J Antimicrob Chemother 2014; 69: 2202 – 2209 doi:10.1093/jac/dku112 Advance Access publication 15 April 2014
Temporal trends in the discontinuation of first-line antiretroviral therapy Alejandro Gonzalez-Serna1*, Keith Chan1, Benita Yip1, William Chau1, Rachel McGovern1, Hasina Samji1, Viviane Dias Lima1,2, Robert S. Hogg1,3 and Richard Harrigan1,2 1
BC Centre for Excellence on HIV/AIDS, Vancouver V6Z1Y6, BC, Canada; 2Division of AIDS, Department of Medicine, University of British Columbia, Vancouver V6T1Z4, BC, Canada; 3Faculty of Health Sciences, Simon Fraser University, Burnaby V5A1S6 BC, Canada *Corresponding author. Tel: +1-604-806-8645; Fax: +1-604-806-9463; E-mail: [email protected]
Objectives: The aim of this study was to describe the rates and predictors of discontinuing first-line antiretroviral therapy in the different eras of treatment over a nearly 20 year period initiated in British Columbia between 1992 and 2010. Methods: All naive adults who started antiretroviral therapy (first-line antiretroviral therapy) at any hospital or clinic in British Columbia (Canada) in 1992 – 2010 were included in this population-based retrospective cohort study. We were primarily interested in whether the era of treatment (1992 – 95, 1996 – 2000, 2001 – 05 and 2006 – 10) was associated with discontinuation (stopping or switching of initial treatment) within 3 years of starting therapy. Weibull survival analysis was used to model the era of treatment and its association with time to discontinuation. Results: The study included 7901 patients. Overall, the probability of discontinuing at 12, 24 and 36 months of treatment was 52%, 68% and 76%, respectively. In the adjusted model, variables associated with discontinuing were earlier treatment era, younger age, low adherence and lower baseline CD4 count. Regarding the 2006 –10 period, the probability of discontinuing at 12, 24 and 36 months was 36%, 47% and 53%, respectively. In the adjusted model, the variables associated with discontinuation were younger age, female gender, AIDS-defining illnesses at baseline, low adherence and a protease inhibitor (PI)-based regimen. Conclusions: Discontinuation rates of first-line therapy have decreased over time, but are still quite high even for the latest drug combinations. In the most recent era, younger women on a PI regimen and those not achieving optimal adherence had the highest risk of discontinuing first-line antiretroviral therapy. Keywords: HIV-1, ART, treatment, risk
Introduction HIV is a chronic disease that requires life-long therapy and the prognosis of HIV-infected patients starting highly active antiretroviral therapy (HAART) greatly depends on treatment optimization.1 – 3 To better understand the determinants of treatment changes over time it is important to optimize the initial HAART regimen and evaluate trends in the incidence of discontinuation (stopping or switching of initial treatment). Several studies analysing rates of discontinuation reported that a high percentage of patients (40% – 60%) underwent discontinuation of their initial regimen within the first year of treatment.4 – 7 Over recent years HAART strategies have evolved dramatically, offering high efficacy, less toxicity and reduced pill burden and dosing frequency.8 Accordingly, some studies found improvements in virological and immunological outcomes and pointed to time trends towards lower rates of discontinuation.9 – 11 A recent study reported that the risk of multiple treatment modifications
(three or more changes in any of the antiretroviral components in the regimen) had decreased over time.12 However, no studies have found evidence of a significant difference in the rate of discontinuation of the first antiretroviral therapy (ART) (discontinuation for any reason) according to calendar period of starting therapy.12 – 15 It is possible that time period length and/or number of patients included in those studies were not big enough to find a significant difference in the rate of discontinuation, so studies analysing these different eras of treatment are still lacking. Therefore, the aim of this study was to describe the rates and predictors of discontinuing first-line ART in the different eras of treatment over a nearly 20 year period initiated in British Columbia between 1992 and 2010.
Methods Samples and records for all naive HIV-positive patients aged ≥19 years starting ART between September 1992 and September 2010 at any
# The Author 2014. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: [email protected]
2202
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
Received 12 December 2013; returned 27 January 2014; revised 7 March 2014; accepted 17 March 2014
Temporal trends in the discontinuation of first-line antiretroviral therapy
Statistical analysis Comparisons across eras were carried out using the x2 test or Fisher’s exact test for categorical variables and the Wilcoxon rank-sum test for continuous variables. Kaplan– Meier methods were used to graphically display the probability of discontinuing regimen after starting ART. Weibull survival analysis was used to model the era of treatment and its association with time to discontinuation. Variables that were significant at P, 0.05 in the unadjusted analyses were candidates for inclusion in the multivariate model. A confounder model was fitted for era using the methods of Maldonado and Greenland.16 All analyses were performed using SAS software version 9.3 (SAS Institute, Cary, NC, USA).
Ethics statement This study was approved by the Committee on Human Research and the University of British Columbia/ Providence Health Care Research Ethics Board.
Results A total of 7901 patients were included in the study. The median age was 39 years (IQR 33 – 46 years) and 83% were male. The
median CD4 cell count at ART initiation was 220 cells/mm3 (IQR 110 – 350 cells/mm3 ) and the median HIV RNA was 4.9 log10 copies/mL (IQR 4.4 –5.0 log10 copies/mL). A total of 1855 patients (23.5%) started treatment in pre-HAART era, 2406 patients (30.5%) in 1996 – 2000, 1533 patients (19.4%) in 2001 – 05 and 2107 patients (26.7%) in 2006 – 10. Patients’ characteristics at baseline stratified by era of treatment are listed in Table 1, where all characteristics were statistically different (P, 0.001). Due to the missing values for ethnicity, risk status and HCV, these variables were not included in the statistical model and are only explored in Table 1. All analyses involving viral load characteristics were limited to the HAART eras (n¼6046), because in the pre-HAART era (n ¼ 1855) viral loads were not available in most patients, and comparisons of discontinuation based on virological suppression were limited to the mid-HAART and late HAART eras (n¼3640). Overall, 5490 (69.5%) patients discontinued therapy during follow-up, of whom 3659 (66.6%) were switches and 1831 (33.4%) were stops. The probability of discontinuation at 12, 24 and 36 months was 0.52 (95% CI 0.51 – 0.53), 0.67 (95% CI 0.66 – 0.69) and 0.76 (95% CI 0.75 – 0.77), respectively (Figure 1a). At 36 months, the probability of switching was higher than that of stopping (0.64 versus 0.23). Based on eras, the probability of discontinuation at 36 months was progressively lower the more recent the era (0.94 in 1992 –96, 0.78 in 1996 – 2000, 0.77 in 2001 – 05 and 0.53 in 2006 – 10, P, 0.001) (Figure 1b). Discontinuation rates in relation to the ART regimen showed that two NRTIs + NNRTI had a lower risk of discontinuation than two NRTIs + boosted or unboosted PI, or other combinations (P, 0.001) (Figure 1c). In the mid-HAART and late HAART eras, 2119 patients (58.2%) discontinued therapy, of whom 1558 (73.5%) were switches and 561 (26.5%) were stops. Of these mid-HAART and late HAART era discontinuations, 1377 (65.0%) were not virologically suppressed (virological failure switch) at the time of discontinuation and 742 (35.0%) were virologically suppressed (probably a tolerability issue). The probability of discontinuation due to virological failure was greater than discontinuation due to tolerability issues through 3 years of follow-up (0.33 versus 0.13 at 12 months and 0.43 versus 0.37 at 36 months) (Figure 1d). During the HAART eras, 3903 (64.6%) patients discontinued therapy, of whom 2628 (67.3%) were switches and 1275 (32.7%) were stops. In unadjusted analysis, most of the variables included were associated with switching or stopping (Table 2). However, in the adjusted model variables associated with switching or stopping were earlier treatment era [mid-HAART (adjusted hazard ratio 1.01, 95% CI 0.94 – 1.09) and late HAART (adjusted hazard ratio 0.61, 95% CI 0.56 – 0.66) compared with the early HAART], female gender (adjusted hazard ratio 1.23, 95% CI 1.14 – 1.33), younger age per decade (adjusted hazard ratio 0.89, 95% CI 0.86 – 0.93), high ART adherence (adjusted hazard ratio 0.39, 95% CI 0.37 – 0.42) and lower baseline CD4 count [200 – 349 CD4 cells/mm 3 (adjusted hazard ratio 0.91, 95% CI 0.84 – 0.99), versus ,200 CD4 cells/mm3]. Regarding the most recent era, i.e. the 2006 – 10 period, the probabilities of discontinuing at 12, 24 and 36 months of treatment were 0.36 (95% CI 0.34 – 0.38), 0.47 (95% CI 0.45 – 0.49) and 0.53 (95% CI 0.51 – 0.55), respectively (Figure 2a). Discontinuation rates in relation to the ART regimen showed
2203
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
hospital or clinic in British Columbia, Canada, were centralized and analysed at the British Columbia Centre for Excellence in HIV/AIDS. Demographic, clinical and laboratory data and information on therapy were collected for all participants. Study was restricted to patients with complete information on age, gender, baseline CD4 and viral load in the HAART era. The primary factor of interest in this population-based retrospective cohort study was the era of starting ART, grouped as: pre-HAART (September 1992 – May 1996); early HAART (June 1996 to December 2000); mid-HAART (January 2001 to December 2005) and late HAART (January 2006 to December 2010). Other covariates of interest included age, gender, detection of hepatitis C virus (HCV) antibodies on at least one occasion during follow-up, baseline CD4 cell count and plasma HIV RNA (copies/mL), risk group, ethnicity, adherence, AIDS status and type of first ART regimen started [one or two nucleoside reverse transcriptase inhibitors (NRTIs), two NRTIs +one non-nucleoside reverse transcriptase inhibitor (NNRTI), two NRTIs +unboosted or boosted protease inhibitor (PI) or another regimen with three or more antiretrovirals]. Measurement of exposure to ART was obtained by prescription refills. Medication adherence, estimated based on antiretroviral prescription refill data during the first year after starting therapy, was calculated by dividing the number of days of medication dispensed by the number of days of follow-up, expressed as a percentage. We investigated the time from ART initiation to discontinuation of the first ART within 3 years of starting therapy. Discontinuation was categorized as either a stop (defined as stopping all drugs in a regimen for .6 months) or a switch (changing any of the antiretroviral components, not the formulation, in the regimen). If a patient was virologically suppressed (defined as the two most recent plasma HIV RNA measurements ,50 copies/mL within 6 months) prior to discontinuation, it was considered a discontinuation ‘probably’ due to a tolerability issue, while those not suppressed were considered as having discontinuation due to virological failure. For the latter purpose only patients from the mid-HAART and late HAART eras were analysed because virological suppression data (plasma HIV RNA ,50 copies/mL) were not available in the pre-HAART and early HAART eras. The main objective was to compare the rate of discontinuation according to era of ART initiation. Follow-up of the patients who did not stop or switch was censored at the death date, date of moving out of British Columbia, date of entering a placebo trial, date of starting supervised therapy interruption, 30 September 2011 (study end date) or 3 years from ART start date, whichever came first.
