Articles Determinants of long-term survival in late HIV presenters: The prospective PISCIS cohort study Raquel Martin-Iguacel,a,b* Juliana Reyes-Uruen ~a,a Andreu Bruguera,a Jordi Aceito n,a Yesika Díaz,a Sergio Moreno-Fornes,a Pere Domingo, Joaquín Burgos-Cibrian, Juan Manuel Tiraboschi, Isik Somuncu Johansen,b Hortensia Alvarez, c d e  f ,g,h Josep M Miro a,i,j,k l Jordi Casabona, and Josep M Llibre, * PISCIS study group a Centre of Epidemiological Studies of HIV/AIDS and STI of Catalonia (CEEISCAT), Health Department, Generalitat de Catalunya, Badalona, Spain b Department of Infectious Diseases, Odense University Hospital, Odense, Denmark c Infectious Diseases Unit, Hospital Universitari de la Santa Creu i Sant Pau, Barcelona, Spain d Department of Infectious Diseases, Hospital Universitari de la Vall d’Hebron, Barcelona, Spain e Department of Internal Medicine and Infectious Diseases, Hospital Universitari de Bellvitge, Hospitalet de Llobregat, Barcelona, Spain f Infectious Diseases Unit, University Hospital of Ferrol, Spain g Hospital Clínic-Institut d’Investigacions Biomediques August Pi i Sunyer, University of Barcelona, Barcelona, Spain h CIBERINFEC, Instituto de Salud Carlos III, Madrid, Spain i Department of Paediatrics, Obstetrics and Gynecology and Preventive Medicine, Universitat Auto noma de Barcelona, Badalona, Spain j Fundacio  Institut D’investigacio  En Ci encies De La Salut Germans Trias I Pujol (IGTP), Badalona, Spain k Spanish Consortium for Research on Epidemiology and Public Health (CIBERESP), Instituto de Salud Carlos III, Madrid, Spain l Infectious Diseases Department and Fight Infections Foundation, University Hospital Germans Trias i Pujol, Badalona, Barcelona, Spain Summary Background Late HIV diagnosis (i.e CD4≤350 cells/µL) is associated with poorer outcomes. However, determinants eClinicalMedicine of long-term mortality and factors influencing immune recovery within the first years after antiretroviral treatment 2022;52: 101600 (ART) initiation are poorly defined. Published online xxx https://doi.org/10.1016/j. eclinm.2022.101600 Methods From PISCIS cohort, we included all HIV-positive adults, two-year survivors after initiating ART between 2005−2019. The primary outcome was all-cause mortality according to the two-year CD4 count. We used Poisson regression. The secondary outcome was incomplete immune recovery (i.e., two-year CD4<500 cells/µL). We used logistic regression and propensity score matching. Findings We included 2,719 participants (16593¢1 person-years): 1441 (53%) late presenters (LP) and 1278 non-LP (1145 non-LP with two-year CD4 count >500 cells/µL, reference population). Overall, 113 patients (4¢2%) died. Mor- tality was higher among LP with two-year CD4 count 200−500 cells/µL (aMRR 1¢95[95%CI:1¢06-3¢61]) or <200 cells/µL (aMRR 4¢59[2¢25-9¢37]). Conversely, no differences were observed in participants with two-year CD4 counts >500 cells/µL, regardless of being initially LP or non-LP (aMRR 1¢05[0¢50-2¢21]). Mortality rates within each two-year CD4 strata were not affected by the initial CD4 count at ART initiation (test-interaction, p = 0¢48). The stronger factor influencing immune recovery was the CD4 count at ART initiation. First-line integrase-inhibitor-(INSTI)-based regimens were associated with reduced mortality compared to other regimens (aMRR 0¢54[0¢31-0¢93]) and reduced risk of incom- plete immune recovery in LP (aOR 0¢70[0¢52-0¢95]). Interpretation Two-year immune recovery is a good early predictor of long-term mortality in LP after surviving the first high-risk 2 years. Nearly half experienced a favorable immune recovery with a life expectancy similar to non-LP. INSTI-based regimens were associated with higher rates of successful immune recovery and better survival com- pared to non-INSTI regimens. Funding Southern-Denmark University, Danish AIDS-foundation, and Region of Southern Denmark. *Corresponding authors. E-mail addresses: raquel@bisaurin.org (R. Martin-Iguacel), jmllibre@flsida.org (J.M. Llibre). www.thelancet.com Vol 52 Month , 2022 1 Articles Copyright Ó 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) Keywords: HIV; Late presenters; Delayed HIV diagnosis; Immune recovery; Immune response; Integrase inhibi- tors; Mortality treatment with CD4 counts ≥350 cells/mL have life Research in context expectancies close to the general population.1−8 Still, despite the enormous advances achieved in the field, Evidence before this study approximately half of PLWH in Western countries are diagnosed at a late disease stage, defined as CD4 Three cohort studies and two randomized controlled tri- cell count below 350 cells/mL or presentation with an als (RCT) have indicated that integrase strand transfer AIDS-defining event, regardless of the CD4 cell inhibitors (INSTI)-based regimens might offer a benefit in immune recovery compared to other ART regimens. count.9 However, these cohort studies included both late and This late HIV diagnosis significantly reduces life non-late presenters, and the last is under-represented expectancy while increasing the frequency of HIV- in RCT. Knowledge on determinants of long-term mor- related comorbidities,10−14 it is associated with higher tality and immune recovery in late presenters is scant. health care costs,15 and onward transmission in the population.16,17 Furthermore, 10−29% of PLWH pre- Added value of this study senting late to care fail to achieve a satisfactory Our results provide new evidence that ART initiation immune recovery despite virologically effective ART. with an INSTI-based regimen is associated with a more These patients, categorized as “immunological non- favorable immune response in late presenters and with responders”, have higher mortality and a higher risk improved long-term survival especially compared to for AIDS- and non-AIDS-related morbidity.12,18,19 protease inhibitors. Various observational studies of late presenters Besides, PLWH surviving the first two years after ART (LP) have reported high mortality rates within the initiation and reaching a CD4 cell count threshold above first year after ART initiation, mainly due to 500 cells/mL achieved similar survival rates to non-late advanced AIDS-defining conditions present at that presenters, irrespective of their baseline CD4 count at time, which tends to decrease after this period.20,21 ART initiation. Thus, ART-associated immune