Gupta RK, Hill A, Sawyer AW, Cozzi-Lepri A, von Wyl V, Yerly S, et al. Virological monitoring and resistance to first-line highly active antiretroviral therapy in adults infected with HIV-1 treated under WHO guidelines: a systematic review and meta-analysis

UCL/MRC Centre for Medical Molecular Virology, University College London Medical School, London, UK.
The Lancet Infectious Diseases (Impact Factor: 22.43). 07/2009; 9(7):409-17. DOI: 10.1016/S1473-3099(09)70136-7
Source: PubMed
ABSTRACT
Antiretroviral-therapy rollout in resource-poor countries is often associated with limited, if any, HIV-RNA monitoring. The effect of variable monitoring on the emergence of resistance after therapy with commonly used drug combinations was assessed by systematic review of studies reporting resistance in patients infected with HIV with a CD4 count of fewer than 200 cells per muL treated with two nucleoside analogues (including a thymidine analogue) and a non-nucleoside reverse transcriptase inhibitor. 8376 patients from eight cohorts and two prospective studies were analysed. Resistance at virological failure to non-nucleoside reverse transcriptase inhibitors at 48 weeks was 88.3% (95% CI 82.2-92.9) in infrequently monitored patients, compared with 61.0% (48.9-72.2) in frequently monitored patients (p<0.001). Lamivudine resistance was 80.5% (72.9-86.8) and 40.3% (29.1-52.2) in infrequently and frequently monitored patients, respectively (p<0.001); the prevalence of at least one thymidine analogue mutation was 27.8% (21.2-35.2) and 12.1% (5.9-21.4), respectively (p<0.001). Genotypic resistance at 48 weeks to lamivudine, nucleoside reverse transcriptase inhibitors (thymidine analogue mutations), and non-nucleoside reverse transcriptase inhibitors appears substantially higher in less frequently monitored patients. This Review highlights the need for cheap point-of-care viral-load tests to identify early viral failures and limit the emergence of resistance.
www.thelancet.com/infection Vol 9 July 2009
409
Review
Introduction
As part of a public-health approach to highly active
antiretroviral therapy (HAART) in resource-poor countries,
WHO currently recommends
1
zidovudine, stavudine,
abacavir, or tenofovir, plus lamivudine or emtricitabine to
form the dual nucleoside reverse transcriptase inhibitor
(NRTI) backbone, combined with one non-nucleoside
reverse transcriptase inhibitor (NNRTI)—either nevirapine
or efavirenz. Although stavudine and zidovudine are not
recommended as fi rst-line therapy in well-resourced
settings because of toxicity concerns,
2,3
these thymidine
analogues along with lamivudine and nevirapine form the
most widely used combinations in resource-poor settings,
because of the availability of a cheap generically produced
xed-dose combination containing stavudine, lamivudine,
and nevirapine.
Although antiretroviral-therapy rollout has yet to reach
all those in need, reductions in morbidity and mortality
have been shown in those treated.
4,5
However, concerns
remain regarding the development of extensive resistance
in the absence of routine virological monitoring,
potentially compromising responses to second-line
therapy. Furthermore, both failure
6
and delayed
switching
7
of NNRTI-containing fi rst-line HAART have
been associated with increased mortality. Perhaps the
greatest concern is the threat of increased transmitted
resistance in the face of continuing incident infection.
In well-resourced settings clinical guidelines
recommend that viral-load testing be done every
3 months, although evidence for this particular frequency
is weak. Some studies in resource-poor settings have
used a less frequent or non-monitoring strategy, although
no direct randomised comparison of monitoring
frequencies has been undertaken. In addition, there has
been limited randomised-clinical-trial data on resistance
associated with the specifi c combinations of zidovudine
or stavudine, lamivudine, and nevirapine or efavirenz.
8
Where randomised clinical trials have been done, patients
who switch study drug are often excluded from resistance
analysis; this design does not adequately refl ect clinical
practice where switches due to tolerability are common.
Given that there is no evidence base for viral
monitoring frequencies, and little trial data on the
generic combinations used to treat the majority of HIV
infections worldwide, we did a systematic review of
studies reporting resistance data on patients starting
HAART using thymidine-analogue and NNRTI-
containing combinations under WHO criteria; we
aimed to compare resistance outcomes across diff ering
virological monitoring intensities.
Methods
Search strategy
We did our meta-analysis in accordance with QUORUM
guidelines.
9
We searched the Medline electronic database
via PubMed and EmBase.com (January, 1994, to
March, 2009). We also searched conference abstracts
(January, 1999–March, 2009) from the following:
Conference on Retroviruses and Opportunistic Infections,
Interscience Conference on Antimicrobial Agents and
Chemotherapy, International AIDS Society, European
AIDS Clinical Society, International Congress on Drug
Virological monitoring and resistance to fi rst-line highly
active antiretroviral therapy in adults infected with HIV-1
treated under WHO guidelines: a systematic review and
meta-analysis
Ravindra K Gupta, Andrew Hill, Anthony W Sawyer, Alessandro Cozzi-Lepri, Viktor von Wyl, Sabine Yerly, Viviane Dias Lima, Huldrych F Günthard,
Charles Gilks, Deenan Pillay
Antiretroviral-therapy rollout in resource-poor countries is often associated with limited, if any, HIV-RNA
monitoring. The eff ect of variable monitoring on the emergence of resistance after therapy with commonly used
drug combinations was assessed by systematic review of studies reporting resistance in patients infected with HIV
with a CD4 count of fewer than 200 cells per μL treated with two nucleoside analogues (including a thymidine
analogue) and a non-nucleoside reverse transcriptase inhibitor. 8376 patients from eight cohorts and two
prospective studies were analysed. Resistance at virological failure to non-nucleoside reverse transcriptase
inhibitors at 48 weeks was 88·3% (95% CI 82·2–92·9) in infrequently monitored patients, compared with 61·0%
(48·9–72·2) in frequently monitored patients (p<0·001). Lamivudine resistance was 80·5% (72·9–86·8) and
40·3% (29·1–52·2) in infrequently and frequently monitored patients, respectively (p<0·001); the prevalence of at
least one thymidine analogue mutation was 27·8% (21·2–35·2) and 12·1% (5·9–21·4), respectively (p<0·001).
Genotypic resistance at 48 weeks to lamivudine, nucleoside reverse transcriptase inhibitors (thymidine analogue
mutations), and non-nucleoside reverse transcriptase inhibitors appears substantially higher in less frequently
monitored patients. This Review highlights the need for cheap point-of-care viral-load tests to identify early viral
failures and limit the emergence of resistance.
Lancet Infect Dis 2009;
9: 409–17
UCL/MRC Centre for Medical
Molecular Virology
(R K Gupta MRCP,
Prof D Pillay MRCPath), and
Epidemiology and Biostatistics
Department of Primary
Care &
Population Sciences
(A Cozzi-Lepri PhD), University
College London Medical School,
London, UK; Pharmacology
Research Laboratories,
University of Liverpool,
Liverpool, UK (A Hill PhD);
MetaVirology Ltd, London, UK
(A W Sawyer PhD); Division of
Infectious Diseases and
Hospital Epidemiology,
University Hospital Zurich,
Zurich, Switzerland
(V von Wyl PhD,
Prof H F Günthard MD);
Laboratory of Virology,
University Hospital Geneva,
Geneva, Switzerland
(S Yerly PhD); BC Centre for
Excellence in HIV/AIDS,
Vancouver, BC, Canada
(V D Lima PhD); and
Department of HIV/AIDS, WHO,
Geneva (Prof C Gilks MD)
Correspondence to:
Dr Ravindra K Gupta, Windeyer
Institute, University College
London Medical School,
Windeyer Building, 46 Cleveland
Street, London W1T 4JF, UK
rgupta2@nhs.net
Page 1
410
www.thelancet.com/infection Vol 9 July 2009
Review
Therapy in HIV Infection, and the International HIV
Drug Resistance Workshop. We constructed two search
strings, as follows, to identify as many relevant studies as
possible: “highly active antiretroviral therapy and
resistance and naive”, and “human immunodefi ciency
virus or HIV and naive”. Each string was then combined
with current US FDA licensed antiretrovirals
10
(zidovudine, stavudine, lamivudine, nevirapine, and
efavirenz) with an “or” term separating them.