JAC
Gonzalez-Serna et al.
Table 1. Baseline characteristics by era (n¼7901) Characteristic
1992 –96 (n¼1855)
1996 –2000 (n ¼2406)
2001 –05 (n¼1533)
2006– 10 (n¼2107)
1668 (89.9) 36 (31 –43)
1994 (82.9) 37 (32 –44)
1196 (78) 41 (35 –48)
1700 (80.7) 42 (35 – 49)
ART discontinuation switch virological failurea switch tolerabilityb switch stop virological failurea stop tolerabilityb stop
1587 (85.6) 1031 (55.6) NA NA 556 (30) NA NA
1784 (74.1) 1070 (44.5) NA NA 714 (29.7) NA NA
1113 (82.6) 802 (52.3) 455 (56.7) 347 (43.3) 311 (20.3) 234 (75.2) 77 (24.8)
1006 (47.7) 756 (35.9) 485 (64.1) 271 (35.9) 250 (11.9) 203 (81.2) 47 (18.8)
396 (21.3) 321 (34.1)
326 (13.5) 1128 (53.4)
278 (18.1) 651 (47.8)
261 (12.4) 742 (39.6)
Ethnicity Caucasian aboriginal other
1102 (82.7) 171 (12.8) 59 (4.4)
1095 (68.6) 388 (24.3) 114 (7.1)
545 (62.5) 201 (23.1) 126 (14.4)
453 (54.5) 236 (28.4) 142 (17.1)
Risk group men who have sex with men heterosexual sex injection drug use
1083 (73.1) 308 (20.8) 438 (28.7)
786 (44.7) 545 (31) 991 (52.8)
350 (34) 376 (36.6) 630 (50.2)
346 (34) 463 (36.1) 762 (45.3)
ART regimen ≤2 NRTIs 2 NRTIs+unboosted PI 2 NRTIs+boosted PI 2 NRTIs+NNRTI other ART combinations
1855 (100) 0 (0) 0 (0) (0) (0)
569 (23.6) 1055 (43.8) 80 (3.3) 555 (23.1) 147 (6.1)
17 (1.1) 104 (6.8) 687 (44.8) 619 (40.4) 106 (6.9)
2 (0.1) 37 (1.8) 1110 (52.7) 911 (43.2) 47 (2.2)
Baseline CD4 (cells/mm3) ,200 200–349 ≥350
775 (41.8) 636 (34.3) 444 (23.9)
846 (35.2) 648 (26.9) 912 (37.9)
959 (62.6) 397 (25.9) 177 (11.5)
855 (40.6) 782 (37.1) 470 (22.3)
Baseline CD4 (cells/mm3), median (IQR)
230 (100– 340)
280 (130 –420)
160 (70 –240)
230 (130 –330)
174 (11.4) 538 (35.1) 821 (53.6)
336 (15.9) 939 (44.6) 832 (39.5)
1187 (76.9) 425 (27.5)
1776 (83.5) 319 (15.0)
909 (59.3)
1405 (66.7)
ADIs at baseline Hepatitis Cc
Baseline viral load (copies/mL) ,10 000 10000– 99999 ≥100000
NA NA NA
270 (11.2) 1005 (41.8) 1131 (47)
Viral load suppressiond during 3 year follow-up within 6 months prior to discontinuation
NA NA
NA NA
High ART adherencee
612 (33.0)
1155 (48.0)
NA, not available. Data are number (%) unless otherwise indicated. All characteristics were significantly different by era (P,0.001). a Discontinuation occurred with detectable viral load suggesting discontinuation for virological failure. b Discontinuation occurred with undetectable viral load suggesting discontinuation for a probable tolerability issue. c HCV antibody positive. d Two consecutive viral loads ,500 copies/mL. e ART adherence in the first year of therapy ≥95%.
that two NRTIs + NNRTI had a lower risk of discontinuation than two NRTIs + boosted or unboosted PI, or other combinations (P,0.001) (Figure 2b). 2204
Again, in the unadjusted model, most of the variables included were associated with switching or stopping. However, in the adjusted model the variables associated with discontinuation
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
Male gender Age (years), median (IQR)
JAC
Temporal trends in the discontinuation of first-line antiretroviral therapy
(a)
(b) 100
Discontinuation Switching Stopping
80
Probability of discontinuation (%)
Probability of discontinuation (%)
100
60 40 20 0
80 60 40
92–96 96–00 01–05 06–10 Log-rank P < 0.001
20 0
6
12
24
18
30
36
0
6
Follow-up time (months) Events
Number at risk 7901
4876
3542
2672
2059
1581
(d) 100
80
60
NRTI × 2, unboosted PI NRTI × 2, boosted PI NRTI × 2, NNRTI Other combination
20
Log-rank P < 0.001 0
Probability of discontinuation (%)
Probability of discontinuation (%)
1855 2406 1533 2107
791 5490
(c) 100
40
12
18
30
36
109 518 364 590
19 267 215 290
24
Follow-up time (months) 841 1493 992 1550
460 1055 740 1287
Number at risk 276 810 582 1004
175 639 463 782
Events 1587 1784 1113 1006
Discontinuation Tolerability discontinuation Virological failure discontinuation
80
60
40
20
0 0
6
12
18
24
30
36
0
6
Follow-up time (months) 1196 1877 2085 2743
742 1349 1478 1307
Number at risk 287 391 525 609 807 1049 808 972 1200 355 502 768
12
18
24
30
36
Follow-up time (months) 227 459 646 249
Events 120 926 250 1144 353 1128 68 2292
Number at risk 3640
2542
2027
1586
Events 1245
954
505 2119
Figure 1. Rates of discontinuation of first-line therapy initiated between 1992 and 2010. (a) Overall probability of switching and/or stopping up to 36 months. The continuous line represents the overall probability of discontinuation, the dashed line represents the probability of switching and the dotted line represents the probability of stopping of first-line therapy initiated between 1992 and 2010. (b) Rates of discontinuation of first-line therapy decreasing over time with advancing era. The probability of discontinuation of first-line therapy up to 36 months is indicated by the continuous black line for the period 1992 – 95, the dashed line for 1996 – 2000, the dotted line for 2001 – 05 and the continuous grey line for 2006 – 10. (c) Rates of discontinuation of first-line therapy depending on ART regimen. The probability of discontinuation of first-line therapy up to 36 months is indicated by the dotted line for the two NRTIs +NNRTI, the dashed line for the two NRTIs +boosted PI, the continuous black line for the two NRTIs +unboosted PI and the continuous grey line for the other combinations. (d) Discontinuation was mainly driven by virological failure (viral load not suppressed at time of discontinuation), mostly during the first year; however, tolerability issues (viral load suppressed at time of discontinuation) increased to a similar level over the third year. The continuous line represents the overall probability of discontinuation, the dashed line represents the probability of discontinuation due to tolerability issues and the dotted line represents the probability of discontinuation due to virological failure of first-line therapy initiated between 2001 and 2010.
were younger age (adjusted hazard ratio 0.93, 95% CI 0.87–0.99), female gender (adjusted hazard ratio 1.64, 95% CI 1.42 – 1.90), AIDS-defining illnesses (ADIs) at baseline (adjusted hazard ratio
1.26, 95% CI 1.05 – 1.50), high ART adherence (adjusted hazard ratio 0.49, 95% CI 0.43 – 0.56) and initial therapy with two NRTIs+ boosted PI (adjusted hazard ratio 1.26, 95% CI 1.10–1.43)
2205
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
0
Gonzalez-Serna et al.