recovery at two years was a good early predictor of long-term Thus, CD4 counts at ART initiation, might not pre- mortality. cisely predict long-term survival in these patients as the estimation might be distorted by this initial Implications of all the available evidence high-mortality period. This study explores the performance of an early The long-term survival prognosis for late presenters can be predictor for long-term survival after ART initiation, assessed upon immune recovery seen two years after ART initiation. Those achieving an early favorable response skipping this initial high-risk time. Furthermore, the improve their life expectancy to the same level as non-late influence of the first-line ART on the clinical trajec- presenters, thus narrowing the gap between late and non- tories of LP is poorly understood. Various studies late presenters. INSTI-based ART is associated with a more have described that first-line ART with integrase favorable immune response and significantly improved strand-transfer inhibitor (INSTI)-based regimens survival, supporting the growing body of evidence in clini- could be associated with an improved immune cal trials. Based on all knowledge available, health care pro- response with greater CD4 gain.22−25 However, how grams should consider using two-year CD4 cell counts these first-line regimens influence adequate immune after ART initiation as an early prognostic factor for long- recovery and long-term survival in LP has not been term survival in late presenters. Also, INSTI-based regimens assessed. should be considered the preferred treatment choice, par- ticularly in late HIV presenters. In this prospective multicentre, ongoing cohort of PLWH in our area, we aimed to investigate whether CD4 cell counts 2 years after ART initiation was a good early predictor of long-term survival in LP com- pared with a non-LP group with an optimal two-year Introduction immune response and explored determinants of sur- HIV-associated morbidity and mortality have vival in these patients, including the potential impact decreased dramatically since the introduction of anti- of INSTI as first-line regimen. Furthermore, we retroviral therapy (ART) in 1996. Today, people liv- assessed the predictors of adequate immune recovery ing with HIV (PLWH) timely diagnosed and starting within this period. 2 www.thelancet.com Vol 52 Month , 2022 Articles Methods year survivors with active follow-up and available CD4 cell count at both timepoints, ART initiation and two Study design and data sources years after treatment start. This is a multicentre, prospective, population-based lon- Patients were classified according to the CD4 cell gitudinal analysis conducted in Catalonia, an autono- count at ART initiation into late HIV presenters (i.e., mous region located in north-easter Spain. On January CD4 count ≤350 cells/mL) and non-LP (i.e., CD4 count 1, 2021, Catalonia had a population of 7¢7 million citi- >350 cells/mL). LP were further stratified according to zens and an estimated HIV prevalence among the adult their CD4 count at two different time points: at ART ini- population of 0¢4%.26 The Catalan healthcare system tiation (<100 cells/mL, 100−199 cells/mL, and 200−350 provides universal, tax-funded healthcare to all citizens. cells/mL) and at 2 years after ART initiation (<200 ART is provided free-of-charge. cells/mL, 200−500 cells/mL, and >500 cells/mL). Non- Data were retrieved from the PISCIS Cohort project LP were also stratified according to their immune recov- (Catalonian and Balearic Islands HIV cohort) and PAD- ery two years after ART initiation (≤500 cells/mL and RIS (Public Data Analysis for Health Research and >500 cells/mL). Innovation Program). For this study, only data from the Catalan PISCIS cohort were used. Briefly, PISCIS is an ongoing, prospective, multicentre, population-based Outcomes cohort ongoing from 1998 that includes all PLWH aged The primary outcome was time to all-cause mortality, ≥16 years followed in one of the 16 collaborating hospi- setting the baseline time at two years after ART initia- tals from Catalonia, representing 84% of all diagnosed tion. We chose a two-year time-point expecting to get rid PLWH.27 Data are updated yearly and include demo- of the bias introduced in the analysis by initial mortality graphics, date of HIV diagnosis, AIDS-defining events, related to baseline conditions in this cohort of PLWH ART, and measurement of CD4 cell count and plasma with advanced HIV infection. The outcome was HIV-RNA over time. PADRIS is a central research-ori- assessed for the different groups of LP based on the ented data repository which gathers, and cross matches CD4 count immune recovery two years after ART initia- real-world health data generated by the public health tion; non-LP with CD4 >500 cells/mL two years after system of Catalonia (SISCAT), provided by the Catalan ART initiation was used as the reference population. Agency for Health Quality and Evaluation (AQuAS). The secondary outcome was incomplete immune The database includes comorbidity data from hospital recovery two years after ART initiation, defined as CD4 discharge diagnoses and primary health care from 2005 cell count ≤500 cells/mL. This was assessed in a nested according to the International Classification of Disease case-control study design. 10th revision (ICD-10).28 Mortality data was obtained cross matching the information with PISCIS and PAD- RIS databases and the national mortality registry. Statistical analysis Continuous variables were described as the median and Ethics approval the interquartile range (IQR), whereas categorical varia- The PISCIS cohort has been approved by the ethics bles were presented as the frequency and percentage committee of the coordinating centre, and patient data over available data. Missing data were not imputed extraction is allowed by the 203/2015 Decree from the because the essential variables for the analysis (i.e., Catalan Health Department. PISCIS data is owned by CD4 count available at ART initiation and 2 years after) each individual patient, and data is eliminated if were among the selection criteria for the study popula- requested by the hospital or by a patient. All data is tion. We used the chi-squared test, the t-test, and the pseudo-anonymized before arriving at the coordinating Mann−Whitney U test to compare variables between center, and