Open Forum Infectious Diseases MAJOR ARTICLE Durability of Efavirenz Compared With Boosted Protease Inhibitor-Based Regimens in Antiretroviral-Naïve Patients in the Caribbean and Central and South America 1 1 1 2 3 4 Yanink Caro-Vega, Pablo F. Belaunzarán-Zamudio, Brenda E. Crabtree-Ramírez, Bryan E. Shepherd, Beatriz Grinsztejn, Marcelo Wolff, 5,6 7 8 2 1 Jean W. Pape, Denis Padgett, Eduardo Gotuzzo, Catherine C. McGowan, and Juan G. Sierra-Madero 1 2 Departmento de Infectología, Instituto Nacional de Ciencias Médicas y Nutrición “Salvador Zubirán,” Mexico City, Mexico; Vanderbilt University School of Medicine, Nashville, Tennessee; 3 4 5 Instituto de Pesquisa Clínica Evandro Chagas, Fundacão Oswaldo Cruz, Rio de Janeiro, Brazil; Fundacion Arriaran, University of Chile School of Medicine, Santiago, Chile; Les Centres GHESKIO, 6 7 8 Port-au-Prince, Haiti; Weill Cornell Medical College, New York, New York; Instituto Hondureño de Seguridad Social, Tegucigalpa, Honduras; Instituto de Medicina Tropical Alexander von Humboldt, Lima, Peru Background. Efavirenz (EFV) and boosted protease inhibitors (bPIs) are still the preferred options for firstline antiretroviral regimens (firstline ART) in Latin America and have comparable short-term efficacy. We assessed the long-term durability and out- comes of patients receiving EFV or bPIs as firstline ART in the Caribbean, Central and South America network for HIV epidemi- ology (CCASAnet). Methods. We included ART-naïve, HIV-positive adults on EFV or bPIs as firstline ART in CCASAnet between 2000 and 2016. We investigated the time from starting until ending firstline ART according to changes of third component for any reason, including toxicity and treatment failure, death, and/or loss to follow-up. Use of a third-line regimen was a secondary outcome. Kaplan-Meier estimators of composite end points were generated. Crude cumulative incidence of events and adjusted hazard ratios (aHRs) were estimated accounting for competing risk events. Results. We included 14 519 patients: 12 898 (89%) started EFV and 1621 (11%) bPIs. The adjusted median years on firstline ART were 4.6 (95% confidence interval [CI], 4.4–4.7) on EFV and 3.8 (95% CI, 3.8–4.0) on bPI (P < .001). Cumulative incidence of firstline ART ending at 10 years of follow-up was 32% (95% CI, 31–33) on EFV and 44% (95% CI, 39–48) on bPI (aHR, 0.88; 95% CI, 0.78–0.97). The cumulative incidence rates of third-line initiation in the bPI-based group were 6% (95% CI, 2.4–9.6) and 2% (95% CI, 1.4–2.2) among the EFV-based group (P < .01). Conclusions. Durability of firstline ART was longer with EFV than with bPIs. EFV-based regimens may continue to be the pre- ferred firstline regimen for our region in the near future due to their high efficacy, relatively low toxicity (especially at lower doses), existence of generic formulations, and ao ff rdability for national programs. Keywords. antiretroviral therapy; durability; HIV; Latin America; nonnucleoside reverse transcriptase inhibitor; protease inhibitor. Life expectancy of people living with HIV (PLWH) increased increased costs or adverse clinical outcomes. Efavirenz (EFV) substantially with the introduction of combined active anti- has been, until now, the preferred option for the third com- retroviral therapy (cART) . Consequently, assessing long- ponent in first antiretroviral regimens (firstline ART) to treat term treatment outcomes and regimen durability has become HIV infection, according to the World Health Organization increasingly relevant. Patients still experience treatment mod- Consolidated Guidelines (WHO)  and several other cART ifications and interruptions for reasons that may include viro- guidelines from Latin American countries [5, 6]. logical failure, adverse events, or poor adherence [2, 3]; and In most clinical trials, before the introduction of integrase treatment modifications or interruptions may be associated with strand transfer inhibitors (INSTIs), efavirenz showed better efficacy or noninferiority in suppressing HIV replication when compared with other regimens [7–9]. Randomized clinical tri- als, however, are typically conducted over 48- to 96-week peri- Received 2 October 2017; editorial decision 22 December 2017; accepted 14 February 2018. ods and evaluate primarily rates of virological suppression in Correspondence: J. G. Sierra-Madero, MD, Departamento de Infectología, Instituto Nacional de Ciencias Médicas y Nutrición, Salvador Zubirán, Vasco de Quiroga # 15, Col. Belisario populations selected with stringent criteria. A recent meta-anal- Domínguez Sección XVI, Delegación Tlalpan; C.p. 14080, CDMX.; México (jsmadero@yahoo. ysis using pooled data of 29 clinical trials evaluated the risk of com). ® death , AIDS progression, and treatment discontinuation Open Forum Infectious Diseases © The Author(s) 2018. Published by Oxford University Press on behalf of Infectious Diseases in ART-naïve patients comparing non-nucleoside reverse tran- Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@ scriptase inhibitors (NNRTIs), including EFV-based regimens, oup.com. DOI: 10.1093/ofid/ofy004 with boosted protease inhibitor (bPI)–based regimens, and it Durability of EFV vs bPI in HIV Patients • OFID • 1 Downloaded from https://academic.oup.com/ofid/article-abstract/5/3/ofy004/4918638 by Ed 'DeepDyve' Gillespie user on 16 March 2018 showed no statistically significant differences between regimens who started their firstline ART between January 1, 2000, in any of the individual and combined outcomes. and June 30, 2016, at sites participating in the Caribbean, Observational studies, in contrast to clinical trials, have the Central and South America network for HIV epidemiology potential to oer va ff luable additional information with respect (CCASAnet) cohort in Argentina, Brazil, Chile, Honduras, to real-life settings, such as durability of the regimen over longer Mexico, Peru, and Haiti. Patients were included if the third periods of time, survival, and loss to follow-up. Cohort studies component of their firstline ART was either EFV or a bPI. Only carried out in Latin America and the Caribbean have evaluated