JAC
antiviral
Journal of Antimicrobial Chemotherapy (2006) 58, 147–153
doi:10.1093/jac/dkl189
Advance Access publication 10 May 2006
Predictors of short-term success of antiretroviral therapy in
HIV infection
Mark Oette1*, Arne Kroidl1, Klaus Göbels1, Antje Stabbert1, Marion Menge1,
Abdurrahman Sagir1, Dieter Kuschak2, Tara O’Hanley1,
Johannes G. Bode1 and Dieter Häussinger1
1
Clinic for Gastroenterology, Hepatology, and Infectious Diseases, University Clinic Düsseldorf,
Moorenstrasse 5, 40225 Düsseldorf, Germany; 2Medical Laboratories, Nordstrasse 44,
40477 Düsseldorf, Germany
Received 29 December 2005; returned 27 February 2006; revised 10 April 2006; accepted 18 April 2006
Objectives: The success of highly active antiretroviral therapy (HAART) in HIV infection may be influenced
by numerous host factors. There is a lack of data presenting a combined assessment of a variety of these
parameters for treatment efficacy in clinical routine practice.
Methods: Different indices of therapeutic drug monitoring (TDM) were evaluated prospectively in the
context of self-reported adherence, health-related quality of life and social determinants, as measured
by a questionnaire.
Results: A total of 210 individuals were studied between 2002 and 2004, 77% were males, mean age was
44 years, mean CD4 count was 336 cells/mm3 and 63% had a viral load <50 copies/mL. In univariate analysis,
baseline viral load, unscheduled drug levels, a 4 h pharmacokinetic profile (PK-P) at a scheduled visit and
self-reported complete adherence within the previous 2 weeks were significantly associated with virological success of HAART at 12 weeks. At 24 weeks, only baseline viral load, the 4 h PK-P and adherence were
significantly associated with HAART efficacy. In multivariate analysis, baseline viral load, adherence,
unscheduled drug levels, trough levels at a visit with appointment as well as the 4 h PK-P were significantly
associated with virological success at 12 weeks. At 24 weeks, only adherence was significantly linked to
outcome. The other parameters were not found to have an impact on treatment efficacy.
Conclusions: TDM and self-reported adherence were independently predictive of short-term HAART
success in this prospective study. Unscheduled drug measurements provided similar diagnostic
information as a 4 h PK-P. Thus, we propose the use of unscheduled drug level monitoring and self-reported
adherence to help identify patients with elevated risk of virological failure.
Keywords: HAART, adherence, TDM, social, quality of life
Introduction
Highly active antiretroviral therapy (HAART) for HIV infection
has led to reduced morbidity and mortality in treated individuals.1
However, in a substantial proportion of patients the effectiveness
of HAART is not sufficient with the consequence of virological
failure and immunological decay.2 One element that has been
identified as essential for a successful HAART combination is
therapeutic drug monitoring (TDM), which is valuable for the
management of insufficient bioavailability of medication, the
determination of drug interactions and the prevention of toxic
effects.3–6 Plasma drug concentrations of antiretroviral compounds have been linked to virological outcome of
HAART.7–9 The application of TDM in daily practice remains
controversial;3 however, it has been described as cost-effective.10
Furthermore, it has been demonstrated that adherence is independently associated with virological failure, clinical progression
and mortality,11–17 and insufficient protease inhibitor (PI) concentrations are linked to low adherence.13,18 A further field of
research into the prediction of short-term success of HAART is
.............................................................................................................................................................................................................................................................................................................................................................................................................................
*Corresponding author. Tel: +49-211-81-18942; Fax: +49-211-81-16294; E-mail: [email protected]
.............................................................................................................................................................................................................................................................................................................................................................................................................................
147
The Author 2006. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved.
For Permissions, please e-mail: [email protected]
Oette et al.
health-related quality of life (HRQL). The beneficial impact of
HAART on HRQL has been demonstrated; however, the influence of HRQL on treatment success remains in question.19–21
Finally, there is evidence that social status has a moderate but
significant impact on the outcome of HAART.22–25
Very few studies exist investigating the influence of these
factors within a combined approach.14,26,27 Moreover, it is
unclear which tools provide the best information for patient management [i.e. pharmacokinetic (PK) parameters trough levels,
unscheduled monitoring or complete PK assessments of TDM
in clinical routine].4,7,14,28,29 The aim of the present study was
to evaluate the effect of TDM in relation to clinical characteristics, adherence, indicators of quality of life and social status
upon short-term success of HAART within a non-selected patient
group.
Methods
The ADDMORE project is an ongoing prospective investigation
evaluating ADherence, Drug MOnitoring and REsistance testing
in HIV-infected patients treated with HAART. We studied an
unselected cohort that underwent TDM while on stable HAART
for more than 3 months. Patients were randomly enrolled in routine
clinical practice in a university out-patient unit specialized for infectious diseases. The local institutional review board approved the
protocol of the study.