Table 2. Predictors of treatment discontinuation in1996–2010 Unadjusted hazard ratio (95% CI)
P value
Adjusted hazard ratio (95% CI)
P value
Era early HAART (1996–2000) mid-HAART (2001– 05) late HAART (2006–10)
1.00 0.92 (0.86, 1.00) 0.53 (0.49, 0.57)
0.037 ,0.001
1.00 1.01 (0.94, 1.09) 0.61 (0.56, 0.66)
0.710 ,0.001
Age per decade
0.81 (0.79, 0.84)
,0.001
0.89 (0.86, 0.93)
,0.001
Gender male female
1.00 1.52 (1.41, 1.64)
ADIs at baseline
1.09 (0.99, 1.19)
0.067
HAART regimen 2 NRTIs+NNRTI 2 NRTIs+unboosted PI 2 NRTIs+boosted PI other combinations
1.00 1.77 (1.62, 1.93) 1.18 (1.09, 1.29) 2.11 (1.92, 2.32)
,0.001 ,0.001 ,0.001
Baseline CD4 (cells/mm3) ,200 200–349 ≥350
1.00 0.85 (0.79, 0.91) 1.10 (1.02, 1.19)
,0.001 0.011
Baseline plasma HIV RNA (copies/mL) ,10 000 10000– 99999 ≥100000
1.00 0.92 (0.83, 1.02) 0.99 (0.90, 1.10)
0.117 0.883
High ART adherencea
0.36 (0.34, 0.39)
,0.001
Variable
,0.001
,0.001 1.00 1.23 (1.14, 1.33)
0.019 0.287
0.39 (0.37, 0.42)
,0.001
All P values are derived from Weibull survival analysis. ART adherence in the first year of therapy ≥95%.
a
or other regimens (adjusted hazard ratio 2.58, 95% CI 1.96–3.39) compared with two NRTIs+ NNRTI (Table 3). In order to determine whether not excluding pregnant women had an effect on the results, we conducted a sensitivity analysis. We assumed that women whose ART regimen contained zidovudine and lasted ,1 year and who had a CD4 count .350 cells/mm3 at therapy initiation had been taking ART solely for the prevention of mother-to-child transmission of HIV. We found 133 women who met these criteria. However, when these women were excluded from the analysis, the results were unchanged.
Discussion In this study we show that rates of first-line therapy discontinuation over a nearly 20 year period decreased over time. However, the probability of stopping and/or switching to a second-line regimen was considerable even with access to the latest era of antiretrovirals and drug combinations. The decreases in rates of discontinuation found in this study challenge previous findings, where similar studies reported unchanging discontinuation rates according to study period.12 – 15,17 It is possible that the study duration, in some cases ,5 years, and/or the low number of patients used in these studies could
2206
explain why a decrease in discontinuation was not found previously. In fact, there is one study, analysing 1189 patients between 1997 and 2007, that found a decreasing risk of multiple treatment modifications (three or more) over time but concluded that first treatment discontinuation was comparable according to calendar period.12 In comparison, our study of 7901 patients over a nearly 20 year period showed a significant decrease in the rates of treatment discontinuation. Moreover, some studies found that discontinuation because of intolerance and/or toxicity was decreasing over time while discontinuation because of therapy simplification was increasing over time. This in part could explain the absence of an overall decreasing trend globally.12 – 15 With the increased availability of newer drugs, drug classes and multidrug-containing regimens, and those with better tolerability, dosing and resistance profiles, the individualization of treatment has led to improved virological and immunological outcomes for HIV-infected patients. This may be one of the primary factors associated with the decrease in discontinuation observed in our study, given the study’s duration and geographic region. All analyses were repeated by defining failure as .500 copies/mL for consistency between the different eras and the results were consistent (data not shown). Overall, the rates of discontinuation of first-line therapy found in our study are slightly higher than those in some previous studies.17 – 21 Injection drug use and
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
1.00 0.91 (0.84, 0.99) 1.04 (0.96, 1.13)
Temporal trends in the discontinuation of first-line antiretroviral therapy
Probability of discontinution (%)
(a)
100
Discontinuation Switching Stopping
80 60 40 20 0
6
12
18
24
30
36
Follow-up time (months) 2107
Probability of discontinution (%)
(b)
1550
Number at risk 1287 1004 782
590
Events 290 1006
100 80 60 40 NRTI × 2, unboosted PI NRTI × 2, boosted PI NRTI × 2, NNRTI Other combination Log-rank P < 0.001
20 0 0
6
12
18
24
30
36
Follow-up time (months) 37 1110 911 49
9 820 697 24
Number at risk 3 3 2 668 522 398 596 463 370 20 16 12
1 315 267 7
Events 1 34 174 566 113 379 2 27
Figure 2. Rates of discontinuation of first-line therapy initiated in the latest period (2006 – 10). (a) Probability of switching and/or stopping up to 36 months. The continuous black line represents the overall probability of discontinuation, the dashed line represents the probability of switching and the dotted line represents the probability of stopping of first-line therapy initiated between 2006 and 2010. (b) Rates of discontinuation of first-line therapy depending on ART regimen. The probability of discontinuation of first-line therapy up to 36 months is indicated by the dotted line for the two NRTIs +NNRTI, the dashed line for the two NRTIs +boosted PI, the continuous black line for the two NRTIs + unboosted PI and the continuous grey line for other combinations.
HCV/HIV co-infection have been associated with decreased adherence and increased treatment discontinuation.22,23 In the adjusted analysis, the risk of discontinuing first-line ART was positively correlated with younger age, female gender, lower CD4 counts, low ART adherence and starting treatment in an earlier era. Several studies have reported that younger HIV patients and those with lower CD4 counts have poorer adherence
and lower rates of virological response.6,14,22,24 However, some studies have found no correlation between discontinuation and CD4 levels,17,25 while others even found a positive relation with higher CD4 levels.7,13 Suboptimal adherence is a well-known driver of therapeutic failure.14 The suboptimal regimens administered during the pre-HAART and early HAART eras, in terms of toxicity/ intolerance and pill burden,19 can account for the high rate of discontinuation found at the earlier timepoints analysed. Previous studies have reported that women are more likely to discontinue than men.7,21,23 Treatment discontinuation in women has been related to higher relative hazards of initial HAART change because of intolerance and/or toxicity and poorer adherence.26,27 Also, female gender, drug injection use and aboriginal ethnicity show some degree of interconnection within British Columbia.28 In addition, other studies have shown that women are less likely to receive ART, tend to initiate treatment at a later stage of disease and suffer more social problems affecting their adherence when compared with men.26,29,30 Furthermore, the higher rate of discontinuation observed in women in our study could be partially explained by the fact that pregnant women were not excluded from the study population. However, when these women were excluded from the analysis, the results were unchanged. It may also be related to the fact that women are more likely to be injection drug users and to be aboriginal. Other social-structural factors, such as housing, income, education and stigma, may disproportionally affect women. These considerations suggest that gender-specific initiatives need to be broadened in order to reduce treatment discontinuation rates among HIV-positive women. Focusing on the most recent calendar period in our study (2006 –10), female gender, low ART adherence and younger age were still related in the adjusted analysis. In addition, during this period, ADIs at baseline and all initial therapies other than two NRTIs + NNRTI were associated with discontinuation. Previous studies reported that patients with ADIs are more likely to discontinue because of drug-related side effects.17 The link to PI-containing regimens found here may be explained by the poor tolerability of PI-based regimens due to a number of associated adverse side effects.31 – 33 However, regimen-related discontinuation should be considered carefully because other confounding factors influence this result. Readers should be cautious when interpreting our results. We do not consistently have data describing the reason for switching or stopping a regimen. Nevertheless, to analyse this matter we inferred the possible reason for discontinuation based on the viral load, suppressed or not, at the time of discontinuation. In agreement with studies cited previously, we found that discontinuations due to virological failure decreased over time while discontinuations due to tolerability issues increased, probably driven by simplification strategies. Ethnicity, risk status and HCV variables were excluded from the analysis due to missing values. The percentages of injection drug users (44.7%) and HCV/ HIV-coinfected patients (45.3%) present in our study were relatively high and could have influenced the overall rate of discontinuation. However, when these variables were included in the model none was related to discontinuation in the adjusted analysis (data not shown). In addition, our measure of adherence was based on prescription refill records and did not directly measure the number of pills that were actually taken. However, estimates of adherence based on prescription refill compliance
2207
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
0
JAC
Gonzalez-Serna et al.