confidentiality is guaranteed in accordance late and non-LP. For polycothomous categorical varia- with the provisions of the Regulation (EU) 2016/679 of bles with a statistically significant chi-square test of the European Parliament and of the Council of 27 April homogeneity, we further provided a significance test of 2016 on the protection of natural persons regarding the the difference between proportions across categories of processing of personal data and on the free movement the variable between LP and non-LP. of such data and the new national Organic Law 3/2018 The survival analysis considered all-cause death, of December 5 on Protection of Personal Data and Digi- loss-to-follow-up, or end of the observation period, tal Rights. whichever occurred first, between the 2-year time point after the date of ART initiation to June 30, 2021. The cumulative survival of PLWH in the different groups of Study population two-year CD4 cell count was compared using the Log- We screened the PISCIS cohort for treatment na€ıve rank test. We used a Poisson regression analysis to con- PLWH ≥18 years, who initiated ART between January 1, trol for the following potential confounders and to 2005, and June 30, 2019. Of these, we included all 2- explore for prognostic factors for survival: age, gender, www.thelancet.com Vol 52 Month , 2022 3 Articles HIV risk group (men who have sex with men (MSM), Role of funding heterosexual men, women, history of injection drug use This work was supported by scholarships from the Uni- (IDU), and other/unknown), region of birth (Spain, versity of Southern Denmark, the Danish AIDS founda- Europe excluding Spain, Africa, America, Asia), two- tion, and Public Regional Funds from the Region of year CD4 cell count recovery (<200, 200−500, and Southern Denmark. The study was investigator-driven >500 cells/mL), two-year HIV viral load (>200 vs ≤ 200 and thus independent of any pharmaceutical company. copies/mL), modified Charlson comorbidity index (CCI) The funding sources were not involved in study design, at two years after ART initiation, prior AIDS-defining data collection, analyses, report writing, or decision to disease at two years after ART initiation, time-updated submit the paper. calendar period (2005−2009, 2010−2014, 2015−2019, according to pre-INSTI-available period, INSTI-available period, and change in guidelines to immediate ART ini- Results tiation,29 respectively), INSTI-based regimen as ART regimen initiated in the first two years, and socioeco- Study participants nomic level based on AQuAS composite socioeconomic From the PISCIS cohort, we identified 4036 PLWH indicator used to classify the primary health areas (no starting ART between 2005 and 2019. Of them, 2719 economic deprivation, mild economic deprivation, and (67¢4%) were eligible for the analysis, giving rise to moderate/severe economic deprivation).30 We provide 16593¢1 person-years of follow-up (median follow-up 5¢ mortality rate (MR) and MR ratios (MRR) and a 95% 8 years [IQR 3¢2-8.9]) (Figure 1). Table S2 summarizes confidence interval (CI). and compares the characteristics of included and CD4 cell counts at ART initiation and two years excluded participants. Of the analysis population, 1441 after were defined as the closest value to the index (53%) were LP and 1,278 were non-LP. date, with a time window of 12 months before and 3 Table 1 summarizes the demographic and clinical months after ART initiation and 3 months before characteristics of individuals included in the analysis and 12 months after for the time point at two years population, according to the defined CD4 count groups. after ART initiation. The CCI was modified accord- Compared with non-LP, LP were more frequently older, ing to the information available on comorbidities heterosexual men, women, individuals with a history of from the PADRIS and excluding AIDS defining IDU, of African origin, with lower educational level and events, which was assessed separately (Table S1, Sup- higher economic deprivation. The percentage of late plementary data). HIV presentation decreased in time from 77% in 2005 We conducted a sensitivity analysis using propensity −2009 to 41% in 2015−2020 (p<0¢001). Among LP, score matching (1:1 nearest-neighbor matching with cal- 27% initiated ART with a CD4 cell count <100 cells/µL, iper 0¢2 on each sample) for gender, age, CD4 cell 21¢8% with 100−199 cells/µL and 51¢6% with 200−350 count, modified CCI, and AIDS-defining event, all at cells/µL, and 13¢2% had an AIDS-defining event at base- ART initiation, to assess the impact of initiation of line. Six hundred thirty-seven (44¢2%) of late presenters INSTI as first- line regimen on long-term mortality. The achieved CD4 cell counts >500 cells/µL two years after application of these probabilities to the study population ART start. The comparator population of non-LP with created a subpopulation in which these potential con- two-year CD4 cell count >500 cells/µL (n =1145) had a founders were more equally distributed across both median two-year CD4 count of 837 (IQR: 697-1030). groups. The same analysis was performed for the whole database population of individuals who initiated ART, regardless of survival in the first two years or availability Mortality risk of CD4 counts at 2 years after ART. Overall, 113 (4¢2%) died during the study period (crude To identify risk factors in the group of LP associated all-cause MR 6¢8/1000 person-years; 95%CI: 5¢7-8¢2). with incomplete immune recovery 2 years after ART ini- Compared with non-LP with two-year CD4 count >500 tiation (defined as CD4 cell count ≤500 cells/mL), we cells/µL (reference population), LP with two-year CD4 performed a logistic regression analysis with propensity count ≤500 cells/µL had higher mortality (adjusted score matching on age (continuous), gender, CD4 cell MRR 1¢95 [1¢06-3¢61] and 4¢59 [2¢25-9¢37] in the groups count at ART initiation (<100 cells/mL, 100−199 cells/ of CD4 200−500 cells/µL and <200 cells/µL, respec- mL, and 200−350 cells/mL)) and modified CCI. We tively). Conversely, no significant differences were assessed the following covariates at ART initiation: age, observed between the adjusted MR of non-LP and LP gender, CD4 cell count, INSTI as first ART regimen, with two-year CD4 count >500 cells/µL (1¢05 [0¢50-2¢ modified CCI, AIDS-defining event within 90 days after 21]) (Table 2 and Table S3). Within