Plasma concentrations of all commercially available PI and nonnucleoside reverse transcriptase inhibitors (NNRTI) were measured.
PK samples were taken at enrolment after informed consent was
obtained. Simultaneously, patients completed a questionnaire to
assess drug adherence, socioeconomic indices and HRQL. During
a second visit 4 weeks later, patients underwent a scheduled PK
assessment to measure trough level and hourly plasma concentrations
up to 4 h after standardized drug ingestion with breakfast with the
exception of efavirenz. PK analysis of efavirenz was performed at the
same time-points as for the other compounds; however, the drug was
taken the night before measurement at the time of regular administration. In patients demonstrating a viral load above 50 copies/mL
under therapy, the scheduled visit was performed 4 weeks after
treatment switch. Concentrations were determined by validated
HPLC, as described previously.30 Concentration thresholds for
trough levels (Ctrough) as well as the lower value of the therapeutic
range were defined as shown in Table 1. The value for ritonavir,
applied as a booster (/r), was not used for the analysis. To create a
comparable classification of the values of PK assessments, the
determined drug levels were defined as sufficient or insufficient.
Concentrations at unscheduled visits and trough levels were regarded
as sufficient if values exceeded the mentioned Ctrough. The 4 h
pharmacokinetic profile (PK-profile) was regarded as sufficient if
all values were above the Ctrough and at least one level of each
measured compound exceeded the lower value of the therapeutic
range.
Demographic and clinical characteristics, CD4 cell count and viral
load at baseline were recorded. The questionnaire consisted of items
that covered the areas of adherence, HRQL and social determinants.
Adherence was assessed considering self-efficacy, evaluating the
belief of the patients to be able to be compliant with the prescribed
combination scheme (yes, no), and correctness of medication intake,
assessed by the question of having forgotten a drug dose within the
last 2 days, 14 days, last weekend or never. Adherence data were
transformed to dichotomous variables for further analysis, dividing
the results into complete and less than complete adherence. HRQL
Table 1. Definition of pharmacokinetic values
Drug
Efavirenz
Nevirapine
Amprenavir
Atazanavir
Lopinavir
Nelfinavir + M8 metabolite
Saquinavir
Trough
level (ng/mL)
Lower value of
the therapeutic
range (ng/mL)
1100
3400
300
150
500
1400
180
1900
5900
5400
2000
3500
3000
2200
was measured using validated methods as described before.31,32 We
selected the questions originally published as numbers 2, 3, 5, 7–18
in the Medical Outcomes Study.32 Answers were grouped to
combined indicators as validated before, resulting in the variables
general, mental and HIV-associated HRQL.32 The ‘general’ variable
consisted of item 1 from the Medical Outcomes Study32 plus the
following question: ‘To what extent is your well-being reduced by
your HIV infection? (very much, much, moderate, not reduced)’. The
‘HIV-associated’ variable consisted of the question as described
before plus items 13e–h from the Medical Outcomes Study.32 The
‘mental’ variable was defined according to the Medical Outcomes
Study.32 School education and professional qualification (each as
completed graduation) were used as social indicators. The analysis
of ordinal variables was performed after transformation to a scale
ranging from 0 to 100, with 0 representing the lower edge of the
scale. Follow-up consisted of viral load measurements at weeks 12
and 24. In addition, we monitored all modifications of HAART
dosage, including those increases due to insufficient drug levels
as well as discontinuations due to drug toxicity or virological failure.
The statistical analysis was performed with SPSS, release 12.1.
Univariate comparisons were applied using the t-test or two-sided
Fisher’s exact test, where appropriate. Multivariate analyses were
performed with the help of logistic regression models. P values <
0.05 were considered significant; no adjustment for multiple testing
was done.