We judged the eligibility of the search results based on
inclusion and exclusion criteria, with subsequent review
of the full article. Reference lists from included studies
were used to identify further publications. Study authors
were contacted where necessary for further data, and, in
the case of the national UK Collaborative HIV Cohort
and UK Drug Resistance Database,
11
Swiss HIV cohort,
12
and the Canadian HOMER cohort,
13
this led to more
detailed sharing of data.
Inclusion criteria
We considered any study in which treatment-naive
adolescent or adult patients (older than 13 years)
chronically infected with HIV-1 received fi rst-line
HAART (zidovudine or stavudine, lamivudine, and
nevirapine or efavirenz) following WHO clinical, CD4,
or both criteria for resource-poor settings. We also
included cohort studies that were able to provide data
on patients with CD4 counts of fewer than
200 cells per μL. Viral-load and genotypic-resistance
data on all patients (including treatment substitutions)
were required up to a maximum of 96 weeks. We
included studies in English from all parts of the world
reporting both public programmes and privately
funded cohort studies. Studies were required to have a
minimum of 36 weeks’ follow-up and a minimum of
50 patients starting HAART, of whom more than 95%
needed to be therapy naive.
Exclusion criteria
The following types of studies were excluded from our
Review: studies with exclusively paediatric populations;
studies on prevention of maternal-to-child transmission;
studies treating patients with HIV-2 infection or acute
HIV-1 infection; studies using non-recommended
Treatment Study type;
predominant
population; HIV
subtype
Number
of
patients
Age Female
(%)
Pre-
HAART
CD4 (cells
per μL)
Pre-
HAART
log viral
load
(copies
per mL)
Follow
up
(weeks)
Frequency
of viral load
monitoring
Viral load
response at
48 weeks
(switch
ignored)
Loss to
follow up
(%)
Ferradini et al
14
(Malawi)
Stavudine or zidovudine,
lamivudine, and nevirapine
Programmatic cohort;
rural adults; subtype C
398 34* 69 114* Not
reported
38 None 84† 12 (3·0)
DART trial team
15,16
(Uganda,
Zimbabwe)
Zidovudine, lamivudine,
and nevirapine
One arm of RCT; urban
adults; subtype A,C,D
300 37* 72·0 100* 5·4‡ 48 None 79§ 7 (2·3)
Kamya et al
17
(Uganda)
stavudine or zidovudine,
lamivudine, and nevirapine
Programmatic cohort;
urban adults and
adolescents; subtype A,D
526 37‡ 69 99* 5·3‡ 48 3–6 months 75† 3 (0·4)
Charles et al
18
(Haiti)
Zidovudine, lamivudine,
and efavirenz or nevirapine
Programmatic cohort;
urban adolescents and
young adults; subtype B
146 Not
reported
66 129* Not
reported
48 None 49¶ 28 (19)
Sungkanuparph
et al
19
|| (Thailand)
Stavudine, lamivudine, and
nevirapine
Programmatic cohort;
urban adults; subtype AE
1700 35‡ 37 Not
reported
Not
reported
80 3–6 months Not reported Not
reported
Marconi et al
20
(South Africa)
Stavudine or zidovudine,
lamivudine, and efavirenz or
nevirapine
Programmatic cohort;
urban and rural adults;
subtype C
3379 37‡ 52·2 Not
reported
Not
reported
43 3–6 months Not reported Not
reported
Laurent et al
21
(Cameroon)
Stavudine, lamivudine, and
nevirapine
ClinOL; urban adults;
subtype A
60 35* 66 118* 5·02* 48 3 months or
less
80†
68§
2 (3·3)
UK CHIC/DRD
11,22
(UK)
Stavudine or zidovudine,
lamivudine, and efavirenz or
nevirapine
National cohort; urban
adults; subtype B
1352 36* 30·0 110·3‡ 5·07* 48 3 months or
less
95** 162
(12)††
Ledergerber et al
12,23
(Switzerland)
Stavudine or zidovudine,
lamivudine, and efavirenz or
nevirapine
National cohort; urban
adults; subtype B
236 39·7‡ 32·2 110‡ 5·7* 48 3 months or
less
81 7 (3·0)††
Harrigan et al
13,24
(Canada)
Stavudine or zidovudine,
lamivudine, and nevirapine
or efavirenz
National cohort; urban
adults; subtype B
288 40* 20·1 120* 5·0* 48 3 months or
less
80¶ 38
(13·2)‡‡
ClinOL=clinical open label single arm study. RCT=randomised controlled trial. UK CHIC/DRD=collaborative HIV cohort/drug resistance database. *Median. †Fewer than 400 copies per mL intent-to-treat (ITT).
‡Mean. §Fewer than 50 copies per mL ITT. ¶Fewer than 50 copies per mL on treatment (OT).
||Sungkanuparph S, Mahidol University, Thailand, personal communication. **Fewer than 400 copies per mL OT.
††27% shown to return after more than 1 year.
25
‡‡Only refers to viral-load test not available between week 24 and week 48.
Table 1: Study characteristics
Page 2
www.thelancet.com/infection Vol 9 July 2009
411
Review
agents (didanosine, delavirdine, zalcitabine, nelfi navir,
indinavir, or full-dose ritonavir); studies not using
lamivudine or emtricitabine; and reviews, case–control
studies, and case reports.
Articles were independently assessed for eligibility
based on the inclusion and exclusion criteria by two
reviewers with a third author available where opinion
diff ered. To assess the degree of heterogeneity between
studies we extracted as much data on study design and
conduct as possible, and have presented this information
in tables 1–3.
Data extraction and quality assessment
We extracted data on the following: study name; authors;
antiretroviral combinations used; year of publication or
presentation; study design categorised according to
Patients with
previous
PMTCT
included?
Patients with baseline
resistance excluded
from resistance
analysis?
Baseline blood tests?
(Exclusion if liver
function test or full
blood count abnormal)
Planned regularity of
nurse visits (per year)
Planned number of
physician
consultations (per
year)
Regular
adherence
support?
Substitutions of NNRTI and
NRTI allowed?
