https://orcid.org/0000-0002-4638-1940
Abstract
HIV-1 replication capacity (RC) of transmitted/founder viruses may influence the further course of HIV-1 infection.
RCs of 355 whole-genome primary HIV-1 isolates derived from samples acquired during acute and recent primary HIV-1 infection (PHI) were determined using a novel high-throughput infection assay in primary cells. The RCs were used to elucidate potential factors that could be associated with RC during PHI.
Increased RC was found to be associated with increased set point viral load (VL), and significant differences in RCs among 13 different HIV-1 subtypes were discerned. Notably, we observed an increase in RCs for primary HIV-1 isolates of HIV-1 subtype B over a 17-year period. Associations were not observed between RC and CD4 count at sample date of RC measurement, CD4 recovery after initiation of antiretroviral treatment, CD4 decline in untreated individuals, and acute retroviral syndrome severity scores.
These findings highlight that RCs of primary HIV-1 isolates acquired during the acute and recent phase of infection are more associated with viral factors, that is set point VL, than with host factors. Furthermore, we observed a temporal increase in RC for HIV-1 subtype B viruses over a period of 17 years.
NCT00537966.
Virus and host factors are known to be associated with surrogate markers of human immunodeficiency virus-1 (HIV-1) pathogenicity in natural infection, such as set point viral load (VL) and CD4 count in untreated HIV-1 infection [1–5]. When investigating effects of viral and host factors on the set point VL, approximately 30%–50% of the variance was attributed to viral factors and approximately 20% to host genetic factors. Regarding host factors, amino acid changes in HLA types B and C have been previously identified, whereas specific virus genomic changes are only starting to be identified [6]. As for viral factors, set point VL and HIV-1 reservoir size are viral traits that have previously been shown to be heritable [5, 7, 8].
Replication capacity (RC) is a viral phenotype that has been measured by various assays in numerous HIV research projects to investigate its potential role as a virulence factor in pathogenicity. Higher RC was associated with lower CD4 counts [9–12], faster CD4 decline [11, 13, 14], higher VL [11, 12, 15], and higher set point VL [16, 17]. However, the majority of those studies investigated how specific regions (gag, pol, env) of the HIV-1 genome contribute to disease progression regarding RC and surrogate markers.
We recently developed a reliable and reproducible high-throughput RC assay to study replication-competent primary HIV-1 isolates from the well-defined patient population of acutely (< 90 days) and recently (90–180 days) HIV-1–infected individuals [18]. In this current study, we focus on potential associations of RC with surrogate pathogenicity markers known to be relevant in the course of HIV-1 infection: CD4 count, set point VL, and clinical symptoms during primary HIV-1 infection (PHI). In addition, we tested viral factors potentially associated with higher RC such as different HIV-1 subtypes and viral diversity. Finally, we investigated whether HIV-1 RC has changed over a period of 17 years.
Using the rich and detailed data collection and biobank of the Zurich Primary HIV Infection study (ZPHI) study, including full genome sequences, we aimed to study the association of RC and viral and host factors in a large patient population. In addition, we wanted to understand the evolution of RC over time, during the shifting stages of the HIV-1 pandemic.
METHODS
Study Population
We included 355 patients enrolled in the ZPHI, an open-label, nonrandomized, observational, multicenter study conducted at the University Hospital Zurich and 2 nontertiary hospitals in Zurich, Switzerland (clinicaltrials.gov NCT00537966) [19–21]. Clinical and laboratory parameters as well as blood samples were collected every 3 months. Plasma and peripheral blood mononuclear cells (PBMCs) were isolated from the blood samples and stored at −80°C for future use.
Definitions: Acute and Recent Infection, Estimated Date of HIV Infection
Acute and recent HIV-1 infection was defined as being diagnosed within 90 days and 91–180 days after the estimated date of HIV infection (EDI), respectively. The EDI was determined based on unambiguous risk behavior, clinical (eg, acute retroviral syndrome) and laboratory data (eg, documented seroconversion, INNO-LIA, p24 antigen, western blot) as published elsewhere [19, 22].
Measurement of HIV-1 RC
The development and optimization of the assay to determine HIV-1 RC used here was previously described in great detail [18]. Briefly, RC was determined based on primary HIV-1 isolates generated from the first available sample acquired from the first clinical visit during acute/recent phase of the infection. Three-way stimulated CD8-depleted PBMCs were infected with the primary HIV-1 isolates in quadruplicates. On days 2, 3, 4, 5, 6, and 7, cell-free virus containing supernatant was transferred to TZM-bl reporter cells. After 24 hours incubation, the TZM-bl cells were lysed and the luciferase expression was measured. In total, we used 24 relative light units (RLU) measurements per patient, 4 replicates per day for day 2 to 7 (Figure 1A). For each day, the median of the 4 replicates was taken. The trajectories of day 2 to 7 of these median values were smoothed with respect to drops in measurements of more than 1 log (Figure 1B). Thereafter, the area under the curve (AUC) of these smoothed curves were used to assign 1 value per patient, further referred to as RC (Figure 1C). Previously, the experiment was repeated twice using PBMCs from 6 different donors, 3 donors combined per experiment, to show the robustness of this assay [18].
