1. Update on HIV-1 viral load blips
Richard E. Nettles and Tara L. Kieffer
Purpose of review Abbreviations
Many patients on highly active antiretroviral therapy with EDTA ethylenediamine tetraacetic acid
undetectable levels of HIV-1 RNA experience viral load HAART highly active antiretroviral therapy
PPT plasma preparation tube
blips. True periods of detectable viremia raise concerns that
regimen potency is inadequate to suppress viral replication
ß 2006 Lippincott Williams & Wilkins
completely, which could lead to the development of 1746-630X
resistance. Because blips are not associated with long-term
clinical or virological failure in most studies, there is Introduction
uncertainty over their clinical significance. Highly active antiretroviral therapy (HAART) is the
Recent findings current standard treatment for patients infected with
Recent data help explain the lack of association between HIV-1. Although this treatment significantly decreases
blips and clinical or virological failure. Many blips are not an viral loads and allows a considerable degree of immune
actual rise in viral load, but instead represent normal reconstitution, it is unable to eradicate the virus com-
biological fluctuations around a mean viral load below pletely because of the existence of long-lived reservoirs,
50 copies/ml as well as statistical variations around the such as within CD4 resting memory T cells. The optimal
detection limit of the viral load assay. Some blips may also response a patient can achieve on HAART is a rapid and
result from laboratory processing artefacts. Therefore, most sustained drop in plasma virus to levels that are below our
blips are not reproducible on duplicate viral load ability to measure with current ultrasensitive assays
measurements. With frequent viral load measurements, (< 50 copies/ml). The goal of therapy is to reach and
there is less correlation between blips and demographic, maintain viral loads below this level to avoid the emer-
treatment, or HIV-associated clinical factors than previously gence of drug resistance. Unfortunately, many patients
reported. Likewise, many blips are often unrelated to experience ‘blips’, intermittent episodes of detectable
intercurrent illnesses, vaccination, non-adherence, or low-level HIV-1 viremia, which resolve spontaneously by
decreases in antiretroviral drug concentrations. Most returning to less than 50 copies/ml without a change in
importantly, new genotypic resistance mutations do not therapy [1–5]. Although there is no consensus on the
develop before, during, or immediately after most blips. maximum duration or magnitude of a blip, episodes of
Summary persistently detectable viremia or episodes of detectable
Despite recent findings suggesting that many blips are viremia of marked magnitude are typically considered
laboratory or statistical aberrations, it remains important to distinct from blips. Mathematical modeling proposes a
differentiate blips from early virological failure or persistent typical blip magnitude of approximately 158 copies/ml
detectable low-level viremia. Once the episode of and a total duration of 20–30 days [6–8]. Blips have
detectable viremia is clearly defined as a blip, however, previously been attributed to periods of poor antiretro-
there should be no cause for clinical concern. viral adherence or bioavailability, vaccination, or sickness
[9,10]. There has been concern that blips could be
Keywords associated with increased viral replication, leading to
AIDS, blips, drug resistance, highly active antiretroviral the development of new drug resistance mutations
therapy, HIV [11,12]. Likewise, there was a fear that blips could allow
for the replenishment of physical or cellular viral reser-
Curr Opin HIV AIDS 1:157–161. ß 2006 Lippincott Williams & Wilkins. voirs, thereby complicating efforts at eradication and
allowing for the cataloging of new resistant variants
Departments of Medicine, Johns Hopkins University School of Medicine, Baltimore
Maryland, USA [13]. Despite the apparent loss of suppression of viral
Correspondence and requests for reprints to Richard E. Nettles, Bristol Myers
load, most studies have shown that blips are not associ-
Squibb, MS E12-16 PO Box 4000, Princeton, NJ 08543, USA ated with long-term clinical failure [1–4]. Recent litera-
Tel: +1 609 252 3719; fax: +1 609 252 7034; e-mail richard.nettles@bms.com ture concerning blips helps explain why most blips do not
Sponsorship: This study received financial support from NIAID grants K08 compromise long-term clinical or virological outcomes.
AI060367 (to R.E.N.).
Current Opinion in HIV and AIDS 2006, 1:157–161 Blips are often not real clinical increases in
viral load
$
Current affiliations: R.E.N. is currently employed by Bristol Myers Squibb; T.L.K. is
Viral load assay variation and laboratory processing arte-
currently employer by Vertex Pharmaceutic. facts can be responsible for many blips without actual
157
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

2. 158 Reservoirs: basic science
significant increases in viral replication. Blips associated three patients (69.6 versus 5.4%, P < 0.0001), respect-
with these phenomena are of limited clinical significance. ively. Low-level viremia was measured in 34 patients
These findings may explain why blips do not compromise (60.7%), which subsequently resolved with the resump-
long-term outcomes in most studies. tion of the use of EDTA tubes, whereas only three (5.4%)
experienced low-level viremia with PPT use that per-
Random biological fluctuation and statistical sisted after EDTA tube use resumed. The authors con-
assay variation cluded that for patients with viral suppression, viral load
Our laboratory recently published results from a prospec- measurements performed on specimens collected and
tive analysis of blips using frequent viral load sampling stored in PPTs result in factitious low-level viremia,
(every 2–3 days) for 3–4 months in a group of 10 HIV-1- and they therefore cautioned against the use of PPTs
infected individuals [14]. Viral loads were measured in for viral load testing. The conclusion of the study by
duplicate in two independent laboratories, both using the Stosor et al. [16] was validated by the producer of PPTs,
same ultrasensitive Roche Amplicor system assay, version who also found discordant results when comparing viral
1.5 (Roche Diagnostic Systems, Inc., Branchburg, New loads from blood collected in PPTs versus EDTA tubes
Jersey, USA). Blips were detected in nine out of 10 [17]. Rainen and colleagues [17] concluded that
patients. Of 713 viral load measurements, 26 (3.6%) were samples collected in PPTs should be centrifuged within
greater than 50 copies/ml, constituting 18 total blips. 6 h of phlebotomy and stored at 48C for no longer than
Nine blips were detected by one laboratory, eight by 72 h, and that PPT plasma should not be frozen in situ in
the other laboratory, and only one was detected by both. order to obtain consistent, concordant viral load results.
Although there was no difference in the sensitivity of
assays used by the two laboratories, concordance was poor Blip genesis
(kappa 4.4%). Using the coefficient of variation of the Many authors continue to consider all blips as actual
viral load assay, it was shown that the proportion of periods of increased HIV viremia. As such, there con-
positive assays was consistent with random variation tinues to be a search for clinical factors that may act as the
around a mean viral load level of 10–20 copies/ml. There- trigger for increased viral replication resulting in blips.
fore, in this study it appeared that many blips resulted
from random assay variations around a mean viral load Demographics, vaccination, and concurrent illness
below 50 copies/ml, rather than clinically significant Previous studies have linked blips to multiple clinical
elevations in viremia. The finding that many blips are factors, such as poor antiretroviral adherence, vaccination,
statistical aberrations probably explains the lack of associ- or sickness. Many of these earlier studies were based on
ation between blips and an increased risk of clinical or infrequent viral load sampling, retrospective chart
virological failure seen in several clinical trials. The major review, and patient self-reported adherence. Therefore,
limitation of the study by Nettles et al. [14] is that the the findings of many of these studies, although important,
sample size was very small (10 patients) and relatively may not be sufficiently precise to define the true relation-
homogenous (most started HAART after reaching low ship of blips to clinical events. In an effort to overcome
CD4 cell nadirs with high viral loads). this limitation, Nettles and colleagues [14] prospec-
tively collected clinical data in addition to the viral load
Blips as a laboratory processing artefact data described previously. They found that when viral
The lack of congruity between blips and the increased load was sampled up to three times a week, blips did not
risk of failure may also be explained if many blips are occur more frequently in any demographic group defined
actually factitiously elevated viral loads. There has been by sex, race, or age. There was no correlation between
concern that viral loads are falsely elevated by the ultra- the frequency of blips and clinical parameters such as
sensitive Roche Amplicor test versions 1.0 or 1.5 (Roche CD4 cell nadir, pretreatment viral load, duration of HIV
Diagnostics, Inc.) from blood collected in plasma prep- infection, duration of virological suppression, or number
aration tubes (PPTs) compared with blood collected in of previous blips. Temporally, blips were not observed
ethylenediamine tetraacetic acid (EDTA) tubes [15]. in relation with intercurrent illnesses or influenza vacci-
Stosor and colleagues [16] recently confirmed that viral nation.
loads may be misleadingly elevated when using PPTs for
the collection of plasma to be used for viral load measure- Blip frequency in relation to timing of initiation of highly
ments. In their study, 56 HAART recipients with three or active antiretroviral therapy
more consecutive undetectable viral loads from plasma Di Mascio and colleagues [18] compared blip frequency
collected in EDTA tubes were identified. There was among patients starting HAART during acute/early
subsequently an 8-month period of PPT use and a infection versus during chronic infection. They found
successive period of 6-month EDTA tube use. Substan- that blip frequency was nearly twofold higher in the
tially more patients experienced viremia with PPTs than chronically infected group compared with the acute/early
during the following period of EDTA tube use; 39 versus infected group (P ¼ 0.001). They concluded that the
Copyright © Lippincott Williams Wilkins. Unauthorized reproduction of this article is prohibited.

3. Update on HIV-1 viral load blips Nettles and Kieffer 159
duration of infection before the initiation of HAART is not temporally associated with blips. In addition, protease
possibly related to the tendency to experience blips. If inhibitor and non-nucleoside reverse transcriptase inhibi-
these findings are confirmed by additional studies, and it tor concentrations in plasma at each timepoint in each
is shown that blips predict an adverse outcome in some patient were measured to determine whether blips were
patients, this could be an argument for the initiation of temporally associated with decreased drug concen-
HAART early in the course of infection. trations. There was no correlation between blips and
insufficient drug concentrations.
Plasmid vaccination
The authors of one study report that the receipt of an HIV Mathematical modeling
plasmid vaccine decreased blip frequency in patients on Jones and Perelson [22] developed a mathematical
HAART [19]. The vaccine was a combination of two HIV- model of intercurrent infection on patients with low viral
DNA constructs, which drive the expression of env/rev and loads on HAART. On the basis of their model, they
gag/pol genes. In that study, three out of five placebo concluded that viral blips with a duration and amplitude
recipients had a blip (10/49 assays) versus one out of 13 similar to those seen by their previous models can be
vaccinees (1/130 assays; P 0.04). Mainly on the basis of generated by opportunistic infections.
this finding, the authors concluded that DNA immuniz-
ation in HAART-treated patients may have therapeutic Blip dynamics
potential. Importantly, it appears that all of the placebo Although there remain some discrepancies between blip
recipients experienced detectable viremia of a marked dynamics, as defined by frequent viral load sampling and
magnitude or duration that many clinicians would con- mathematical modeling, the frequency, amplitude and
sider viral failure rather than a blip, and so the applica- duration of typical blips has been clarified.
bility of that study to this current review is marginal.
Frequent viral load sampling
3-Hydroxy-3-methylglutaryl coenzyme A To define an extremely precise profile of blip frequency,
reductase inhibitors amplitude, and duration, Nettles and colleagues [14]
Based on a report that patients receiving 3-hydroxy-3- used frequent viral load sampling as defined above. Most
methylglutaryl coenzyme A reductase inhibitor therapy blips (15 out of 18) represented isolated measurements
(or ‘statins’) for hyperlipidemia, but not receiving above 50 copies/ml, with the subsequent measurement
HAART, were found coincidentally to have lowered being undetectable. The typical blip was brief (median
HIV-RNA levels, Waters and colleagues [20] looked duration 2.5 days; range 2–11.5 days) and low in magni-
to see if statins lowered the frequency of blips. Using tude (median 79 copies/ml; range 51–201 copies/ml),
a prospective database of 1521 patients on first-line with a clustering of values towards the 50 copies/ml limit
HAART (78 of which were also prescribed a statin) for and only one value greater than 200 copies/ml.