Results
Altogether, 210 patients participated in the study between 2002
and 2004. The baseline characteristics of this population are
described in Table 2. Patients were treated with the following
quantifiable antiretroviral drugs, used as a single or boosted
agent: ritonavir-boosted amprenavir (n = 10), ritonavir-boosted
atazanavir (n = 30), efavirenz (n = 24), ritonavir-boosted lopinavir (n = 78), nelfinavir (n = 17), nevirapine (n = 23), ritonavirboosted saquinavir (n = 4). The following combinations of
compounds were applied: efavirenz and ritonavir-boosted
lopinavir (n = 4), nevirapine and ritonavir-boosted amprenavir
(n = 2), nevirapine and ritonavir-boosted lopinavir (n = 8),
ritonavir-boosted lopinavir and amprenavir (n = 3), ritonavirboosted lopinavir and saquinavir (n = 7), ritonavir-boosted
saquinavir and atazanavir (n = 1). At unscheduled PK measurement, 87.3% of individuals had sufficient drug levels, 12.7%
showed low concentrations. Ctrough on the day of appointment
for PK was above the lower cut-off in 82.9% of cases and below
148
Predictors of successful HAART
Table 2. Baseline characteristics
General
gender
males
females
age
mean (standard deviation)
body mass index
mean kg/m2 (standard deviation)
ethnic origin
Caucasian
African
Asian
Determinants of HIV infection
duration of diagnosis
mean years (standard deviation)
transmission groups
homosexual
heterosexual
endemic region
blood component
iv-drug use
unknown
CDC stage
A
B
C
CD4 cell count
Mean (cells/mm3) (standard deviation)
viral load
mean (copies/mL) (standard deviation)
treatment efficacy
viral load < 50 copies/mL
161 (76.7%)
49 (23.3%)
44.3 (10.4)
23.2 (4.3)
170 (80.9%)
30 (14.3%)
10 (4.8%)
7.7 (5.4)
108
36
34
7
13
12
(51.4%)
(17.1%)
(16.2%)
(3.3%)
(6.2%)
(5.7%)
47 (22.4%)
75 (35.7%)
88 (41.9%)
333.5 (189)
3793 (15112)
132 (62.9%)
this level in 17.1% (n = 36). Of these, 17 had a detectable viral
load at baseline, and 19 had a viral load below the detection limit,
respectively. The 4 h PK-profile showed sufficient concentrations
in 61.9%, whereas in 38.1% the concentration was regarded as
insufficient. In 16.7% of patients, a dose adjustment due to low
concentrations was applied.
Considering adherence, 97.0% of individuals pledged to take
the prescribed medication regularly; 3.0% declined to pledge or
gave no answer. In 88.4% medication intake was considered to be
complete; 11.6% of the patients admitted missing a dose or gave
no answer. HRQL results, after conversion to a metric scale, were
as follows: mean value of the combined variables (–SD) of general HRQL was 65.4 (–21), for mental HRQL it was 66.3 (–21)
and for HIV-associated HRQL it was 69.7 (–22). The social
determinant of school education showed intermediate or high
graduation levels in 40.7% and low or no qualification in
59.3%. Considering professional education, 17.0% of cases
had intermediate or high graduation levels and 83.0% had a
low or no qualification.
All but two patients were monitored for 24 weeks. No discontinuations of HAART occurred, but 11 patients switched
antiretroviral therapy due to side effects or insufficient decline
of viral load within the follow-up period. Univariate analysis
identified the following parameters as being significantly
JAC
antiviral
associated with suppression of viral replication at 12 weeks:
baseline viral load, unscheduled drug levels, the 4 h PK-profile
and completeness of drug intake (see Table 3). At 24 weeks, only
baseline viral load, the PK-profile and correctness of medication
intake were significantly associated with suppression of viral
replication. There was a tendency towards an association of
the trough level of the PK-profile with virological efficacy at
12 weeks that did not reach statistical significance. The assessed
variables of HRQL, social status and the question on self-efficacy
were not predictive for virological success.
In the multivariate analyses, three different models of combinations of confounding variables were tested, which correspond
to the items that were found to be significantly associated with
virological efficacy in univariate analysis (baseline patient characteristics, the different PK assessments and the adherence question on correctness of drug intake) (Tables 4 and 5). Additionally,
the parameters baseline CD4 cell count, body mass index and
dosage modification due to low drug levels were included, as
these were considered to be of significant impact on treatment
outcome. Baseline viral load and each of the PK tests were
significantly associated with virological success of therapy at
12 weeks, but not at 24 weeks. Adherence measured by correctness of drug intake was significantly associated with effective
HAART at 12 and 24 weeks.
No association was found between the PK tests and adherence
indicators or between baseline viral load and PK sampling or
adherence. Furthermore, parameters found not to be significantly
associated with treatment outcome in univariate analysis were
also found not to be associated in two different models of multivariate analysis (data not shown).
In summary, baseline viral load, PK variables and adherence
measured by completeness of drug intake in the near past were
significantly associated with virological efficacy at 12 weeks.
With the exception of Ctrough (non-significant in univariate analysis), this result was valid both in univariate and multivariate
analyses. At 24 weeks, the 4 h PK-profile (in univariate analysis)
and the adherence question on medication intake (in both analyses) were associated with success. The other tested variables
did not show a significant association with virological outcome.