Raw RLU HIV-1 RC observations. A, All replicates for the 355 patient samples for each day of observation. There were 4 replicates per day per patient. B, Replication kinetics measured as the median value over the 4 replicates for each patient and day. A and B, The x-axis depicts the day when the replication measurement was acquired postinfection, and the y-axis represents the RC value in RLU in log scale. C, The violin plot depicts the distribution of the derived AUC values in log scale for the 355 samples. The light box indicates the median and interquartile range of the AUC values, the vertical line the 95% confidence interval. Abbreviations: AUC, area under the curve; HIV-1, human immunodeficiency virus-1; RC, replication capacity; RLU, relative light unit.
Average Pairwise Diversity
The average pairwise diversity (APD) was determined using a method previously developed [23, 24]. In this method, the APD was determined based on the third-codon positions of the gag or pol regions of sequences generated by next-generation sequencing (NGS). For this study, third-codon positions of near full-length genomes sequenced with NGS were used [23, 25].
Set Point Viral Load
Set point VL was determined by taking the median of all RNA measurements of antiretroviral therapy (ART)-naive patients sampled starting 30 days after the upper range of the EDI [26]. Due to potentially high VLs occurring within acute infection, the upper range plus 30 days and all values of the EDI (30 days post-EDI) was used when available to exclude HIV-1 RNA measurements in the acute peak of viremia. We chose to consider HIV-1 RNA measurements already after 30 days based on the study by Robb et al showing that set point VL is reached rapidly after infection [26].
For determining set point VL after structured treatment interruption, patients were selected when they had undergone treatment interruption for at least 90 days, and the RNA laboratory values were selected starting 30 days after treatment interruption. Set point VL was then determined using the same protocol described above.
CD4 Recovery Rate and Decline 12 Months After Sample Date
To determine if CD4 recovery or decline correlates with RC of the primary HIV-1 isolates, we performed 2 different analyses. For CD4 recovery, we utilized all available CD4 values after the start of ART up to 12 months after sample date of RC measurement. For CD4 decline before the start of ART, we utilized all available CD4 values up to 12 months after sample date, or up to the start of ART, whichever date comes first. Hence, patients starting ART within the first 12 months could contribute to both analyses, that is CD4 decline while untreated and CD4 recovery after treatment initiation. In both analyses, only patients with CD4 measurements spanning at least 30 days were included.
For each analysis, we attempted 2 approaches for determining CD4 recovery and decline. In the first approach, termed “difference approach,” the difference between the first and last included CD4 value was calculated, then divided by the time difference in days. In the second approach, termed “regression approach,” we performed a linear regression using all included CD4 counts to estimate a slope for each patient individually.
Acute Retroviral Syndrome Severity Scores
The acute retroviral syndrome severity scores (ARSSS) were defined and assessed in [27]. Briefly, the score, ranging between 0 and 10 points, is based on 6 variables of specified weight: fever (1 point), elevated liver enzymes (1 point), thrombocytopenia (1 point), age >50 years (1 point), inpatient treatment (3 points), and severe neurological symptoms (3 points). The ARSSS was shown to correlate with the surrogate markers of HIV-1 disease progression, namely CD4+ T-cell count and HIV-1 RNA levels, as well as set point VL.
Statistical Analysis
The statistical analysis was performed using R. Linear regressions were estimated using the lm function in the stats package, using predict for estimating confidence intervals. The logarithm of RC was chosen as explanatory variable when comparing RC with CD4 at sample date of RC measurement (in square-root), CD4 recovery rate (2 approaches), CD4 decline (2 approaches), set point VL, ARSSS, and APD. HIV-1 RC and HIV-1 subtype (B or non-B) and dichotomized ARSSS (high or low) were compared using 2-sided t tests. For the time trend analysis, RC was taken as the dependent variable and the calendar year as the explanatory variable.
RESULTS
Basic Characteristics of the Study Population
The majority of the study population was male (n = 338, 95%), of white ethnicity (n = 298, 84%), and infected with HIV-1 subtype B (n = 254, 71.5%). Men having sex with men (MSM) and heterosexual transmission were the most prominent transmission groups: 78% (n = 276) and 18% (n = 62), respectively. The year of infection ranged from 2002 to 2019. Most patients underwent early ART (<90 days from EDI) (n = 228, 64%), while only 34 patients were ART naive and remained without treatment for longer than 12 months.