lopinavir/ritonavir and atazanavir/ritonavir were considered the reasons for change of firstline regimen in the short term qualifying bPIs for this study; 15 patients starting darunavir/ [2, 11] and long term , the prevalence of use of third-line regi- ritonavir were excluded. mens , and clinical outcomes such as death, loss to follow-up, and virological suppression aer 1 ft year of cART initiation Definitions and Statistical Analyses [13, 14]. Nevertheless, long-term outcomes on firstline ART and We compared the durability of EFV-based or bPI-based first- other outcomes that are seldom measured in these studies, such line ART by assessing the time from treatment initiation as the probability of requiring a third-line regimen, may have to the end of firstline ART. We defined the end of firstline significant economic and public health relevance. Therefore, in ART as the occurrence of any of the following: (1) changes this study, we compared the durability of firstline ART according in third component for any reason (eg, virological or treat- to the third component in a cohort from Latin America and the ment failure [TF], toxicity, or drug interactions); (2) inter- Caribbean, and explored the effect of the use of EFV vs bPI as a ruption of treatment longer than 180 days; (3) death; or (4) third component of firstline ART on the subsequent need of a loss to follow-up (LTFU), whichever occurred first. Detailed third-line regimen over a long period of time. definitions of specific events including virological failure and LTFU are given in Figure 1. Time in follow-up started on the METHODS date of firstline ART initiation and ended at the date of the first event of interest or the database closing date. (Database Study Design and Study Population closing dates differed by study site and are included in the In this observational, retrospective cohort study, we included Supplemental Material.) ART-naïve adults (18 years or older), living with HIV infection, End of the ﬁrst antiretroviral regimen was deﬁned as the occurrence of any of the following, whichever occurred ﬁrst: 1. Changes in the third component Virological or treatment failure recorded as a reason for ART change; or changes due to reasons such as toxicity, drug-interaction, or unknown reasons which do not include only NRTI changes; or documented virological failure deﬁned as: (1) HIV RNA levels never decreased to <400 copies/mL after 6 months of ART; or (2) Two consecutive measurements of HIV RNA levels >400 copies/mL after a measurement <400 copies/mL; or (3) A single measurement of HIV-RNA levels >1000 copies/mL after a measurement <400 copies/mL. OR 2. Treatment interruption longer than 180 days OR 3. Death OR 4. Lost to follow-up No visits in the year before the closing date of the dataset at each site. Treatment failure was deﬁned as any change in the third component of their ﬁrst ART regimen for patients receiving care in Haiti, where HIV RNA measurement is not routinely performed. Figure 1. Operational definition of end of the first antiretroviral treatment regimen. 2 • OFID • Caro-Vega et al Downloaded from https://academic.oup.com/ofid/article-abstract/5/3/ofy004/4918638 by Ed 'DeepDyve' Gillespie user on 16 March 2018 e m Th edian time on the firstline ART was estimated as the of at least 2 of the above-mentioned drugs in the selected third time corresponding to the 50th percentile taken from a covari- regimen. ate-adjusted Kaplan-Meier survival curve. Kaplan Meier curves In secondary analyses, we performed a cross-sectional ana- were estimated adjusting for covariates at firstline ART initia- lysis at different time points aer ft firstline ART initiation (1, 3, tion using inverse probability of treatment weights, defined as 5, and 10 years) and took snapshots of patient status (remain- the inverse of the predicted probability of the patient getting ing on firstline ART, LTFU, ART interrupted, change of third their observed regimen (EFV vs bPI). These weights were pro- component, or death) by firstline ART group in subgroups of duced by tfi ting a logistic regression model for the probability patients who started ART early enough to reach the potential of starting a bPI regimen as firstline ART based on study site, follow-up time. For example, for the analysis looking at patient date of ART initiation, sex, probable route of HIV transmission, status aer 1 ft year, we only included patients who started ART age at ART initiation, CD4 at ART initiation, and nucleoside at least 1 year before the database closing date for their site, so reverse transcriptase inhibitor (NRTI) backbone of the regimen that all included patients had the chance of being in follow-up at . We evaluated the balance of baseline covariables among least 1 year. When death or LTFU occurred before a given time regimens in the weighted sample estimating standardized mean point, those events were considered the outcome at that time differences; results are available in Figure S8 (Supplementary point even if the patient had presented a prior end of firstline Material). Analyses were repeated among those who were late ART event. A sensitivity analysis was done excluding patients ART initiators (LI, CD4 <200 cells/µL or AIDS-defining event from Haiti because viral load measurements were not routinely [ADE]) and among those who were non-LI (CD4 ≥200 cells/µL performed at that site. and no previous ADE). In secondary analysis, we stratified the Ethical Considerations bPI group by type of PI (ATV/r: atazanavir/ritonavir vs LPV/r: Ethical approval was obtained from institutional review boards lopinavir/ritonavir) to explore whether any differences with at each study site and at Vanderbilt University Medical Center. EFV could be attributed to either. We also performed analyses that did not include death and RESULTS LTFU as part of the outcome, but treated death and LTFU as Characteristics of Study Population competing events. In this second analysis, cumulative incidence We included data on 14 519 patients: 12 898 (89%) started of changing/interrupting firstline ART was estimated, account- ART with an EFV-based regimen and 1621 (11%) with