the individual strata ART initiation. We provide the odds ratio (OR) and a of immune-recovery (two-year CD4 cell count >500, 95% CI. >350−500, 200−350, <200 cells/µL), MR were not sig- All analyses were conducted using the STATA soft- nificantly different when stratifying by initial CD4 cell ware (v.16; Stata Corp, College Station, TX, USA). count at ART initiation (>350, 200−350, 100−199, 4 www.thelancet.com Vol 52 Month , 2022 Articles Figure 1. Flowchart of cohort construction. <100 cells/µL) (test for interaction p = 0¢48) (Figure 2, 06-1¢12], respectively). However, the sensitivity analysis Table S4). HIV viral load >200 copies/mL two years with propensity score matching, conducted to address after ART initiation was also associated with increased the potential channeling bias, did show a significant mortality risk (1¢78 [0¢99-3¢19]). Other factors associated benefit of INSTI versus non-INSTI regimens (0¢48 [0¢ with increased mortality according to the adjusted anal- 25-0¢90]) (Table S5). We did a second sensitivity analysis ysis included a modified CCI score at ART start ≥4 ver- including all individuals starting ART regardless of survival sus 0 and HIV risk group (i.e., heterosexual men and the first two years or availability of CD4 cell count two years IDU versus MSM). after ART. In line with the previous analysis, the sensitivity INSTI-based regimens as first-line ART were associ- analysis showed that INSTI-based first-line regimens were ated with lower mortality than non-INSTI regimens (0¢ associated with a survival benefit when compared to prote- 54 [0¢31-0¢93]). There was an improved survival trend ase inhibitor (PI) (0¢41 [0¢23-0¢72]) but not versus non- when considering LP and non-LP separately, but it was nucleoside reverse transcriptase inhibitors (NNRTI (0¢88 not statistically significant (0¢61 [0¢33-1¢13] and 0¢25 [0¢ [0¢48-1¢61]), Table S5). www.thelancet.com Vol 52 Month , 2022 5 6 Articles Study population according to CD4 cell count at ART initiation Study population according to immune recovery 2 years after ART initiation Non-late Late p-value Late presenters Non-late presenters presenters n =1441 presenters n =1278 n =1441 (47¢0%) (53¢0%) CD4 cell count (cells/µ) CD4 cell count (cells/µ) >500 >350-500 ≥200-350 <200 >500 cells/µL n =637 (44¢2%) n =374 n =304 n =126 n =1145 (26¢0%) (21¢1%) (8¢7%) Male, n (%) 1122 (87¢8) 1202 (83¢4) 0¢001 549 (86¢2) 314 (84¢0) 238 (78¢3) 101 (80¢2) 1022 (89¢3)) Age at ART initiation (years), median (IQR) 35 (29-42) 38 (32-45) <0¢001 36 (30-43) 39 (32-46) 41 (35-48) 40 (34-48) 35 (29-42) Birth area, n (%) 0¢015 Spain 598 (46¢8) 743 (51¢6) 0¢013 330 (51¢8) 180 (48¢1) 163 (53¢6) 70 (55¢6) 531 (46¢4) Europe 124 (9¢7) 97 (6¢3) 0¢005 47 (7¢4) 31 (8¢3) 15 (4¢9) 4 (3¢2) 111 (9¢7) Africa 43 (3¢4) 72 (5¢0) 0¢035 18 (2¢8) 20 (5¢4) 22 (7¢2) 12 (9¢5) 36 (3¢1) America 249 (19¢5) 289 (20¢1) 0¢71 141 (22¢1) 81 (21¢7) 49 (16¢1) 18 (14¢3) 221 (19¢3) Asia 14 (1¢1) 15 (1¢0) 0¢89 10 (1¢6) 3 (0¢8) 0 2 (1¢6) 14 (1¢2) Unknown 250 (19¢6) 225 (15¢6) 0¢007 91 (14¢3) 59 (15¢8) 55 (18¢1) 20 (15¢9) 232 (20¢3) HIV risk group, n (%) <0¢001 MSM 910 (71¢2) 742 (51¢5) <0¢001 415 (65¢2) 198 (52¢9) 94 (30¢9) 35 (27¢8) 842 (73¢5) Heterosexual men 127 (9¢9) 266 (18¢5) <0¢001 85 (13¢3) 65 (17¢4) 85 (28¢0) 31 (24¢6) 109 (9¢5) Women 128 (10¢0) 195 (13¢5) 0¢005 74 (11¢6) 51 (13¢6) 54 (17¢8) 16 (12¢7) 103 (9¢0) IDU 56 (4¢4) 131 (9¢1) <0¢001 26 (4¢1) 38 (10¢2) 36 (11¢8) 31 (24¢6) 44 (3¢8) Unknown 57 (4¢5) 107 (7¢4) 0¢001 37 (5¢8) 22 (5¢9) 35 (11¢5) 13 (10¢3) 47 (4¢1)) ART- regimen in the first 2 years, n (%) NNRTI-based regimen 570 (44¢6) 723 (50¢2) 0¢004 338 (53¢1) 197 (52¢7) 131 (43¢1) 57 (45¢2) 508 (44¢4) PI-based regimen 350 (27¢4) 629 (43¢7) <0¢001 240 (37¢7) 164 (43¢9) 147 (48¢4) 78 (61¢9) 300 (26¢2) INSTI-based regimen 544 (42¢6) 435 (30¢2) <0¢001 198 (31¢1) 109 (29¢1) 83 (27¢3) 45 (35¢7) 508 (44¢4) Viral load at HIV diagnosis (log10 copies/mL), median (IQR) 4¢8 (3¢9-4¢9) 4¢9 (4¢7-5¢7) <0¢001 4¢9 (4¢7-5¢7) 4¢9 (4¢0-5¢7) 4¢9 (4¢7-5¢7) 4¢9 (4¢7-5¢8) 4¢8 (3¢9-4¢9) CD4 cell count at ART initiation (cells/ µL), n (%) <100 cells/µL - 383 (26¢6) 67 (17¢5) 92 (24¢0) 133 (34¢7) 91 (23¢8) - 100-199 cells/µL - 314 (21¢8) 109 (34¢7) 97 (30¢9) 88 (28¢0) 20 (6¢4) - 200-350 cells/µL - 744 (51¢6) 461 (62¢0) 185 (24¢9) 83 (11¢2) 15 (2¢0) - >350-500 cells/µL 613 (48¢0) - - - - - 509 (44¢5) >500 cells/µL 665 (52¢0) - - - - 636 (55¢6) Viral load at ART initiation (log10 copies/mL), median (IQR) 4¢4 (3¢6-4¢9) 4¢9 (4¢0-5¢5) <0¢001 4¢9 (4¢3-5¢3) 4¢8 (3¢9-5¢4) 4¢9 (3¢9-5¢7) 4¢9 (4¢1-5¢7) 4¢4 (3¢7-4¢9) <0¢001 www.thelancet.com Vol 52 Month , 2022 CD4 cell count 2 years after ART initiation (cells/µL), median (IQR) 802 (632-998) 469 (320-627) - - - - - Detectable viral load 2 years after ART initiation (>200 copies/mL), n (%) 61 (4¢8) 94 (6¢5) 0¢049 25 (3¢9) 15 (4¢0) 24 (7¢9) 30 (23¢8) 35 (3¢1) Unavailable viral load 2 years after ART initiation 113 (8¢8) 81 (5¢6) 0¢001 34 (5¢3) 21 (5¢6) 18 (5¢9) 8 (6¢4) 109 (9¢5) Individuals with AIDS defining events at ART initiationa, n (%) 20 (1¢6) 190 (13¢2) <0¢001 49 (7¢7) 47 (12¢6) 56 (18¢4) 38 (30¢2) 14 (1¢2) Individuals with AIDS defining events in the first 2 years beyond ART initiation, n (%) 1 (0¢1) 14 (1¢0) 0¢002 5 (0¢8) 3 (0¢8) 3 (1¢0) 3 (2¢4) 1 (0¢1) Modified Charlson comorbidity score at ART initiation, n (%) 0¢080 0 1158 (90¢6) 1262 (87¢6) 583 (91¢5) 328 (87¢7) 256 (84¢2) 95 (75¢4) 1041 (90¢9) 1 76 (6¢0) 108 (7¢5) 32 (5¢0) 25 (6¢7) 31 (10¢2) 20 (15¢9) 67 (5¢9) 2-3 36 (2¢8) 60 (4¢2) 20 (3¢1) 18 (4¢8) 14 (4¢6) 8 (6¢4) 30 (2¢6) ≥4 8 (0¢6) 11 (0¢8) 2 (0¢3) 3 (0¢8) 3 (1¢0) 3 (2¢4) 7 (0¢6) Table 1 (Continued) www.thelancet.com Vol 52 Month , 2022 Study population according to CD4 cell count at ART initiation Study population according to immune recovery 2 years after ART initiation Non-late Late p-value Late presenters Non-late presenters presenters n =1441 presenters n =1278 n =1441 (47¢0%) (53¢0%) CD4 cell count (cells/µ) CD4 cell count (cells/µ) >500 >350-500 ≥200-350 <200 >500 cells/µL n =637 (44¢2%) n =374 n =304 n =126 n =1145 (26¢0%) (21¢1%) (8¢7%) Modified Charlson comorbidity score at 2 years after ART initiation, n (%) <0¢001 0 1094 (85¢6) 1146 (79¢5) 0¢012 551 (86¢5) 301 (80¢5) 221 (72¢7) 73 (57¢9) 988 (86¢3) 1 107 (8¢4) 158 (11¢0) 0¢11 52 (8¢2) 39 (10¢4) 42 (13¢8) 25 (19¢8) 90 (7¢9) 2-3 62 (4¢9) 106 (7¢4) 0¢058 29 (4¢6) 26 (7¢0) 32 (10¢5) 19 (15¢1) 53 (4¢6) ≥4 15 (1¢2) 31 (2¢2) 0¢67 5 (0¢8) 8 (2¢1) 9 (3¢0) 9 (7¢1) 14 (1¢2) Calendar time of ART initiation, n (%) <0¢001 2005-2009 180 (23¢4) 588 (76¢6) <0¢001 232 (36¢4) 148 (39¢6) 147 (48¢4) 61 (48¢1) 141 (12¢3) 2010-2014 651 (54¢5) 543 (45¢5) 0¢001 264 (41¢4) 145 (38¢8) 94 (30¢9) 40 (31¢8)) 586(51¢2) 2015-2019 447 (59¢1) 306 (41¢0) <0¢001 141 (22¢1) 81 (21¢7) 63 (20¢7) 25 (19¢8) 418 (36¢5) Educational level, n (%) <0¢001 None or primary education only 226 (39¢2) 350 (60¢8) <0¢001 134 (28¢3) 90 (33¢6) 92 (45¢1) 34 (46¢0) 190 (21¢7) Secondary education 396 (51¢0) 380 (49¢0) 0¢008 187 (39¢5) 106 (39¢6) 64 (31¢4) 23 (31¢1) 361 (41¢2) University 350 (55¢7) 278 (44¢3) <0¢001 153 (32¢3) 67 (25¢0) 42 (20¢6) 16 (21¢6) 319 (36¢4) Unknown 8 (40¢0) 12 (60¢0) 0¢53 0 5 (1¢9) 6 (2¢9) 1 (1¢4) 7 (0.8) Income, n (%) 0¢010 No economic deprivation 678 (53¢1) 691 (48¢0) 0¢008 337 (52¢9) 176 (47¢1) 126 (41¢5) 52 (41¢3) 613 (53¢5) Mild economic deprivation 215 (16¢8) 309 (21¢4) 0¢002 129 (20¢3) 71 (19¢0) 77 (25¢3) 32 (25¢4) 184 (16¢1) Moderate/severe economic deprivation 353 (27¢6) 399 (27¢7) 0¢97 152 (23¢9) 118 (31¢6) 93 (30¢6) 36 (28¢6) 319 (27¢9) Unknown 32 (2¢5) 42 (2¢9) 0¢51 19 (3¢0) 9 (2¢4) 8 (2¢6) 6 (4¢8) 29 (2¢5) Death 2 year after ART initiation, n (%) 23 (1¢8) 90 (6¢3) <0¢001 14 (2¢2) 25 (6¢7) 22 (7¢2) 29 (23¢0) 17 (1¢5) Table 1: Baseline characteristics. ART, antiretroviral therapy; INSTI, integrase strand transfer inhibitor; IQR, interquartile range (percentiles 25th and 75th); MSM, men who have sex with men; IDU, injection drug use; NNRTI, non-nucleoside reverse transcriptase inhibitors; PI, protease inhibitor. a (before or within 90 days from ART initiation). Articles 7 Articles Table 2: Crude and adjusted mortality in treatment-naive individuals surviving the first 2 years after ART initiation. ART, antiretroviral therapy; INSTI, integrase strand transfer inhibitor; MSM, men who have sex with men; IDU, injection drug use. a MRR, mortality rate ratio. b aMRR, adjusted mortality rate ratio. The multivariate analysis was adjusted for age at baseline, transmission mode, CD4 cell count recovery at 2 years after ART initiation, HIV viral load > 200 c/ml vs ≤200 c/ml at 2 years, INSTI-based regimen as first-line regimen in the first 2 years, modified Charlson comorbid- ity score at baseline, history of AIDS-defining event at baseline, and calendar time (time-updated). The log-rank test revealed significant differences baseline covariates in both groups of INSTI and non- between survival curves for all-cause mortality (Figure 3). INSTI before and after propensity score matching is However, while LP with CD4 ≤500 cells/µL at two years presented in table S8. When including calendar time in had remarkably lower survival, the curve of LP with the model, we found the same trend, but the results did CD4 >500 overlapped with that of non-LP. As a sensi- not reach statistical significance (0¢76 [0¢52-1¢13]). The tivity analysis, we excluded patients with HIV viral load same analysis including also non-LP with CD4 cell >200 copies/mL or unavailable (n =349), with similar counts <500 cells/µL at ART initiation and using a results (Figure S1). The long-term survival analysis com- threshold of two-year CD4 cell count ≥700 cells/µL for paring non-LP achieving CD4 cell count >500−699 a satisfactory immune recovery, showed a similar bene- and those with ≥700 cells/µL did not reveal significant fit of first-line INSTI regimens also incorporating calen- differences between groups (0¢57 [0¢21-1¢51]) (Tables S6 dar time in the model (0¢74 [0¢57-0¢97]) (Table S9). and S7). Discussion Risk factors for incomplete immune recovery In this population-based prospective cohort study with Table 3 summarizes the results of univariate and multi- 2719 PLWH initiating their ART between 2005 and variate logistic regression analyses after propensity 2019 and surviving the first two years, 53% of them score matching assessing incomplete immune recovery were diagnosed at a late stage (i.e., CD4 count ≤350 (i.e., two-year CD4 counts ≤500 cells/µL) in LP. Lower cells/µL). Despite this delayed ART initiation in these initial CD4 counts at ART initiation strongly influenced patients presenting late to care, 44% achieved an opti- the likelihood of incomplete immune recovery at two mal immune recovery with a CD4 cell count >500 years, together with older age and modified CCI score 1. cells/µL at two years. LP achieving two-year CD4 cell First-line INSTI-based therapy was associated with 30% count >500 cells/µL had comparable long-term mortal- lower odds of incomplete immune recovery in LP ity rates to non-LP, irrespective of their initial CD4 cell (adjusted OR 0¢70 [0¢52-0¢95]). The distribution of count. CD4 cell count recovery two years after ART 8 www.thelancet.com Vol 52 Month , 2022 Articles Figure 2. Mortality rates upon immune recovery 2 years after initiation of antiretroviral therapy stratified by nadir CD4 cell count at ART initiation. Figure 3. Kaplan-Meier curves for overall survival by immune recovery 2 years after antiretroviral therapy (ART) initiation (non-late pre- senter with CD4 counts >500 cells/µL versus late presenters with CD4 counts >500 cells/µL, late presenters with CD4 counts >350-500 cells/µL, late presenters with CD4 counts 200-350 cells/µL and late presenters with CD4 counts <200 cells/µL at 2 years after ART initiation. initiation was a useful early predictor of long-term mor- score at ART initiation significantly increased the risk tality. Exposure to an INSTI-based ART regimen in the for incomplete two-year immune recovery. first two years was associated with a significantly better Although LP have an overall increased mortality, ear- two-year immune recovery and better long-term sur- lier studies have shown that the risk of death and/or vival. A lower CD4 cell count, older age, and worse CCI AIDS is highest in the first year after HIV diagnosis www.thelancet.com Vol 52 Month , 2022 9 Articles and decreases hereafter.20,21 There is less knowledge on had an approximately double mortality risk compared to the determinants of long-term survival after this initial non-LP. high-risk period. Furthermore, many of the studies eval- Our findings in immunological non-responders are uating life expectancy in LP include patients initiating in accordance with results from the COHERE cohort,37 ART before the year 2012, thus exposed to older