The primary HIV-1 isolates were obtained during the acute (n = 283, 80%) and recent (n = 72, 20%) phase of infection. The EDI ranged from 12 to 180 days before sample date, with a median of 44 days. The sample population was largely represented by samples taken within 0–60 days after infection (n = 235, 66.3%). A detailed overview of baseline characteristics is shown in Table 1.
Patient Demographics and Sample Information
| Characteristic | Value |
|---|---|
| Total No. | 355 |
| Sex | |
| Male | 338 (95.2) |
| Female | 17 (4.8) |
| Birth year, median (range) | 1973 (1932–1995) |
| Ethnicity | |
| White | 298 (83.9) |
| Black | 16 (4.5) |
| Hispano-American | 28 (7.9) |
| Asian | 5 (1.4) |
| Unknown or other | 8 (2.3) |
| HIV-1 subtype | |
| B | 254 (71.5) |
| CRF01_AE | 30 (8.5) |
| A | 19 (5.4) |
| C | 12 (3.4) |
| CRF02_AG | 10 (2.8) |
| D | 3 (0.8) |
| Other | 30 (8.5) |
| Transmission group | |
| MSM | 276 (77.7) |
| Heterosexual | 62 (17.5) |
| IDU | 9 (2.5) |
| Other | 8 (2.3) |
| Year of infection, median (range) | 2009 (2002–2019) |
| ART treatment start | |
| Very early, <90 d | 228 (64.2) |
| Early, >90 d | 52 (14.6) |
| Late, >180 d | 16 (0.05) |
| ART naive | 34 (0.1) |
| Samples | |
| Acute | 283 (79.7) |
| Recent | 72 (20.3) |
| Estimated days since infection, median (range) | 44 (12–180) |
| Sample timing | |
| < 30 d after EDI | 87 (24.6) |
| 30–60 d after EDI | 148 (41.7) |
| 60–90 d after EDI | 54 (15.2) |
| > 90 d after EDI | 66 (18.6) |
| CD4 cell count at sample date, median (range) | 375 (87–1304) |
| CD4 cell count, cells/µL | |
| <200 | 38 (11) |
| 200–500 | 212 (59.7) |
| >500 | 95 (27.5) |
| Log viral load/mL, median (range) | 12.5 (5.4–18.2) |
| Days in culture, median (range) | 19 (5–96) |
| TCID50 /mL, median (range) | 5.2 (3.4–8.2) |
| Characteristic | Value |
|---|---|
| Total No. | 355 |
| Sex | |
| Male | 338 (95.2) |
| Female | 17 (4.8) |
| Birth year, median (range) | 1973 (1932–1995) |
| Ethnicity | |
| White | 298 (83.9) |
| Black | 16 (4.5) |
| Hispano-American | 28 (7.9) |
| Asian | 5 (1.4) |
| Unknown or other | 8 (2.3) |
| HIV-1 subtype | |
| B | 254 (71.5) |
| CRF01_AE | 30 (8.5) |
| A | 19 (5.4) |
| C | 12 (3.4) |
| CRF02_AG | 10 (2.8) |
| D | 3 (0.8) |
| Other | 30 (8.5) |
| Transmission group | |
| MSM | 276 (77.7) |
| Heterosexual | 62 (17.5) |
| IDU | 9 (2.5) |
| Other | 8 (2.3) |
| Year of infection, median (range) | 2009 (2002–2019) |
| ART treatment start | |
| Very early, <90 d | 228 (64.2) |
| Early, >90 d | 52 (14.6) |
| Late, >180 d | 16 (0.05) |
| ART naive | 34 (0.1) |
| Samples | |
| Acute | 283 (79.7) |
| Recent | 72 (20.3) |
| Estimated days since infection, median (range) | 44 (12–180) |
| Sample timing | |
| < 30 d after EDI | 87 (24.6) |
| 30–60 d after EDI | 148 (41.7) |
| 60–90 d after EDI | 54 (15.2) |
| > 90 d after EDI | 66 (18.6) |
| CD4 cell count at sample date, median (range) | 375 (87–1304) |
| CD4 cell count, cells/µL | |
| <200 | 38 (11) |
| 200–500 | 212 (59.7) |
| >500 | 95 (27.5) |
| Log viral load/mL, median (range) | 12.5 (5.4–18.2) |
| Days in culture, median (range) | 19 (5–96) |
| TCID50 /mL, median (range) | 5.2 (3.4–8.2) |
Data are No. (%) except where indicated.