Discussion
Combined analyses of factors influencing the outcome of
HAART are scarce. Aside from trials on specific regimens, several studies have addressed aspects such as baseline clinical
characteristics, TDM and adherence as predictors of therapeutic
success (see discussion in Reference 14). However, only a minority of investigations used a combined analysis for detecting independent factors.8,33 Thus, the purpose of our study was to assess
the significance of a variety of clinical characteristics and different means of HAART monitoring on virological success of
treatment. This was a prospective study, which tested parameters
that have previously been shown to have an impact on HAART
efficacy. Individuals were enrolled in daily practice in a nonselected manner. Altogether, 210 individuals were studied, predominantly middle-aged normal-weight Caucasian males with
homosexual HIV transmission. Applied regimens consisted of
modern combinations including NNRTI or boosted PI. The cohort
was moderately advanced in the course of HIV disease, more than
half of the cases were treated sufficiently with HAART. All
149
Oette et al.
Table 3. Treatment outcome I, univariate analysis
Viral load at 12 weeks
Parameter
Viral load (baseline)
<50 copies/mL
>50 copies/mL
CD4 cells (baseline)
mean cells/mm3
(SD)
Unscheduled drug level
sufficient
low
Scheduled trough level
sufficient
low
4 h PK-profile
sufficient
low
Adherence: self efficacy
sufficient
low
Adherence: med. intake
always
not always
General HRQL
mean
(SD)
Mental HRQL
mean
(SD)
HIV-assoc. HRQL
mean
(SD)
School education
high
low
Professional education
high
low
<50 copies/mL
>50 copies/mL
85.4%
43.2%
14.6%
56.8%
333
(179)
Viral load at 24 weeks
P value
<50 copies/mL
>50 copies/mL
P value
<0.01
80.0%
50.8%
20.0%
49.2%
<0.01
343
(218)
0.75
342
(191)
331
(190)
0.72
73.8%
47.8%
26.2%
52.2%
0.02
71.2%
50.0%
28.8%
50.0%
0.09
72.9%
55.9%
27.1%
44.1%
0.06
69.9%
70.4%
30.1%
29.6%
1.0
76.4%
59.7%
23.6%
40.3%
0.02
75.4%
60.9%
24.6%
39.1%
0.05
70.3%
50.0%
29.7%
50.0%
0.37
71.5%
50.0%
28.5%
50.0%
0.36
71.9%
50.0%
28.1%
50.0%
0.05
73.7%
47.6%
26.3%
52.4%
0.02
66.4
(21)
64.5
(20)
0.57
66.8
(20)
64.6
(23)
0.56
67.2
(21)
64.8
(21)
0.50
68.0
(21)
64.3
(20)
0.30
70.4
(22)
69.7
(22)
0.84
70.4
(23)
69.9
(26)
0.89
69.5%
63.2%
30.5%
36.8%
0.61
70.1%
68.8%
29.9%
31.1%
1.0
69.3%
68.9%
30.7%
31.1%
1.0
74.1%
64.9%
25.9%
35.1%
0.22
SD: standard deviation.
Bold: significant association.
determined parameters were classified in a manner that allowed a
normalized comparison. We found sufficient drug levels and
good adherence indices in the vast majority of the patients,
matching the results of several previous investigations.34 Mean
quality of life indices ranked as good for this population. The
majority of cases belonged to groups with lower social determinants.
We identified baseline viral load, unscheduled drug level, a 4 h
PK-profile and the question on correctness of medication intake
in the near past as the most important predictors of virological
success. Clinical status, CD4 cell count, indicators of social status and HRQL were not associated with virological outcome.
Baseline viral load is well-known to be of significant impact
on clinical and virological progression,35 and, not surprisingly,
our data are in concordance with this fact. The more important
finding is the significant association of different strategies of PK
sampling and adherence estimation with suppression of viral
replication at 12 weeks, and, to a lesser extent, at 24 weeks.
This association is independent of the other assessed parameters
that did not influence HAART efficacy in the described context.
Thus, our data provide an insight into the strength of association
of different clinical aspects with short-term virological outcome.