Table 3. Predictors of treatment discontinuation in 2006– 10 Unadjusted hazard ratio (95% CI)
P value
Adjusted hazard ratio (95% CI)
Age per decade
0.85 (0.80, 0.91)
,0.001
0.93 (0.87, 0.99)
Gender male female
1.00 1.97 (1.71, 2.26)
ADIs at baseline
1.31 (1.10, 1.57)
,0.001
1.26 (1.05, 1.50)
0.013
HAART regimen 2 NRTIs+NNRTI 2 NRTIs+boosted PI other combination
1.00 1.28 (1.12, 1.45) 3.01 (2.30, 3.95)
,0.001 ,0.001
1.00 1.26 (1.10, 1.43) 2.58 (1.96, 3.39)
,0.001 ,0.001
Baseline CD4 (cells/mm3) ,200 200–349 ≥350
1.00 0.73 (0.63, 0.84) 0.99 (0.85, 1.17)
,0.001 0.946
Baseline HIV RNA (copies/mL) ,10 000 10000– 99999 ≥100000 High ART adherencea
1.00 0.81 (0.68, 0.96) 0.87 (0.73, 1.04) 0.44 (0.39, 0.49)
0.017 0.129 ,0.001
0.49 (0.43, 0.56)
,0.001
Variable
,0.001
P value 0.016 ,0.001
1.00 1.64 (1.42, 1.90)
have been independently associated with clinical and biological endpoints.34 Nonetheless, we can assume relatively good adherence in our cohort based on recent studies analysing adherence to ART in British Columbia.22 Our cohort consists of individuals who access a universal, publicly funded healthcare system where medically necessary services, including HAART, are provided free of charge. Consequently, even the most marginalized participants in our cohort were within the reach of medical and social services and were deemed adequately prepared to adhere to their treatment regimen. Accordingly, the universal nature of the Canadian healthcare system gave us the opportunity to explore discontinuation rates without the biases introduced by access to care directly related to income or health insurance coverage. In conclusion, although we have shown that the rates of discontinuation to first-line therapy are decreasing over time, the probability of discontinuing the first-line regimen is still high even for the latest treatment era and drug combinations and may be related to social-structural barriers not measured here. Focusing on the most recent era, younger women on PI regimens and those not achieving optimal adherence run the risk of discontinuing first-line ART and should be closely monitored and provided with support.
Acknowledgements We would like to thank the staff from the British Columbia Centre for Excellence in HIV/AIDS for their assistance and commitment to maintain a state-of-the-art database and our patients for participating in the study.
2208
Funding This work was supported by the Canadian Institutes of Health Research for the GlaxoSmithKline (GSK)—CIHR Research Chair awarded to R. H.
Transparency declarations None to declare.
References 1 Deeks SG, Hecht FM, Swanson M et al. HIV RNA and CD4 cell count response to protease inhibitor therapy in an urban AIDS clinic: response to both initial and salvage therapy. AIDS 1999; 13: F35– 43. 2 Mocroft A, Phillips AN, Miller V et al. The use and response to second-line protease inhibitor regimens: results from the EuroSIDA Study. AIDS 2001; 15: 201– 9. 3 Hammer SM, Saag MS, Schechter M et al. International AIDS Society-USA panel. JAMA 2006; 296: 827– 43. 4 Yuan Y, L’Italien G, Mukherjee J et al. Determinants of discontinuation of initial highly active antiretroviral therapy regimens in a US HIV-infected patient cohort. HIV Med 2006; 7: 156– 62. 5 Althoff KN, Justice AC, Gange SJ et al. Virologic and immunologic response to HAART, by age and regimen class. AIDS 2010; 24: 2469 –79. 6 O’Brien ME, Clark RA, Besch L et al. Patterns and correlates of discontinuation of the initial HAART regimen in an urban outpatient cohort. J Acquir Immune Defic Syndr 2003; 34: 407–14. 7 Elzi L, Marzolini C, Furrer H et al. Treatment modification in human immunodeficiency virus-infected individuals starting combination
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
All P values are derived from Weibull survival analysis. a ART adherence in the first year of therapy ≥95%.
Temporal trends in the discontinuation of first-line antiretroviral therapy
antiretroviral therapy between 2005 and 2008. Arch Intern Med 2010; 170: 57 –65. 8 Camacho R, Teo´filo E. Antiretroviral therapy in treatment-naı¨ve patients with HIV infection. Curr Opin HIV AIDS 2011; 6 Suppl 1: S3 –11. 9 Oette M, Schu¨lter E, Rosen-Zvi M et al. Efficacy of antiretroviral therapy switch in HIV-infected patients: a 10-year analysis of the EuResist Cohort. Intervirology 2012; 55: 160– 6. 10 The Pursuing Later Treatment Option II (PLATO II) project team for the Collaboration of Observational HIV Epidemiological Research Europe (COHERE) Group. Trends in virological and clinical outcomes in individuals with HIV-1 infection and virological failure of drugs from three antiretroviral drug classes: a cohort study. Lancet Infect Dis 2012; 12: 119–27.
12 Helleberg M, Kronborg G, Larsen CS et al. Decreasing rate of multiple treatment modifications among individuals who initiated antiretroviral therapy in 1997 – 2009 in the Danish HIV Cohort Study. Antivir Ther 2013; 18: 345–54. 13 Vo TT, Ledergerber B, Keiser O et al. Durability and outcome of initial antiretroviral treatments received during 2000 – 2005 by patients in the Swiss HIV Cohort Study. J Infect Dis 2008; 197: 1685– 94. 14 Cicconi P, Cozzi-Lepri A, Castagna A et al. Insights into reasons for discontinuation according to year of starting first regimen of highly active antiretroviral therapy in a cohort of antiretroviral-naı¨ve patients. HIV Med 2010; 11: 104–13. 15 Mocroft A, Phillips AN, Soriano V et al. Reasons for stopping antiretrovirals used in an initial highly active antiretroviral regimen: increased incidence of stopping due to toxicity or patient/physician choice in patients with hepatitis C coinfection. AIDS Res Hum Retroviruses 2005; 21: 743–52. 16 Maldonado G, Greenland S. Simulation study of confounder-selection strategies. Am J Epidemiol 1993; 138: 923– 36. 17 Prosperi MC, Fabbiani M, Fanti I et al. Predictors of first-line antiretroviral therapy discontinuation due to drug-related adverse events in HIV-infected patients: a retrospective cohort study. BMC Infect Dis 2012; 12: 296. 18 Hart E, Curtis H, Wilkins E et al. National review of first treatment change after starting highly active antiretroviral therapy in antiretroviralnaı¨ve patients. HIV Med 2007; 8: 186– 91. 19 Vella S, Schwartla¨nder B, Sow SP et al. The history of antiretroviral therapy and of its implementation in resource-limited areas of the world. AIDS 2012; 26: 1231 –41. 20 Van Roon EN, Verzijl JM, Juttmann JR et al. Incidence of discontinuation of highly active antiretroviral therapy (HAART) and its determinants. J Acquir Immune Defic Syndr 1999; 20: 290–4.
21 d’Arminio Monforte A, Cozzi Lepri A, Rezza G et al. Insights into the reasons for discontinuation of the first highly active antiretroviral therapy (HAART) regimen in a cohort of antiretroviral naı¨ve patients. AIDS 2000; 14: 499– 507. 22 O’Neil CR, Palmer AK, Coulter S et al. Factors associated with antiretroviral medication adherence among HIV-positive adults accessing highly active antiretroviral therapy (HAART) in British Columbia, Canada. J Int Assoc Physicians AIDS Care (Chic) 2012; 11: 134–41. 23 Werb D, Milloy MJ, Kerr T et al. Injection drug use and HIV antiretroviral therapy discontinuation in a Canadian setting. AIDS Behav 2013; 17: 68 –73. 24 Rudy BJ, Murphy DA, Harris DR et al. Patient related risks for non adherence to antiretroviral therapy among HIV-infected youth in the United States: a study of prevalence and interactions. AIDS Patient Care STDS 2009; 23: 185– 94. 25 Mocroft A, Youle M, Moore A et al. Reasons for modification and discontinuation of antiretrovirals: results from a single treatment centre. AIDS 2001; 15: 185– 94. 26 Puskas CM, Forrest JI, Parashar S et al. Women and vulnerability to HAART non-adherence: a literature review of treatment adherence by gender from 2000 to 2011. Curr HIV/AIDS Rep 2011; 8: 277– 87. 27 Nicastri E, Leone S, Angeletti C et al. Sex issues in HIV-1-infected persons during highly active antiretroviral therapy: a systematic review. J Antimicrob Chemother 2007; 60: 724–32. 28 Duncan KC, Reading C, Borwein AM et al. HIV incidence and prevalence among aboriginal peoples in Canada. AIDS Behav 2011; 15: 214–27. 29 Gebo KA, Fleishman JA, Conviser R et al. Racial and gender disparities in receipt of highly active antiretroviral therapy persist in a multistate sample of HIV patients in 2001. J Acquir Immune Defic Syndr 2005; 38: 96– 103. 30 Monforte AD, Anderson J, Olczak A. What do we know about antiretroviral treatment of HIV in women? Antivir Ther 2013; 18: 27– 34. 31 Riddler SA, Haubrich R, DiRienzo AG et al. Class-sparing regimens for initial treatment of HIV-1 infection. N Engl J Med 2008; 358: 2095– 106. 32 MacArthur RD, Novak RM, Peng G et al. A comparison of three highly active antiretroviral treatment strategies consisting of nonnucleoside reverse transcriptase inhibitors, protease inhibitors, or both in the presence of nucleoside reverse transcriptase inhibitors as initial therapy (CPCRA 058 FIRST Study): a long-term randomised trial. Lancet 2006; 368: 2125 –35. 33 Hui DY. Effects of HIV protease inhibitor therapy on lipid metabolism. Prog Lipid Res 2003; 42: 81– 92. 34 Hogg RS, Heath K, Bangsberg D et al. Intermittent use of triplecombination therapy is predictive of mortality at baseline and after 1 year of follow-up. AIDS 2002; 16: 1051 –8.
2209
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
11 Willig JH, Westfall AO, Mugavero M et al. Effect of persistency of first-line HIV antiretroviral therapy on clinical outcomes. AIDS Res Hum Retroviruses 2013; 29: 698– 703.