Abbreviations: ART, antiretroviral therapy; EDI, estimated date of HIV infection; HIV, human immunodeficiency virus; IDU, injection drug user; MSM, men who have sex with men; TCID50, 50% tissue culture infectious dose.
Patient Demographics and Sample Information
| Characteristic | Value |
|---|---|
| Total No. | 355 |
| Sex | |
| Male | 338 (95.2) |
| Female | 17 (4.8) |
| Birth year, median (range) | 1973 (1932–1995) |
| Ethnicity | |
| White | 298 (83.9) |
| Black | 16 (4.5) |
| Hispano-American | 28 (7.9) |
| Asian | 5 (1.4) |
| Unknown or other | 8 (2.3) |
| HIV-1 subtype | |
| B | 254 (71.5) |
| CRF01_AE | 30 (8.5) |
| A | 19 (5.4) |
| C | 12 (3.4) |
| CRF02_AG | 10 (2.8) |
| D | 3 (0.8) |
| Other | 30 (8.5) |
| Transmission group | |
| MSM | 276 (77.7) |
| Heterosexual | 62 (17.5) |
| IDU | 9 (2.5) |
| Other | 8 (2.3) |
| Year of infection, median (range) | 2009 (2002–2019) |
| ART treatment start | |
| Very early, <90 d | 228 (64.2) |
| Early, >90 d | 52 (14.6) |
| Late, >180 d | 16 (0.05) |
| ART naive | 34 (0.1) |
| Samples | |
| Acute | 283 (79.7) |
| Recent | 72 (20.3) |
| Estimated days since infection, median (range) | 44 (12–180) |
| Sample timing | |
| < 30 d after EDI | 87 (24.6) |
| 30–60 d after EDI | 148 (41.7) |
| 60–90 d after EDI | 54 (15.2) |
| > 90 d after EDI | 66 (18.6) |
| CD4 cell count at sample date, median (range) | 375 (87–1304) |
| CD4 cell count, cells/µL | |
| <200 | 38 (11) |
| 200–500 | 212 (59.7) |
| >500 | 95 (27.5) |
| Log viral load/mL, median (range) | 12.5 (5.4–18.2) |
| Days in culture, median (range) | 19 (5–96) |
| TCID50 /mL, median (range) | 5.2 (3.4–8.2) |
| Characteristic | Value |
|---|---|
| Total No. | 355 |
| Sex | |
| Male | 338 (95.2) |
| Female | 17 (4.8) |
| Birth year, median (range) | 1973 (1932–1995) |
| Ethnicity | |
| White | 298 (83.9) |
| Black | 16 (4.5) |
| Hispano-American | 28 (7.9) |
| Asian | 5 (1.4) |
| Unknown or other | 8 (2.3) |
| HIV-1 subtype | |
| B | 254 (71.5) |
| CRF01_AE | 30 (8.5) |
| A | 19 (5.4) |
| C | 12 (3.4) |
| CRF02_AG | 10 (2.8) |
| D | 3 (0.8) |
| Other | 30 (8.5) |
| Transmission group | |
| MSM | 276 (77.7) |
| Heterosexual | 62 (17.5) |
| IDU | 9 (2.5) |
| Other | 8 (2.3) |
| Year of infection, median (range) | 2009 (2002–2019) |
| ART treatment start | |
| Very early, <90 d | 228 (64.2) |
| Early, >90 d | 52 (14.6) |
| Late, >180 d | 16 (0.05) |
| ART naive | 34 (0.1) |
| Samples | |
| Acute | 283 (79.7) |
| Recent | 72 (20.3) |
| Estimated days since infection, median (range) | 44 (12–180) |
| Sample timing | |
| < 30 d after EDI | 87 (24.6) |
| 30–60 d after EDI | 148 (41.7) |
| 60–90 d after EDI | 54 (15.2) |
| > 90 d after EDI | 66 (18.6) |
| CD4 cell count at sample date, median (range) | 375 (87–1304) |
| CD4 cell count, cells/µL | |
| <200 | 38 (11) |
| 200–500 | 212 (59.7) |
| >500 | 95 (27.5) |
| Log viral load/mL, median (range) | 12.5 (5.4–18.2) |
| Days in culture, median (range) | 19 (5–96) |
| TCID50 /mL, median (range) | 5.2 (3.4–8.2) |
Data are No. (%) except where indicated.
Abbreviations: ART, antiretroviral therapy; EDI, estimated date of HIV infection; HIV, human immunodeficiency virus; IDU, injection drug user; MSM, men who have sex with men; TCID50, 50% tissue culture infectious dose.
The Influence of Viral Factors on RC
Association Between APD, EDI, and RC
APD scores were determined for the full-length sequences of 306/355 primary HIV-1 isolates after passing quality controls (read length, coverage, etc.). The median APD score was 0.002 (interquartile range [IQR] = 0.0015–0.0049), with a lower median score for isolates acquired early after the EDI (Figure 2A). Increased RC was significantly associated with higher diversity (n = 306, slope = 0.064, P value = .032; Figure 2B); however, not after correcting the regression model for EDI and HIV-1 subtype (slope = 0.036, P value = .25).
Analysis of average pairwise diversity score in relation to RC of primary HIV-1 isolates, and EDI. A, The box plot displays the average pairwise diversity scores of primary HIV-1 isolates in relation to the corresponding EDI. The x-axis depicts the EDI in days, and the y-axis represents the average pairwise diversity score. The mean AUC value is depicted by the black line located in each box, the box represents the range from the lower to the upper quartile, and the range of the diversity scores is depicted with the whiskers above and below the boxes with values removed by more than 1.5 times the interquartile range from the box depicted as circles. B, The scatter plot depicts the average pairwise diversity scores of 306 primary HIV-1 isolates compared to the corresponding RCs. The x-axis represents the RC as AUC values in log scale, and the y-axis represents the average pairwise diversity in log scale. The solid line represents the fitted OLS regression and the dashed lines the 95% confidence intervals. Abbreviations: AUC, area under the curve; EDI, estimated date of HIV infection; HIV-1, human immunodeficiency virus-1; OLC, ordinary least squares; RC, replication capacity.
Differences in RC Among HIV-1 Subtypes
We observed a large variation in the RC distributions among all HIV-1 subtypes (Figure 3). Notably, HIV-1 subtype B had the highest RC (Figure 3). In contrast, the RC of the HIV-1 recombinant forms CRF01_AE, CRF12_BF, and CRF20_BG exhibited low RC. The median RC was 5.64 (IQR = 5.06–6.19) for HIV-1 subtype B and 4.82 (IQR = 4.43–5.56) for non-B (P value < .001) Table 2.
Mean Replication Capacity (RC) Value and the Number of Observations for the 10 HIV-1 Subtypes
| HIV-1 Subtype | Mean RC | No. |
|---|---|---|
| A | 4.95 | 19 |
| B | 5.60 | 254 |
| C | 5.19 | 12 |
| CRF01_AE | 4.77 | 30 |
| CRF02_AG | 5.34 | 10 |
| CRF12_BF | 4.69 | 2 |
| CRF20_BG | 4.41 | 3 |
| D | 5.24 | 3 |
| F | 5.64 | 7 |
| G | 5.05 | 5 |
| HIV-1 Subtype | Mean RC | No. |
|---|---|---|
| A | 4.95 | 19 |
| B | 5.60 | 254 |
| C | 5.19 | 12 |
| CRF01_AE | 4.77 | 30 |
| CRF02_AG | 5.34 | 10 |
| CRF12_BF | 4.69 | 2 |
| CRF20_BG | 4.41 | 3 |
| D | 5.24 | 3 |
| F | 5.64 | 7 |
| G | 5.05 | 5 |
Mean Replication Capacity (RC) Value and the Number of Observations for the 10 HIV-1 Subtypes
| HIV-1 Subtype | Mean RC | No. |
|---|---|---|
| A | 4.95 | 19 |
| B | 5.60 | 254 |
| C | 5.19 | 12 |
| CRF01_AE | 4.77 | 30 |
| CRF02_AG | 5.34 | 10 |
| CRF12_BF | 4.69 | 2 |
| CRF20_BG | 4.41 | 3 |
| D | 5.24 | 3 |
| F | 5.64 | 7 |
| G | 5.05 | 5 |
| HIV-1 Subtype | Mean RC | No. |
|---|---|---|
| A | 4.95 | 19 |
| B | 5.60 | 254 |
| C | 5.19 | 12 |
| CRF01_AE | 4.77 | 30 |
| CRF02_AG | 5.34 | 10 |
| CRF12_BF | 4.69 | 2 |
| CRF20_BG | 4.41 | 3 |
| D | 5.24 | 3 |
| F | 5.64 | 7 |
| G | 5.05 | 5 |
Distribution of the RCs of primary HIV-1 isolates relative to HIV-1 subtype. A, The boxplot depicts the RCs of HIV-1 subtypes with 10 or more primary HIV-1 isolates within a given subtype category The x-axis represents the HIV-1 subtype; n represents the number of primary HIV-1 isolates corresponding to each HIV-1 subtype. The y-axis depicts the AUC values in log-scale. The mean AUC value is depicted by the black line located in each box, the box represents the range from the lower to the upper quartile, and the range of AUC values is depicted with the whiskers above and below the boxes with values removed by more than 1.5 times the interquartile range from the box depicted as circles. Abbreviations: AUC, area under the curve; HIV-1, human immunodeficiency virus-1; RC, replication capacity.
Change in RC of Primary HIV-1 Isolates Over a Period of 17 Years
The 355 primary HIV-1 isolates were generated from patient samples over a period of 17 years (Figure 4). The RC of the 355 HIV-1 primary isolates did not significantly change from 2002 to 2019 (slope = 0.0008, P value = .92; Figure 4A). However, when restricting to subtype B isolates, the main subtype transmitted continuously throughout Switzerland in the MSM community [28], there was a significant increase in RC (slope = 0.0223, P value = .043; Figure 4B).
RC of primary HIV-1 isolates by year of infection, for all HIV-1 subtypes (A) and HIV-1 subtype B (B). The x-axes represent the year of infection and the y-axes represent the RC as AUC values in log scale. The solid line represents the fitted OLS regression and the dashed lines the 95% confidence interval. Abbreviations: AUC, area under the curve; HIV-1, human immunodeficiency virus-1; RC, replication capacity.
Effect of RC on Surrogate Markers of Pathogenicity of HIV Infection
Association Between RC and Set Point VL
We observed that high RC was significantly associated with higher set point VL (slope = 0.163, P = .046, n = 212; Figure 5A). Interestingly, the association remained significant (slope = 0.242, P = .019) when restricting to HIV-1 subtype B (n = 156) (Figure 5B), and after correcting for the host demographic factors age and sex (slope = 0.192, P = .007). In addition, we compared the set point VL for the patients that underwent structured treatment interruption to the corresponding RC values and observed that high RC was significantly associated with higher set point VL (n = 71, P = .016; Figure 5C).
Association between set point viral load and replication capacity of primary HIV-1 isolates for all HIV-1 subtypes (A), HIV-1 subtype B (B), and for patients that underwent structured treatment interruption (C). The x-axes represent the replication capacity measured as AUC values in log scale, and the y-axes represent the set point viral load value in log scale. The solid lines represent the fitted OLS regression and the dashed lines the 95% confidence intervals. Abbreviations: AUC, area under the curve; HIV-1, human immunodeficiency virus; OLC, ordinary least squares.
Comparison of CD4 Values at Sample Date and RC of Primary HIV-1 Isolates
We investigated the association between CD4 count at the sample date of RC measurement and RC of 280 primary HIV-1 isolates (Figure 6A). We observed a negative correlation between the CD4 count at sample date and RC, although the association was not statistically significant (slope = −0.52, P = .14).
Association of RC to CD4 count at sample date of RC measurement, CD4 recovery, and CD4 decline. A, The scatter plot displays the linear regression between CD4 count at sample date and RC of primary HIV-1 isolates. The x-axis represents the RC measured as AUC values in log-scale, and the y-axis CD4 count in log-scale. B and C, The scatter plots depict the relationship between the slope of the CD4 recovery (B) or the slope of CD4 decline (C) and RC. The x-axes in both plots represents the RC measured as AUC values in log-scale. The y-axes represent estimated slopes of the CD4 recovery of ART treated patients (B) and the CD4 decline in ART-naive patients (C). For both graphs, the slope was measured as the difference between the first and the last included CD4 value. The solid lines represent the fitted OLS regression and the dashed lines the 95% confidence intervals. Abbreviations: AUC, area under the curve; HIV-1, human immunodeficiency virus-1; RC, replication capacity.
CD4 Cell Recovery 12 Months After the EDI in ART Treated Patients
For the first approach (difference approach, see “Methods” section), we observed no significant association between CD4 cell recovery and RC (n = 277, slope = 0.042, P value = 0.532; Figure 6B). The same was observed when using the second approach (regression approach; Supplementary Figure 2A). Furthermore, we restricted the sample population to patients that underwent early ART treatment (started within 90 days of EDI) and compared the CD4 cell recovery to RC. We did not observe a significant association for both approaches (Supplementary Figure 3A and B).
CD4 Cell Decline 12 Months After EDI in ART-Naive Patients
There was no significant association between CD4 cell decline and RC (difference approach; n = 47, slope = 0.158, P value = .529; Figure 6C). Similar results were obtained for the regression approach (Supplementary Figure 2B).
Comparison of Acute Retroviral Syndrome Severity Scores to RC
The median ARSSS was 2 (IQR = 1–3) for the 334 patients included in this analysis (Supplementary Figure 4). There was no significant association between ARSSS and RC (slope = 0.006, P value = .771) or after dichotomizing the ARSSS using 3 as cutoff (t test; P = 1).
DISCUSSION
In this study, we performed a systematic analysis of RC from 355 primary HIV-1 isolates obtained early after HIV-1 infection from individuals enrolled in the ZPHI over a 17-year period. The major findings were (1) higher RC was associated with higher set point VL, (2) there was an increase of RC over a 17-year period for HIV-1 subtype B isolates, and (3) RC was higher for subtype B isolates compared to non-B isolates.
In line with other research groups, increased RC was significantly associated with increased set point VL [11, 16, 17]. In addition, high set point VL was significantly associated with high RC in patients that underwent structured treatment interruption. This could suggest that RC has predictive power and that the variants that have an originally high RC are the ones that replicate after stopping treatment. Furthermore, high set point VL is known to be a useful predictor for fast HIV-1 disease progression [29]. There was no association found between RC and the previously developed and validated ARSSS [27, 30]. This could be explained by the fact that in vivo the host immune response is responsible for a large part of the ARS, and thus cannot be captured by this ex vivo virus growth assay alone.
HIV-1 is a rapidly evolving virus, and it has been highly debated whether it is evolving to become more or less virulent over time (reviewed in [31]). Our study shows an increase in RC over time in HIV-1 subtype B samples. As higher RC was significantly associated with higher set point VL, a predictor for disease progression, this might be an indication for increasing virulence of HIV-1 subtype B. A previous study from the Swiss HIV Cohort Study showed that there was stable virulence over a period of 2 decades [32]. However, this analysis was performed 14 years ago and the HIV-1 epidemic in Switzerland has changed over time with the majority of HIV-infected individuals successfully treated with ART and virally suppressed [33]. Currently, due to immediate start of ART in Switzerland, set point VL cannot be determined in the majority of patients. Other studies based on set point VL have suggested that there may be an increase of virulence over time [34, 35]. What could explain the increase in virulence observed in our study is that we primarily studied very early isolates, essentially transmitter/founder viruses. An increase in RC of this unique group of well-documented early isolates might be lost in a large population of combined viruses of unknown EDI and potential chronic virus isolates. Since 2010, the vast majority of patients in Switzerland, particularly in Zurich where the ZPHI study takes place, were treated rapidly after diagnosis and independent of the CD4 count. This finding could illustrate that by increasing the proportion of ART-treated and suppressed individuals over time, there could be a selection for viruses with increased virulence in the diminishing pool of remaining transmitters. In line with our study finding, a modelling prediction study suggested universal ART could potentially lead to an increase in virulence [36]. Furthermore, a recent study investigating the evolution of the HIV-1 envelope glycoprotein for subtype B in 40 patients over 3 time periods found that envelopes from patients in the more recent years were more infectious and had faster viral entry than the envelopes obtained at the start of the HIV-1 pandemic [37].
Furthermore, in the univariable analysis, higher viral diversity was associated with higher RC. Previous studies have hypothesized that higher RC is indicative of infection due to multiple HIV-1 variants [38]. In our study, the association of viral diversity and RC was not significant after correcting for HIV-1 subtype and EDI, suggesting a confounding effect. Interestingly, we observed outliers well above the median APD score of 0.002 and outside of the upper and lower range of diversity scores for the first 60 days of infection (Figure 2A). This could suggest that a small portion of the early isolates consist of multiple variants, in line with previous findings [38]. We did not observe a significant correlation between RC and the early isolates with high diversity, which suggests that high RC is not a feature of multiple variant transmission. A study by Villabona-Arenas et al found an increased chance of multiple variant transmission during the acute stage of infection across both MSM and heterosexual transmission groups compared to transmission during the chronic stage of infection [39]. However, another study reported that multiple particles of the same variant are transmitted rather than multiple variants [40], because an acute infection is associated with a homogeneous viral population within the infected individual. Although our sample population consisted mainly of MSM and heterosexual transmissions that potentially occurred during the acute or recent phase of infection due to early start of ART, our observations do not suggest an increase in multiple variant transmission. We plan to further explore diversity of the sequence populations of our primary HIV-1 isolates in more depth.
We observed significant differences among 13 HIV-1 subtypes, as well as HIV-1 subtype B having significantly higher RC compared to non-B subtypes. Of the HIV-1 subtypes with 10 or more primary HIV-1 isolates within a given subtype category, we observed the following order of mean RC from lowest to highest: CRF01_AE < A < C < CRF02_AG < B. Similar findings have been observed by other research groups [41, 42]. Given that our primary HIV-1 isolates consist of the whole HIV-1 genome, these RC variations could be attributed to differences in the genome, such as in the Gag protease region [43], reverse transcriptase [44], protease [45], Vif [46], Nef [47], Env [48], or in the LTR region [49].