ART which also found increased mortality in patients with regimens.6,31−35 The available evidence seems to indi- CD4 ≤200 cells/µL compared to those above this cate that the association between CD4 cell count at ART threshold after three years of successful suppressive initiation and mortality fades over the years as patients ART (adjusted hazard ratio 2¢60 [95%CI: 1¢86−3¢61]). survive. An ART-CC cohort study of PLWH initiating Similarly, the Italian MASTER and the Dutch ATHENA ART between 1996−2001, found that there were no dif- cohorts reported higher mortality risks and non-AIDS- ferences in long-term survival after 5 years among those defining events in a subpopulation of immunological starting ART with CD4 cell count 200−350 cells/µL or non-responders.12,18 Participants in these studies initi- >500 cells/µL.31 Similarly, a Spanish cohort found that ated their ART in earlier calendar years (1996 to 2011 in five-year survivors without history of AIDS-defining and 1998-2009, respectively), with older ART regi- events, there was no survival differences between those mens. A recent published study found that in an overall starting ART with CD4 count >350 cells/µL or with 200 cohort of PLWH initiating ART, five-year survivors with −350 cells/µL.21 A cohort study from UK including five-year CD4 <200 cells/µL remained with the worse patients from 2000−2010 showed that LP improved prognosis.32 their life span if they achieved a good immune response Several studies have described a better CD4 count and suppressed HIV-1 viral load. In fact, those achieving recovery in patients exposed to INSTI-based regimens, a good early immune response, with one-year CD4 cell mainly versus efavirenz-based regimens.22−25,38−40 Ini- count ≥350 cells/mL and virological suppression, experi- tiation of ART with a INSTI-based regimen was associ- enced a life expectancy close to the background popula- ated with a higher CD4 count increase at three years tion.36 However, in this study, half of the patients compared to efavirenz in the SINGLE and STARTMRK started ART before 2005 and no individual in the study phase 3 randomized studies.23,38,39 However, no differ- was exposed to INSTI. Existing evidence supports ences were observed between Dolutegravir and Daruna- that current CD4 count correlates strongly with vir in the FLAMINGO and SYMTRI randomized mortality15,33,34 and some studies have shown that trials.41,42 A recent collaborative cohort showed a more PLWH with current CD4 count ≥500 cells/mL and with- favourable immunologic response, defined as 25% out history of IDU have life expectancies close to the increase in CD4 count or as reaching ≥750 cells/mL at background population.2 We investigated if CD4 cell one year, with INSTI- vs PI- or NNRTI-based regi- count at 2-two years could be used as an early predictor mens.40 This study, however, included both treatment- of survival. Our results support that after this initial na€ıve and treatment-experienced individuals, did not high-risk period of 2 years after HIV diagnosis in LP, assess the effect of INSTI on long-term mortality nor likely associated with advanced AIDS-related medical the specific effect on immune responses in the subpop- conditions and complications present at diagnosis, two- ulation of LP, and residual confounding could not be year CD4 count recovery is a strong independent early fully excluded. In our analysis, incomplete immune predictor for long-term all-cause mortality, indepen- recovery was defined as CD4 cell count <500 cells/mL dently of the CD4 cell count at ART initiation. Actually, two years after ART, given that this threshold was asso- 44% of the two-year survivors successfully achieved a ciated with increased mortality compared with non-LP good immune recovery, with two-year CD4 cell counts in our analysis. Choosing a higher threshold of >700 >500 cells/µL. Importantly, the mortality rate in this cells/mL lead to similar results both in successful subgroup of patients did not differ from that of non-LP immune recovery and mortality. We found 30% lower with optimal immune response (i.e., CD4 cell counts > odds of incomplete immune recovery in individuals 500 cells/µL with a median 2-years CD4 837 [IQR: 697- starting INSTI-based regimens, both in LP and in indi- 1030]), thus narrowing the survival gap between late viduals with an initial CD4 count ≤500 cells/mL, and and non-LP. Only 8¢7% of LP in our cohort remained approximately 46% decrease in long-term mortality in with CD4 cell count <200 cells/µL at two years. These the whole study population. patients with immune discordance exhibited the highest To our knowledge, this is the first study to identify mortality rates, approximately 5 times greater than non- the exposure to INSTI-based regimen first-line ART LP. Of note, this group had a high percentage of detect- with a significantly higher probability of reaching a suc- able two-year viremia (24%), suggesting that increased cessful immune recovery with a significant impact on mortality rates and immune discordance might be par- long-term mortality in LP. Nonetheless, the strongest tially explained by ART non-adherence, aside from their independent risk factor for incomplete immune recov- low baseline CD4 count. Patients achieving CD4 cell ery was lower CD4 cell count at ART initiation, reinforc- counts between 200 and 500 cells/µL at two years still ing the importance of timely HIV diagnosis. 10 www.thelancet.com Vol 52 Month , 2022 Articles Table 3: Variables associated with incomplete immune restoration, defined as CD4 cell count ≤500 cells/µL, 2 years after ART initiation, in individuals with CD4 cell count ≤350 cells/µL at ART initiation. AIDS, Acquired Immune Deficiency Syndrome; ART, antiretroviral therapy; INSTI, integrase strand transfer inhibitor. a Multivariate analysis adjusted for age (categorical), gender, CD4 cell count at ART initiation, INSTI-based regimen, Charlson comorbidity score at ART initia- tion, and AIDS-defining event at ART initiation. Propensity score weighting on age, gender, CD4 cell count at ART initiation, and Charlson comorbidity index at ART initiation. Our study is strengthened by the prospective, popu- because initial (n =223) or two-year CD4 counts (n lation-based, multicentre HIV-cohort design, including =708) were not available. Of the last group, 41% (n 84% of all PLWH followed in Catalonia, and the