Lower Incidence of HIV-1 Blips Observed During Integrase Inhibitor–Based Combination Antiretroviral Therapy

Suzan Dijkstra, MD; L. Marije Hofstra, MD, PhD; Tania Mudrikova, MD, PhD; Annemarie M. J. Wensing, MD, PhD; Patrick G. A. Oomen, MD; Andy I. M. Hoepelman, MD, PhD; Berend J. van Welzen, MD, PhD


J Acquir Immune Defic Syndr. 2022;89(5):575-582. 

In This Article


In this longitudinal analysis of a large, real-life cohort of PLWH treated with contemporary cART regimens, viral blips were frequently encountered. Their incidence differed significantly between the 3 cART anchors, with INSTI-based treatment courses showing the lowest blip rate, followed by NNRTI-based courses. The blip incidence rate in PIs was more than twice as high as the one observed in INSTIs. In addition, we found that blips were of predictive value for the occurrence of PLLV. Finally, we observed that most blips led to an increased clinical burden.

To the best of our knowledge, this is the first study to evaluate the incidence of blips among all 3 current cART anchors. In general, blips are frequently observed during HIV treatment, although various rates are reported in different cohorts (9%–52%).[5] Although study comparisons are complicated by the large variation in blip definitions and sampling frequency, the overall incidence rate of 3.46/100 person-years of follow-up in our cohort was somewhat low compared with the approximate rate of 10/100 person-years observed in other recent cohort studies (in which 96% of blips met our definition with the 499-copy upper limit).[5,32] The higher incidence in these studies may have been influenced by the inclusion of nonadherent PLWH and predominantly PI-based/NNRTI-based treatment.

In this study, the lower incidence of blips during INSTI-based cART persisted after controlling for an extensive number of cofactors such as zenith VL and lowest available CD4+ count. Moreover, regarding PI-based and NNRTI-based cART, our data confirm previous findings that blips are more frequently observed in PLWH using PIs.[5,7,20,33] For instance, Sörstedt et al,[20] who retrospectively assessed 735 PLWH, found that PLWH treated with a boosted PI had an increased odds for experiencing blips compared with PLWH treated with NNRTIs, even after adjusting for baseline VL and year of measurement. Furthermore, in accordance with other reports,[7–10] they observed that PLWH with blips were more likely to demonstrate residual viremia than PLWH with a consistently suppressed VL.[20] This study adds to this existing evidence in that the newer INSTI-based cART was shown to be associated with an even lower incidence of blips than NNRTIs, although both anchor groups exert their antiretroviral effect preintegration and may therefore be expected to similarly influence HIV-1 virion production.

However, to date, no consensus has been reached regarding the blip consequences. It remains poorly understood whether they represent innocuous occurrences[15] or increased viremia permitting resistance and replenishment of viral reservoirs.[14,18] Furthermore, the definition of blips is heterogeneous, potentially leading to misclassification. An explanation for their occurrence above the presently low, 50-copy limit can be sought in the residual viremia below this level. The current understanding of viral dynamics is that PLWH with suppressed infection experience variations around a differing, undetectable mean VL. Because higher levels of this residual viremia require a smaller rise to reach the 50-copy limit, the occurrence of blips is likely indicative of the level of residual viremia.[9] Therefore, it is intuitively attractive to assume that the lower blip incidence for INSTIs indicates lower levels of residual viremia as well, as was previously reported.[27] Whether these lower levels would result from better antiretroviral potency, tissue penetration, tolerability, or barriers to resistance remains to be fully elucidated.

In addition to the association between blips and cART anchors, both higher zenith VL and shorter time since ART initiation were found to be correlated with a higher blip incidence. These factors are generally consistent with previous research[5,7,12] and add a further argument for an underlying biological source rather than random fluctuations or assay variability. The proposed mechanism for blips that is most widely accepted is the intermittent release of virions from a latent reservoir.[14] Because both the zenith VL and time since ART initiation are associated with the reservoir size,[34] the correlations found in this study lend support to this explanation. This theory is further reinforced by the previous finding that ART initiation early in the course of infection, when the reservoir has yet to be established, is associated with a lower blip incidence.[31,35] In this study, we did not find a significant impact of Fiebig stage on blip incidence, but ART initiation during acute infection in our cohort was rare.