a bPI- ing for competing events, both for the overall cohort and among based regimen. Of these, 920 (57%) started with LPV/r and LI and non-LI. We used a proportional subdistribution hazard 701 (43%) with ATV/r. Patients were followed for a median of regression model to assess the association between clinical and 3.4 years (interquartile range [IQR], 1.5–6.2 years). The median demographic variables on the time to ending firstline ART time in follow-up was slightly longer for the EFV-based group due to changes/interruptions in the third component . We (3.44 years; IQR, 1.5–6.3 years) than for the bPI-based group included sex, age, year of ART initiation, probable route of HIV (3.16 years; IQR, 1.4–5.6 years; P < .001). Demographic and transmission, CD4 count at ART initiation, ADE at ART initi- baseline clinical characteristics are summarized in Table 1. ation, and NRTI backbone of the regimen as covariables and Patients started on EFV-based regimens were on average 2 years stratified by site. CD4 count at ART initiation was the closest older and more likely to be male (76% vs 59%) than patients measurement to the date of ART initiation within 180 days starting a bPI-based regimen. Median CD4 count at ART initi- prior to, or 30 days following, ART start. We calculated e-values ation was lower among patients starting an EFV-based regimen to measure the strength of association on the risk ratio scale (175 vs 217 cells/µL); 61% of patients starting an EFV-based that an unmeasured confounder would need to have with both regimen were late ART initiators (LI) compared with 49% start- the regimen and the change/interruption of third component, ing a bPI-based regimen. There were also differences in the dis- conditional on the measured covariates, to explain away the tribution of the type of NRTI backbone between those starting observed association . EFV- vs bPI-based regimens. The proportion of patients initi- We performed a separate, unadjusted analysis to estimate the ating EFV- vs bPI-based regimens differed between study sites percentage of patients that required the initiation of a third- (Table 1). Comparisons stratifying the bPI regimens are pre- line ART regimen at any time point during follow-up, and the sented in Supplementary Table S1. Briefly, patients started on cumulative incidence of this event by firstline ART accounting ATV/r were older, more frequently male, had higher baseline for the competing events of death and LTFU. The initiation of a CD4 counts, and were more likely to be started with TDF- and third-line regimen was defined as follows: (1) 2 previous ART ABC-based regimens than their counterparts. regimen changes due to treatment failure and at least 1 of the drugs in the selected third regimen was etravirine, raltegravir, Durability of First ART Regimen maraviroc, tipranavir, dolutegravir, or darunavir; or (2) 1 pre- The adjusted median time to the end of the first ART regi- vious ART regimen change due to treatment failure and the use men for any reason was 4.6 years (95% confidence interval Durability of EFV vs bPI in HIV Patients • OFID • 3 Downloaded from https://academic.oup.com/ofid/article-abstract/5/3/ofy004/4918638 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Table 1. Summary of Patient Demographics and Clinical Characteristics by Third Component of First ART Regimen EFV-Based Boosted PI Combined Characteristics (n = 12 898) (n = 1621) (n = 14 519) P Age at ART initiation, y 37 (30–45) 35 (29–44) 37 (30–45) <.001 Male, n (%) 9864 (76) 956 (59) 10 820 (75) <.001 175 (66–289) 217 (79–351) 178 (67–295) <.001 CD4 at ART initiation, cells/µL Missing CD4 at ART, n (%) 1208 (9) 203 (13) 1411 (10) Late ART initiation, n (%) 7863 (61) 804 (49) 8667 (60) <.001 Probable route of HIV transmission, n (%) <.001 Heterosexual 3874 (30) 665 (41) 4539 (31) MSM 4100 (32) 529 (33) 4629 (32) Other 130 (1) 25 (2) 155 (1) Unknown 4794 (37) 402 (25) 5196 (36) NRTI backbone, n (%) <.001 3TC/ABC 483 (4) 126 (8) 609 (4) 3TC/AZT 8234 (64) 858 (53) 9092 (63) 3TC/D4T 744 (6) 60 (4) 804 (6) 3TC/TDF 1754 (14) 351 (22) 2105 (14) FTC/TDF 1683 (13) 226 (14) 1909 (13) Patients per site, n (%) <.001 Argentina 1680 (13) 345 (21) 2025 (14) Brazil 2258 (18) 507 (31) 2765 (19) Chile 1109 (9) 186 (11) 1295 (9) Haiti 3824 (30) 234 (14) 4058 (28) Honduras 800 (6) 26 (2) 826 (6) Mexico 873 (7) 145 (9) 1018 (7) Peru 2354 (18) 178 (11) 2532 (17) Year of ART initiation 2009 (2007–2012) 2011 (2008–2013) <.001 Continuous variables are reported as medians (interquartile range). CCASAnet participating centers per country are: Hospital Fernandez and Centro Medico Huesped in Buenos Aires, Argentina; Instituto de Pesquisa Clinica Evandro Chagas, Fundacão Oswaldo Cruz in Rio de Janeiro, Brazil; Fundación Arriarán in Santiago, Chile; Le Groupe Haitien d’Etude du Sarcome de Kaposi et des Infections Opportunistes in Port-au-Prince, Haiti; Instituto Hondureño de Seguridad Social and Hospital Escuela Universitario in Tegucigalpa, Honduras; Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán in Mexico City, Mexico; and Instituto de Medicina Tropical Alexander von Humboldt in Lima, Peru. Abbreviations: 3TC, lamivudine; ABC, abacavir; AZT, zidovudine; D4T, stavudine; EFV, efavirenz; FTC, emtricitabine; late ART Initiation, CD4 <200 or AIDS-defining event at ART initiation; MSM, men having sex with men; NRTI, nucleoside reverse transcriptase inhibitor; PI, protease inhibitor; TDF, tenofovir. [CI], 4.4–4.7 years) in the EFV-based group vs 3.8 years (95% aer AR ft T initiation were 32% (95% CI, 31%–33%) for those started on EFV and 44% (95% CI, 39%–48%) for those started CI, 3.8–4.0 years; P < .001) in the bPI-based group (Figure 2 on a bPI (P < .001). Among non-LI, cumulative incidence at A). In stratified analysis, the median durations of first ART 10 years was 31% (95% CI, 29%–33%) for EFV and 48% (95% were 4.0 years (95% CI, 3.3–4.5 years) on ATV/r and 2.9 years CI, 37%–59%) for bPI (P < .01); among LI, cumulative inci- (95% CI, 2.4–3.2 years) on LPV/r. Among non-LI, the adjusted dence at 10 years was 33% (95% CI, 32%–34%) for EFV and 42% median duration of the first ART regimen