long- =281) corresponded to individuals whose two-year blood term observation time. We had access to high-quality, samples had to be collected during the COVID-19 pan- real-world health data generated by the public health demic years 2020-2021, where hospital closure for non- system providing additional information on comorbid- COVID-19 patients prevented access to perform these ities and mortality. We compared a fully characterized blood analyses. Of those missing CD4 cell count at ART group of LP regarding their CD4 count at ART initiation initiation, most (n =137, 61%) belonged to the initial cal- and two-years after with non-LP with optimal immune endar period 2005-2009. Thirty-one (3¢8%) of the response (i.e. two-year CD4 >500 cells/µL, median excluded individuals died after ART initiation with a CD4 count 837 (IQR: 697-1030)) also including more high proportion having a history of IDU (58%). This recent calendar years and treatments. A propensity population differs from our overall study population score matching helped us to obtain more comparable being probably more prone to non-adherence to care treatment risk groups in the cohort regarding INSTI- and ART. Overall, these differences in the excluded pop- exposure, reducing possible channeling bias. ulation represent small absolute numbers and it is On the other hand, some limitations must be consid- unlikely that it would have affected our results. Third, ered. First, the definition of LP in our study included the availability and prescribing practices of INSTI have only CD4 cell count at ART initiation ≤ 350 cells/mL varied over time. We added calendar time to our models and not AIDS-defining events as we focused in CD4 to adjust for these differences over time. However, we responses. However, few non-LP experienced an AIDS- cannot rule out bias, especially residual channelling defining event (n =20), so including these in the group bias. When assessing the risk of incomplete immune of LP would not have significantly change the results. recovery (two-year CD4 cell count ≤ 500 cells/mL) in LP Second, we excluded 817 individuals from the analysis we found 30% lower odds when initiating ART with an www.thelancet.com Vol 52 Month , 2022 11 Articles INSTI-based regimen. This trend was maintained when The study protocol, the statistical codebook and we also adjusted for calendar time, however, the results dofiles for the analysis can be requested from RMI did not reach statistical significance, possibly due to (raquel@bisaurin.org). lack of power. Other residual confounders such as viral resistance, lifestyle factors, or comorbidities other than the ones included were not examined, though these fac- Declaration of interests tors would be unlikely to have a significant impact on All authors have completed the ICMJE uniform disclosure the results. Forth, the size of the study cohort is little form at www.icmje.org/coi_disclosure.pdf and declare: no with regard to a rare event like death, so risk of type II support from any organization for the submitted work; no errors cannot be excluded. Finally, the lack of access to financial relationships with any organizations that might cause of death information is a limitation for the have an interest in the submitted work in the previous detailed interpretation of the mortality figures. three years; no other relationships or activities that could In conclusion, two-year CD4 count was inversely corre- appear to have influenced the submitted work. lated with the mortality risk, and LP with a two-year CD4 RMI has received consulting honoraria and/or research cell count above 500 cells/µL, irrespective of their initial grants from GSK, MSD, and Lundbeck, outside the sub- CD4 cell count, had the same long-term survival as non- mitted work. RMI has received payment for travel and con- LP. Immune recovery at this time point could be used as gress assistance from MSD, outside the present work. an early prediction of long-term survival. Despite being JMM has received consulting honoraria and/or diagnosed at a late stage with severe immune damage, LP research grants from AbbVie, Angelini, Contrafect, will regain the possibility of achieving a similar prognosis Cubist, Genentech, Gilead Sciences, Jansen, Lysovant, as non-LP if there is an early optimal immune recovery. In Medtronic, MSD, Novartis, Pfizer, and ViiV Healthcare, our cohort of LP, INSTI-based regimens were associated outside the submitted work. with a significantly improved two-year immune recovery JML has received payments or honoraria for lectures, and lower long-term mortality, thus being a modifiable fac- presentations or speaker’s bureaus from Janssen-Cilag, tor to improve immune response and survival in these Gilead Sciences, and ViiV Healthcare, outside of the patients. Our data support positioning INSTI-based ART present work. as preferred regimens, particularly in late HIV presenters, though further studies are needed to confirm these find- ings. Whether our results can be extrapolated to dolutegra- Acknowledgements vir-based two-drug regimens will also need to be explored We thank the patients who participated in this cohort further. and our colleagues in the clinical departments for their Finally, our findings highlight the need for intensi- continued contribution and dedication. We would like fied public health efforts to improve timely HIV diagno- to thank Gerard Carot-Sans for providing medical writ- sis and the need to continue with studies to identify ing support with manuscript preparation. modifiable determinants for immune recovery to We also want to thank PADRIS and the Programme for improve survival in LP. the Prevention, Control and Care for HIV, Sexually Trans- mitted Diseases and Viral Hepatitis of the Ministry of Health of the Government of Catalonia for their support Contributors and the Danish AIDS foundation, the University of South- J Aceiton, Y Dıaz, S Moreno-Fornes had access to the ern Denmark, the Region of Southern Denmark and raw data and performed the initial data management. R Odense University Hospital for financial support. Martin-Iguacel conducted the research and analysed the We hereby declare that the manuscript is an honest, data. R Martin-Iguacel wrote the first draft of the manu- accurate, and transparent account of the study being script under the supervision of JM Llibre. All authors reported; that no important aspects of the study have contributed to the interpretation of the results and revi- been omitted; and that any discrepancies from the study sion of the manuscript and gave their final approval to as originally planned have been explained. the manuscript. Supplementary materials Data sharing statement Supplementary material associated with this article can The data collected for this study are available from the be found in the online version at doi:10.1016/j. Centre for Epidemiological Studies of Sexually Trans- eclinm.2022.101600. mitted Diseases and HIV/AIDS in Catalonia (CEEIS- CAT), the coordinating centre of the PISCIS cohort study and from each of the collaborating hospitals upon References 1 Gueler A, Moser A, Calmy A, et al. Life expectancy in HIV-positive request. Requests can be made via https://pisciscohort. persons in Switzerland: matched comparison with general popula- org/contacte/. tion. Aids. 2017;31:427–436. 