Ferradini et al
14
(Malawi) Yes No baseline resistance
testing reported
Yes (not reported) Bi-monthly until
stabilisation
Monthly until
stabilisation
Yes Yes: nevirapine for efavirenz and
zidovudine for stavudine
DART trial team
15,16
(Uganda, Zimbabwe)
Yes No, one patient had
NNRTI resistance
Yes (yes) 4 4 Yes Yes: tenofovir or abacavir for
nevirapine or zidovudine
Kamya et al
17
(Uganda) Yes No resistance in 4/7, other
3 patients not available
Yes (not reported) 12 4 Yes Yes: nevirapine for efavirenz and
zidovudine for stavudine
Charles et al
18
(Haiti) Yes No baseline resistance
testing performed
Yes (not reported) 12 6 Yes Yes: efavirenz for nevirapine
Sungkanuparph et al
19
*
(Thailand)
No No baseline resistance
testing reported
Yes (not reported) 2–4 once stabilised 2–4 once stabilised Yes Yes: no details given
Marconi et al
20
(South Africa)
Yes No baseline resistance
testing reported
Yes (not reported) Not reported Not reported Yes Yes
Laurent et al
21
(Cameroon)
Yes No Yes (not reported) 12 Not reported Yes Yes: efavirenz for nevirapine
UK CHIC/DRD
11,22
(UK) Yes No Yes (yes) 5 4 Yes Yes
Ledergerber et al
12,23
(Switzerland)
Yes No Yes (yes) 4 4 Yes Yes
Harrigan et al
13,24
(Canada)
Yes No Yes (not reported) 4 4 Yes Yes: nevirapine for efavirenz and
zidovudine for stavudine
PMTCT=prevention of mother to child transmission. UK CHIC/DRD=collaborative HIV cohort/drug resistance database. *Sungkanuparph S, Mahidol University, Thailand, personal communication.
Table 2: Study eligibility, exclusions, and patient management details
Resistance test population
(viral load [copies per mL])
Number of virological
failures (%)
GA (% virological
failures)
Number (% GA) with drug resistance mutations
Major NNRTI
mutation
M184V/I Any thymidine analogue
mutation
K65R No major
mutations
Infrequent or no monitoring
Ferradini et al
14
(Malawi) Single (>1000) 52 (13·4) 50 (96) 47 (94) 38 (76) 6 (12) 5 (10) 3 (6)
DART trial team
15,16
(Uganda, Zimbabwe)
Single (>1000) 35 (11·7) 32 (91) 23 (72) 23 (72) 9 (28) 0* 5 (16)
Kamya et al
17
(Uganda) Single (>400) 59 (11·2) 7 (NA) 7 (100) 7 (100) 2 (29) 0 0
Charles et al
18
(Haiti) Single (>1000) 32 (22) 29 (91) 23 (79) 21 (72) 9 (31) 0* 4 (14)
Sungkanuparph et al
19
(Thailand)
Two (>1000) ·· 98 (NA) 90 (92) 87 (89) 36 (37) 6 (6) 5 (5)
Marconi et al
20
(South Africa) Single (>1000) 147 (4·3) 115 (78·2) 90 (78) 74 (64) 37 (32) 3 (3) 19 (17)
Frequent monitoring
Laurent et al
21
(Cameroon) Single (>1000) 10 (17) 10 (100) 3 (30) 2 (20) 0 0 7 (70)
UK CHIC/DRD
11,22
(UK) Single (>1000) 93 (6·9) 58 (48) 39 (67) 23 (40) 10 (17) 2 (3) 18 (27)
Ledergerber et al
12,23
(Switzerland)
Single (>1000) 9 (3·8) 6 (67) 4 (67) 4 (67) 1 (17) 0 2 (33)
Harrigan et al
13,24
(Canada) Single (>1000) 28 (9·7) 20 (71) 11 (55) 9 (45) 3 (15) 2 (10) 8 (40)
··=not reported or not defi ned. GA=number of genotypes successfully analysed. NA=not applicable. *All patients on zidovudine-containing regimens (zidovudine prevents selection of K65R mutation).
†Sungkanuparph S, Mahidol University, Thailand, personal communication.
Table 3: Virological and resistance data
Page 3
412
www.thelancet.com/infection Vol 9 July 2009
Review
whether cohort, part of a clinical trial, or patients were
randomised to particular treatment; participant charac-
teristics (age, proportion of females); baseline median or
mean CD4 count (cells per μL); median or mean log
10
viral
load (copies per mL); intention-to-treat (ITT) viral sup-
pression rates (ITT switch ignored where available);
defi nition of virological failure for resistance testing; and
numbers of virological failures with viral load greater than
1000 copies per mL on at least one occasion and number
of genotypes successfully obtained from this group.
Furthermore, genotypic resistance data for virological
failures were extracted and these were categorised by
treatment-emergent major NNRTI mutations (at positions
100, 103, 106, 108, 181, 188, 190, and 225), or three measures
of NRTI resistance (the M184V/I mutation, at least one
thymidine analogue mutation [M41L, D67N, K70R, L210W,
T215Y/F, K219Q/E], and the K65R mutation). Resulting
from the nature of clinical practice and that genotypic
analysis does not always immediately follow virological
failure, a window of 6 months before to 12 months after
was created around the date of virological failure (viral
loads above 1000 copies per mL) to identify eligible
genotypes from UK, Canadian, and Swiss cohort studies.
We also collected data on the quality of study design as
follows: inclusion of patients previously receiving
prevention of mother-to-child transmission (PMTCT)
prophylaxis, inclusion of patients with baseline resistance,
availability of baseline blood tests and subsequent
exclusion of patients with signifi cant abnormalities,
regularity of clinic and doctor consultations (per year),
and availability of regular adherence support.
Statistical analyses
Estimates of group means (and SEs) of baseline variables
(age, CD4, log
10
viral load) were calculated using inverse-
variance weights. Studies were stratifi ed for analysis into
those with intensive monitoring (more frequently than
every 12 weeks), and those with infrequent (less frequently
than every 12 weeks) or no monitoring. The main analysis
was a comparison of proportions with the following
classes of resistance mutations across monitoring groups:
M184V, at least one thymidine analogue mutation, K65R,
and one or more major NNRTI mutations. The two main
populations for analysis were genotypes (the total number
of patients with genotypic resistance data available after
virological failure) and ITT (the total number of patients
randomised).
Where only a subset of virological failures was
sequenced, the genotypes analysed rates were calculated
but such studies were not included in the ITT analysis.
In studies with no viral-load monitoring the testing was
done retrospectively on stored plasma, taken at a median
of 38 weeks for the Malawi study and at 48 weeks in the
development of antiretroviral therapy in Africa nevirapine
or abacavir (DART-NORA) study. Although resistance at
36–48 weeks was reported in these studies, the time from
rst virological failure to resistance testing was not
reported. Therefore total time of viraemia could not be
calculated.
Next, to make group comparisons between monitoring
groups, we used inverse-variance weighting and
calculation of eff ective sample size to create robust
estimates of the frequency of major NNRTI mutations,
thymidine analogue mutations, M184I/V, and K65R
accompanied by exact 95% CIs for each of the groups.
With this method, fi rst the 95% confi dence limits were
produced for each of the individual studies. The exact
interval (also called the Clopper-Pearson interval) was
used instead of the normal approximation, since the
latter can give confi dence limits below zero or above
100%. The overall estimate for the group of studies in
each monitoring group was calculated using the
weighting W=1/V, with V being the variance of the study
estimate, calculated from V=P×(1–P)/N, where P=R/N
(P=proportion with resistance mutation, R=number with
resistance mutation, N=genotype analysis population).
Finally the CI for the group estimate was found using the
concept of eff ective sample size.
26
We also undertook
sensitivity analyses to assess the robustness of our
results. Statistical analyses were performed using
Microsoft Excel version 2004.
Results
We intentionally designed as broad a search strategy as
possible, to avoid omitting potentially important studies
from the analysis. Of the 2235 abstracts identifi ed,
122 full-text papers or posters were obtained and assessed
further (fi gure 1). The majority of excluded abstracts were
studies on antiretroviral therapy without resistance data.