was 4.7 years (95% (95% CI, 37%–47%) for bPI (P < .01) (Figure 3). The adjusted CI, 4.5–5.0 years) for those who started with an EFV-based reg- hazard ratio for changing or interrupting the third component imen vs 3.4 years (95% CI, 3.2–3.6 years) for those who started among patients starting EFV vs bPI was 0.88 (95% CI, 0.78– a bPI-based regimen (P < .01) (Figure 2B). The median time 0.97; P = .016). This point estimate and 95% CI correspond with on first ART was 4.1 years (95% CI, 3.2–6.2 years) on ATV/r e-values of 1.41 and 1.17. Other covariables, including sex, age, and 3.2 years (95% CI, 2.7–3.9 years) on LPV/r. Among LI, the men having sex with men (MSM) route of HIV transmission, median time to the end of the first ART regimen was 4.4 years and CD4 count at firstline ART, were also independently asso- (95% CI, 4.1–4.6 years) for the EFV-based group and 3.6 years ciated with the incidence of changing or interrupting the third (95% CI, 3.6–3.9 years) among those in the bPI-based group component (Table 2). The results of the analysis stratifying bPI (P = .37) (Figure 2C). The median time on first ART was regimen in ATV/r and LPV/r are shown in Table S3. 3.8 years (95% CI, 3.1–4.9 years) on ATV/r and 2.2 years (95% CI, 1.8–3.1 years) on LPV/r. See Figure S1 in the Supplementary Reasons for Ending Firstline ART Material for stratified comparative analysis. The distribution of the reasons for ending the first ART reg- Treating death and loss to follow-up as competing events, the imen differed between the EFV and bPI groups (P < .001) (Table 3). There were 6914/12 898 (53%) events of firstline crude cumulative incidences of ending firstline ART 10 years 4 • OFID • Caro-Vega et al Downloaded from https://academic.oup.com/ofid/article-abstract/5/3/ofy004/4918638 by Ed 'DeepDyve' Gillespie user on 16 March 2018 All, (n=14 519) Non LI, (n = 5852) LI, (n = 8667) 1.0 1.0 1.0 P <.01 P < .01 P = .36 0.8 0.8 0.8 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 0.0 0.0 0.0 02468 10 0246 810 02468 10 Years from start of ART Years from start of ART Years from start of ART Boosted Pl-based, (n =1621) Boosted Pl-based, (n= 817) Boosted Pl-based, (n = 804) EFV-based, (n =12 898) EFV-based, (n = 5035) EFV-based, (n = 7863) Figure 2. Adjusted probability of first antiretroviral regimen (firstline ART) termination in the overall cohort, stratified based on stage of HIV-associated disease at firstline ART start (non-LI vs LI), by group of first treatment regimen (efavirenz vs boosted protease inhibitor). Abbreviations: EFV, efavirenz; LI, patients initiating with CD4 <200 cells/µL or AIDS-defining event; Non-LI, patients initiating with CD4 ≥200 cells/µL and no AIDS-defining event; PI, protease inhibitor . All, (n =14 519) NLI, (n= 5852) LI, (n = 8667) 0.6 0.6 0.6 0.5 0.5 0.5 0.4 0.4 0.4 0.3 0.3 0.3 0.2 0.2 0.2 0.1 0.1 0.1 0.0 0.0 0.0 02468 10 0 246 810 0 246 810 Years from start of ART Years from start of ART Years from start of ART Change EFV Death EFV LTFU PI LTFU EFV Change PI Death PL Figure 3. Cumulative incidence for change of third component of firstline ART using death and loss to follow-up as competing events by treatment group (EFV vs bPI) in the overall cohort (A), and in analysis stratified by stage of HIV-associated disease at firstline ART initiation, for non-late ART initiators (B) or late ART initiators (C) over time. Abbreviations: bPI, boosted protease inhibitor; Change, includes changes in the third component of firstline ART for any reason or interruptions of treatment; EFV, efavirenz; firstline ART, first antiretroviral regimen; LI, patients initiating with CD4 <200 cells/µL or AIDS-defining event; LTFU, lost to follow-up, defined as no visits in the year before the closing date of data set for each site; Non-LI, patients initiating with CD4 >200 cells/µL and no AIDS-defining event. Durability of EFV vs bPI in HIV Patients • OFID • 5 Downloaded from https://academic.oup.com/ofid/article-abstract/5/3/ofy004/4918638 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Probability of outcome Cumulative incidence Probability of outcome Cumulative incidence Cumulative incidence Probability of outcome Table 2. Adjusted Hazard Ratio for Cumulative Incidence of Changes of The distribution of different reasons for ending firstline ART Third Component by First Antiretroviral Regimen at 1, 3, 5, and 10 years is shown in Figure 4. e Th percentage of patients remaining on firstline ART was nearly 5% higher in the aHR (95% CI) P EFV group than in the bPI group during the first 5 years, but EFV vs boosted PI 0.88 (0.78–0.97) .016 only 1% at 10 years (Figure 4). Year of firstline ART 0.99 (0.97–1.00) .05 Male 0.79 (0.72–0.86) <.001 Need of Third-Line Regimen Age (per 10 y) 0.81 (0.78–0.84) <.001 In the analysis of the need to start a third-line regimen, we Route of HIV transmission observed 218 events of third-line initiation: 185 in the EFV- MSM vs heterosexual 0.78 (0.71–0.86) <.001 based group (1.43%) and 33 in the bPI-based group (2.03%; Other vs heterosexual 1.06 (0.79–1.41) .70 P = .07). Using competing events, the cumulative incidence of Unknown vs heterosexual 0.89 (0.77–1.04) .15 CD4 count at ART initiation 0.99 (0.98–0.99) .016 third-line initiation in the bPI-based group was 6% (95% CI, ADE at ART initiation 2.4%–9.6%), and among the EFV-based group it was 2% (95% ADE vs unknown 0.94 (0.83–1.06) .36 CI, 1.4%–2.2%; P < .01) (Figure 5). Non-ADE vs unknown 0.91 (0.82–1.01) .09 NRTI backbone Sensitivity Analysis 3TC/ABC vs 3TC/AZT 0.89 (0.75–1.07) .22 In a sensitivity analysis excluding data of patients receiving care 3TC/D4T vs 3TC/AZT 1.09 (0.94–1.26) .24 in Haiti because they did not routinely measure HIV-1 RNA, FTC/TDF vs 3TC/AZT 0.74 (0.64–0.86) <.001 we included a total of 10 461 patients: 9074 (86.74%) in the 3TC/TDF vs 3TC/AZT 0.82 (0.72–0.93) <.001 EFV group and 1387 (13.26%) in the bPI group. The results Abbreviations: 3TC, lamivudine; ABC, abacavir; ADE, AIDS-defining event; AHR, adjusted were similar to the main analysis, except that the durability of hazard ratio; AZT, zidovudine; D4T, stavudine; EFV, efavirenz; firstline ART, firstline