The results are comparable to those from a study that assessed
similar factors in a HAART-naive population treated with
indinavir.14 For the clinician, the parameters unscheduled PK
analysis and the scheduled 4 h PK-profile as well as the adherence estimation by correctness of drug intake were valuable
predictors of success. The other means of monitoring were
150
JAC
Predictors of successful HAART
antiviral
Table 4. Treatment outcome II, multivariate analysis: impact of
different parameters on virological success at 12 weeks
Table 5. Treatment outcome III, multivariate analysis: impact of
different parameters on virological success at 24 weeks
Parameter
Parameter
Model 1
Model 2
Viral load at baseline (for each 1000 copies/mL)
OR [95% CI] 0.95 [0.9–0.99] 0.94 [0.9–0.98]
P value
0.04
0.01
CD4 cells at baseline (for each 100 cells/mm3)
OR [95% CI] 0.93 [0.8–1.1] 0.90 [0.8–1.1]
P value
0.45
0.24
Body mass index (for every kg/m2)
OR [95% CI] 0.99 [0.9–1.1] 0.99 [0.9–1.1]
P value
0.73
0.88
Dose modification (yes versus no)
OR [95% CI] 1.30 [0.5–3.3] 1.79 [0.7–4.7]
P value
0.57
0.24
Adherence: intake of medication (yes versus no)
OR [95% CI] 2.56 [0.9–7.0] 3.63 [1.4–9.4]
P value
0.07
0.01
Unscheduled drug level (sufficient versus low)
OR [95% CI] 2.72 [1.1–7.0]
P value
0.04
Scheduled trough level (sufficient versus low)
OR [95% CI]
3.06 [1.3–7.3]
P value
0.01
4 h PK-profile (sufficient versus low)
OR [95% CI]
P value
Model 3
0.94 [0.9–0.98]
0.01
0.92 [0.8–1.1]
0.32
1.0 [0.9–1.1]
0.96
1.56 [0.6–4.0]
0.35
2.73 [1.1–7.0]
0.04
2.11 [1.1–4.2]
0.04
Model 1
Model 2
Viral load at baseline (for each 1000 copies/mL)
OR [95% CI] 0.97 [0.9–1.0] 0.97 [0.9–1.0]
P value
0.07
0.07
CD4 cells at baseline (for each 100 cells/mm3)
OR [95% CI] 0.99 [0.8–1.2] 1.01 [0.8–1.2]
P value
0.92
0.88
Body mass index (for every kg/m2)
OR [95% CI] 1.05 [0.95–1.2] 1.05 [0.96–1.2]
P value
0.36
0.30
Dose modification (yes versus no)
OR [95% CI] 1.32 [0.5–3.3] 1.32 [0.5–3.4]
P value
0.56
0.56
Adherence: intake of medication (yes versus no)
OR [95% CI] 2.96 [1.1–8.1] 3.35 [1.3–8.6]
P value
0.04
0.01
Unscheduled drug level (sufficient versus low)
OR [95% CI] 2.47 [0.8–7.4]
P value
0.11
Scheduled trough level (sufficient versus low)
OR [95% CI]
1.20 [0.4–3.2]
P value
0.72
4 h PK-profile (sufficient versus low)
OR [95% CI]
P value
Model 3
0.97 [0.9–1.0]
0.09
1.03 [0.9–1.2]
0.75
1.07 [0.97–1.2]
0.19
1.42 [0.6–3.5]
0.45
3.05 [1.2–8.0]
0.02
1.93 [0.9–4.0]
0.08
OR: odds ratio for being below the detection limit of 50 c/mL.
95% CI: 95% confidence interval.
Bold: significant association.
OR: odds ratio for being below the detection limit of 50 c/mL.
95% CI: 95% confidence interval.
Bold: significant association.
inferior in predicting outcome. Thus, we present simple diagnostic tools that may improve HAART efficacy by identification of patients with a risk of short-term failure. Finally, HRQL
and social determinants had no detectable influence on outcome.
In conclusion, we identified easy-to-assess variables with significant impact on HAART efficacy in a prospective cohort representing routine clinical practice. Furthermore, we were able to
identify parameters with little effect on treatment results. This
may help to improve management of patients treated with
HAART.
The debate about practical tools for assessing drug levels is
not finished.4,7,36 We showed that TDM is capable of predicting
virological outcome, a fact that has previously been demonstrated.8,9,37 The new aspect highlighted by our results is the
independent association of PK sampling with efficacy after
adjustment for a number of important confounders such as baseline virological and immunological values, adherence, HRQL and
social indices. This is valid especially for the 4 h PK-profile,
which produced the clearest association with virological success.
The information gained by a PK-profile may thus provide the best
estimation of future efficacy of HAART. However, this measurement is only achievable with substantial effort by both the
clinician as well as the patient, who has to present to the treatment centre at appointment fasting and must remain for the whole
morning. Thus, the PK-profile may not be the most feasible tool
to measure bioavailability of drugs in daily practice.
It is notable that an unscheduled drug level was almost as
clearly associated with virological outcome as the complete 4 h
profile in a scheduled PK assessment. Other retrospective trials
showed similar results without consideration of different confounders.26,28,38 Untimed drug levels were sensitive identifiers
of individuals who had been adherent to 60% or less of medication doses in a cross-sectional study.29 In another study, a low
drug level on random testing was significantly associated with
poor clinical response to therapy.28 Moreover, in our study
untimed drug levels were more valuable in predicting outcome
than trough levels at scheduled appointments. As the unscheduled
measurement is an easily applicable and cheap method, it may be
of considerable value in detecting patients with a high risk of
short-term virological failure. Reasons for this association may
include reduced adherence, poor absorption of drugs or individual
metabolic disposition. PK results were not associated with
adherence in our dataset, a finding previously described for
indinavir-based therapy.39 In another study, unscheduled drug
levels were associated only with very low adherence.29 Further
trials demonstrated a moderate association of PI levels and adherence.18,40,41 Thus, a complete understanding of the connection of
these parameters remains elusive. As we could not find a substantial association of adherence with PK levels or PK data with
baseline viral load, the assessed parameters are more likely to
serve as risk factors for future treatment failure than they are able
to explain treatment efficacy at a certain time point.
