of KIR3DL1 (c), or between patients who did and did not carry the KIR3DL1*004 allele (d). The compound genotype KIR3DS1+/Bw4-80Ile+ did not significantly influence post-STI viral load (e), as so polymorphisms in HLA complex P5 (HCP-5) and HLA-C −35 (f, g).

We next analysed the impact of allelic diversity within the KIR3DL1 locus in Bw4-positive patients. Of 125 KIR3DL1-positive patients, 84 tested positive for at least one Bw4 antigen. We found no difference between patients carrying KIR3DL1 alleles with high (*h/*x) and low (*l/*l) surface expression (median increase 2.91 and 2.71 log copies/ml, respectively; P = 0.57; Fig. 2c). Equally, the presence of the KIR3DL1*004 allele – which in conjunction with Bw4 has been shown to delay the progression to AIDS – had no significant impact on post-STI viral loads (median increase 2.65 vs. 2.91 log copies/ml, respectively; P = 0.58; Fig. 2d).

The activating receptor KIR3DS1 – which segregates as an allele of KIR3DL1 – was contained in 45 patients’ genotypes (35%), of which 13 also carried Bw4Ile. The presence of KIR3DS1 with Bw4Ile has been shown to delay progression to AIDS [25]. In our setting, we found no difference in the rise in viral load between KIR3DS1+/Bw4-80Ile+ patients (median increase 2.65 log copies/ml) and patients who did not carry either KIR3DS1 or Bw4-80Ile or both (median increase 2.91 log copies/ml; P = 0.81; Fig. 2e).

Finally, we analysed the impact of the SNPs in HCP5 and in HLA-C −35. Nine patients (7%) carried one G allele in the HCP5 locus, and all remaining patients were homozygous for the wild-type T-allele. The median viral load was lower in patients with HCP5-G (median 2.76 log copies/ml) compared with HCP5-TT homozygous patients (median 2.85 log copies/ml). This difference was, however, not statistically significant (P = 0.90; Fig. 2f). At the HLA-C −35 locus, 79 patients (61%) were homozygous for the major T-allele and seven patients (5%) were homozygous carriers of the protective C allele, whereas the remaining 44 patients (34%) carried one copy of each allele. We found a lower post-STI viral load in homozygous carriers of the protective C allele (median 2.33 log copies/ml) compared with heterozygous patients (median 2.91 log copies/ml), and homozygous carriers of the T allele (median 2.81 log copies/ml). However, this difference did not reach statistical significance (P = 0.74; Fig. 2g).

Multivariable and categorical analysis

To account for the possibility of an interaction between variables predicting HIV viral load evolution after STI, we used multivariable generalized linear models to analyse the impact of pretreatment viral load, the duration of STI and genotype. Results are summarized in Table 2. Importantly, the protective effects of both Bw4-80Thr and Bw4-80Ile were maintained in the analyses adjusted for other covariates including time of STI and pretreatment set-point viral load.

Table 2. Multivariable analysis
Risk factorChange in post-STI viral load (log copies/ml)95% CIP
  1. CI, confidence interval; HLA, human leucocyte antigen; HCP-5, HLA complex P5; KIR, killer cell immunoglobulin-like receptor; STI, structured treatment interruption.
HLA Bw4   
No Bw4 (baseline)
Bw4-80Thr−0.82−1.27 to −0.36<0.001
Bw4-80Ile−1.15−1.61 to −0.70<0.001
Wild type (baseline)
Mutated−0.79−1.52 to −0.060.04
HLA-C −35   
TT (baseline)
CT−0.02−0.37 to +0.420.90
CC−0.12−0.91 to +0.660.74
Pretreatment viral load   
Increase per 1 log copies/ml+0.19+0.02 to 0.350.03
STI duration   
Increase per day+0.01+0.002 to 0.0140.006
KIR3DL1 (Bw4+ only)   
*h/*y vs. *l/*x+0.43−0.05 to +0.900.08
*004 present vs. absent−0.22−0.76 to −0.110.41
KIR3DS1 (Bw4-80Ile+ only)   
Present vs. absent+0.00−0.48 to +0.480.99

Using a predefined cut-off of a post-STI viral load copy number of 1000 copies/ml, the frequency of patients able to control viral replication increased from 39% of Bw4-negative patients to 53% of Bw4-80Thr patients to 65% of Bw4-80Ile patients (P = 0.02). None of the other polymorphisms analysed showed any significant impact in this analysis.


Previous studies have identified a number of genetic factors affecting viral load at diagnosis of HIV infection and the interval from seroconversion to the development of AIDS [10, 11, 26]. STI has been advocated as a therapeutic strategy in HIV-infected patients. Although a minority of patients in STI trials were able to suppress viral replication off ART, this approach has largely been abandoned, after randomized studies had shown increases in complications following STI when compared with patients treated continuously [4]. A genetic profile identifying patients with a higher likelihood of being able to suppress viral replication might point towards pathways involved in the control of viral replication and may renew interest in STI.

Our study found that an HLA-B allele containing the Bw4 public epitope conferred statistically significant protection regarding the rise in viral load after treatment interruption. No effect of KIR3DL1 alleles – which act as receptors for HLA-Bw4 – on post-STI viral load was detected. This may be a consequence of the relatively small sample size or be an indication that HLA-Bw4-related effects are the results of T-cell- rather than NK-cell-mediated immunity to HIV-1. Similarly, polymorphisms in HCP5 and in HLA-C −35 did not significantly influence post-STI viral loads in this analysis. However, the number of patients carrying the respective protective alleles was low in this study, which may preclude a definitive appraisal. One further drawback inherent to the design of this study is that only patients requiring treatment were included, which may select against HIV ‘elite suppressors’.

Importantly, the impact of Bw4 on viral load after STI operated independently from pretreatment viral loads, indicating a prognostic power additional to that of pretreatment set-point viral load. Other factors already known to influence viral replication during and after treatment of HIV infection include viral ‘fitness’ parameters such as its sensitivity to cytokines [27], viral diversity before initiation of ART [28], and parameters of humoral immunity such as anti-p24 antibody titres [8]. Interestingly, neither the quantity nor the quality of HIV-specific CD8 T-cell responses has previously been found to be predictive for the ability to control HIV replication after STI [23, 29].

In conclusion, we show that the presence of HLA-Bw4 significantly impacts on the control of viral load after STI during chronic HIV infection. Whether the increased capacity to suppress HIV-1 replication associated with HLA-Bw4 warrants reappraisal of STI as a treatment option in selected patient populations depends on the findings of future studies.


We thank the patients for participating in the SHCS and in these treatment interruption trials, the nurses and physicians at the various SHCS centres for excellent patient care, the SHCS data centre in Lausanne for data management and Marie Christine Francioli for administrative assistance.

This study was financed in the framework of the Swiss HIV Cohort Study, supported by the Swiss National Science Foundation (SNF grant #33CS30-134277). M.S and C.H. were supported by the Swiss National Science Foundation (grants # PP00P3_128461/1 and 31003A_135677, respectively).