JAC
Temporal trends in the discontinuation of first-line antiretroviral therapy Alejandro Gonzalez-Serna1*, Keith Chan1, Benita Yip1, William Chau1, Rachel McGovern1, Hasina Samji1, Viviane Dias Lima1,2, Robert S. Hogg1,3 and Richard Harrigan1,2 1
BC Centre for Excellence on HIV/AIDS, Vancouver V6Z1Y6, BC, Canada; 2Division of AIDS, Department of Medicine, University of British Columbia, Vancouver V6T1Z4, BC, Canada; 3Faculty of Health Sciences, Simon Fraser University, Burnaby V5A1S6 BC, Canada *Corresponding author. Tel: +1-604-806-8645; Fax: +1-604-806-9463; E-mail: [email protected]
Objectives: The aim of this study was to describe the rates and predictors of discontinuing first-line antiretroviral therapy in the different eras of treatment over a nearly 20 year period initiated in British Columbia between 1992 and 2010. Methods: All naive adults who started antiretroviral therapy (first-line antiretroviral therapy) at any hospital or clinic in British Columbia (Canada) in 1992 – 2010 were included in this population-based retrospective cohort study. We were primarily interested in whether the era of treatment (1992 – 95, 1996 – 2000, 2001 – 05 and 2006 – 10) was associated with discontinuation (stopping or switching of initial treatment) within 3 years of starting therapy. Weibull survival analysis was used to model the era of treatment and its association with time to discontinuation. Results: The study included 7901 patients. Overall, the probability of discontinuing at 12, 24 and 36 months of treatment was 52%, 68% and 76%, respectively. In the adjusted model, variables associated with discontinuing were earlier treatment era, younger age, low adherence and lower baseline CD4 count. Regarding the 2006 –10 period, the probability of discontinuing at 12, 24 and 36 months was 36%, 47% and 53%, respectively. In the adjusted model, the variables associated with discontinuation were younger age, female gender, AIDS-defining illnesses at baseline, low adherence and a protease inhibitor (PI)-based regimen. Conclusions: Discontinuation rates of first-line therapy have decreased over time, but are still quite high even for the latest drug combinations. In the most recent era, younger women on a PI regimen and those not achieving optimal adherence had the highest risk of discontinuing first-line antiretroviral therapy. Keywords: HIV-1, ART, treatment, risk
Introduction HIV is a chronic disease that requires life-long therapy and the prognosis of HIV-infected patients starting highly active antiretroviral therapy (HAART) greatly depends on treatment optimization.1 – 3 To better understand the determinants of treatment changes over time it is important to optimize the initial HAART regimen and evaluate trends in the incidence of discontinuation (stopping or switching of initial treatment). Several studies analysing rates of discontinuation reported that a high percentage of patients (40% – 60%) underwent discontinuation of their initial regimen within the first year of treatment.4 – 7 Over recent years HAART strategies have evolved dramatically, offering high efficacy, less toxicity and reduced pill burden and dosing frequency.8 Accordingly, some studies found improvements in virological and immunological outcomes and pointed to time trends towards lower rates of discontinuation.9 – 11 A recent study reported that the risk of multiple treatment modifications
(three or more changes in any of the antiretroviral components in the regimen) had decreased over time.12 However, no studies have found evidence of a significant difference in the rate of discontinuation of the first antiretroviral therapy (ART) (discontinuation for any reason) according to calendar period of starting therapy.12 – 15 It is possible that time period length and/or number of patients included in those studies were not big enough to find a significant difference in the rate of discontinuation, so studies analysing these different eras of treatment are still lacking. Therefore, the aim of this study was to describe the rates and predictors of discontinuing first-line ART in the different eras of treatment over a nearly 20 year period initiated in British Columbia between 1992 and 2010.
Methods Samples and records for all naive HIV-positive patients aged ≥19 years starting ART between September 1992 and September 2010 at any
# The Author 2014. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: [email protected]
2202
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
Received 12 December 2013; returned 27 January 2014; revised 7 March 2014; accepted 17 March 2014
Temporal trends in the discontinuation of first-line antiretroviral therapy
Statistical analysis Comparisons across eras were carried out using the x2 test or Fisher’s exact test for categorical variables and the Wilcoxon rank-sum test for continuous variables. Kaplan– Meier methods were used to graphically display the probability of discontinuing regimen after starting ART. Weibull survival analysis was used to model the era of treatment and its association with time to discontinuation. Variables that were significant at P, 0.05 in the unadjusted analyses were candidates for inclusion in the multivariate model. A confounder model was fitted for era using the methods of Maldonado and Greenland.16 All analyses were performed using SAS software version 9.3 (SAS Institute, Cary, NC, USA).
Ethics statement This study was approved by the Committee on Human Research and the University of British Columbia/ Providence Health Care Research Ethics Board.
Results A total of 7901 patients were included in the study. The median age was 39 years (IQR 33 – 46 years) and 83% were male. The
median CD4 cell count at ART initiation was 220 cells/mm3 (IQR 110 – 350 cells/mm3 ) and the median HIV RNA was 4.9 log10 copies/mL (IQR 4.4 –5.0 log10 copies/mL). A total of 1855 patients (23.5%) started treatment in pre-HAART era, 2406 patients (30.5%) in 1996 – 2000, 1533 patients (19.4%) in 2001 – 05 and 2107 patients (26.7%) in 2006 – 10. Patients’ characteristics at baseline stratified by era of treatment are listed in Table 1, where all characteristics were statistically different (P, 0.001). Due to the missing values for ethnicity, risk status and HCV, these variables were not included in the statistical model and are only explored in Table 1. All analyses involving viral load characteristics were limited to the HAART eras (n¼6046), because in the pre-HAART era (n ¼ 1855) viral loads were not available in most patients, and comparisons of discontinuation based on virological suppression were limited to the mid-HAART and late HAART eras (n¼3640). Overall, 5490 (69.5%) patients discontinued therapy during follow-up, of whom 3659 (66.6%) were switches and 1831 (33.4%) were stops. The probability of discontinuation at 12, 24 and 36 months was 0.52 (95% CI 0.51 – 0.53), 0.67 (95% CI 0.66 – 0.69) and 0.76 (95% CI 0.75 – 0.77), respectively (Figure 1a). At 36 months, the probability of switching was higher than that of stopping (0.64 versus 0.23). Based on eras, the probability of discontinuation at 36 months was progressively lower the more recent the era (0.94 in 1992 –96, 0.78 in 1996 – 2000, 0.77 in 2001 – 05 and 0.53 in 2006 – 10, P, 0.001) (Figure 1b). Discontinuation rates in relation to the ART regimen showed that two NRTIs + NNRTI had a lower risk of discontinuation than two NRTIs + boosted or unboosted PI, or other combinations (P, 0.001) (Figure 1c). In the mid-HAART and late HAART eras, 2119 patients (58.2%) discontinued therapy, of whom 1558 (73.5%) were switches and 561 (26.5%) were stops. Of these mid-HAART and late HAART era discontinuations, 1377 (65.0%) were not virologically suppressed (virological failure switch) at the time of discontinuation and 742 (35.0%) were virologically suppressed (probably a tolerability issue). The probability of discontinuation due to virological failure was greater than discontinuation due to tolerability issues through 3 years of follow-up (0.33 versus 0.13 at 12 months and 0.43 versus 0.37 at 36 months) (Figure 1d). During the HAART eras, 3903 (64.6%) patients discontinued therapy, of whom 2628 (67.3%) were switches and 1275 (32.7%) were stops. In unadjusted analysis, most of the variables included were associated with switching or stopping (Table 2). However, in the adjusted model variables associated with switching or stopping were earlier treatment era [mid-HAART (adjusted hazard ratio 1.01, 95% CI 0.94 – 1.09) and late HAART (adjusted hazard ratio 0.61, 95% CI 0.56 – 0.66) compared with the early HAART], female gender (adjusted hazard ratio 1.23, 95% CI 1.14 – 1.33), younger age per decade (adjusted hazard ratio 0.89, 95% CI 0.86 – 0.93), high ART adherence (adjusted hazard ratio 0.39, 95% CI 0.37 – 0.42) and lower baseline CD4 count [200 – 349 CD4 cells/mm 3 (adjusted hazard ratio 0.91, 95% CI 0.84 – 0.99), versus ,200 CD4 cells/mm3]. Regarding the most recent era, i.e. the 2006 – 10 period, the probabilities of discontinuing at 12, 24 and 36 months of treatment were 0.36 (95% CI 0.34 – 0.38), 0.47 (95% CI 0.45 – 0.49) and 0.53 (95% CI 0.51 – 0.55), respectively (Figure 2a). Discontinuation rates in relation to the ART regimen showed
2203
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
hospital or clinic in British Columbia, Canada, were centralized and analysed at the British Columbia Centre for Excellence in HIV/AIDS. Demographic, clinical and laboratory data and information on therapy were collected for all participants. Study was restricted to patients with complete information on age, gender, baseline CD4 and viral load in the HAART era. The primary factor of interest in this population-based retrospective cohort study was the era of starting ART, grouped as: pre-HAART (September 1992 – May 1996); early HAART (June 1996 to December 2000); mid-HAART (January 2001 to December 2005) and late HAART (January 2006 to December 2010). Other covariates of interest included age, gender, detection of hepatitis C virus (HCV) antibodies on at least one occasion during follow-up, baseline CD4 cell count and plasma HIV RNA (copies/mL), risk group, ethnicity, adherence, AIDS status and type of first ART regimen started [one or two nucleoside reverse transcriptase inhibitors (NRTIs), two NRTIs +one non-nucleoside reverse transcriptase inhibitor (NNRTI), two NRTIs +unboosted or boosted protease inhibitor (PI) or another regimen with three or more antiretrovirals]. Measurement of exposure to ART was obtained by prescription refills. Medication adherence, estimated based on antiretroviral prescription refill data during the first year after starting therapy, was calculated by dividing the number of days of medication dispensed by the number of days of follow-up, expressed as a percentage. We investigated the time from ART initiation to discontinuation of the first ART within 3 years of starting therapy. Discontinuation was categorized as either a stop (defined as stopping all drugs in a regimen for .6 months) or a switch (changing any of the antiretroviral components, not the formulation, in the regimen). If a patient was virologically suppressed (defined as the two most recent plasma HIV RNA measurements ,50 copies/mL within 6 months) prior to discontinuation, it was considered a discontinuation ‘probably’ due to a tolerability issue, while those not suppressed were considered as having discontinuation due to virological failure. For the latter purpose only patients from the mid-HAART and late HAART eras were analysed because virological suppression data (plasma HIV RNA ,50 copies/mL) were not available in the pre-HAART and early HAART eras. The main objective was to compare the rate of discontinuation according to era of ART initiation. Follow-up of the patients who did not stop or switch was censored at the death date, date of moving out of British Columbia, date of entering a placebo trial, date of starting supervised therapy interruption, 30 September 2011 (study end date) or 3 years from ART start date, whichever came first.