antire- troviral regimen; FTC, emtricitabine; MSM, men having sex with men; NRTI, nucleoside the EFV-based regimen was statistically longer in both LI and reverse transcriptase inhibitor; PI, protease inhibitor; TDF, tenofovir. non-LI. Details are shown in the Supplementary Material. DISCUSSION ART termination in the EFV group: 821 (12%) were deaths, In this large, observational cohort study of people living with 3040 (44%) losses to follow-up, 2894 (42%) changes in HIV and receiving care in Latin America, we found that in ART- third component, and 159 (2%) ART interruptions. In the naïve patients starting treatment, the median duration of EFV- bPI-based group, we observed 928/1621 (57%) events: 77 based regimens was almost a year longer than boosted PI–based (8%) were deaths, 343 (37%) losses to follow-up, 491 (53%) regimens. A lower incidence of changes in the third component changes in third component, and 17 (2%) ART interruptions. for EFV regimens for any reason, rather than the number of The proportion of changes of the third component as firstline deaths and losses to follow-up, seems to explain this difference. ART ending events due to treatment failure was significantly This difference was determined mainly by the significantly higher for bPI-based regimens than for regimens contain- longer duration of EFV-based regimens than those receiving ing EFV in the overall cohort, and for both LI and non-LI lopinavir/ritonavir, and to a lesser extent by differences with (Table 3). Table 3. Number of Patients Ending First Antiretroviral Regimen by Reason of Ending in the Cohort and Stratifying by late ART Initiation All Non-LI LI EFV-Based Boosted PI EFV-Based Boosted PI EFV-Based Boosted PI (n = 12 898), n (%) (n = 1621), n (%) (n = 5035), n (%) (n = 817), n (%) (n = 7863), n (%) (n = 804), n (%) Patients ending firstline 6914 (53) 928 (57) 2378 (42) 449 (50) 4536 (58) 479 (59) ART Reason of ending Death 821 (12) 77 (8) 149 (6) 15 (3) 672 (15) 62 (13) LTFU 3040 (44) 343 (37) 1190 (50) 184 (41) 1850 (41) 159 (33) Changes of third 2894 (42) 491(53) 978 (41) 238 (53) 1916 (42) 253 (53) component Treatment failure 1866 (27) 340 (37) 595 (25) 179 (40) 1271 (28) 161 (34) Other reasons 1028 (15) 155 (17) 383 (16) 59 (13) 645 (14) 92 (19) Percentages for reason of ending are relative to the total number of ending firstline ART. Abbreviations: EFV, efavirenz; firstline ART, first antiretroviral regimen; LI, patients initiating with CD4 <200 cells/µL or AIDS-defining event; LTFU, lost to follow-up, defined as no visits in the year before the closing date of data set in each site; Non-LI, patients initiating with CD4 ≥200 cells/µL and no AIDS-defining event; PI, protease inhibitor. Changes of third component include changes for any reason; in this table, we show the changes due to treatment failure, which could be documented virological failures and virological or treatment failures recorded as reason for change. Other reasons include toxicity and drug interactions, and changes with reason unknown. ART interruption is defined as interruptions longer than 180 days. 6 • OFID • Caro-Vega et al Downloaded from https://academic.oup.com/ofid/article-abstract/5/3/ofy004/4918638 by Ed 'DeepDyve' Gillespie user on 16 March 2018 1 year (n=13 958) 3 years (n =11 687) P < .001 P < .001 1.00 1.00 0.75 0.75 0.50 0.50 0.25 0.25 0.00 0.00 PI (n = 1571) EFV (n =12 387) PI (n = 1262) EFV (n =10 425) ART regimen ART regimen 5 years (n= 7962) 10 years (n = 2173) P < .001 P<.92 1.00 1.00 0.75 0.75 0.50 0.50 0.25 0.25 0.00 0.00 PI (n = 806) EFV (n = 7156) PI (n = 84) EFV (n= 2089) ART regimen ART regimen Outcome Remaining on ﬁrstline ART LTFU Death Change of third component ART interrupted Figure 4. Distribution of individual outcomes by first antiretroviral regimen at 1, 3, 5, and 10 years after ART initiation. Abbreviations: ART interrupted, includes interrup- tions longer than 180 days; Change of third component, includes changes in the third component of firstline ART for any reason; EFV, efavirenz; firstline ART, first antiretroviral regimen; LTFU, lost to follow-up, defined as no visits in the year before the closing date of data set for each site; PI, protease inhibitor. boosted atazanavir. In the main analysis, patients starting EFV by our group reporting that patients who had started bPI reg- without advanced HIV disease had a significantly longer dur- imens were overrepresented as a group among those starting ability of firstline ART. Among patients initiating ART with second-line regimens in CCASAnet centers . advanced disease (CD4 <200 cells/µL or ADE), those starting e diff Th erence in durability among PLWH starting ART with EFV also had a longer durability of firstline ART, although the advanced HIV-associated disease was not statistically signifi- association was no longer statistically significant. At differ- cant, even though clinical trials in this population have found ent time points during follow-up, the proportion of patients significant differences . In a sensitivity analysis excluding remaining in firstline ART was significantly higher among Haiti, however, firstline EFV-based regimens had a longer dur - patients who started EFV. In addition, third-line regimen use ability in both LI and non-LI (see the Supplementary Materials). among subjects starting ART with EFV-based regimens during This discrepancy might be explained by clinical practice at the complete follow-up was significantly lower than among those center in Haiti, where rather than routine viral load measure- starting a bPI-based regimen. ment, CD4 count and clinical status were used to monitor ART This study is consistent with previous clinical trials, obser - efficacy and to direct changes in regimen. In addition, bPI use vational cohort studies, and meta-analyses that separately have in Haiti was extremely unusual, and our analyses may not have compared the short-term efficacy of EFV- vs bPI-based regi- sufficiently adjusted for confounding variables. Other factors mens to suppress HIV replication and time to regimen dis- such as differences in study populations and estimands  and continuation or