Of the full-text articles obtained, ten were selected for
inclusion. Common reasons for exclusion of full texts
and posters were: exclusion of patients who switched
10 articles eligible for inclusion and further data extraction
122 full-text articles or posters obtained for full review
2235 citations from all sources using search strategies
112 excluded after review
23 no CD4 stratification of <200 cells per μL
or exclusion for drug switches
10 triple nucleoside studies
20 non-recommended regimens or dosing
15 inadequate genotypic resistance data
12 non-naive patients
8 cross sectional stu dies
6 review articles
5 dose-finding or pharmacokinetic studies
7 sample size too small (<50 patients)
6 resistance follow-up period too short
2113 citations excluded after
preliminary screen
Figure 1: Summary of study search and selection
Page 4
www.thelancet.com/infection Vol 9 July 2009
413
Review
from the study drug, lack of resistance data for patients
starting HAART with a CD4 count of fewer than
200 cells per μL, non-recommended regimen or dosing,
mixing of naive and experienced patient data, and use of
triple nucleoside-analogue combinations. Clinical trials
generally did not meet the criteria since they did not
provide resistance data stratifi ed by a CD4 count of fewer
than 200 cells per μL; furthermore, they tended to exclude
patients from the resistance analysis if they switched
study drug for tolerability reasons. This last point is
important since we primarily aimed to examine resistance
in all patients starting WHO-recommended combinations
and to arrive at realistic estimates that could accurately
inform policy.
Characteristics of eligible studies
The studies included were done and presented or
published between January, 2001, and March, 2009. Their
characteristics are summarised in table 1. The majority of
cohort studies in resource-limited settings used the
generic-fi xed-dose combination stavudine, lamivudine,
and nevirapine, and, less commonly, zidovudine,
lamivudine, and nevirapine; stavudine, lamivudine, and
efavirenz; or zidovudine, lamivudine, and efavirenz.
Exact numbers of patients taking each combination were
not available since some studies used a mixture of
combinations.
8376 patients, from eight cohorts and two prospective
studies
14–24
(Sungkanuparph S, Mahidol University,
Infrequent or no monitoring
Ferradini et al
14
DART trial team
15,16
Kamya et al
17
Charles et al
18
Sungkanuparph et al
19*
Marconi et al
21
Summary
Frequent monitoring
Laurent et al
21
UK CHIC/DRD
11,22
Ledergerber et al
12,23
Harrigan et al
13,24
Summary
Major NNRTI mutation
Percentage of genotypes analysed with resistance
10 20 30 40 50 60 70 80 90 100
0
10 20 30 40 50 60 70 80 90 1000
Percentage of genotypes analysed with resistance
Estimated %
(95% CI)
M184V
94 (84–98)
72 (53–86)
100 (59–100)
79 (60–92)
91 (85–96)
78 (70–85)
88 (82–93)
30 (7–65)
67 (54–79)
67 (22–96)
55 (32–77)
61 (49–72)
p<0·001
76 (62–87)
81 (64–93)
100 (59–100)
72 (53–87)
89 (81–94)
64 (55–73)
81 (73–87)
20 (3–56)
40 (27–53)
67 (22–96)
45 (23–68)
40 (29–52)
p<0·001
Infrequent or no monitoring
Ferradini et al
14
DART trial team
15,16
Kamya et al
17
Charles et al
18
Sungkanuparph et al
19*
Marconi et al
21
Summary
Frequent monitoring
Laurent et al
21
UK CHIC/DRD
11,22
Ledergerber et al
12,23
Harrigan et al
13,24
Summary
Any thymidine analogue mutations Estimated %
(95% CI)
Estimated %
(95% CI)
K65R
12 (5–24)
32 (16–50)
29 (4–71)
31 (15–51)
37 (27–47)
32 (24–42)
28 (21–35)
0 (0–29)
17 (9–29)
17 (0–64)
15 (3–38)
12 (6–21)
p<0·001
10 (3–22)
0 (0–41)
6 (2–13)
3 (1–7)
4 (1–9)
0 (0–29)
3 (0–12)
0 (0–46)
4 (0–11)
3 (0–11)
p=0·8
Estimated %
(95% CI)
10 20 30 40 50 60 70 80 90 100
0
10 20 30 40 50 60 70 80 90 1000
Figure 2: Frequency of resistance at 48 weeks using genotypes analysed data
Frequent (3 monthly) and infrequent (less than 3 monthly) monitoring is compared. Each forest plot represents a resistance mutation category (major NNRTI,
M184V/I, thymidine analogue mutations, and K65R) and summary estimate calculated using inverse-variance weighting. Diamonds represent weighted estimates
and bars 95% CIs. Studies with purely zidovudine-based HAART were excluded from K65R analysis because of antagonism between drug and mutation.
*Sungkanuparph S, Mahidol University, Thailand, personal communication.
Page 5
414
www.thelancet.com/infection Vol 9 July 2009
Review
Thailand, personal communication) were studied from
Uganda (n=526), Thailand (n=1700), Haiti (n=146),
Malawi (n=389), South Africa (n=3379), Zimbabwe and
Uganda (n=300), Cameroon (n=60), UK (n=1352),
Switzerland (n=236), and Canada (n=288). Most studies
reported resistance no longer than 48 weeks after
initiation, and therefore similar comparative data were
extracted from UK, Canadian, and Swiss cohorts with
patients with starting CD4 counts of fewer than
200 cells per μL. Sequencing-based resistance assays
were used in all included studies.
Heterogeneity among studies
The median follow-up ranged from 38 to 80 weeks. Rates
of patients lost to follow up were low in the antiretroviral-
therapy studies in resource-poor counties, with the
exception of Charles and colleagues,
18
where 41% of the
patients were below the age of 20 years and half of these
under 16 years. This study from Haiti also reported
poorer virological-response rates than other studies
(although they were in keeping with data in adolescents
from the USA). However, the frequency of resistance
mutations at 48 weeks in viral failures was similar to the
other predominantly adult studies with infrequent or no
monitoring.
We also noted that the number of genotype analyses
successfully done as a proportion of eligible viral loads
(greater than 1000 copies per mL) was also higher in the
less intensively monitored patients (around 90% vs
48–71% in the frequently monitored cohorts). This may
be attributable to the more complete collection and
storage of samples for retrospective testing in studies
without real-time monitoring, by contrast with the data
from routine clinical practice in the Swiss and UK
cohorts. Secondly, patients failing therapy with very high
viral loads under real-time monitoring are often not
genotypically analysed, since poor adherence is assumed
by clinicians. This might explain the lower rates of
successful genotypic analysis in cohorts in high-income
countries, although any bias introduced would tend to
overestimate resistance in these patients.
Table 2 shows potential sources of bias in studies
meeting our eligibility criteria. None of the studies
excluded patients who switched drugs, and all provided
psychosocial support. Patients with a history of PMTCT
were uncommon across studies, and were included in
the analyses. Having documented key data regarding
potential confounders and fi nding little diff erence across
studies, we felt that a meta-analysis would be appropriate
and not substantially biased. All studies, although not
randomised clinical trials (resistance testing was
retrospective), did not highly select patients, further
justifying the analysis done.
After studies were stratifi ed into two groups by viral-load-
monitoring intensity (infrequent or none vs frequent),
baseline characteristics were compared. Median age was
36·0 and 37·0 years, respectively (p=0·39); mean baseline
CD4 was 106·8 versus 118·7 cells per L (p=0·52) and mean
log
10
viral load was 5·3 and 5·1, respectively (p=0·11).