JAC
Gonzalez-Serna et al.
Table 1. Baseline characteristics by era (n¼7901) Characteristic
1992 –96 (n¼1855)
1996 –2000 (n ¼2406)
2001 –05 (n¼1533)
2006– 10 (n¼2107)
1668 (89.9) 36 (31 –43)
1994 (82.9) 37 (32 –44)
1196 (78) 41 (35 –48)
1700 (80.7) 42 (35 – 49)
ART discontinuation switch virological failurea switch tolerabilityb switch stop virological failurea stop tolerabilityb stop
1587 (85.6) 1031 (55.6) NA NA 556 (30) NA NA
1784 (74.1) 1070 (44.5) NA NA 714 (29.7) NA NA
1113 (82.6) 802 (52.3) 455 (56.7) 347 (43.3) 311 (20.3) 234 (75.2) 77 (24.8)
1006 (47.7) 756 (35.9) 485 (64.1) 271 (35.9) 250 (11.9) 203 (81.2) 47 (18.8)
396 (21.3) 321 (34.1)
326 (13.5) 1128 (53.4)
278 (18.1) 651 (47.8)
261 (12.4) 742 (39.6)
Ethnicity Caucasian aboriginal other
1102 (82.7) 171 (12.8) 59 (4.4)
1095 (68.6) 388 (24.3) 114 (7.1)
545 (62.5) 201 (23.1) 126 (14.4)
453 (54.5) 236 (28.4) 142 (17.1)
Risk group men who have sex with men heterosexual sex injection drug use
1083 (73.1) 308 (20.8) 438 (28.7)
786 (44.7) 545 (31) 991 (52.8)
350 (34) 376 (36.6) 630 (50.2)
346 (34) 463 (36.1) 762 (45.3)
ART regimen ≤2 NRTIs 2 NRTIs+unboosted PI 2 NRTIs+boosted PI 2 NRTIs+NNRTI other ART combinations
1855 (100) 0 (0) 0 (0) (0) (0)
569 (23.6) 1055 (43.8) 80 (3.3) 555 (23.1) 147 (6.1)
17 (1.1) 104 (6.8) 687 (44.8) 619 (40.4) 106 (6.9)
2 (0.1) 37 (1.8) 1110 (52.7) 911 (43.2) 47 (2.2)
Baseline CD4 (cells/mm3) ,200 200–349 ≥350
775 (41.8) 636 (34.3) 444 (23.9)
846 (35.2) 648 (26.9) 912 (37.9)
959 (62.6) 397 (25.9) 177 (11.5)
855 (40.6) 782 (37.1) 470 (22.3)
Baseline CD4 (cells/mm3), median (IQR)
230 (100– 340)
280 (130 –420)
160 (70 –240)
230 (130 –330)
174 (11.4) 538 (35.1) 821 (53.6)
336 (15.9) 939 (44.6) 832 (39.5)
1187 (76.9) 425 (27.5)
1776 (83.5) 319 (15.0)
909 (59.3)
1405 (66.7)
ADIs at baseline Hepatitis Cc
Baseline viral load (copies/mL) ,10 000 10000– 99999 ≥100000
NA NA NA
270 (11.2) 1005 (41.8) 1131 (47)
Viral load suppressiond during 3 year follow-up within 6 months prior to discontinuation
NA NA
NA NA
High ART adherencee
612 (33.0)
1155 (48.0)
NA, not available. Data are number (%) unless otherwise indicated. All characteristics were significantly different by era (P,0.001). a Discontinuation occurred with detectable viral load suggesting discontinuation for virological failure. b Discontinuation occurred with undetectable viral load suggesting discontinuation for a probable tolerability issue. c HCV antibody positive. d Two consecutive viral loads ,500 copies/mL. e ART adherence in the first year of therapy ≥95%.
that two NRTIs + NNRTI had a lower risk of discontinuation than two NRTIs + boosted or unboosted PI, or other combinations (P,0.001) (Figure 2b). 2204
Again, in the unadjusted model, most of the variables included were associated with switching or stopping. However, in the adjusted model the variables associated with discontinuation
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
Male gender Age (years), median (IQR)
JAC
Temporal trends in the discontinuation of first-line antiretroviral therapy
(a)
(b) 100
Discontinuation Switching Stopping
80
Probability of discontinuation (%)
Probability of discontinuation (%)
100
60 40 20 0
80 60 40
92–96 96–00 01–05 06–10 Log-rank P < 0.001
20 0
6
12
24
18
30
36
0
6
Follow-up time (months) Events
Number at risk 7901
4876
3542
2672
2059
1581
(d) 100
80
60
NRTI × 2, unboosted PI NRTI × 2, boosted PI NRTI × 2, NNRTI Other combination
20
Log-rank P < 0.001 0
Probability of discontinuation (%)
Probability of discontinuation (%)
1855 2406 1533 2107
791 5490
(c) 100
40
12
18
30
36
109 518 364 590
19 267 215 290
24
Follow-up time (months) 841 1493 992 1550
460 1055 740 1287
Number at risk 276 810 582 1004
175 639 463 782
Events 1587 1784 1113 1006
Discontinuation Tolerability discontinuation Virological failure discontinuation
80
60
40
20
0 0
6
12
18
24
30
36
0
6
Follow-up time (months) 1196 1877 2085 2743
742 1349 1478 1307
Number at risk 287 391 525 609 807 1049 808 972 1200 355 502 768
12
18
24
30
36
Follow-up time (months) 227 459 646 249
Events 120 926 250 1144 353 1128 68 2292
Number at risk 3640
2542
2027
1586
Events 1245
954
505 2119
Figure 1. Rates of discontinuation of first-line therapy initiated between 1992 and 2010. (a) Overall probability of switching and/or stopping up to 36 months. The continuous line represents the overall probability of discontinuation, the dashed line represents the probability of switching and the dotted line represents the probability of stopping of first-line therapy initiated between 1992 and 2010. (b) Rates of discontinuation of first-line therapy decreasing over time with advancing era. The probability of discontinuation of first-line therapy up to 36 months is indicated by the continuous black line for the period 1992 – 95, the dashed line for 1996 – 2000, the dotted line for 2001 – 05 and the continuous grey line for 2006 – 10. (c) Rates of discontinuation of first-line therapy depending on ART regimen. The probability of discontinuation of first-line therapy up to 36 months is indicated by the dotted line for the two NRTIs +NNRTI, the dashed line for the two NRTIs +boosted PI, the continuous black line for the two NRTIs +unboosted PI and the continuous grey line for the other combinations. (d) Discontinuation was mainly driven by virological failure (viral load not suppressed at time of discontinuation), mostly during the first year; however, tolerability issues (viral load suppressed at time of discontinuation) increased to a similar level over the third year. The continuous line represents the overall probability of discontinuation, the dashed line represents the probability of discontinuation due to tolerability issues and the dotted line represents the probability of discontinuation due to virological failure of first-line therapy initiated between 2001 and 2010.
were younger age (adjusted hazard ratio 0.93, 95% CI 0.87–0.99), female gender (adjusted hazard ratio 1.64, 95% CI 1.42 – 1.90), AIDS-defining illnesses (ADIs) at baseline (adjusted hazard ratio
1.26, 95% CI 1.05 – 1.50), high ART adherence (adjusted hazard ratio 0.49, 95% CI 0.43 – 0.56) and initial therapy with two NRTIs+ boosted PI (adjusted hazard ratio 1.26, 95% CI 1.10–1.43)
2205
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
0
Gonzalez-Serna et al.