modification. Overall, treatment failure due to selection bias might also contribute to this discrepancy. virological failure or treatment discontinuation is more com- Our study adds to previous knowledge by showing that, over mon in patients starting ART with bPI regimens than those a 10-year period, in day-to-day routine care, starting treatment starting with EFV [18–21]. Nonetheless, information derived with EFV as the third component of a combined ART regimen from clinical trials is limited because of the short-term fol- had advantages over selected boosted PIs in terms of regimen low-up [19, 20], small numbers of participants [18, 19, 21], and durability in a cohort of patients from 7 Latin American and restrictive inclusion criteria [18–20] that may not allow for wide Caribbean countries. The longer durability of EFV-based regi- applicability in all contexts of routine clinical care . Our mens in this study, explained largely by differences with boosted results are also consistent with previous observations published lopinavir, was attributed to lower rates of treatment failure, which Durability of EFV vs bPI in HIV Patients • OFID • 7 Downloaded from https://academic.oup.com/ofid/article-abstract/5/3/ofy004/4918638 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Percentage of patients Percentage of patients Percentage of patients Percentage of patients regimens for reasons not fully captured in our analyses. It is 0.6 possible that unmeasured variables (ie, confounders) associated with both the choice of firstline ART and subsequent events 0.5 may explain the differences between EFV- and bPI-based regi- 0.4 mens seen in this study. For instance, young women were over- represented among people starting boosted lopinavir-based 0.3 regimens, which were recommended as the firstline regimen for reproductive-age women during the study period in several 0.2 countries [28–32]. While we controlled for sex and age, we were not able to account for pregnancy status at ART initiation, but 0.1 there were only 28 women with end of pregnancy registered as a reason for ending bPI (data not shown), which makes it 0.0 unlikely to have had a strong confounding effect. In addition, 02468 10 because of limited data, we were unable to include tubercu- Years from start of ART losis information in our analyses. In the primary analysis, an unmeasured confounder with an association with both expos- Death PI Death EFV LTFU PI ure and outcome of at least 1.41 (for the point estimate to be LTFU EFV Third-line PI Third-line EFV 1) or 1.17 (for the upper limit of the 95% CI to include 1) would be needed to explain away the observed association. It is quite Figure 5. Cumulative incidence for use of a third-line ART regimen using death possible that such an unmeasured confounder exists, which is a and loss to follow-up as competing events by first regimen treatment group (efa- virenz vs boosted protease inhibitor). Abbreviations: bPI, boosted protease inhibi- limitation of our study. tor; EFV, efavirenz; LTFU, lost to follow-up, defined as no visits in the year before Finally, our study leaves unanswered questions that warrant the closing date of data set in each site; Third-line, use of a third-line regimen, further investigation. Due to the limited number and follow-up defined as (1) 2 previous ART regimen changes due to treatment failure and at least 1 of these drugs in the selected third regimen, etravirine, raltegravir, maraviroc, of patients started on INSTI, we did not compare the durabil- tipranavir, dolutegravir, or darunavir; or (2) 1 previous ART regimen change due to ity of this drug class in our study, even though these regimens treatment failure and the use of at least 2 of the above-mentioned drugs in the are currently recommended as preferred initial ART regimens selected third regimen. in high-income countries, and by the WHO as an alternative to EFV . While our results support continuing use of EFV- could be due to lower toxicity, lower rates of virological failure, based regimens as firstline ART over bPIs, particularly boosted or a combination of both. We consider that our definitions of LPV/r, they do not provide any information comparing EFV durability, which includes a combination of several possible with INSTI. However, they suggest that in cases where EFV outcomes, and of the secondary outcome (need of a third-line is not a valid option, skipping past bPI to Integrase inhibi- regimen) are relevant from a public health and programmatic tor–based therapy may be warranted. Despite the preference perspective because of the increased cost of switching a failing of INSTI in firstline regimens in high-income countries , ART regimen and the use of broader options of active drugs in EFV-based regimens may continue to be preferred for our salvage regimens [12, 24]. In addition, we observed a small but region in the future, due to their high efficacy, relatively low statistically significant reduction in the subsequent need to use toxicity (especially at lower doses) , the existence of generic third-line drugs in the EFV-treated group. The small difference formulations, and most importantly, ao ff rdability for national might be explained partially by the lack of access to third-line programs. drugs in the region during the study period or to practice in the region where patients failing a bPI are more likely to escalate to CONCLUSIONS third-line therapy than those failing EFV. We acknowledge that these results may not be fully repre- In summary, in our study, patients initiating ART with EFV- sentative due to the limited sample of centers and the unique based regimens had longer durability of their first ART regimen characteristics of the participating sites in CCASAnet, which and eventually required the use of a third-line ART regimen less are primarily academic centers. While we have discussed the frequently than patients starting ART with bPI-based regimens. issue of representativeness in other analyses by our group, we Our results are relevant because in our region, as in other devel- are uncertain how our results would vary had a wider variety oping countries where the consolidated WHO recommenda- of centers been included. Nonetheless, previous reports from tions for ART use are currently followed, EFV-based regimens other local and multinational groups in the region usually have are still the preferred options for first ART regimen. Our results been consistent with results reported from our centers [11, 13, strengthen and support current WHO recommendations, as 25–27]. Moreover, patients were likely started on bPI-based well as most regional national guidelines. 