Previous studies have found contradicting evidence when examining the relationship between blips and virologic failure.[5,8,12,19–21] In our population, no association between the two could be established, although confidence intervals were broad. We did, however, observe a higher incidence of PLLV among PLWH experiencing blips. This relationship was also demonstrated in some previous studies that included consecutive VLs ≥ 50 copies/mL in their definition of virologic failure.[5,20,32,36] The observation that blips were associated with PLLV but not failure is notable because PLLV has previously been found to increase the risk of subsequent failure.[6,9] However, the occurrence of PLLV or virologic failure was relatively infrequent in our population,[5,6,8] especially considering the long mean follow-up of 5.4 years. These low frequencies most likely result from our exclusion of nonadherent PLWH. It is conceivable that nonadherence would lead to multiple episodes of viremia in a single individual and thus a relationship between those episodes. Moreover, viral rebounds during appropriate treatment may carry a different risk of viral escape than rebounds found in populations with frequent or long-term treatment interruptions. Alternatively, the lack of an association between blips and failure in our cohort could indicate a role for random assay variability as the underlying mechanism for some blips. However, considering the association with PLLV, our results suggest that PLWH might warrant closer monitoring when blips are encountered. Furthermore, although most studies and clinicians agree that isolated blips do not compromise long-term virologic control,[37] they have the potential to generate uncertainty and increased health care costs independent of their prognostic value, given that a detectable VL can only retrospectively be classified as a blip once resuppressed. In PLWH experiencing a blip, the VL was often retested after several weeks, ahead of the regularly scheduled follow-up, especially when blip magnitude was high. Intensification of therapy monitoring is inevitable to distinguish PLWH experiencing blips from those experiencing PLLV or virologic failure. However, because these extra outpatient visits, VL measurements, and, occasionally, other laboratory tests increase costs and psychological burden, it is desirable to strive for the lowest number of blips.

The major strength of this study is the analysis of a large amount of data derived from a real-world cohort with up to 10 years of follow-up, which allowed for rigorous assessment of medical records and collection of exact dates of cART modifications and information on adherence. Nevertheless, a limitation to our study that should be considered stems from its retrospective design. Observational studies are at risk of unmeasured confounding. For example, time to treatment may have been shorter in more recent treatment courses. Although the risk of confounding bias was mitigated by controlling for a multitude of factors, including markers for the timing of ART initiation, a risk of residual confounding by factors not included in the analyses (eg, drug–drug interactions, undocumented nonadherence, or an association with certain comorbidities) remains. Furthermore, as with any observational study, PLWH were treated in routine medical practice, and the potential for differential selection of cART regimens existed. The development of new antiretroviral drugs, changing treatment policies, and new data influencing the choice for certain drugs in certain patient categories may have influenced several baseline parameters. For instance, physicians may have preferentially prescribed PIs in higher-risk PLWH because PIs had a higher genetic barrier to resistance than did other anchors before the arrival of the newer INSTIs.[38] We believe that the exclusion of treatment courses with detectable viremia resulting from nonadherence has strengthened our study because the higher rate of nonadherence in PI recipients could potentially have biased the data toward a higher blip incidence for PIs. Although INSTIs are relatively new, as reflected by the shorter mean follow-up in this anchor group, the statistical model was chosen to account for different follow-up periods. Finally, we cannot completely exclude an impact of the 20 censored treatment courses. However, considering clinical characteristics of these specific courses and the robustness demonstrated in the general analysis of this large cohort, we believe it is unlikely that this significantly influenced our results.

In conclusion, this study demonstrates that the incidence of viral blips among PLWH is strongly associated with the type of cART anchor used, the lowest rate being observed during INSTI-based regimens. Although the debate regarding their nature is ongoing, blips were associated with an increased clinical burden but not virologic failure in our cohort. Understanding the underlying mechanism of blips requires further research and long-term observational studies to assess their impact on clinical outcomes.