Table 2. Predictors of treatment discontinuation in1996–2010 Unadjusted hazard ratio (95% CI)
P value
Adjusted hazard ratio (95% CI)
P value
Era early HAART (1996–2000) mid-HAART (2001– 05) late HAART (2006–10)
1.00 0.92 (0.86, 1.00) 0.53 (0.49, 0.57)
0.037 ,0.001
1.00 1.01 (0.94, 1.09) 0.61 (0.56, 0.66)
0.710 ,0.001
Age per decade
0.81 (0.79, 0.84)
,0.001
0.89 (0.86, 0.93)
,0.001
Gender male female
1.00 1.52 (1.41, 1.64)
ADIs at baseline
1.09 (0.99, 1.19)
0.067
HAART regimen 2 NRTIs+NNRTI 2 NRTIs+unboosted PI 2 NRTIs+boosted PI other combinations
1.00 1.77 (1.62, 1.93) 1.18 (1.09, 1.29) 2.11 (1.92, 2.32)
,0.001 ,0.001 ,0.001
Baseline CD4 (cells/mm3) ,200 200–349 ≥350
1.00 0.85 (0.79, 0.91) 1.10 (1.02, 1.19)
,0.001 0.011
Baseline plasma HIV RNA (copies/mL) ,10 000 10000– 99999 ≥100000
1.00 0.92 (0.83, 1.02) 0.99 (0.90, 1.10)
0.117 0.883
High ART adherencea
0.36 (0.34, 0.39)
,0.001
Variable
,0.001
,0.001 1.00 1.23 (1.14, 1.33)
0.019 0.287
0.39 (0.37, 0.42)
,0.001
All P values are derived from Weibull survival analysis. ART adherence in the first year of therapy ≥95%.
a
or other regimens (adjusted hazard ratio 2.58, 95% CI 1.96–3.39) compared with two NRTIs+ NNRTI (Table 3). In order to determine whether not excluding pregnant women had an effect on the results, we conducted a sensitivity analysis. We assumed that women whose ART regimen contained zidovudine and lasted ,1 year and who had a CD4 count .350 cells/mm3 at therapy initiation had been taking ART solely for the prevention of mother-to-child transmission of HIV. We found 133 women who met these criteria. However, when these women were excluded from the analysis, the results were unchanged.
Discussion In this study we show that rates of first-line therapy discontinuation over a nearly 20 year period decreased over time. However, the probability of stopping and/or switching to a second-line regimen was considerable even with access to the latest era of antiretrovirals and drug combinations. The decreases in rates of discontinuation found in this study challenge previous findings, where similar studies reported unchanging discontinuation rates according to study period.12 – 15,17 It is possible that the study duration, in some cases ,5 years, and/or the low number of patients used in these studies could
2206
explain why a decrease in discontinuation was not found previously. In fact, there is one study, analysing 1189 patients between 1997 and 2007, that found a decreasing risk of multiple treatment modifications (three or more) over time but concluded that first treatment discontinuation was comparable according to calendar period.12 In comparison, our study of 7901 patients over a nearly 20 year period showed a significant decrease in the rates of treatment discontinuation. Moreover, some studies found that discontinuation because of intolerance and/or toxicity was decreasing over time while discontinuation because of therapy simplification was increasing over time. This in part could explain the absence of an overall decreasing trend globally.12 – 15 With the increased availability of newer drugs, drug classes and multidrug-containing regimens, and those with better tolerability, dosing and resistance profiles, the individualization of treatment has led to improved virological and immunological outcomes for HIV-infected patients. This may be one of the primary factors associated with the decrease in discontinuation observed in our study, given the study’s duration and geographic region. All analyses were repeated by defining failure as .500 copies/mL for consistency between the different eras and the results were consistent (data not shown). Overall, the rates of discontinuation of first-line therapy found in our study are slightly higher than those in some previous studies.17 – 21 Injection drug use and
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
1.00 0.91 (0.84, 0.99) 1.04 (0.96, 1.13)
Temporal trends in the discontinuation of first-line antiretroviral therapy
Probability of discontinution (%)
(a)
100
Discontinuation Switching Stopping
80 60 40 20 0
6
12
18
24
30
36
Follow-up time (months) 2107
Probability of discontinution (%)
(b)
1550
Number at risk 1287 1004 782
590
Events 290 1006
100 80 60 40 NRTI × 2, unboosted PI NRTI × 2, boosted PI NRTI × 2, NNRTI Other combination Log-rank P < 0.001
20 0 0
6
12
18
24
30
36
Follow-up time (months) 37 1110 911 49
9 820 697 24
Number at risk 3 3 2 668 522 398 596 463 370 20 16 12
1 315 267 7
Events 1 34 174 566 113 379 2 27
Figure 2. Rates of discontinuation of first-line therapy initiated in the latest period (2006 – 10). (a) Probability of switching and/or stopping up to 36 months. The continuous black line represents the overall probability of discontinuation, the dashed line represents the probability of switching and the dotted line represents the probability of stopping of first-line therapy initiated between 2006 and 2010. (b) Rates of discontinuation of first-line therapy depending on ART regimen. The probability of discontinuation of first-line therapy up to 36 months is indicated by the dotted line for the two NRTIs +NNRTI, the dashed line for the two NRTIs +boosted PI, the continuous black line for the two NRTIs + unboosted PI and the continuous grey line for other combinations.
HCV/HIV co-infection have been associated with decreased adherence and increased treatment discontinuation.22,23 In the adjusted analysis, the risk of discontinuing first-line ART was positively correlated with younger age, female gender, lower CD4 counts, low ART adherence and starting treatment in an earlier era. Several studies have reported that younger HIV patients and those with lower CD4 counts have poorer adherence
and lower rates of virological response.6,14,22,24 However, some studies have found no correlation between discontinuation and CD4 levels,17,25 while others even found a positive relation with higher CD4 levels.7,13 Suboptimal adherence is a well-known driver of therapeutic failure.14 The suboptimal regimens administered during the pre-HAART and early HAART eras, in terms of toxicity/ intolerance and pill burden,19 can account for the high rate of discontinuation found at the earlier timepoints analysed. Previous studies have reported that women are more likely to discontinue than men.7,21,23 Treatment discontinuation in women has been related to higher relative hazards of initial HAART change because of intolerance and/or toxicity and poorer adherence.26,27 Also, female gender, drug injection use and aboriginal ethnicity show some degree of interconnection within British Columbia.28 In addition, other studies have shown that women are less likely to receive ART, tend to initiate treatment at a later stage of disease and suffer more social problems affecting their adherence when compared with men.26,29,30 Furthermore, the higher rate of discontinuation observed in women in our study could be partially explained by the fact that pregnant women were not excluded from the study population. However, when these women were excluded from the analysis, the results were unchanged. It may also be related to the fact that women are more likely to be injection drug users and to be aboriginal. Other social-structural factors, such as housing, income, education and stigma, may disproportionally affect women. These considerations suggest that gender-specific initiatives need to be broadened in order to reduce treatment discontinuation rates among HIV-positive women. Focusing on the most recent calendar period in our study (2006 –10), female gender, low ART adherence and younger age were still related in the adjusted analysis. In addition, during this period, ADIs at baseline and all initial therapies other than two NRTIs + NNRTI were associated with discontinuation. Previous studies reported that patients with ADIs are more likely to discontinue because of drug-related side effects.17 The link to PI-containing regimens found here may be explained by the poor tolerability of PI-based regimens due to a number of associated adverse side effects.31 – 33 However, regimen-related discontinuation should be considered carefully because other confounding factors influence this result. Readers should be cautious when interpreting our results. We do not consistently have data describing the reason for switching or stopping a regimen. Nevertheless, to analyse this matter we inferred the possible reason for discontinuation based on the viral load, suppressed or not, at the time of discontinuation. In agreement with studies cited previously, we found that discontinuations due to virological failure decreased over time while discontinuations due to tolerability issues increased, probably driven by simplification strategies. Ethnicity, risk status and HCV variables were excluded from the analysis due to missing values. The percentages of injection drug users (44.7%) and HCV/ HIV-coinfected patients (45.3%) present in our study were relatively high and could have influenced the overall rate of discontinuation. However, when these variables were included in the model none was related to discontinuation in the adjusted analysis (data not shown). In addition, our measure of adherence was based on prescription refill records and did not directly measure the number of pills that were actually taken. However, estimates of adherence based on prescription refill compliance
2207
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
0
JAC
Gonzalez-Serna et al.