8 • OFID • Caro-Vega et al Downloaded from https://academic.oup.com/ofid/article-abstract/5/3/ofy004/4918638 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Cumulative incidence first antiretroviral regimen at 7 sites in the Caribbean, Central and South America Supplementary Data Network. J Acquir Immune Defic Syndr 2016; 71:102–10. Supplementary materials are available at Open Forum Infectious Diseases 4. WHO. The use of antirretroviral drugs for treating and preventing HIV infection. online (http://ofid.oxfordjournals.org): results of the stratified analysis Available at: http://www.who.int/hiv/pub/arv/arv-2016/en/. Accessed 23 August for Boosted-PI as ATV/r and LPV/r vs EFV and the sensitivity analysis to evaluate long-term outcomes among naïve patients initiating with EFV- 5. CENSIDA. Guía de Manejo Antirretroviral de las Personas con VIH. Available at: based and Boosted-PI regimes, excluding Haiti. http://www.censida.salud.gob.mx/interior/guiasmanuales.html. Accessed 4 July 6. SADI. Consenso argentino de terapía antirretroviral 2014–2015. Available Acknowledgments at: https://www.dropbox.com/s/xol78s0j5mkmdf2/consenso%202014–2015. Authors’ contributions. Study concept and design: Y.C.V., P.F.B.Z., pdf ?dl=0. Accessed 4th July 2017. B.C.R., J.S.M., B.E.S. Acquisition of data: B.C.R., B.G., M.W., J.W.P., D.P., 7. Daar ES, Tierney C, Fischl MA, et al; AIDS Clinical Trials Group Study A5202 E.G., C.M. Team. Atazanavir plus ritonavir or efavirenz as part of a 3-drug regimen for initial Analysis and interpretation of data: Y.C.V., P.F.B.Z., B.E.S. Drafting of the treatment of HIV-1. Ann Intern Med 2011; 154:445–56. manuscript: Y.C.V., P.F.B.Z., B.C.R., J.S.M. Critical revision of the manu- 8. Echeverría P, Negredo E, Carosi G, et al. Similar antiviral efficacy and tolera- bility between efavirenz and lopinavir/ritonavir, administered with abacavir/ script for important intellectual content: Y.C.V., P.F.B.Z., B.C.R., B.E.S., B.G., lamivudine (Kivexa), in antiretroviral-naïve patients: a 48-week, multicentre, M.W., J.W.P., D.P., E.G., C.M., J.S.M. Statistical analysis: Y.C.V., P.F.B.Z., randomized study (Lake Study). Antiviral Res 2010; 85:403–8. B.E.S. All the authors approved the final version of the manuscript and 9. Miró JM, Manzardo C, Pich J, et al; Advanz Study Group. Immune reconstitution agreed to be accountable for all aspects of the work in ensuring that ques- in severely immunosuppressed antiretroviral-naive HIV type 1-infected patients tions related to the accuracy or integrity of any part of the work are appro- using a nonnucleoside reverse transcriptase inhibitor-based or a boosted protease priately investigated and resolved. inhibitor-based antiretroviral regimen: three-year results (The Advanz Trial): a Financial support. This work was supported by the National Institutes randomized, controlled trial. AIDS Res Hum Retroviruses 2010; 26:747–57. of Heatlh–funded Caribbean, Central and South America network for 10. Borges ÁH, Lundh A, Tendal B, et al. Nonnucleoside reverse-transcriptase inhib- HIV epidemiology (CCASAnet), a member cohort of the International itor- vs ritonavir-boosted protease inhibitor-based regimens for initial treatment Epidemiologic Databases to Evaluate AIDS (leDEA; U01AI069923). This of HIV Infection: a systematic review and metaanalysis of randomized trials. Clin Infect Dis 2016; 63:268–80. award is funded by the following institutes: Eunice Kennedy Shriver National 11. Tuboi SH, Schechter M, McGowan CC, et al. Mortality during the first year Institute of Child Health and Human Development (NICHD), Office of The of potent antiretroviral therapy in HIV-1-infected patients in 7 sites through- Director (OD), National Institutes of Health, National Institute of Allergy out Latin America and the Caribbean. J Acquir Immune Defic Syndr 2009; and Infectious Diseases (NIAID), National Cancer Institute (NCI), and the 51:615–23. National Institute of Mental Health (NIMH). 12. Cesar C, Shepherd BE, Jenkins CA, et al; Caribbean, Central and South America Potential coni fl cts of interest. Y.C.V., P.F.B.Z., B.E.S., B.G., J.W.P., D.P., Network for HIV Epidemiology (CCASAnet). Use of third line antiretroviral E.G., and C.M. declare that they have no competing interests. J.S.M. has therapy in Latin America. PLoS One 2014; 9:e106887. received lecture fees, sponsorship, and honoraria from Gilead, Stendhal, 13. Angriman F, Belloso WH, Sierra-Madero J, et al. Clinical outcomes of first-line Abbvie, ViiV, Janssen, and MSD. M.W. has received congress attendance antiretroviral therapy in Latin America: analysis from the LATINA retrospective cohort study. Int J STD AIDS 2016; 27:118–26. sponsorships from Gilead and MSD. B.C.R. has received lecture fees, spon- 14. Cesar C, Koethe JR, Giganti MJ, et al; Caribbean, Central and South America sorship, and honoraria from Gilead, Stendhal, Abbvie, Janssen, and MSD. Network for HIV epidemiology (CCASAnet) and North American AIDS Cohort e C Th CASAnet includes the following sites: Fundación Huesped, Collaboration on Research and Design (NA-ACCORD). Health outcomes among Argentina: Pedro Cahn, Carina Cesar, Valeria Fink, Omar Sued, Emanuel HIV-positive Latinos initiating antiretroviral therapy in North America versus Dell’Isola, Hector Perez, Jose Valiente, Cleyton Yamamoto. Instituto Nacional Central and South America. J Int AIDS Soc 2016; 19:20684. de Infectologia-Fiocruz, Brazil: Beatriz Grinsztejn, Valdilea Veloso, Paula 15. Austin PC, Stuart EA. Moving towards best practice when using inverse probabil- Luz, Raquel de Boni, Sandra Cardoso Wagner, Ruth Friedman, Ronaldo ity of treatment weighting (IPTW) using the propensity score to estimate causal Moreira. Universidade Federal de Minas Gerais, Brazil: Jorge Pinto, Flavia treatment effects in observational studies. Stat Med 2015; 34:3661–79. Ferreira, Marcelle Maia. Universidade Federal de São Paulo, Brazil: Regina 16. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a com- peting risk. J Am Stat Assoc 1999; 94:496–509. Celia de Menezes Succi, Daisy Maria Machado, Aida de Fátima Barbosa 17. VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introduc- Gouvêa. Fundación Arriarán, Chile: Marcelo Wolff, Claudia P. Cortes, Maria ing the E-value. Ann Intern Med 2017; 167(4):268–74. Fernanda Rodriguez, Gladys Allende. Les Centres GHESKIO, Haiti: Jean 18. Riddler SA, Haubrich R, DiRienzo AG, et al; AIDS Clinical Trials Group Study William Pape, Vanessa Rouzier, Adias Marcelin, Christian Perodin. Hospital A5142 Team. Class-sparing regimens for initial treatment of HIV-1 infection. N Escuela Universitario, Honduras: Marco Tulio Luque. Instituto Hondureño Engl J Med 2008; 358:2095–106. de Seguridad Social, Honduras: Denis Padgett. Instituto Nacional de 19. Sierra-Madero J, Villasis-Keever A, Méndez P, et al. Prospective, randomized, Ciencias Médicas y Nutrición Salvador Zubirán, México: Juan open label trial of Efavirenz vs Lopinavir/Ritonavir in HIV+ treatment-naive sub- Sierra Madero, Brenda Crabtree Ramirez, Pablo F. Belaunzaran- jects with CD4+<200 cell/mm3 in Mexico. J Acquir Immune Defic Syndr 2010; Zamudio, Yanink Caro Vega. Instituto de Medicina Tropical Alexander von 53:582–8. 20. Juday T, Grimm K, Zoe-Powers A, et al. A retrospective study of HIV antiretro- Humboldt, Peru: Eduardo Gotuzzo, Fernando Mejia, Gabriela Carriquiry. viral treatment persistence in a commercially insured population in the United Vanderbilt University Medical Center, USA: Catherine C. McGowan, Bryan States. AIDS Care 2011; 23:1154–62. E. Shepherd, Timothy Sterling, Karu Jayathilake, Anna K. Person, Peter 21. Esposito A, Floridia M, d’Ettorre G, et al. Rate and determinants of treatment F. Rebeiro, Mark Giganti, Jessica Castilho, Stephany N. Duda, Fernanda response to different antiretroviral combination strategies in subjects presenting Maruri, Hilary Vansell. at HIV-1 diagnosis with advanced disease. BMC Infect Dis 2011; 11:341. 22. López-Martínez A, O’Brien NM, Caro-Vega Y, et al. Different baseline character- References istics and different outcomes of HIV-infected patients receiving HAART through 1. Palella FJ Jr, Delaney KM, Moorman AC, et al. Declining morbidity and mortality clinical trials compared with routine care in Mexico. J Acquir Immune Defic among patients with advanced human immunodeficiency virus infection. HIV Syndr 2012; 59:155–60. Outpatient Study Investigators. N Engl J Med 1998; 338:853–60. 23. Hernán MA, Hernández-Díaz S. Beyond the intention-to-treat in comparative 2. Cesar C, Shepherd BE, Krolewiecki AJ, et al; Caribbean, Central and South effectiveness research. Clin Trials 2012; 9:48–55. America Network for HIV Research (CCASAnet) Collaboration of the 24. Caro JJ, O’Brien JA, Migliaccio-Walle K, Raggio G. Economic analysis of ini- International Epidemiologic Databases to Evaluate AIDS (IeDEA) Program. tial HIV treatment. Efavirenz- versus indinavir-containing triple therapy. Rates and reasons for early change of first HAART in HIV-1-infected patients in Pharmacoeconomics 2001; 19:95–104. 7 sites throughout the Caribbean and Latin America. PLoS One 2010; 5:e10490. 25. Crabtree-Ramírez B, Caro-Vega Y, Shepherd BE, et al; CCASAnet Team. Cross- 3. Wolff M, Shepherd BE, Cortés C, et al; Caribbean, Central and South America sectional analysis of late HAART initiation in Latin America and the Caribbean: Network for HIV Epidemiology. Clinical and virologic outcomes after changes in late testers and late presenters. PLoS One 2011; 6:e20272. Durability of EFV vs bPI in HIV Patients • OFID • 9 Downloaded from https://academic.oup.com/ofid/article-abstract/5/3/ofy004/4918638 by Ed 'DeepDyve' Gillespie user on 16 March 2018 26. Celi AP, Greco M, Martinez E, et al; for the Latin American HIV Workshop Study humana. Available at: http://www.who.int/hiv/pub/guidelines/peru_art.pdf. Group. Presentation to care with advanced HIV disease is still a problem in Latin Accessed 27 November 2017. America. J Int AIDS Soc 2016; 19(Suppl 2S1):35. 31. MINSAL. Chile. Guía Clínica Sindrome de Inmunodeficiencia adquirida VIH/ 27. Magis-Rodríguez CL, Villafuerte-García A, Cruz-Flores RA, Uribe-Zúñiga P. SIDA. Available at: http://www.who.int/hiv/pub/guidelines/chile_art.pdf. Inicio tardío de terapia antirretroviral en México. Salud Publica Mex 2015; Accessed 27 November 2017. 57(Suppl 2): S127–S34. 32. MSPP. Manuel de Normes de Prise en Charge Clinique et Thérapeutique des 28. Ministério da Saúde. Recomendacoes para Terapia Antiretroviral en Adultos Adultes et Adolescents Vivant avec le VIH. Available at: http://www.who.int/hiv/ infectados pelo HIV. Available at: http://www.who.int/hiv/pub/guidelines/bra- pub/guidelines/haiti_art.pdf. Accessed 27 November 2017. zil_art.pdf. Accessed 27 November 2017. 33. WHO. Transition to new antiretrovirals in HIV programmes. Available at: http:// 29. Ministerio de Salud. Manejo de los pacientes adultos con infección por VIH. apps.who.int/iris/bitstream/10665/255888/1/WHO-HIV-2017.20-eng.pdf ?ua=1. Available at: http://www.who.int/hiv/pub/guidelines/argentina_art.pdf. Accessed Accessed 27 November 2017. 27 November 2017. 34. Carey D, Puls R, Amin J, et al. Efficacy and safety of efavirenz 400 mg daily versus 30. Ministerio de Salud. Peru. Norma técnica para el tratamiento Antiretroviral de 600 mg daily: 96-week data from the randomised, double-blind, placebo-con- gran actividad-TARGA en adultos infectados por el virus de la inmunodeficiencia trolled, non-inferiority ENCORE1 study. Lancet Infect Dis 2015; 15:793–802. 10 • OFID • Caro-Vega et al Downloaded from https://academic.oup.com/ofid/article-abstract/5/3/ofy004/4918638 by Ed 'DeepDyve' Gillespie user on 16 March 2018
Open Forum Infectious Diseases – Oxford University Press
Published: Mar 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera