Antiretroviral Therapy Within 2 Years of Human Immunodeficiency Virus Acquisition Is Associated With Fewer Viral Blips: A Retrospective Analysis of More Than 20 Years of Data From the US Military HIV Natural History Study

Abstract

Background

Viral blips have been associated with larger reservoir size and slower decay. Earlier antiretroviral therapy (ART) initiation may decrease the risk of blips.

Methods

We analyzed participants from the US Military HIV Natural History Study with an estimated human immunodeficiency virus (HIV) seroconversion date, viral suppression ≤400 copies/mL within 1 year after starting ART, and at least 3 HIV RNA measurements after suppression. A blip was defined as HIV RNA 401–1000 copies/mL preceded and followed by HIV RNA ≤400 copies/mL without changing ART. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for factors potentially associated with the time from viral suppression to first blip.

Results

From 1996 through 2022, among 1413 participants on stable suppressive ART, 88 (6.2%) had at least one blip, including 68 (77.3%) with only a single blip. The overall incidence was 1.2 blips per 100 person-years (95% CI: .9–1.4). In multivariable modeling, ART initiation within 24 months of estimated HIV acquisition was independently associated with decreased hazard of viral blips compared with ART initiation after more than 24 months (0–6 months HR: 0.29 and 95% CI: .18–.48; 6–12 months HR: 0.43 and 95% CI: .31–.59; 12–24 months HR: 0.46 and 95% CI: .35–.60).

Conclusions

Participants who initiated ART within 2 years of HIV acquisition had lower hazard of blips, potentially reflecting smaller reservoir size and suggesting reservoir plasticity that extends beyond the acute phase of HIV.

Graphical Abstract

acquired immunodeficiency syndrome, antiretroviral agents, anti-HIV agents, viremia, reservoir

In the setting of otherwise suppressive antiretroviral therapy (ART), transient low-level viremias, or viral blips, are common [1]. They can be concerning for both patients and healthcare providers and may lead to increased resource utilization, such as more frequent viral load testing. Most blips, especially those <200 copies/mL, appear to represent random biological fluctuations, statistical variations, and artifacts of human immunodeficiency virus (HIV) RNA assay variability [2–4]. In some cases, blips may result from poor ART adherence [5, 6] or drug–drug interactions that reduce ART bioavailability [7]. In other cases, blips may represent bursts of HIV replication and activation of viral reservoirs or may result from immune activation with expansion and contraction of latently infected cells [8–10]. Blips have been associated with larger size and slower decay of the reservoir of cells that harbor proviral HIV DNA [11].

Long-lived HIV reservoirs represent the major barrier to HIV cure [12, 13]. Despite rapid seeding of HIV reservoirs, there appears to be some early plasticity, with similar HIV DNA levels observed 6–12 months after ART initiation during any of the earliest stages of acute HIV (Fiebig stages I–V) [14] or in children with perinatally acquired HIV who initiate ART within up to 1 year after acquisition [15]. Compared with ART initiation during chronic HIV, initiation during acute HIV results in limited establishment of HIV reservoirs [16–19]. Some people with HIV (PWH) who start ART during acute or early HIV demonstrate delayed rebound viremia after treatment interruption, and approximately 5%–15% show sustained viral control [19–22]. ART initiation during acute HIV has also been associated with fewer blips on ART [23] and reduced chronic inflammation [24, 25]. However, it is unclear when the period of HIV reservoir plasticity ends and just how long after HIV acquisition “early” ART initiation continues to have a beneficial impact on the reservoir and improve the potential to achieve HIV remission.

While guidelines have shifted to encourage ART initiation at HIV diagnosis [26–28], ART initiation during acute HIV remains exceedingly rare and mostly limited to research settings [29]. In clinical care settings, a more achievable timeline for ART is within the first 3–12 months after HIV acquisition, considering that annual HIV screening is recommended for many vulnerable populations [30–32] and that every 3 months is standard for patients prescribed preexposure prophylaxis [33]. Understanding the impact of ART initiation within this window after HIV acquisition is critical in the current “test-and-start” era. If ART initiation within this window reduces HIV reservoirs or improves clinical outcomes, the guideline-recommended frequency of HIV screening to facilitate earlier ART may be reexamined and candidate populations considered for HIV remission studies may be expanded.

Measuring HIV reservoirs directly requires samples that are not routinely available in clinical practice, but viral loads are routinely monitored for all PWH, enabling evaluation of viral blips as an indirect measure of stochastic reactivation of latent HIV reservoirs [8, 9]. We hypothesized that ART initiation within 1 year of HIV acquisition would be associated with decreased frequency of viral blips compared with later ART initiation.

METHODS

Participants

Since 1986, the ongoing US Military HIV Natural History Study (NHS) has enrolled military personnel and medical beneficiaries with HIV who are aged ≥18 years. Demographic, laboratory, and clinical data are systematically collected every 6–12 months. CD4 count, HIV RNA, and other laboratory parameters are assessed according to local standards using routine clinical assays. These analyses included NHS participants who had an estimated HIV seroconversion date (negative test followed by positive test), accrued observation time after 1996 when ART became widely available in the United States, achieved viral suppression ≤400 copies/mL within 1 year after starting ART (confirmed by 2 measurements at least 14 days apart), maintained viral suppression for at least 6 months from the first documented HIV RNA ≤400 copies/mL, and had at least 3 HIV RNA measurements after achieving viral suppression. Participants prescribed monotherapy or dual therapies that would be considered obsolete by modern standards were included in analyses only if they achieved viral suppression and satisfied other inclusion criteria.

All participants provided written informed consent prior to enrollment. Institutional review boards at the Uniformed Services University, Bethesda, Maryland, and at all participating sites approved the study.

Viral Blips

After viral suppression ≤400 copies/mL was achieved, a blip was defined as any HIV RNA 401–1000 copies/mL immediately preceded and followed by HIV RNA ≤400 copies/mL without a change in ART. A threshold of 400 copies/mL was used because this represented the limit of HIV RNA quantification for the least sensitive of the many HIV RNA assays used in the study. Multiple measurements within 30 days that would otherwise satisfy blip criteria were considered a single blip.

Timeliness of ART Initiation

US military service members are routinely tested for HIV at least every 2 years, enabling the calculation of relatively precise seroconversion windows for many NHS participants [34, 35]. The date of HIV acquisition was estimated as the midpoint between the dates of the last documented negative and first documented positive HIV test. ART history was extracted from electronic medical records and confirmed through participant interviews. The date of estimated HIV acquisition was subtracted from the date of ART initiation to calculate the timing of ART initiation.

Covariates

Race and sex were categorized by self-report. Age was calculated using the birth date on official government identification. Time-updated ART regimen, CD4 count, HIV RNA, history of hepatitis B, and history of hepatitis C were extracted from the electronic medical record. Pharmacy records were leveraged to assess adherence using the medication possession ratio (MPR), which was calculated as the sum of days supplied divided by the number of days between refills. An MPR >90% was considered indicative of good adherence and was calculated once per participant over the entire observation period.

Statistical Analyses

Demographic and clinical characteristics were summarized using medians with interquartile ranges (IQRs) for continuous variables and counts with percentages for categorical variables. Kruskal–Wallis and χ² or Fisher exact tests were used to compare continuous and categorical variables, respectively, between participants with and without blips.

For time-to-event analyses, baseline was established after 6 consecutive months of viral suppression ≤400 copies/mL. Because multiple blips were uncommon, analyses focused on time to the first blip after viral suppression was achieved. Observation time was censored at last study visit, viral failure (any HIV RNA >1000 copies/mL), or HIV RNA 401–1000 copies/mL that did not meet the definition of blip (including persistent low-level viremia, measurement immediately preceding viral failure, and measurement immediately preceding loss to follow-up).

Cox proportional hazard models were used to estimate hazard ratios and associated 95% confidence intervals (CIs) for factors potentially associated with blips. Covariates with potentially significant associations (P < .20) in unadjusted models were considered for inclusion in the multivariable model, and adjusted hazard ratios (aHRs) were calculated. Further variable downselection was performed to avoid multicollinearity. Because a complete-case approach was used, participant visits with any missing data were excluded from the multivariable model. Multivariable modeling was repeated using the same independent variables in sensitivity analyses that excluded female participants since they represented a small minority of the population and redefined blips using HIV RNA 51–1000 copies/mL, reflecting a more modern and inclusive definition for blips but also restricting HIV RNA data to assessments made using clinical assays with a lower limit of quantification of ≤50 copies/mL.

All tests were 2-sided with P < .05 considered statistically significant. All analyses were conducted using SAS 9.4 (SAS Institute Inc, Cary, NC).

RESULTS

Study Population

Between 1 January 1996 and 1 December 2022, 6451 participants were enrolled in the NHS, including 4229 with an estimated date of HIV seroconversion, of whom 3784 had observation time in 1996 or later. Of these, 3143 started ART (including old monotherapy or dual therapies), 1986 achieved viral suppression ≤400 copies/mL within 1 year of ART initiation, 1870 maintained viral suppression for at least 6 months after their first HIV RNA ≤400 copies/mL, and 1413 had at least 3 HIV RNA assessments after first achieving viral suppression on ART. Among the 1413 participants included in these analyses, 88 (6.2%) experienced viral blips during observation. Among the 88 participants with blips, 68 (77.3%) experienced a single blip, 13 (14.8%) experienced 2 blips, 4 (4.5%) experienced 3 blips, and 3 (3.4%) experienced 4 blips. The overall incidence was 1.2 blips per 100 person-years (95% CI: .9–1.4).

Characteristics of the study population are summarized in Table 1. The median age at HIV diagnosis was 29.2 years (IQR, 24.9–35.4) and did not differ between participants with and without blips (31.1 and IQR, 24.7–37.7 vs 29.1 and IQR, 24.9–35.3; P = .2194). However, participants with blips were older at ART initiation than participants without blips (median, 34.9 and IQR, 30.1–39.3 vs 31.6 and IQR, 26.3–37.7; P = .0057) and were diagnosed with HIV during earlier calendar years (1997: IQR, 1994–2000 vs 2007: IQR, 1999–2012; P < .0001). Participants with blips had a longer median time from estimated HIV acquisition to ART initiation (33.9 months: IQR, 16.5–67.5 vs 16.6 months: IQR, 9.7–41.4; P < .0001) and a longer median time from estimated HIV acquisition to first viral suppression <400 copies/mL on ART (40.0 months: IQR, 23.9–72.0 vs 22.5 months: IQR, 15.0–47.8; P < .0001).

Table 1.

Study Population Characteristics at the Time of Antiretroviral Therapy Initiation by Service Members and Their Medical Beneficiaries in the US Military HIV Natural History Study, Stratified by the Presence or Absence of Viral Blips After Antiretroviral Therapy Initiation

Characteristic	Overall (1413 Participants, 7559 Person-Years)	Participants With Blips (88 Participants, 177 Person-Years)	Participants With No Blips (1325 Participants, 7381 Person-Years)	P Value
Time from HIV acquisition to ART initiation, m				<.0001
≥24	568 (40.2)	55 (62.5)	513 (38.7)
12–24	403 (28.5)	22 (25.0)	381 (28.8)
6–12	287 (20.3)	5 (5.7)	282 (21.3)
0–6	155 (11.0)	6 (6.8)	149 (11.2)
Age, y				.0015
18–29	603 (42.7)	21 (23.9)	582 (43.9)
30–39	554 (39.2)	49 (55.7)	505 (38.1)
40–49	205 (14.5)	13 (4.8)	192 (14.5)
≥50	51 (3.6)	5 (5.7)	46 (3.5)
Sex				.1324
Male	1361 (96.3)	82 (93.2)	1279 (96.5)
Female	52 (3.7)	6 (6.8)	46 (3.5)
Race				.9529
White/Caucasian	562 (39.8)	35 (39.8)	527 (39.8)
Black/African American	594 (42.0)	38 (43.2)	556 (42.0)
Hispanic/Other	257 (18.2)	15 (17.0)	242 (18.3)
ART regimen				<.0001
Nonnucleoside reverse transcriptase inhibitor	643 (45.5)	25 (28.4)	618 (46.6)
Integrase strand transferase inhibitor	307 (21.7)	5 (5.7)	302 (22.8)
Boosted PI	151 (10.7)	9 (10.2)	142 (10.7)
Unboosted PI	234 (16.6)	39 (44.3)	195 (14.7)
Triple nucleoside reverse transcriptase inhibitor	8 (0.6)	1 (1.1)	7 (0.5)
Mono- or old dual-therapy	3 (0.2)	0 (0.0)	3 (0.2)
Other	67 (4.7)	9 (10.2)	58 (4.4)
CD4 count, cells/mm3				.0307
0–200	136 (9.6)	15 (17.0)	121 (9.1)
201–350	433 (30.6)	24 (27.3)	409 (30.9)
351–500	406 (28.7)	25 (28.4)	381 (28.8)
≥501	370 (26.2)	15 (17.0)	355 (26.8)
Unknown/Missing	68 (4.8)	9 (10.2)	59 (4.5)
HIV RNA, copies/mL				.3545
<10 000	372 (26.3)	27 (30.7)	345 (26.0)
10 000–99 999	667 (47.2)	34 (38.6)	633 (47.8)
≥100 000 copies/mL	316 (22.4)	20 (22.7)	296 (22.3)
Unknown/Missing	58 (4.1)	7 (8.0)	51 (3.8)
ART adherencea				<.0001
MPR <90%	549 (38.9)	64 (72.7)	485 (36.6)
MPR ≥90%	833 (59.0)	20 (22.7)	813 (61.4)
Unknown/Missing	31 (2.2)	4 (4.5)	27 (2.0)
History of hepatitis Bb				<.0001
No	1051 (74.4)	47 (53.4)	1004 (75.8)
Yes	294 (20.8)	31 (35.2)	263 (19.8)
Missing	68 (4.8)	10 (11.4)	58 (4.4)
History of hepatitis Cb				.2727
No	1314 (93.0)	76 (86.4)	1238 (93.4)
Yes	37 (2.6)	4 (4.5)	33 (2.5)
Missing	62 (4.4)	8 (9.1)	54 (4.1)

These analyses included participants in the US Military HIV Natural History Study who had an estimated HIV seroconversion date, accrued observation time after 1996, achieved viral suppression ≤400 copies/mL within 1 year after starting ART, maintained viral suppression for at least 6 months, and had at least 3 HIV RNA measurements after achieving viral suppression. χ² or Fisher exact tests were used to compare characteristics between participants with and without blips 401–1000 copies/mL. Statistically significant comparisons (P < .05) are highlighted in bold.

Abbreviations: ART, antiretroviral therapy; HIV, human immunodeficiency virus; MPR, medication possession ratio; PI, protease inhibitor.

^aART adherence was calculated from ART initiation to event/last visit.

^bHistory of hepatitis B and/or C was ascertained by medical record review at the time of ART initiation.

Temporal Trends

The frequency of routine HIV RNA testing decreased over time (Figure 1A). An average of 3 HIV RNA assessments per participant were performed in 1997; testing frequency peaked in 2000 at 3.7 tests per participant and then steadily declined to just 1 test per participant in 2022. The number of unique participant visits generally increased over time until 2020 through 2022, when visits were mostly paused due to the coronavirus disease 2019 pandemic.

Figure 1.

Annual trends in viral load testing and viral blip incidence. A, The number of viral load tests performed (red/right bars), number of participant visits completed (blue/left bars), and average testing rate calculated from these numbers (tests per visit, green line) for calendar years 1997–2022. B, The incidence of blips during these same calendar years, calculated as number of blips per 100 person-years of observation time. The following years have the same incidence rate (IR): 2003, 2005, and 2017 (IR = 0.5 per 100 person-years), 2007 and 2012 (IR = 0.4 per 100 person-years), 2008 and 2010 (IR = 1.2 per 100 person-years), and 2016 and 2018 (IR = 0.2 per 100 person-years).

The incidence of blips followed a similar pattern (Figure 1B), with a peak of 12.5 blips per 100 person-years (95% CI: 7.7–20.4) in 1999, downtrending to 0.9 blips per 100 person-years (95% CI: .2–3.6) in 2006, and plateauing at around that rate until 2014. Blip incidence dropped to around 0.0–0.6 blips per 100 person-years through 2020 (0.6 blips per 100 person-years; 95% CI: .1–4.4), which was the last year that a blip was observed.

Participants with blips tended to have a higher number of viral load tests per year (median, 4.4; IQR, 3.2–6.8 vs 2.5; IQR, 2.0–3.3; P < .0001).

Factors Associated With Viral Blips

In multivariable modeling that included 1310 participants with complete data, there was a decreased hazard of viral blips among participants who initiated ART within 0–6 months of estimated HIV acquisition (aHR, 0.29; 95% CI: .18–.48), 6–12 months (aHR, 0.43; 95% CI: .31–.59), or 12–24 months (aHR, 0.46; 95% CI: .35–.60) compared with ART initiation after more than 24 months (Table 2, Figure 2). Independent associations with race were also observed, with lower hazard of viral blips among those who identified as Black/African American (aHR, 0.64; 95% CI: .53–.77) and Hispanic/other (aHR, 0.48; 95% CI: .35–.67) compared with White/Caucasian. Compared with ART regimens that contained a nonnucleoside reverse transcriptase inhibitor, the lowest hazard of blips was seen with integrase strand transferase inhibitors (INSTIs; aHR, 0.44; 95% CI: .28–.69), and the highest hazard was seen with unboosted protease inhibitors (aHR, 3.23; 95% CI: 2.59–4.03). Consistent ART adherence was protective against blips (MPR ≥90%: aHR, 0.25; 95% CI: .20–.31 compared with MPR <90%). Hazard of blips was higher in participants with a history of hepatitis B (aHR, 1.22; 95% CI: 1.02–1.46) and lower in participants with a history of hepatitis C (aHR, 0.38; 95% CI: .22–.63). Associations between the timing of ART initiation and hazard of blips were consistent across sensitivity analyses (Table 3).

Figure 2.

Kaplan–Meier survival curves for factors associated with time to first viral blip. Kaplan–Meier curves were generated to visually explore factors that were identified as having significant associations with time to first HIV RNA 401–1000 copies/mL in the multivariable model. These included (A) time from estimated HIV aquisition to ART initiation; (B) age; (C) sex; (D) race; (E) ART regimen; (F) ART adherence; (G) history of hepatitis B; and (H) history of hepatitis C. Abbreviations: ART, antiretroviral therapy; HIV, human immunodeficiency virus; INSTI, integrase strand transferase inhibitor; NNRTI, nonnucleoside reverse transcriptase inhibitor; NRTI, nucleoside reverse transcriptase inhibitor; PI, protease inhibitor.

Table 2.

Factors Associated With Decreased Time to First Viral Blip

Characteristic	Unadjusted HR (95% CI)	Adjusted HR (95% CI)
Time from HIV to ART initiation, m
≥24	Reference	…
12–24	0.30 (.24–.37)	0.46 (.35–.0)
6–12	0.21 (.16–.29)	0.43 (.31–.59)
0–6	0.37 (.27–.51)	0.29 (.18–.48)
Age, y
18–29	Reference	…
30–39	1.12 (.86–1.45)	0.59 (.44–.78)
40–49	2.09 (1.61–2.71)	1.00 (.74–1.35)
≥50	3.18 (3.48–2.36)	1.39 (.99–1.95)
Sex
Male	Reference	…
Female	1.52 (1.06–2.17)	0.94 (.55–1.58)
Race
White/Caucasian	Reference	…
Black/African American	0.57 (.48–.67)	0.64 (.53–.77)
Hispanic/Other	0.45 (.35–.58)	0.48 (.35–.67)
ART regimen (time-updated)
Nonnucleoside reverse transcriptase inhibitor	Reference	…
Integrase strand transferase inhibitor	0.23 (.15–.37)	0.44 (.28–.69)
Boosted PI	1.98 (1.50–2.61)	1.94 (1.43–2.63)
Unboosted PI	6.13 (5.05–7.44)	3.23 (2.59–4.03)
Triple nucleoside reverse transcriptase inhibitor	6.20 (4.37–8.80)	2.80 (1.84–4.28)
Mono- or old dual-therapy	1.26 (.18–8.98)	1.34 (.18–9.72)
Other	5.55 (4.22–7.31)	2.82 (2.10–3.79)
CD4 count (time-updated), cells/mm3
0–200	Reference	…
201–350	1.33 (.75–2.34)	1.64 (.86–3.11)
351–500	0.83 (.48–1.43)	1.28 (.69–2.39)
≥501	0.41 (.24–.70)	0.83 (.45–1.52)
HIV RNA at ART initiation, copies/mL
<10 000	Reference	…
10 000–99 999	0.70 (.59–.84)	1.20 (.99–1.45)
≥100 000	0.52 (.41–.66)	0.89 (.70–1.15)
ART adherencea
MPR <90%	Reference	…
MPR ≥90%	0.14 (.12–.17)	0.25 (.20–.31)
History of hepatitis B (time-updated)b
No	Reference	…
Yes	2.51 (2.15–2.94)	1.22 (1.02–1.46)
History of hepatitis C (time-updated)b
No	Reference	…
Yes	0.65 (.39–1.06)	0.38 (.22–.63)

Unadjusted and adjusted HRs with 95% CIs were calculated using Cox proportional hazards models. The adjusted model included all listed variables. A complete-case approach was used, and 1319 participants were included in the multivariable model, of whom 76 experienced a blip. Statistically significant HRs (P < .05) are in bold.

Abbreviations: ART, antiretroviral therapy; CI, confidence interval; HIV, human immunodeficiency virus; HR, hazard ratio; MPR, medication possession ratio; PI, protease inhibitor.

^aART adherence was calculated as a single measurement from ART initiation to event/last visit.

^bHistory of hepatitis B and/or C was ascertained by medical record review at each study visit and updated to reflect the date of diagnosis.

Table 3.

Associations Between Time From Human Immunodeficiency Virus Acquisition to Antiretroviral Therapy Initiation and Viral Blips in Sensitivity Analyses Excluding Female Participants and Redefining Blips as 51–1000 Copies/mL

Characteristic	Excluding Female Participants Adjusted HR (95% CI)	Blips 51–1000 copies/mL Adjusted HR (95% CI)
Time from human immunodeficiency virus to antiretroviral therapy initiation, m
≥24	Reference	Reference
12–24	0.47 (.35–.61)	0.77 (.66–.90)
6–12	0.45 (.32–.61)	0.74 (.62–.89)
0–6	0.31 (.19–.50)	0.52 (.39–.70)

Abbreviations: CI, confidence interval; HR, hazard ratio.

As in the primary multivariable analysis, HRs with 95% CIs were calculated using Cox proportional hazards modeling and a complete-case approach. In 1 sensitivity analysis, the analytic population was restricted to 1266 male participants, of whom 72 (5.7%) experienced a blip with HIV RNA 401–1000 copies/mL. This analysis was adjusted for age, race, antiretroviral therapy (ART) regimen, CD4 count, HIV RNA at ART initiation, ART adherence, history of hepatitis B, and history of hepatitis C. In another sensitivity analysis, the analytic population was restricted to 804 participants with HIV RNA results obtained from clinical assays with a lower limit of quantification of ≤50 copies/mL, of whom 167 (20.8%) experienced a blip with HIV RNA 51–1000 copies/mL. This analysis was adjusted for the same potential confounding variables and also for sex. Statistically significant HRs (P < .05) are in bold.

DISCUSSION

We found the lowest hazard of viral blips when ART was initiated within 6 months of estimated HIV acquisition, but the apparent protective effect of earlier ART on viral blips persisted with ART initiation up to 2 years after HIV acquisition. This both affirms prior reports of the beneficial effects of ART initiation during acute or early HIV [16–25], essentially the first few months after HIV acquisition, and suggests that some benefit of “early” ART initiation persists for up to 2 years after HIV acquisition. Prior research has suggested that ART initiation within 1 year of HIV acquisition is associated with reduced reservoir size [11] but did not explicitly evaluate the 2-year timeframe. ART initiation within 1–2 years of HIV acquisition is readily achievable in behaviorally vulnerable populations that undergo routine screening for HIV. If additional research suggests other benefits of treatment during these timeframes, then the research definition of “early-treated” HIV could be reconsidered and more PWH might become candidates for therapeutic studies that leverage and augment early treatment, including studies of interventions designed to achieve HIV remission.

Since INSTI-based therapy is the first-line therapy listed in most major guidelines in the modern era, it is reassuring that INSTI-based regimens carried the lowest hazard of blips in our study. This confirms findings from at least 1 prior report from a cohort of more than 1600 PWH in the Netherlands [36]. INSTIs are significantly more potent than other antiretroviral medications, leading to a faster reduction in viral load after ART initiation and decreased residual viremia [37]. These properties could plausibly help INSTIs to restrict reservoir formation, particularly when used as initial ART [38]. However, investigations into the effects of INSTI-based regimens on HIV reservoir size have shown inconsistent benefits [39–41]. If INSTI-based therapies reduce the establishment of integrated forms among total HIV DNA, this could be a mechanism by which they reduce viral blips. Unsurprisingly, consistent ART adherence was protective against blips [42].

Due to universal access to medical care in the US military, the NHS is uniquely positioned to evaluate biological effects of race while mitigating confounding by socioeconomic and healthcare access disparities that are common elsewhere. White/Caucasian participants in our study had the highest hazard of blips. Prior studies have shown that White/Caucasian PWH are less likely to be spontaneous controllers of HIV compared with Black/African American PWH [43].

We found that history of viral hepatitis was associated with blips, with opposite directions of association for hepatitis C and hepatitis B. Though mechanisms are unclear, several studies have shown that coinfection with hepatitis C is associated with lower peripheral blood HIV RNA in the setting of untreated HIV [44, 45]. The reverse relationship has been observed with hepatitis B coinfection [45]. Viral hepatitis was relatively uncommon in this cohort, and the associations that we observed warrant further investigation in cohorts that are better powered to evaluate the effects of these comorbid conditions.

Comparisons across the existing literature on viral blips are complicated by the variety of definitions used. Our study used a lower threshold of 400 copies/mL to define blips, but others used thresholds as low as 50 copies/mL, 20 copies/mL, or even HIV RNA detectable but not quantifiable. Our choice was largely pragmatic. Over the long history of the NHS, a variety of HIV RNA assays have been used, and the least sensitive of these had a lower limit of quantification of 400 copies/mL. However, these higher-magnitude blips are also the most likely to reflect true physiologic phenomena rather than inter- and intra-assay variability. Poor correlation has been described between HIV RNA assays at viral loads <200 copies/mL [46, 47]. Up to 15% of samples with an HIV RNA <50 copies/mL on 1 assay show an HIV RNA >50 copies/mL on another assay [2, 3]. In one study, only 19% of blips in the range of 50 to 100 copies/mL were reproduced on a second test of the same sample [4]. Across assays, an HIV-1 RNA measured >200 copies/mL is predictive of subsequent confirmed viral failure, but viral failure is not uniformly predicted by measurements <200 copies/mL [47, 48]. Risk of later viral rebound is particularly pronounced with blips ≥500 copies/mL [1, 49]. Nonetheless, our primary findings were robust to a sensitivity analysis using a lower threshold of 50 copies/mL in a restricted subgroup of the study population with appropriate assay data.

It is common in clinical practice to repeat viral load testing when a blip is observed, which could partially explain our finding that participants with blips tended to have a higher number of viral load tests per year. However, it is also possible that blips were more likely to be detected when viral load testing was more frequent, such as during the earlier years of observation in this study.

While we evaluated blips as an indirect marker of the HIV reservoir, additional research is needed to directly measure the reservoir in PWH who were treated within the first few months to years of estimated HIV acquisition. Prior evidence suggests that ART initiation before the development of anti-HIV antibody responses may be the most effective in preventing reservoir formation [50] and viral blips [23] but could not be evaluated in this study. The NHS is overwhelmingly comprised of younger males, so results may not be generalizable to other populations. Some characteristics of the cohort have changed over time, such as its composition by race and sex. Since calendar year was colinear with ART regimen, it could not be included in multivariable analyses. Also, there may be residual confounding, particularly since the frequency of viral load measurements also changed over time. Adherence was assessed only once per participant across all observation time, so fluctuations in adherence were not captured in modeling and may have contributed to blips. Many NHS participants were ineligible for inclusion in these analyses for reasons such as failure to achieve viral suppression within 1 year. This is influenced by the long history of the cohort, which includes periods when less potent and efficacious ART regimens were common, but also potentially introduces some selection bias to the analyzed population.

Several of the associations observed in this study were consistent with prior studies of viral blips, viral set point, and/or HIV reservoir size. While previous studies have suggested that ART initiation during acute HIV may be associated with lower incidence of viral blips, our findings suggest that the window of opportunity to intervene with ART and achieve a beneficial effect in preventing blips may extend to 2 years after HIV acquisition. This potentially widens the aperture for study populations that researchers may consider for clinical trials of new therapies designed to leverage “early treatment’ to facilitate HIV remission. Additional research is needed to define the timeframe within which other beneficial effects of “early” ART may be observed.

Notes

Author Contributions. T. A. C. conceptualized the analyses and drafted the manuscript. H. C. H., X. W., and X. C. were responsible for data curation, conducted statistical analyses, and directly accessed and verified the underlying data. B. G., J. H. P., and R. J. O. contributed to the interpretation of data. C. M. B., J. M. B., J. M. Y., D. T. L., and A. G. collected clinical cohort data and contributed to the study design and analysis. B. K. A. collected clinical cohort data, contributed to the study design and analysis, and oversaw project administration as principal investigator of the US Military HIV Natural History Study. All authors contributed to the interpretation of results, critically reviewed and edited the manuscript, and had final responsibility for the decision to submit for publication.

Acknowledgments. We thank the following additional Infectious Disease Clinical Research Program (IDCRP) HIV Working Group members for help developing the study, collecting and reviewing study data, and ensuring effective protocol operations and ongoing success:

Brooke Army Medical Center, Fort Sam Houston, TX: CPT C. Bettger, COL E. Markelz, K. Mende, S. Merritt, T. Merritt, T. Sjoberg; Madigan Army Medical Center, Joint Base Lewis McChord, WA: C. Baker, S. Chambers, R. Colombo, C. Schofield; National Institute of Allergy and Infectious Diseases, Bethesda, MD: J. Powers; Naval Medical Center Portsmouth, Portsmouth, VA: S. Banks, CDR C. Joya, T. Lalani, R. Tant; Naval Medical Center San Diego, San Diego, CA: C. Brandt, N. Kirkland, LCDR D. Krauth, CDR A. Lane, LCDR P. Wisniewski; Tripler Army Medical Center, Honolulu, HI: LTC N. Copeland, MAJ E. Ewers; Uniformed Services University of the Health Sciences, Bethesda, MD: T. Burgess, M. Horn, E. Parmelee, B. Roth, N. Schvey, D. Tribble, S. Waggoner; Walter Reed Army Institute of Research, Silver Spring, MD: J. Kim, S. Peel, M. Rao; Walter Reed National Military Medical Center, Bethesda, MD: I. Barahona, MAJ B. Custer, C. Decker, LCDR L. Gilbert, MAJ G. Justin, LT A. Maier, P. Paalzow, COL R. Ressner. We are indebted to the study team of IDCRP clinical research coordinators, laboratory personnel, and administrative and data management staff for their efforts, dedication, and contributions to the success of this research. We appreciate the assistance of Jason Thean Kit Ooi in preparing a graphical abstract to summarize this work.

Disclaimer. The views expressed here are those of the authors and should not be construed to represent the positions of the US Army; the Department of Defense (DoD); the Henry M. Jackson Foundation for the Advancement of Military Medicine, Inc (HJF); the Uniformed Services University of the Health Sciences; the National Institutes of Health or the Department of Health and Human Services; the Defense Health Agency; the departments of the Army, Navy or Air Force; and US government. Mention of trade names, commercial products, or organizations does not imply endorsement by the US government. The investigators have adhered to the policies for protection of human research participants as prescribed in Army Regulation 70–25.

Financial support. This work was supported, in part (IDCRP-000–37), by the Infectious Disease Clinical Research Program (IDCRP), a Department of Defense (DoD) program executed through the Uniformed Services University of the Health Sciences, Department of Preventive Medicine and Biostatistics, through a cooperative agreement with Henry M. Jackson Foundation for the Advancement of Military Medicine (HJF). This project has been funded by the National Institute of Allergy and Infectious Diseases, National Institutes of Health, under Inter-Agency Agreement Y1-AI-5072, and the Defense Health Program, US DoD, under award HU0001190002. This work has also been funded, in part, with federal funds from the National Institutes of Health Clinical Center and the National Cancer Institute, National Institutes of Health, under contract 75N91019D00024 task order 75N91019F00130. This work was also supported by other agreements between HJF and the DoD (W81XWH-18–2–0040, HT9425–24–3–0004).

Author notes

Presented in part: 31st Conference on Retroviruses and Opportunistic Infections, Denver, CO, 3–6 March 2024.

Potential conflicts of interest. The authors report no potential conflicts of interest.