151
Oette et al.
The question about correctness of medication intake in the last
14 days turned out to be of substantial benefit for treatment
efficacy at 12 and 24 weeks. This is in line with published
data.11,15,16,33,42 Adherence measured by self-administered questionnaires has been shown to correlate with methods such as
direct and indirect pill count12 and was an independent risk factor
for viral rebound in other studies.35,43 Owing to a broader evaluation of confounding variables, we were able to show that adherence plays a more important role for short-term treatment
outcome than do the above mentioned parameters. Moreover,
self-reported accuracy of drug intake was a better predictor of
success than the personal estimation of future adherence.
Several biases of our results must be considered. First, the
studied patient group is diverse in respect to the large number of
different HAART combinations. Thus, the information on specific characteristics of each combination used in antiretroviral
therapy and especially class-specific effects was lost due to
this population approach. As such, the results of this investigation
cannot be compared with other trials that studied only single
defined combinations of drugs. However, our study was designed
to represent contemporary clinical reality. Consequently, while
the administered regimens show a large variety of combinations
and the conclusions of the whole study may not be valid for each
individual combination, the facts may be regarded as representative of antiretroviral therapy in daily practice. Second, a substantial subgroup had non-suppressive HAART at baseline. This
mixture of cases is unique in our dataset and again underscores
the validity of the results for clinical routine, where HAART
combinations without suppressed viral replication are common.
Third, our study aimed at modelling easy-to-apply parameters as
tools for the clinician. To this end we used a normalized classification scheme for interindividual comparisons. By ignoring
drug-specific effects in defined patient subgroups, we gained an
estimation of the effects of different monitoring methods on a
population basis. This led to the characterization of the value of
different diagnostic applications rather than the analysis of singleagent-characteristics. Moreover, the administered regimens present a cross-sectional description of currently applied HAART
concepts. Fourth, the lack of an association between PK results
and baseline viral load shows that the applied parameters may
only serve as risk factors for future development rather than
monitoring tools for a current clinical situation. Accordingly, a
number of individuals with insufficient drug levels at baseline
show virological control at study endpoints. Thus, patients
demonstrating unfavourable results should undergo intensive
counselling in order to improve the likelihood of effective control
of viral replication. Fifth, classification PK levels in the 4 h
profile were quite generous, as a relatively low profile could
still be regarded as sufficient. We attempted to define PK levels
that would be applicable to daily practice. Sixth, our findings are
valid only for short-term efficacy of antiretroviral therapy. More
information over longer follow-up periods is necessary to gain a
better understanding of the relevance of the studied parameters.
Finally, by only recording increases in dosing, we did not take
into account the important application of TDM to identify cases
with toxic drug levels. While this parameter has been shown to be
of significant value for improving administration of HAART,6 we
may have underestimated its effect in our investigation. The low
number of treatment discontinuations or switches indicates, however, that this factor may have a low impact on the results. Taken
together, the discussed limitations imply a need for cautious
interpretation of the results, but do not outweigh the validity
of the data within daily practice. Thus, our results are of
value for clinical routine.
In conclusion, we identified unscheduled PK measurement and
self-reported adherence as independently associated with shortterm virological success of HAART in routine clinical practice.
By considering these aspects, the clinician is able to apply
diagnostic tools of identifying individuals with a possible risk
of treatment failure with reasonable effort and low expense.
Interventions for better adherence and diagnostics for reasons
of suboptimal bioavailability of drugs may follow. Besides
baseline viral load, the variety of other studied factors could
not be associated with effective HAART.
Acknowledgements
Dr M. O. and Dr A. K. contributed equally to this study. We wish to
thank Jutta Rüdiger and Lasse Kajala for help in specimen handling.
The study was presented in part at the Europeans AIDS Conference 2005 in Dublin44 and the 10th German and 16th Austrian
AIDS Congress 2005 in Vienna.45
Transparency declarations
The laboratory costs were financed, in part, by the companies
Abbott, Boehringer-Ingelheim, GlaxoSmithKline and HoffmannLa Roche.
References
1. Egger M, May M, Chene G et al. Prognosis of HIV-1-infected
patients starting highly active antiretroviral therapy: a collaborative
analysis of prospective studies. Lancet 2002; 360: 119–29.
2. Bartlett JA, DeMasi R, Quinn J et al. Overview of the effectiveness
of triple combination therapy in antiretroviral-naive HIV-1 infected adults.
AIDS 2001; 15: 1369–77.
3. Boffito M, Acosta E, Burger D et al. Current status and future prospects of therapeutic drug monitoring and applied clinical pharmacology
in antiretroviral therapy. Antivir Ther 2005; 10: 375–92.
4. Clevenbergh P, Mouly S, Sellier P et al. Improving HIV infection
management using antiretroviral plasma drug levels monitoring: a clinician’s point of view. Curr HIV Res 2004; 2: 309–21.
5. Piscitelli SC, Gallicano KD. Interactions among drugs for HIV
and opportunistic infections. N Engl J Med 2001; 344: 984–96.
6. Rendón AL, Núnez M, Jiménez-Nácher I et al. Clinical benefit of
interventions driven by therapeutic drug monitoring. HIV Med 2005;
6: 360–5.
7. Aarnoutse RE, Schapiro JM, Boucher CA et al. Therapeutic drug
monitoring: an aid to optimising response to antiretroviral drugs? Drugs
2003; 63: 741–53.
8. Burger DM, Hoetelmans RM, Hugen PW et al. Low plasma concentrations of indinavir are related to virological treatment failure in HIV-1infected patients on indinavir-containing triple therapy. Antivir Ther 1998;
3: 215–20.
9. Powderly WG, Saag MS, Chapman S et al. Predictors of
optimal virological response to potent antiretroviral therapy. AIDS 1999;
13: 1873–80.
10. Touw DJ, Neef C, Thomson AH et al. Cost-effectiveness of
therapeutic drug monitoring: a systematic review. Ther Drug Monit
2005; 27: 10–7.
11. Mannheimer S, Friedland G, Matts J et al. The consistency of adherence to antiretroviral therapy predicts biologic outcomes for human
152
Predictors of successful HAART
immunodeficiency virus-infected persons in clinical trials. Clin Infect Dis
2002; 34: 1115–21.
12. Bangsberg DR, Hecht FM, Charlebois ED et al. Adherence to
protease inhibitors, HIV-1 viral load, and development of drug resistance
in an indigent population. AIDS 2000; 14: 357–66.
13. Duong M, Piroth L, Peytavin G et al. Value of patient self-report and
plasma human immunodeficiency virus protease inhibitor level as
markers of adherence to antiretroviral therapy: relationship to virologic
response. Clin Infect Dis 2001; 33: 386–92.
14. Duval X, Mentre F, Lamotte C et al. Indinavir plasma concentration
and adherence score are codeterminant of early virologic response in
HIV-infected patients of the APROCO Cohort. Ther Drug Monit 2005;
27: 63–70.
15. Paterson DL, Swindells S, Mohr J et al. Adherence to protease
inhibitor therapy and outcomes in patients with HIV infection. Ann Intern
Med 2000; 133: 21–30.
16. Raboud JM, Harris M, Rae S et al. Impact of adherence on duration
of virological suppression among patients receiving combination antiretroviral therapy. HIV Med 2002; 3: 118–24.
17. Bangsberg DR, Perry S, Charlebois ED et al. Non-adherence
to highly active antiretroviral therapy predicts progression to AIDS.
AIDS 2001; 15: 1181–3.
18. Murri R, Ammassari A, Gallicano K et al. Patient-reported nonadherence to HAART is related to protease inhibitor levels. J Acquir Immune
Defic Syndr 2000; 24: 123–8.
19. Ammassari A, Trotta MP, Murri R et al. Correlates and predictors of
adherence to highly active antiretroviral therapy: overview of published
literature. J Acquir Immune Defic Syndr 2002; 31 Suppl 3: S123–7.
20. Goujard C, Bernard N, Sohier N et al. Impact of a patient education
program on adherence to HIV medication: a randomized clinical trial.
J Acquir Immune Defic Syndr 2003; 34: 191–4.
21. Le Moing V, Chene G, Carrieri MP et al. Clinical, biologic, and
behavioral predictors of early immunologic and virologic response in
HIV-infected patients initiating protease inhibitors. J Acquir Immune
Defic Syndr 2001; 27: 372–6.
22. Dray-Spira R, Lert F. Social health inequalities during the course
of chronic HIV disease in the era of highly active antiretroviral therapy.
AIDS 2003; 17: 283–90.
23. Gordillo V, del Amo J, Soriano V et al. Sociodemographic and
psychological variables influencing adherence to antiretroviral therapy.
AIDS 1999; 13: 1763–9.
24. McFarland W, Chen S, Hsu L et al. Low socioeconomic status is
associated with a higher rate of death in the era of highly active
antiretroviral therapy, San Francisco. J Acquir Immune Defic Syndr
2003; 33: 96–103.
25. Wood E, Montaner JS, Chan K et al. Socioeconomic status,
access to triple therapy, and survival from HIV-disease since 1996.