Table 3. Predictors of treatment discontinuation in 2006– 10 Unadjusted hazard ratio (95% CI)
P value
Adjusted hazard ratio (95% CI)
Age per decade
0.85 (0.80, 0.91)
,0.001
0.93 (0.87, 0.99)
Gender male female
1.00 1.97 (1.71, 2.26)
ADIs at baseline
1.31 (1.10, 1.57)
,0.001
1.26 (1.05, 1.50)
0.013
HAART regimen 2 NRTIs+NNRTI 2 NRTIs+boosted PI other combination
1.00 1.28 (1.12, 1.45) 3.01 (2.30, 3.95)
,0.001 ,0.001
1.00 1.26 (1.10, 1.43) 2.58 (1.96, 3.39)
,0.001 ,0.001
Baseline CD4 (cells/mm3) ,200 200–349 ≥350
1.00 0.73 (0.63, 0.84) 0.99 (0.85, 1.17)
,0.001 0.946
Baseline HIV RNA (copies/mL) ,10 000 10000– 99999 ≥100000 High ART adherencea
1.00 0.81 (0.68, 0.96) 0.87 (0.73, 1.04) 0.44 (0.39, 0.49)
0.017 0.129 ,0.001
0.49 (0.43, 0.56)
,0.001
Variable
,0.001
P value 0.016 ,0.001
1.00 1.64 (1.42, 1.90)
have been independently associated with clinical and biological endpoints.34 Nonetheless, we can assume relatively good adherence in our cohort based on recent studies analysing adherence to ART in British Columbia.22 Our cohort consists of individuals who access a universal, publicly funded healthcare system where medically necessary services, including HAART, are provided free of charge. Consequently, even the most marginalized participants in our cohort were within the reach of medical and social services and were deemed adequately prepared to adhere to their treatment regimen. Accordingly, the universal nature of the Canadian healthcare system gave us the opportunity to explore discontinuation rates without the biases introduced by access to care directly related to income or health insurance coverage. In conclusion, although we have shown that the rates of discontinuation to first-line therapy are decreasing over time, the probability of discontinuing the first-line regimen is still high even for the latest treatment era and drug combinations and may be related to social-structural barriers not measured here. Focusing on the most recent era, younger women on PI regimens and those not achieving optimal adherence run the risk of discontinuing first-line ART and should be closely monitored and provided with support.
Acknowledgements We would like to thank the staff from the British Columbia Centre for Excellence in HIV/AIDS for their assistance and commitment to maintain a state-of-the-art database and our patients for participating in the study.
2208
Funding This work was supported by the Canadian Institutes of Health Research for the GlaxoSmithKline (GSK)—CIHR Research Chair awarded to R. H.
Transparency declarations None to declare.
References 1 Deeks SG, Hecht FM, Swanson M et al. HIV RNA and CD4 cell count response to protease inhibitor therapy in an urban AIDS clinic: response to both initial and salvage therapy. AIDS 1999; 13: F35– 43. 2 Mocroft A, Phillips AN, Miller V et al. The use and response to second-line protease inhibitor regimens: results from the EuroSIDA Study. AIDS 2001; 15: 201– 9. 3 Hammer SM, Saag MS, Schechter M et al. International AIDS Society-USA panel. JAMA 2006; 296: 827– 43. 4 Yuan Y, L’Italien G, Mukherjee J et al. Determinants of discontinuation of initial highly active antiretroviral therapy regimens in a US HIV-infected patient cohort. HIV Med 2006; 7: 156– 62. 5 Althoff KN, Justice AC, Gange SJ et al. Virologic and immunologic response to HAART, by age and regimen class. AIDS 2010; 24: 2469 –79. 6 O’Brien ME, Clark RA, Besch L et al. Patterns and correlates of discontinuation of the initial HAART regimen in an urban outpatient cohort. J Acquir Immune Defic Syndr 2003; 34: 407–14. 7 Elzi L, Marzolini C, Furrer H et al. Treatment modification in human immunodeficiency virus-infected individuals starting combination
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
All P values are derived from Weibull survival analysis. a ART adherence in the first year of therapy ≥95%.
Temporal trends in the discontinuation of first-line antiretroviral therapy
antiretroviral therapy between 2005 and 2008. Arch Intern Med 2010; 170: 57 –65. 8 Camacho R, Teo´filo E. Antiretroviral therapy in treatment-naı¨ve patients with HIV infection. Curr Opin HIV AIDS 2011; 6 Suppl 1: S3 –11. 9 Oette M, Schu¨lter E, Rosen-Zvi M et al. Efficacy of antiretroviral therapy switch in HIV-infected patients: a 10-year analysis of the EuResist Cohort. Intervirology 2012; 55: 160– 6. 10 The Pursuing Later Treatment Option II (PLATO II) project team for the Collaboration of Observational HIV Epidemiological Research Europe (COHERE) Group. Trends in virological and clinical outcomes in individuals with HIV-1 infection and virological failure of drugs from three antiretroviral drug classes: a cohort study. Lancet Infect Dis 2012; 12: 119–27.
12 Helleberg M, Kronborg G, Larsen CS et al. Decreasing rate of multiple treatment modifications among individuals who initiated antiretroviral therapy in 1997 – 2009 in the Danish HIV Cohort Study. Antivir Ther 2013; 18: 345–54. 13 Vo TT, Ledergerber B, Keiser O et al. Durability and outcome of initial antiretroviral treatments received during 2000 – 2005 by patients in the Swiss HIV Cohort Study. J Infect Dis 2008; 197: 1685– 94. 14 Cicconi P, Cozzi-Lepri A, Castagna A et al. Insights into reasons for discontinuation according to year of starting first regimen of highly active antiretroviral therapy in a cohort of antiretroviral-naı¨ve patients. HIV Med 2010; 11: 104–13. 15 Mocroft A, Phillips AN, Soriano V et al. Reasons for stopping antiretrovirals used in an initial highly active antiretroviral regimen: increased incidence of stopping due to toxicity or patient/physician choice in patients with hepatitis C coinfection. AIDS Res Hum Retroviruses 2005; 21: 743–52. 16 Maldonado G, Greenland S. Simulation study of confounder-selection strategies. Am J Epidemiol 1993; 138: 923– 36. 17 Prosperi MC, Fabbiani M, Fanti I et al. Predictors of first-line antiretroviral therapy discontinuation due to drug-related adverse events in HIV-infected patients: a retrospective cohort study. BMC Infect Dis 2012; 12: 296. 18 Hart E, Curtis H, Wilkins E et al. National review of first treatment change after starting highly active antiretroviral therapy in antiretroviralnaı¨ve patients. HIV Med 2007; 8: 186– 91. 19 Vella S, Schwartla¨nder B, Sow SP et al. The history of antiretroviral therapy and of its implementation in resource-limited areas of the world. AIDS 2012; 26: 1231 –41. 20 Van Roon EN, Verzijl JM, Juttmann JR et al. Incidence of discontinuation of highly active antiretroviral therapy (HAART) and its determinants. J Acquir Immune Defic Syndr 1999; 20: 290–4.
21 d’Arminio Monforte A, Cozzi Lepri A, Rezza G et al. Insights into the reasons for discontinuation of the first highly active antiretroviral therapy (HAART) regimen in a cohort of antiretroviral naı¨ve patients. AIDS 2000; 14: 499– 507. 22 O’Neil CR, Palmer AK, Coulter S et al. Factors associated with antiretroviral medication adherence among HIV-positive adults accessing highly active antiretroviral therapy (HAART) in British Columbia, Canada. J Int Assoc Physicians AIDS Care (Chic) 2012; 11: 134–41. 23 Werb D, Milloy MJ, Kerr T et al. Injection drug use and HIV antiretroviral therapy discontinuation in a Canadian setting. AIDS Behav 2013; 17: 68 –73. 24 Rudy BJ, Murphy DA, Harris DR et al. Patient related risks for non adherence to antiretroviral therapy among HIV-infected youth in the United States: a study of prevalence and interactions. AIDS Patient Care STDS 2009; 23: 185– 94. 25 Mocroft A, Youle M, Moore A et al. Reasons for modification and discontinuation of antiretrovirals: results from a single treatment centre. AIDS 2001; 15: 185– 94. 26 Puskas CM, Forrest JI, Parashar S et al. Women and vulnerability to HAART non-adherence: a literature review of treatment adherence by gender from 2000 to 2011. Curr HIV/AIDS Rep 2011; 8: 277– 87. 27 Nicastri E, Leone S, Angeletti C et al. Sex issues in HIV-1-infected persons during highly active antiretroviral therapy: a systematic review. J Antimicrob Chemother 2007; 60: 724–32. 28 Duncan KC, Reading C, Borwein AM et al. HIV incidence and prevalence among aboriginal peoples in Canada. AIDS Behav 2011; 15: 214–27. 29 Gebo KA, Fleishman JA, Conviser R et al. Racial and gender disparities in receipt of highly active antiretroviral therapy persist in a multistate sample of HIV patients in 2001. J Acquir Immune Defic Syndr 2005; 38: 96– 103. 30 Monforte AD, Anderson J, Olczak A. What do we know about antiretroviral treatment of HIV in women? Antivir Ther 2013; 18: 27– 34. 31 Riddler SA, Haubrich R, DiRienzo AG et al. Class-sparing regimens for initial treatment of HIV-1 infection. N Engl J Med 2008; 358: 2095– 106. 32 MacArthur RD, Novak RM, Peng G et al. A comparison of three highly active antiretroviral treatment strategies consisting of nonnucleoside reverse transcriptase inhibitors, protease inhibitors, or both in the presence of nucleoside reverse transcriptase inhibitors as initial therapy (CPCRA 058 FIRST Study): a long-term randomised trial. Lancet 2006; 368: 2125 –35. 33 Hui DY. Effects of HIV protease inhibitor therapy on lipid metabolism. Prog Lipid Res 2003; 42: 81– 92. 34 Hogg RS, Heath K, Bangsberg D et al. Intermittent use of triplecombination therapy is predictive of mortality at baseline and after 1 year of follow-up. AIDS 2002; 16: 1051 –8.
2209
Downloaded from http://jac.oxfordjournals.org/ at University of North Dakota on June 3, 2015
11 Willig JH, Westfall AO, Mugavero M et al. Effect of persistency of first-line HIV antiretroviral therapy on clinical outcomes. AIDS Res Hum Retroviruses 2013; 29: 698– 703.
JAC