Resistance at virological failure to NNRTI (as a
proportion of those with a viral load greater than
1000 copies per mL having a resistance test, also defi ned
as genotypes analysed) in patients virologically monitored
at intervals of more than 3 months or not at all was
88·3% (95% CI 82·2–92·9), compared with 61·0%
(48·9–72·2%) in more frequently monitored patients
(p<0·001). Lamivudine resistance was 80·5% (72·9–86.8)
and 40·3% (29·1–52·2), respectively (p<0·001). The
frequency of at least one thymidine analogue mutation
was 27·8% (21·2–35·2) and 12·1% (5·9–21·4),
respectively (p<0·001). The frequency of the K65R
mutation conferring reduced susceptibility to didanosine,
abacavir, and tenofovir was 4·1% (1·4–9·2) and 3·3%
(0·4–11·3), respectively (p=0·8). Figure 2 illustrates this
genotype analysis data, and ITT results in fi gure 3 show
the same patterns. We also assessed the eff ect of excluding
the Thai study
19
(Sungkanuparph S, Mahidol University,
Thailand, personal communication) from the analysis
since its median follow up was 80 weeks; there was little
diff erence in genotype analysis or ITT (data not shown)
weighted estimates. In a further sensitivity analysis we
also stratifi ed studies into three groups: no monitoring,
3-monthly monitoring, and 6-monthly monitoring. We
found that the resistance rates to NNRTI and lamivudine
in the 6-monthly and no monitoring groups were similar,
and both statistically signifi cantly higher than the
3-monthly group (data not shown).
Discussion
This is the largest analysis of resistance following public
health rollout of thymidine analogue, lamivudine, and
NNRTI-based HAART (by far the most commonly used
Any thymidine analogue mutations
NNRTI clinical trials
Cohort (frequent monitoring)
Cohort (infrequent or no monitoring)
M184V/I mutation
NNRTI clinical trials
Cohort (frequent monitoring)
Cohort (infrequent or no monitoring)
Major NNRTI
NNRTI clinical trials
Cohort (frequent monitoring)
Cohort (infrequent or no monitoring)
K65R mutation
NNRTI clinical trials
Cohort (frequent monitoring)
Cohort (infrequent or no monitoring)
0% 1% 2% 3%
Percentage of all patients starting HAART with resistance at 48 weeks (95% CI)
4% 5% 6% 7% 8%
Figure 3: Intention-to-treat resistance in patients undergoing frequent (3 monthly) or infrequent (less than
3 monthly) viral load monitoring or in NNRTI clinical trials
Estimates calculated by inverse-variance weighting; bars are 95% CIs.
Page 6
www.thelancet.com/infection Vol 9 July 2009
415
Review
regimens in resource-poor settings). Given the caveats of
such an analysis, genotypic resistance to lamivudine,
NRTIs (thymidine analogue mutations), and NNRTIs
appears substantially higher in those with virological
failure in cohorts of fi rst-line NNRTI-based HAART for
which monitoring of viral load is relatively infrequent or
absent, compared with more intensively monitored
cohorts. The most likely reason for the diff erence seen at
48 weeks is time left on HAART after initial virological
failure, thus highlighting the potential importance of
identifi cation of early viral failure. Targeted interventions
started soon after the fi rst viral failure under frequent
monitoring, would be expected to result in resuppression
of virus replication and lower the probability of detection
of resistance in such groups of patients (including
frequently monitored patients in the African setting, as
demonstrated by the data from Cameroon).
21
Interestingly, the cohorts monitored infrequently (every
3–6 months) had similar rates of 48-week resistance to
NNRTI and lamivudine to those in cohorts not monitored
at all. This may be because a single detectable viral load
at 24 weeks would generally not result in intervention
such as adherence counselling or switch before the
48 week point in a resource-poor setting where reporting
and feedback of results might be delayed. On the basis of
these data, one might suggest that frequent viral
monitoring is particularly useful in the fi rst few months
of HAART.
Randomised clinical trial resistance data for zidovudine
or stavudine with lamivudine and nevirapine is available
from the COMBINE and 2NN studies.
27–30
Thymidine
analogues with lamivudine and efavirenz as the NNRTI
have been studied in a number of randomised clinical
trials, usually as comparator NRTIs to newer agents.
31–36
Since randomised clinical trials include the most highly
selected and most intensively monitored patients, we
wondered how their resistance rates would compare with
cohorts over 48 weeks. We calculated inverse variance-
weighted resistance estimates using data collected from
these trials in a recent meta-analysis
37
and compared
them with our data. We found that frequently monitored
patients in cohort settings have similar resistance
mutation rates (in genotype analyses) to clinical trials for
NNRTI (52·6%) and lamivudine (35·1%) at 48 weeks,
although the frequency of thymidine analogue mutations
was lower in clinical trials (2·7%). The most likely
explanation for the lower rate of thymidine analogue
mutations would be more intensive monitoring in
clinical-trial settings with exclusion from the study soon
after viral failure, thus providing less time for thymidine
analogue mutation accumulation. Comparison of ITT
resistance (fi gure 3) showed similar relations.
The 48-week frequency of at least one thymidine
analogue mutation in the context of no or infrequent
monitoring is 25%, roughly twice that in frequently
monitored patients. Although this might compromise
the nucleoside analogue component of second-line
therapy, it might be that the boosted protease inhibitor
component represents suffi cient antiviral potency to
resuppress viraemia in the short term at least.
Unfortunately, more detailed analysis of the risk of this
compromise is not possible, since limited details of the
number and characteristics of thymidine analogue
mutations are available from the studies included in our
analysis. European
38
and South African
39
data suggest
similar rates of accumulation of thymidine analogue
mutations at a population level.
Limitations of this study
Our search of published work was intentionally made as
broad as possible to capture all pertinent studies, and
therefore also resulted in a high rate of exclusion. The
potential for selection bias does exist, although the lack
of heterogeneity in resistance rates from both
government-funded and privately-funded antiretroviral
therapy programmes in resource-poor settings argues
against this.
Given the paucity of available data, both observational
cohorts and a study arm from a randomised clinical trial
were included in our analysis. However, we thoroughly
evaluated potential causes of biased resistance estimates
across studies, and came to the conclusion that it was
methodologically sound to perform meta-analysis on this
dataset. Mainly because the two trials, DART-NORA and
the Cameroon study by Laurent and colleagues,
21
were
not highly selective in terms of CD4 counts or the criteria
outlined in table 2. Importantly, resistance following
PMTCT would not account for the large diff erences in
resistance frequencies seen, given that the numbers of
such patients were small (4% of viral failures in South
African study
20
and 3% in DART
15,16
). Similarly baseline
resistance, where assessed, was uncommon across less
frequent monitoring studies (consistent with published
data) and therefore would not explain higher resistance
rates in this group. The possibility also exists that studies
with viral-load testing might have provided more eff ective
adherence support, despite employing similar
frequencies of clinic visits as those studies with less
frequent viral-load testing.
An important potential source of bias is that most
patients in the frequent-monitoring group were from
well-resourced settings. However, recent data from South
Africa argue against this possibility. Hoff man and
colleagues
39
studied adults starting NNRTI-based HAART
and found that at fi rst viral-load failure the frequencies of
resistance to NNRTI, lamivudine, and thymidine
analogue mutations were 61%, 34%, and 3% respectively.
These data from patients in Africa is very similar to
frequencies seen in both intensively monitored cohorts
from well-resources countries in our study and clinical
trials.
37
Furthermore, Hoff man and colleagues found that
6 months after fi rst viral failure the frequencies in the
same group of patients had increased to 87%, 70%, and
20% respectively. These are similar to our data for non-
Page 7
416
www.thelancet.com/infection Vol 9 July 2009
Review
monitored or infrequently monitored patients,
implicating persistent viraemia as the key determinant of
this large increase over a short period of time.
First-line combinations with low risk of resistance at
time of virological failure will of course be of benefi t within
the resource-poor world, where virological monitoring
might not exist and limited second-line therapies are
available. However, a recent modelling study by Phillips
and colleagues
40
reported only a small eff ect of regular
viral-load monitoring on long-term survival compared with
clinical monitoring alone in patients starting stavudine,
lamivudine, and nevirapine in resource-poor settings
where boosted protease inhibitor containing second-line
therapy was available, despite assuming extensive
resistance at time of switch. Thus, in a context of limited
rst-line and second-line treatment options, the overall
benefi t conferred by antiretroviral therapy might outweigh
the presence of substantial levels of resistance at time of
clinical failure of fi rst-line therapy. However, despite
assuming baseline resistance following PMTCT, this
study
40
did not directly assess the eff ect of higher rates of
transmitted resistance on outcomes.
Following the fi ndings from the analysis presented,
rigorous modelling of the likely eff ects on transmitted
resistance is needed, in combination with eff ective
surveillance. There are early reports suggesting
transmitted resistance directly following national
antiretroviral-therapy-rollout programmes.
41,42
A future
scenario where rates of baseline resistance to NNRTI
with or without lamivudine are high might need
reconsideration of fi rst-line treatment regimen to
maintain both individual and population level benefi ts of
HAART.
Viral-load tests appropriate for resource-poor settings
are a research priority and have been reviewed.
43
Unfortunately, currently available methods including
ultrasensitive p24 assays and reverse-transcriptase-
activity assays still require basic laboratory
infrastructure, and point-of-care tests are needed. Such
testing would also be important for pregnant women
on HAART to ensure viral suppression in the latter
stages of gestation, and also to verify that a patient with
clinical or immunological failure is indeed virologically
failing before therapy switch is made. Importantly,
viral-load testing should not be at the expense of access
to antiretroviral therapy and associated infrastructure.
Conclusions
In the absence of an evidence base for use of viral-load
testing in the public-health rollout of antiretroviral
therapy, this Review represents the fi rst attempt to
examine the resistance implications of diff ering
virological-monitoring strategies. Our data show large,
indisputable diff erences in frequency of major resistance
mutations in virologically failing patients under
infrequent versus frequent viral-load monitoring, and
thus supports the use of sequential drug combinations
with non-overlapping resistance profi les in the context of
limited monitoring. The current WHO recommendations
for resource-poor settings have attempted to formulate
second-line regimens with this in mind. The clinical
implications of our fi ndings will be determined by studies
of second-line therapy following failure of fi rst-line
therapy with limited or no monitoring.
For the present, universal access to HAART remains
the priority. Lack of viral-load monitoring seems to be
associated with resistance in the vast majority of those
with viral failure after the fi rst year of HAART, and
therefore early identifi cation of viral failure within this
period might be the optimal way to employ point-of-care
viral-load testing. Alternatively, use of well-tolerated
HAART regimens associated with lower rates of
resistance might be a future option.
Confl icts of interest
AH has received consultancy payments from Tibotec. RG has received
educational travel grants from Gilead Sciences, Bristol-Myers Squibb,
and Boehringer Ingelheim. HFG has been an advisor to or has
received unrestricted research and travel grants from
GlaxoSmithKline, Abott, Novartis, Boehringer Ingelheim, Roche,
Pfi zer, Tibotec, Merck, and Bristol-Myers Squibb. DP has acted as a
consultant for Bristol-Myers Squibb, Johnson & Johnson, Boehringer
Ingelheim, Roche, and Gilead Sciences. RG is funded as a Wellcome
Trust Clinical Research Training Fellow. UK CHIC is funded by the
UK Medical Research Council (G0600337). The Swiss HIV Cohort
study is supported by the Swiss National Science Foundation (SNF #
3345-062041). Further support was provided by SNF # 3247B0-112594
to HFG and SY, SHCS project 470 and 528, and by a further research
grant of the Union Bank of Switzerland in the name of a donor (to
HFG). The funding sources had no role in the conception, design,
and execution of this study.
Acknowledgments
We would like to thank Richard Harrigan for his collaborative eff orts on
behalf of the HOMER study and Somnuek Sungkanuparph for
providing further data on the Thai study. We would also like to thank
Andrew Phillips for critical review of the manuscript. We also thank
UCLH/UCL Comprehensive Biomedical Research Centre for support.
The research leading to these results has recieved funding from the
European Community’s Seventh Framework Programme (FP7/2007–
2013) under the project: Collaborative HIV and Anti-HIV Drug
Resistance Network (CHAIN)—grant agreement no 223131. Members of
the UK CHIC steering committee, the UK HIV Drug Resistance
Database Steering Committee, and the Swiss HIV Cohort Study are
listed in the webappendix.
References
1 Gilks CF, Crowley S, Ekpini R, et al. The WHO public-health
approach to antiretroviral treatment against HIV in resource-
limited settings. Lancet 2006; 368: 505–10.
2 Gazzard B, Bernard AJ, Boffi to M, et al. British HIV
Association (BHIVA) guidelines for the treatment of HIV-infected
adults with antiretroviral therapy (2006). HIV Med 2006; 7: 487–503.
3 Hammer SM, Saag MS, Schechter M, et al. Treatment for adult
HIV infection: 2006 recommendations of the International AIDS
Society-USA panel. JAMA 2006; 296: 827–43.
4 Mermin J, Were W, Ekwaru JP, et al. Mortality in HIV-infected
Ugandan adults receiving antiretroviral treatment and survival of
their HIV-uninfected children: a prospective cohort study. Lancet
2008; 371: 752–59.
5 Jahn A, Floyd S, Crampin AC, et al. Population-level eff ect of
HIV on adult mortality and early evidence of reversal after
introduction of antiretroviral therapy in Malawi. Lancet 2008;
371: 1603–11.
6 Hogg RS, Bangsberg DR, Lima VD, et al. Emergence of drug
resistance is associated with an increased risk of death among
patients fi rst starting HAART. PLoS Med 2006; 3: e356.
Search Strategy and
selection criteria
Theses are described
in detail in the
Methods section
See Online for webappendix
Page 8
www.thelancet.com/infection Vol 9 July 2009
417
Review
7 Petersen ML, van der Laan MJ, Napravnik S, Eron JJ, Moore RD,
Deeks SG. Long-term consequences of the delay between virologic
failure of highly active antiretroviral therapy and regimen
modifi cation. AIDS 2008; 22: 2097–106.
8 Siegfried NL, Van Deventer PJ, Mahomed FA, Rutherford GW.
Stavudine, lamivudine and nevirapine combination therapy for
treatment of HIV infection and AIDS in adults.
Cochrane Database Syst Rev 2006; 2: CD004535.
9 Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF.
Improving the quality of reports of meta-analyses of randomised
controlled trials: the QUOROM statement. Lancet 1999;
354: 1896–900.
10 US Food and Drug Administration. Drugs used in the treatment of
HIV infection. http://www.fda.gov/oashi/aids/virals.html
(accessed May 19, 2009).
11 The UK Collaborative HIV Cohort Steering Committee. The
creation of a large UK-based multicentre cohort of HIV-infected
individuals: the UK collaborative HIV cohort (UK CHIC) study.
HIV Medicine 2004; 5: 115–24.
12 Ledergerber B, Egger M, Opravil M, et al. Clinical progression and
virological failure on highly active antiretroviral therapy in HIV-1
patients: a prospective cohort study. Lancet 1999; 353: 863–68.
13 Harrigan PR, Hogg RS, Dong WW, et al. Predictors of HIV drug-
resistance mutations in a large antiretroviral-naive cohort initiating
triple antiretroviral therapy. J Infect Dis 2005; 191: 339–47.
14 Ferradini L, Jeannin A, Pinoges L, et al. Scaling up of highly active
antiretroviral therapy in a rural district of Malawi: an eff ectiveness
assessment. Lancet 2006; 367: 1335–42.
15 DART trial team. Twenty-four-week safety and tolerability of
nevirapine vs abacavir in combination with zidovudine/lamivudine
as fi rst-line antiretroviral therapy: a randomized double-blind trial
(NORA). Trop Med Int Health 2008; 13: 6–16.
16 Ndembi N, Pillay D, Goodall R, et al. Diff erences in the dynamics of
viral rebound and evolution of resistance between CBV/NVP and
CBV/ABC uncovered in the absence of viral load monitoring in real
time: NORA substudy of the DART trial. 15th conference on
retroviruses and opportunistic infections; Boston, USA;
Feb 3–6, 2008. Abstract 889.
17 Kamya MR, Mayanja-Kizza H, Kambugu A, et al. Predictors of long-
term viral failure among Ugandan children and adults treated with
antiretroviral therapy. J Acquir Immune Defi c Syndr 2007; 46: 187–93.
18 Charles M, Noel F, Leger P, et al. Survival, plasma HIV-1 RNA
concentrations and drug resistance in HIV-1-infected Haitian
adolescents and young adults on antiretrovirals.
Bull World Health Organ 2008; 86: 970–77.
19 Sungkanuparph S, Manosuthi W, Kiertiburanakul S, Piyavong B,
Chumpathat N, Chantratita W. Options for a second-line
antiretroviral regimen for HIV type 1-infected patients whose initial
regimen of a fi xed-dose combination of stavudine, lamivudine, and
nevirapine fails. Clin Infect Dis 2007; 44: 447–52.
20 Marconi VC, Sunpath H, Lu Z, et al. Prevalence of HIV-1 drug
resistance after failure of a fi rst highly active antiretroviral therapy
regimen in KwaZulu Natal, South Africa. Clin Infect Dis 2008;
46: 1589–97.
21 Laurent C, Kouanfack C, Koulla-Shiro S, et al. Eff ectiveness and
safety of a generic fi xed-dose combination of nevirapine, stavudine,
and lamivudine in HIV-1-infected adults in Cameroon: open-label
multicentre trial. Lancet 2004; 364: 29–34.
22 Phillips AN, Dunn D, Sabin C, et al. Long term probability of
detection of HIV-1 drug resistance after starting antiretroviral
therapy in routine clinical practice. AIDS 2005; 19: 487–94.
23 von Wyl V, Yerly S, Böni J, et al. Emergence of HIV-1 drug
resistance in previously untreated patients initiating combination
antiretroviral treatment: a comparison of diff
erent regimen types.
Arch Intern Med 2007; 167: 1782–90.
24
Tam LW, Hogg RS, Yip B, Montaner JS, Harrigan PR, Brumme CJ.
Performance of a World Health Organization fi rst-line regimen
(stavudine/lamivudine/nevirapine) in antiretroviral-naïve
individuals in a Western setting. HIV Med 2007; 8: 267–70.
25 Mocroft A, Kirk O, Aldins P, et al. Loss to follow-up in an
international, multicentre observational study. HIV Med 2008;
9: 261–69.
26 Armitage P, Berry G. Statistical methods in medical research,
3rd edn. Oxford: Blackwell Scientifi c Publications, 1994.
27 van Leth F, Phanuphak P, Ruxrungtham K, et al. Comparison of
rst-line antiretroviral therapy with regimens including nevirapine,
efavirenz, or both drugs, plus stavudine and lamivudine: a
randomised open-label trial, the 2NN Study. Lancet 2004;
363: 1253–63.
28 Hall D, van Leth F, Schere J, et al. Genotypic analysis of reverse
transcriptase in treatment-naïve HIV-1 patients treated with
lamivudine, stavudine and nevirapine and/or efavirenz.
11th conference on retroviruses and opportunistic infections;
San Francisco, USA; Feb 8–11, 2004. Abstract 694.
29 Podzamczer D, Ferrer E, Consiglio E, et al. A randomized clinical
trial comparing nelfi navir or nevirapine associated to zidovudine/
lamivudine in HIV-infected naive patients (the Combine Study).
Antivir Ther 2002; 7: 81–90.
30 Ferrer E, Podzamczer D, Arnedo M, et al. Genotype and phenotype
at baseline and at failure in human immunodefi ciency virus-
infected antiretroviral-naive patients in a randomized trial
comparing zidovudine and lamivudine plus nelfi navir or
nevirapine. J Infect Dis 2003; 187: 687–90.
31 Gallant JE, DeJesus E, Arribas JR, et al. Tenofovir DF, emtricitabine,
and efavirenz vs zidovudine, lamivudine, and efavirenz for HIV.
N Engl J Med 2006; 354: 251–60.
32 Gallant JE, Staszewski S, Pozniak AL, et al. Effi cacy and safety of
tenofovir DF vs stavudine in combination therapy in antiretroviral-
naive patients: a 3-year randomized trial. JAMA 2004; 292: 191–201.
33 DeJesus E, Herrera G, Teofi lo E, et al. Abacavir versus zidovudine
combined with lamivudine and efavirenz, for the treatment of
antiretroviral-naive HIV-infected adults. Clin Infect Dis 2004;
39: 1038–46.
34 DeJesus E, McCarty D, Farthing CF, et al. Once-daily versus twice-daily
lamivudine, in combination with zidovudine and efavirenz, for the
treatment of antiretroviral-naive adults with HIV infection:
a randomized equivalence trial. Clin Infect Dis 2004; 39: 411–18.
35 Vavro C, McCarty D, Shortino D, Hetherington S. Genotypic
analysis of HIV-1 from subjects experiencing virologic breakthrough
while taking 3TC QD vs 3TC BID, EFV and ZDV. 42nd interscience
conference on antimicrobial agents and chemotherapy; San Diego,
California; Sept 27–30, 2002. Abstract H-2052.
36 Cameron W, da Silva B, Arribas J,
et al. A two-year randomized
controlled clinical trial in antiretroviral-naïve subjects using
lopinavir/ritonavir (LPV/r) monotherapy after initial induction
treatment compared to an efavirenz (EFV) 3-drug regimen (Study
M03-613). 16th international AIDS conference; Toronto, Canada;
Aug 13–18, 2006. Abstract THLB0201.
37 Gupta RK, Hill A, Sawyer AW, Pillay D. Emergence of drug
resistance in HIV-1 infected patients after fi rst-line highly active
antiretroviral therapy (HAART): a systematic review of clinical
trials. Clin Infect Dis 2008; 47: 712–22.
38 Cozzi-Lepri A, Ruiz L, Loveday C, et al. Thymidine analogue
mutation profi les: factors associated with acquiring specifi c profi les
and their impact on the virological response to therapy. Antivir Ther
2005; 10: 791–802.
39 Hoff mann C, Charalambous S, Ledwaba J, et al. The rate of
developing ART resistance during HIV viraemia on HAART in
South Africa. 16th conference on retroviruses and opportunistic
infections; Montreal, Canada; Feb 8–11, 2009. Abstract 656.
40 Phillips AN, Pillay D, Miners AH, et al. Outcomes from monitoring
of patients on antiretroviral therapy in resource-limited settings
with viral load, CD4 cell count, or clinical observation alone:
a computer simulation model. Lancet 2008; 371: 1443–51.
41 Barth RE, Wensing AM, Tempelman HA, Moraba R, Schuurman R,
Hoepelman AI. Rapid accumulation of nonnucleoside reverse
transcriptase inhibitor-associated resistance: evidence of
transmitted resistance in rural South Africa. AIDS 2008;
22: 2210–12.
42 Apisarnthanarak A, Jirayasethpong T, Sa-nguansilp C, et al.
Antiretroviral drug resistance among antiretroviral-naïve persons
with recent HIV infection in Thailand. HIV Med 2008; 9: 322–25.
43 Fiscus SA, Cheng B, Crowe SM, et al. HIV-1 viral load assays for
resource-limited settings. PLoS Med 2006; 3: e417.
Page 9
  • Source
    • "WHO has recently recommended monitoring patients by measuring plasma HIV RNA level, or viral load testing, which can reduce transmitted drug resistance (TDR) if implemented at regular intervals (every 6 or 12 monthly). viral load testing can reduce the emergence of HIV drug resistance by early identification of patients with virological failure, prompting intensified adherence counselling and switch to second-line ART as necessary, thereby minimizing emergence of HIV drug resistance [6,7]. Second, prompt switching to a protease-inhibitor (PI)-based second-line regimen of individuals experiencing virological failure has been associated with a reduced risk for drug resistance [5,8]. "
    [Show abstract] [Hide abstract] ABSTRACT: Introduction Earlier antiretroviral therapy (ART) initiation reduces HIV-1 incidence. This benefit may be offset by increased transmitted drug resistance (TDR), which could limit future HIV treatment options. We analyze the epidemiological impact and cost-effectiveness of strategies to reduce TDR. Methods We develop a deterministic mathematical model representing Kampala, Uganda, to predict the prevalence of TDR over a 10-year period. We then compare the impact on TDR and cost-effectiveness of: (1) introduction of pre-therapy genotyping; (2) doubling use of second-line treatment to 80% (50–90%) of patients with confirmed virological failure on first-line ART; and (3) increasing viral load monitoring from yearly to twice yearly. An intervention can be considered cost-effective if it costs less than three times the gross domestic product per capita per quality adjusted life year (QALY) gained, or less than $3420 in Uganda. Results The prevalence of TDR is predicted to rise from 6.7% (interquartile range [IQR] 6.2–7.2%) in 2014, to 6.8% (IQR 6.1–7.6%), 10.0% (IQR 8.9–11.5%) and 11.1% (IQR 9.7–13.0%) in 2024 if treatment is initiated at a CD4 <350, <500, or immediately, respectively. The absolute number of TDR cases is predicted to decrease 4.4–8.1% when treating earlier compared to treating at CD4 <350 due to the preventative effects of earlier treatment. Most cases of TDR can be averted by increasing second-line treatment (additional 7.1–10.2% reduction), followed by increased viral load monitoring (<2.7%) and pre-therapy genotyping (<1.0%). Only increasing second-line treatment is cost-effective, ranging from $1612 to $2234 (IQR $450-dominated) per QALY gained. Conclusions While earlier treatment initiation will result in a predicted increase in the proportion of patients infected with drug-resistant HIV, the absolute numbers of patients infected with drug-resistant HIV is predicted to decrease. Increasing use of second-line treatment to all patients with confirmed failure on first-line therapy is a cost-effective approach to reduce TDR. Improving access to second-line ART is therefore a major priority.
    Full-text · Article · Dec 2014 · Journal of the International AIDS Society
  • Source
    • "In summary, our study shows that preexisting K103N, Y181C or M184V variants only have a very low positive predictive value for ART-naive patients in routine clinical settings wherein the rate of virologic failure is very low as it is the case for the SHCS [34]. Hence, the number of possibly preventable virologic failures in patients harbouring minority NNRTI-resistant HIV-1 variants was small. "
    [Show abstract] [Hide abstract] ABSTRACT: Objective:The presence of minority nonnucleoside reverse transcriptase inhibitor (NNRTI)-resistant HIV-1 variants prior to antiretroviral therapy (ART) has been linked to virologic failure in treatment-naive patients.Design:We performed a large retrospective study to determine the number of treatment failures that could have been prevented by implementing minority drug-resistant HIV-1 variant analyses in ART-naive patients in whom no NNRTI resistance mutations were detected by routine resistance testing.Methods:Of 1608 patients in the Swiss HIV Cohort Study, who have initiated first-line ART with two nucleoside reverse transcriptase inhibitors (NRTIs) and one NNRTI before July 2008, 519 patients were eligible by means of HIV-1 subtype, viral load and sample availability. Key NNRTI drug resistance mutations K103N and Y181C were measured by allele-specific PCR in 208 of 519 randomly chosen patients.Results:Minority K103N and Y181C drug resistance mutations were detected in five out of 190 (2.6%) and 10 out of 201 (5%) patients, respectively. Focusing on 183 patients for whom virologic success or failure could be examined, virologic failure occurred in seven out of 183 (3.8%) patients; minority K103N and/or Y181C variants were present prior to ART initiation in only two of those patients. The NNRTI-containing, first-line ART was effective in 10 patients with preexisting minority NNRTI-resistant HIV-1 variant.Conclusion:As revealed in settings of case-control studies, minority NNRTI-resistant HIV-1 variants can have an impact on ART. However, the implementation of minority NNRTI-resistant HIV-1 variant analysis in addition to genotypic resistance testing (GRT) cannot be recommended in routine clinical settings. Additional associated risk factors need to be discovered.
    Full-text · Article · Jul 2014 · AIDS (London, England)
  • Source
    • "Among patients who achieved a suppressed HIV RNA level, the majority of initial HIV RNA measurements of at least 400 copies/ml were followed by a consecutive HIV RNA measurement of less than 400 copies/ml, and we were unable to detect any effect of delayed switch on mortality among patients with nonconfirmed virologic rebound. These observations support the use of a confirmatory HIV RNA level for switching, at least at the lower threshold for suppression [37]. Patients in our cohort initiated ART at low CD4 þ cell counts. "
    [Show abstract] [Hide abstract] ABSTRACT: This paper describes a targeted maximum likelihood estimator (TMLE) for the parameters of longitudinal static and dynamic marginal structural models. We consider a longitudinal data structure consisting of baseline covariates, time-dependent intervention nodes, intermediate time-dependent covariates, and a possibly time-dependent outcome. The intervention nodes at each time point can include a binary treatment as well as a right-censoring indicator. Given a class of dynamic or static interventions, a marginal structural model is used to model the mean of the intervention-specific counterfactual outcome as a function of the intervention, time point, and possibly a subset of baseline covariates. Because the true shape of this function is rarely known, the marginal structural model is used as a working model. The causal quantity of interest is defined as the projection of the true function onto this working model. Iterated conditional expectation double robust estimators for marginal structural model parameters were previously proposed by Robins (2000, 2002) and Bang and Robins (2005). Here we build on this work and present a pooled TMLE for the parameters of marginal structural working models. We compare this pooled estimator to a stratified TMLE (Schnitzer et al. 2014) that is based on estimating the intervention-specific mean separately for each intervention of interest. The performance of the pooled TMLE is compared to the performance of the stratified TMLE and the performance of inverse probability weighted (IPW) estimators using simulations. Concepts are illustrated using an example in which the aim is to estimate the causal effect of delayed switch following immunological failure of first line antiretroviral therapy among HIV-infected patients. Data from the International Epidemiological Databases to Evaluate AIDS, Southern Africa are analyzed to investigate this question using both TML and IPW estimators. Our results demonstrate practical advantages of the pooled TMLE over an IPW estimator for working marginal structural models for survival, as well as cases in which the pooled TMLE is superior to its stratified counterpart.
    Full-text · Article · Jun 2014 · Journal of Causal Inference
Show more