AIDS 2002; 16: 2065–72.
26. Antinori A, Cozzi-Lepri A, Ammassari A et al. Relative prognostic
value of self-reported adherence and plasma NNRTI/PI concentrations
to predict virological rebound in patients initially responding to HAART.
Antivir Ther 2004; 9: 291–6.
27. Reynolds NR, Testa MA, Marc LG et al. Factors influencing
medication adherence beliefs and self-efficacy in persons naive to antiretroviral therapy: a multicenter, cross-sectional study. AIDS Behav 2004;
8: 141–50.
28. Alexander CS, Asselin JJ, Ting LS et al. Antiretroviral concentrations in untimed plasma samples predict therapy outcome in a population
with advanced disease. J Infect Dis 2003; 188: 541–8.
JAC
antiviral
29. Liechty CA, Alexander CS, Harrigan PR et al. Are untimed antiretroviral drug levels useful predictors of adherence behavior? AIDS 2004;
18: 127–9.
30. Kuschak D, Mauss S, Schmutz G et al. Simultaneous determination
of the new HIV protease inhibitor lopinavir (ABT 378) and of indinavir (1),
amprenavir, saquinavir, ritonavir (ABT 538)(2) and nelfinavir (3) in human
plasma by gradient HPLC. Clin Lab 2001; 47: 471–7.
31. Wachtel T, Piette J, Mor V et al. Quality of life in persons with human
immunodeficiency virus infection: measurement by the Medical Outcomes
Study instrument. Ann Intern Med 1992; 116: 129–37.
32. Wu AW, Hays RD, Kelly S et al. Applications of the Medical
Outcomes Study health-related quality of life measures in HIV/AIDS.
Qual Life Res 1997; 6: 531–54.
33. Nieuwkerk PT, Sprangers MA, Burger DM et al. Limited patient
adherence to highly active antiretroviral therapy for HIV-1 infection in
an observational cohort study. Arch Intern Med 2001; 161: 1962–8.
34. Lucas GM, Wu AW, Cheever LW. Adherence to antiretroviral therapy: an update of current concepts. Curr HIV/AIDS Rep 2004; 1: 172–80.
35. Le Moing V, Chene G, Carrieri MP et al. Predictors of virological
rebound in HIV-1-infected patients initiating a protease inhibitorcontaining regimen. AIDS 2002; 16: 21–9.
36. Back DJ, Khoo SH, Gibbons SE et al. Therapeutic drug monitoring
of antiretrovirals in human immunodeficiency virus infection. Ther Drug
Monit 2000; 22: 122–6.
37. Castagna A, Gianotti N, Galli L et al. The NIQ of lopinavir is predictive of a 48-week virological response in highly treatment-experienced
HIV-1-infected subjects treated with a lopinavir/ritonavir-containing
regimen. Antivir Ther 2004; 9: 537–43.
38. Yasuda JM, Miller C, Currier JS et al. The correlation between
plasma concentrations of protease inhibitors, medication adherence
and virological outcome in HIV-infected patients. Antivir Ther 2004; 9:
753–61.
39. Alcoba M, Cuevas MJ, Perez-Simon MR et al. Assessment of
adherence to triple antiretroviral treatment including indinavir: role of
the determination of plasma levels of indinavir. J Acquir Immune Defic
Syndr 2003; 33: 253–8.
40. Duran S, Peytavin G, Carrieri P et al. The detection of nonadherence by self-administered questionnaires can be optimized by
protease inhibitor plasma concentration determination. AIDS 2003; 17:
1096–9.
41. Hugen PW, Burger DM, Aarnoutse RE et al. Therapeutic drug
monitoring of HIV-protease inhibitors to assess noncompliance. Ther
Drug Monit 2002; 24: 579–87.
42. Haubrich RH, Little SJ, Currier JS et al. The value of patient-reported
adherence to antiretroviral therapy in predicting virologic and immunologic
response. California Collaborative Treatment Group. AIDS 1999; 13:
1099–107.
43. Carrieri P, Cailleton V, Le Moing V et al. The dynamic of adherence
to highly active antiretroviral therapy: results from the French National
APROCO cohort. J Acquir Immune Defic Syndr 2001; 28: 232–9.
44. Kroidl A, Menge M, Oette M et al. Evaluation of different strategies
of therapeutic drug monitoring (TDM) for response to antiretroviral treatment and virologic outcome in HIV. 10th European AIDS Conference,
Dublin, 2005. Abstract PE4.2/9.
45. Kroidl A, Stabbert A, Oette M et al. Therapeutic drug monitoring in
unscheduled plasma samples predict therapy outcome in HIV-positive
patients. Eur J Med Res 2005; 10 Suppl 2: 27, Abstract V9.
153