Although CD4 count at the sample date of RC measurement, CD4 recovery after initiation of ART, and CD4 decline in ART-naive patients has been found to be associated with viral fitness and RC [10, 12], we and others have not observed these findings [41]. These inconsistent findings could be influenced by varying experimental formats, sample size, and restricted viral genomes being analyzed. Surprisingly, there are few studies with a large sample size utilizing the whole HIV-1 genome (primary HIV-1 isolates) to determine RC, and even fewer utilizing primary HIV-1 isolates derived from patient samples acquired during early and/or recent PHI. This could be due to the labor-intensive nature of deriving primary HIV-1 isolates, and limited opportunities to obtain patient samples during early PHI. Another potential reason could be the time point the sample was taken, whether the individual was ART naive or started an ART regimen when the sample was acquired, or whether there were any other illnesses that the patient may have had when the blood sample was acquired leading to an increased VL. Furthermore, it has been observed that the season of the year can affect the CD4 cells circulating in the blood due to allergies and seasonal illnesses, such as influenza [50].
Although we had a large sample population of very well characterized PHI patients and used a robust high-throughput replication assay to determine the RCs of 355 primary HIV-1 isolates, one limitation of this study is that for the time trend analysis, the power per time point for time (in years) was low. Another limitation is that determination of set point VL was not feasible for every patient due to the early start of treatment. However, we were able to determine the set point VL for the patients who underwent structured treatment interruption, which allowed us to determine the set point VL for 80% of our sample population and further confirm the significant association between high set point VL and high RC. A further limitation of this study is that most included participants were white MSM infected with HIV-1 subtype B, and hence characteristics such as sex, transmission group, and ethnicity could not be studied in detail.
In conclusion, we utilized RCs of 355 primary HIV-1 isolates of very well documented acute and recently HIV-1 infected patients to determine potential factors influenced by HIV-1 RC. We elucidated that RC is associated with viral factors, such as set point VL, HIV-1 subtype, and viral diversity. We found no association with host CD4 cell counts (at sample date and recovery) or the ARSSS, the latter reflecting severity of acute retroviral syndrome during PHI. Finally, we detected an increased trend in RCs of HIV-1 subtype B viruses over a 17-year period.
Supplementary Data
Supplementary materials are available at The Journal of Infectious Diseases online. Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.
Notes
The initial ZPHI protocol was approved by the ethics committee of Zurich in September 2007 (KEK-ZH-Nr, EK-1452). An amendment of the protocol was approved by the same ethics committee in October 2015 (KEK-ZH-Nr. EK-1452). Prior to the inclusion in the ZPHI study, the patients provided written informed consent
Acknowledgments. We are very appreciative to all the patients who participate in the Zurich Primary HIV-1 Infection (ZPHI) study, and the physicians and Christina Grube for excellent patient care.
Author contributions. A. E. R., K. K., H. K., R. K., K. J. M., and H. F. G. conceived and designed the study. A. E. R., H. K., K. N., and C. L. performed experiments. A. E. R. and K. K. analyzed the data. K. K. generated all graphs and performed all statistical analysis. S. E. C. provided the pipeline for generating consensus sequences and M. Z. provided the average pairwise diversity scores. D. L. B. provided the acute retroviral syndrome severity scores and contributed to the clinical aspects of the study. A. E. R. wrote the first draft of the manuscript. All authors read, reviewed, and approved the final manuscript.
Financial support. This work was supported by the Swiss National Science Foundation (grant number 179 571 to H. F. G.).
Potential conflicts of interest. K. J. M. has received travel grants and honoraria from Gilead Sciences, Roche Diagnostics, GlaxoSmithKline, Merck Sharp & Dohme, Bristol-Myers Squibb, ViiV and Abbott; and the University of Zurich received research grants from Gilead Science, Novartis, Roche, and Merck Sharp & Dohme for studies in which K. J. M. serves as principal investigator, and advisory board honoraria from Gilead Sciences. D. L. B. received honoraria for advisory boards from Gilead, MSD, and ViiV outside of the submitted work. H. F. G. has received unrestricted research grants from Gilead Sciences; fees for data and safety monitoring board membership from Merck; consulting/advisory board membership fees from Gilead Sciences, Merck, ViiV, Janssen and Novartis; and grants the Swiss National Science Foundation, National Institutes of Health and the Yvonne Jacob Foundation.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed
References
Author notes
A. E. R. and K. K. contributed equally.