at least 3 months, they found no significant decrease in
blip frequency in those patients prescribed a statin. Mathematical modeling
Mathematical modeling predicts an entirely different
Non-adherence blip profile. Previously, mathematical models proposed
It has long been assumed that many blips result from a mean blip amplitude of 158 Æ 132 copies/ml and a blip
periods of poor adherence to HAART. This assumption duration of 20–30 days [6–8]. Di Mascio and colleagues
was not supported by a careful assessment of adherence [23] have recently refined their mathematical model and
in 128 patients, using pill count, self-report and the applied it to the same population of patients as their
Medication Event Monitoring System (MEMS; AAR- previous publication to conclude that blips extend for a
DEX Ltd., Switzerland), an electronic pill bottle cap period of approximately 3 weeks.
that records each time a patient opens a pill bottle
[21]. Twenty-five per cent of the patients experienced Immune response to blips
a blip. Among those patients who experienced a blip, In one study, the effect of blips on the immune system
there was no significant decrease in adherence before the was compared with patients with persistent undetectable
blip. Patients who experienced a blip had higher rates of viral loads. Karlsson and colleagues [24] categorized
overall adherence (P ¼ 0.046). There were no significant patients on HAART as having sustained undetectable
differences in the dose-timing error between those who levels of viremia or intermittent detectable viremia
experienced a blip and those who did not. The authors of 50–1000 copies/ml (blips). The viral load was
concluded that blips do not appear to be associated with measured every 4 months in that study. Patients with
non-adherence. intermittent viremia had an HIV-specific CD8 T-cell
response statistically significantly greater in magnitude
Similar results were found by Nettles and colleagues and breadth compared with those with sustained suppres-
[14], in which patient-reported non-adherence was sion of viremia (P 0.001 for magnitude and breadth). In
Copyright © Lippincott Williams Wilkins. Unauthorized reproduction of this article is prohibited.

4. 160 Reservoirs: basic science
particular, patients with blips had greater immune loads that are below 50 copies/ml, rather than clinically
responses to gag, pol, env, and nef. In addition, HIV- significant elevations in the level of viral replication.
specific cellular immune response to the HIV proteins Most blips are not surrogate markers for periods of
were generally higher at those timepoints when blips non-adherence, immune system activation, inadequate
were detected (P ¼ 0.057). The study is intriguing; how- drug concentrations, or HAART impotency, and there-
ever, it was somewhat limited by the infrequent viral fore should not serve as such. Care should be exercised in
load measurements. the interpretation of data generated from studies of blips
with infrequent viral load measurements or viral loads
Blips and antiretroviral drug resistance measured from blood collected by PPTs. Despite these
The authors of two studies have previously reported recent findings, it remains clinically important to differ-
finding new resistance mutations in association with blips entiate blips from early virological failure or persistent
[11,12]. Those studies were compromised by having a detectable low-level viremia. Studies defining the mag-
limited assessment of preexisting mutations and limited nitude and duration of viremia necessary for the per-
sensitivity of genotyping assays. In a large prospective mission of viral replication, leading to the development
study incorporating an ultrasensitive genotyping method of new drug resistance mutations and for the replenish-
before, during, and after blips and performing extensive ment of physical or cellular viral reservoirs, are urgently
baseline genotyping of the virus residing in the plasma needed. However, once an episode of detectable viremia
and the resting CD4 T-cell reservoir [14], no such link is clearly defined as a blip, a low level measurement
was found. During the course of that study, nearly 1000 (50–200 copies/ml) with a subsequent and immediate
viral clones were genotyped, including approximately return to an undetectable measurement, it should not
150 clones obtained during blip timepoints. New be a cause of clinical concern.
mutations were not seen before, during or immediately
after blips. Patients with more resistance mutations did
not experience a greater number of blips. In addition, References and recommended reading
Papers of particular interest, published within the annual period of review, have
viruses present during blips were not phylogenetically been highlighted as:
distinct from non-blip samples, indicating that viral evol- of special interest
of outstanding interest
ution was probably not greater during blip timepoints. Additional references related to this topic can also be found in the Current
World Literature section in this issue (p. 180).
Blips effect on clinical and 1 Havlir DV, Bassett R, Levitan D, et al. Prevalence and predictive value of
virological outcome intermittent viremia with combination hiv therapy. JAMA 2001; 286:171–
179.
Most studies have shown that blips are not associated
2 Raboud JM, Rae S, Woods R, et al. Consecutive rebounds in plasma viral load
with long-term clinical failure or virological failure. Havlir are associated with virological failure at 52 weeks among HIV-infected
and colleagues [1] first made this observation on the basis patients. AIDS 2002; 16:1627–1632.
of data from two clinical trials in which patients received 3 Mira JA, Macias J, Nogales C, et al. Transient rebounds of low-level viraemia
among HIV-infected patients under HAART are not associated with virological
indinavir plus two nucleosides. The study showed that or immunological failure. Antiviral Ther 2002; 7:251–256.
blips were associated with higher steady-state levels of 4 Sklar PA, Ward DJ, Baker RK, et al. Prevalence and clinical correlates of HIV
viremia, but were not associated with virological failure viremia (‘blips’) in patients with previous suppression below the limits of
quantification. AIDS 2002; 16:2035–2041.
for up to 4.5 years (P ¼ 0.53).
5 Greub G, Cozzi-Lepri A, Ledergerber B, et al. Intermittent and sustained low-
level HIV viral rebound in patients receiving potent antiretroviral therapy. AIDS
Martinez and colleagues [25 ] essentially reproduced the 2002; 16:1967–1969.
findings of Havlir and colleagues [1] recently in patients 6 Di Mascio M, Markowitz M, Louie M, et al. Viral blip dynamics during highly
active antiretroviral therapy. J Virol 2003; 77:12165–12172.
treated with non-nucleoside reverse transcriptase inhibi-
7 Percus JK, Percus OE, Markowitz M, et al. The distribution of viral blips
tor-based HAART. They found that eight out of 43 observed in HIV-1 infected patients treated with combination antiretroviral
patients (18%) experienced a blip during 18 months of therapy. Bull Math Biol 2003; 65:263–277.
follow-up. Patients with blips had a significantly lower 8 Di Mascio M, Ribeiro RM, Markowitz M, et al. Modeling the long-term control
of viremia in HIV-1 infected patients treated with antiretroviral therapy. Math
CD4 cell count after 12 and 18 months than did those who Biosci 2004; 188:47–62.
had not experience a blip. As in previous studies, blip 9 Gunthard HF, Wong JK, Spina CA, et al. Effect of influenza vaccination on viral
occurrence was not associated with virological failure. replication and immune response in persons infected with human immuno-
deficiency virus receiving potent antiretroviral therapy. J Infect Dis 2000;
This is reassuring, given the low genetic barrier of resist- 181:522–531.
ance for non-nucleoside reverse transcriptase inhibitors 10 Kolber MA, Gabr AH, De La RA, et al. Genotypic analysis of plasma HIV-1
compared with protease inhibitors. RNA after influenza vaccination of patients with previously undetectable viral
loads. AIDS 2002; 16:537–542.
11 Cohen Stuart JW, Wensing AM, Kovacs C, et al. Transient relapses (‘blips’) of
Conclusion plasma HIV RNA levels during HAART are associated with drug resistance.
J Acquir Immune Defic Syndr 2001; 28:105–113.
Among patients on HAART with suppression of HIV-1
12 Easterbrook PJ, Ives N, Waters A, et al. The natural history and clinical
viremia, most blips represent laboratory artefacts or nor- significance of intermittent viraemia in patients with initial viral suppression
mal biological and statistical variations around mean viral to 400 copies/ml. AIDS 2002; 16:1521–1527.
Copyright © Lippincott Williams Wilkins. Unauthorized reproduction of this article is prohibited.

5. Update on HIV-1 viral load blips Nettles and Kieffer 161
13 Ramratnam B, Mittler JE, Zhang L, et al. The decay of the latent reservoir of 20 Waters L, Stebbing J, Jones R, et al. The effect of statins on hiv rebound and
replication-competent HIV-1 is inversely correlated with the extent of residual blips. J Acquir Immune Defic Syndr 2005; 39:637–638.
viral replication during prolonged anti-retroviral therapy. Nat Med 2000; 6: There was no significant decrease in blip frequency in those patients prescribed a
82–85. statin in this study.
14 Nettles RE, Kieffer TL, Kwon P, et al. Intermittent HIV-1 viremia (blips) and
drug resistance in patients receiving HAART. JAMA 2005; 293:817–829. 21 Miller LG, Golin CE, Liu H, et al. No evidence of an association between
This study defines blip dynamics, clinical associations, and genotypic resistance transient HIV viremia (‘blips’) and lower adherence to the antiretroviral
consequences with frequent viral load sampling of 10 patients. Most importantly, medication regimen. J Infect Dis 2004; 189:1487–1496.
no new resistance mutations before, during, or immediately after blips were The authors conclude that blips do not appear to be associated with non-
detected. adherence. This finding is consistent with the conclusion of others that many
15 Squires K, Lazzarin A, Gatell JM, et al. Comparison of once-daily atazanavir blips are related to laboratory or statistical aberrations rather than clinical events
with efavirenz, each in combination with fixed-dose zidovudine and lamivudine, generating increased viral replication.
as initial therapy for patients infected with HIV. J Acquir Immune Defic Syndr
2004; 36:1011–1019. 22 Jones LE, Perelson AS. Opportunistic infection as a cause of transient viremia
16 Stosor V, Palella FJ, Berzins B, et al. Transient viremia in HIV-infected patients in chronically infected HIV patients under treatment with HAART. Bull Math
and vacutainer plasma preparation tubes. Clin Infect Dis 2005; 41:1671– Biol 2005; 67:1227–1251.
1674. A mathematical model demonstrates that blips can be generated by opportunistic
This study demonstrates that viral load measurements performed on specimens infections.
collected and stored in PPTs result in factitious low-level viremia. The interpretation
of viral loads originating from PPTs should be done with caution in both clinical and 23 Di Mascio M, Percus JK, Percus OE, et al. Duration of an intermittent episode
research settings. of viremia. Bull Math Biol 2005; 67:885–900.
17 Rainen L, Salimnia H, Fairfax M, et al. Sample handling parameters affecting A mathematical model demonstrates that blips may extend for a period of
performance of BD vacutainer1 PPTTM and plus K2EDTA tubes with the approximately 3 weeks.
Roche AMPLICORTM and COBAS AMPLICORTM HIV-1 MONITOR1 test,
v1.5 UltraSensitive and Standard specimen processing procedures. BD 24 Karlsson AC, Younger SR, Martin JN, et al. Immunologic and virologic
Diagnostics, Franklin Lakes, NJ, USA. www.bd.com. Released 3 March 2005. evolution during periods of intermittent and persistent low-level viremia. AIDS
This report defines the PPT handling conditions necessary to obtain consistent, 2004; 18:981–989.
concordant viral load results. Patients with intermittent viremia had an HIV-specific CD8 T-cell response
18 Di Mascio M, Markowitz M, Louie M, et al. Dynamics of intermittent viremia statistically significantly greater in magnitude and breadth compared with those
during highly active antiretroviral therapy in patients who initiate therapy during with sustained suppression of viremia.
chronic versus acute and early human immunodeficiency virus type 1 infec-
tion. J Virol 2004; 78:10566–10573.
25 Martinez V, Marcelin AG, Morini JP, et al. HIV-1 intermittent viraemia in patients
Blip frequency was nearly twofold higher in a chronically HIV-infected group treated by non-nucleoside reverse transcriptase inhibitor-based regimen.
compared with an acute/early HIV-infected group. This study may be seen by
AIDS 2005; 19:1065–1069.
some as another reason to initiate HAART early in the course of infection.
This study shows that the occurrence of blips was not associated with virological
19 MacGregor RR, Boyer JD, Ugen KE, et al. Plasmid vaccination of stable HIV- failure for patients on non-nucleoside reverse transcriptase inhibitor-based
positive subjects on antiviral treatment results in enhanced CD8 T-cell immunity HAART. This is another confirmation that blips do not predict future failure,
and increased control of viral ‘blips’. Vaccine 2005; 23:2066–2073. regardless of the HAART regimen.
Copyright © Lippincott Williams Wilkins. Unauthorized reproduction of this article is prohibited.