12 www.thelancet.com Vol 52 Month , 2022 Articles 2 Lewden C, Bouteloup V, De Wit S, et al. All-cause mortality in in CoRIS: a prospective multicentre cohort study. Lancet HIV. treated HIV-infected adults with CD4 ≥500/mm3 compared with 2020;7:e565–e573. the general population: evidence from a large European observa- 23 Walmsley SL, Antela A, Clumeck N, et al. Dolutegravir plus tional cohort collaboration. Int J Epidemiol. 2012;41:433–445. Abacavir−Lamivudine for the treatment of HIV-1 infection. N Engl 3 Marcus JL, Leyden WA, Alexeeff SE, et al. Comparison of overall J Med. 2013;369:1807–1818. and comorbidity-free life expectancy between insured adults with 24 Blanco JR, Alejos B, Moreno S. Impact of dolutegravir and efavir- and without HIV Infection, 2000-2016. JAMA Netw open. 2020;3: enz on immune recovery markers: results from a randomized clini- e207954. cal trial. Clin Microbiol Infect. 2018;24:900–907. 4 Marcus JL, Chao CR, Leyden WA, et al. Narrowing the gap in life 25 Edwards JK, Cole SR, Hall HI, Mathews WC, Moore RD, Eron JJ. expectancy between HIV-infected and HIV-uninfected individuals Virologic suppression and CD4 cell count recovery after initiation with access to care. JAIDS J Acquir Immune Defic Syndr. of raltegravir- or efavirenz- containing HIV treatment regimens. 2016;73:39–46. AIDS. 2018;32:261–266. 5 Lohse MD P Nicolai, Ann-Brit Eg Hansen M, Pedersen MD P Gitte, 26 Statistical Institute of Catalonia. Available at: https://www.idescat. et al. Survival of Persons with and without HIV Infection in cat/pub/?id=aec& n= 245&lang=en. Accessed 1 January 2021. Denmark. Ann Intern Med Artic. 2013. 27  Casabona J, Esteve A, et al. Caracterısticas clinicoepide- Jaen A, 6 Samji H, Cescon A, Hogg RS, et al. Closing the gap: increases in miologicas y tendencias en el tratamiento antirretroviral de life expectancy among treated HIV-positive individuals in the una cohorte de pacientes con infeccion por el virus de la inmu- United States and Canada. PLoS One. 2013;8:e81355. nodeficiencia humana. Cohorte PISCIS. Med Clin (Barc). 7 May M, Gompels M, Delpech V, et al. Impact of late diagnosis and 2005;124:525–531. treatment on life expectancy in people with HIV-1: UK Collabora- 28 Government of Catalonia. Public program of data analysis for tive HIV Cohort (UK CHIC) Study. BMJ. 2011;343:1–11. health research and innovation in Catalonia −PADRIS−. AQuAS. 8 Edwards JK, Cole SR, Breger TL, et al. Mortality among persons Barcelona. Available https//aquas.gencat.cat/ca/ambits/analitica- entering hiv care compared with the general U.S. population: an dades/padris/. Accessed 20 March 2022. observational study. Ann Intern Med. 2021;174:1197–1206. 29 Lundgren J, Babiker AG, Fred G, et al. Initiation of antiretroviral 9 HIV/AIDS surveillance in Europe 2018 (2017 data). European Cen- therapy in early asymptomatic HIV infections. The insight start tre for Disease Prevention and Control (ECDC). Available at: study group. N Engl J Med. 2015;373:795–807. https://www.ecdc.europa.eu/en/publications-data/hivaids-surveil 30 Colls C, Mias M, Garcıa-altes A. A deprivation index to reform the lance-europe-2018-2017-data. Accessed 26 July 2022. financing model of primary care in Catalonia (Spain). Gac Sanit. 10 Mocroft A, Furrer HJ, Miro JM, et al. The incidence of AIDS- 2018;34:44–50. defining illnesses at a current CD4 count ≥200 cells/mL in the 31 May MT, Vehreschild JJ, Trickey A, et al. Mortality according to CD4 post-combination antiretroviral therapy era. Clin Infect Dis. count at start of combination antiretroviral therapy among HIV- 2013;57:1038–1047. infected patients followed for up to 15 years after start of treatment: col- 11 Rava M, Domınguez-Domınguez L, Bisbal O, et al. Late presenta- laborative cohort study. Clin Infect Dis. 2016;62:1571–1577. tion for HIV remains a major health issue in Spain: results from a 32 Bijker R, Kiertiburanakul S, Kumarasamy N, et al. Survival after multicenter cohort study, 2004−2018. PLoS One. 2021;16:2004– long-term ART exposure: findings from an Asian patient popula- 2018. tion retained in care beyond 5 years on ART. Antivir Ther. 12 van Lelyveld SFL, Gras L, Kesselring A, et al. Long-term complica- 2021;25:131–142. tions in patients with poor immunological recovery despite virolog- 33 CJ H, KL F, V J, et al. Changing predictors of mortality over time ical successful HAART in Dutch ATHENA cohort. AIDS. from cART start: implications for care Christopher. J Acquir 2012;26:465–474. Immune Defic Syndr. 2011;58:269–276. 13 Moore David M, Harris Ross, Lima Viviane, et al. Effect of baseline 34 McManus H, O’Connor CC, Boyd M, et al. Long-term survival in CD4 cell counts on the clinical significance of short-term immuno- HIV positive patients with up to 15 years of antiretroviral therapy. logic response to antiretroviral therapy in individuals with virologic PLoS One. 2012;7. https://doi.org/10.1371/journal.pone.0048839. suppression. Bone. 2014;23:1–7. 35 Lima VD, Eyawo O, Ma H, et al. The impact of scaling-up com- 14 Mocroft A, Laut K, Reiss P, et al. Where is the greatest impact of bination antiretroviral therapy on patterns of mortality among uncontrolled HIV infection on AIDS and non-AIDS events in HIV-positive persons in British Columbia, Canada. J Int AIDS HIV? AIDS. 2018;32:205–215. Soc. 2015;18:1–10. 15 Guaraldi G, Zona S, Menozzi M, et al. Late presentation increases 36 May MT, Gompels M, Delpech V, et al. Impact on life expectancy of risk and costs of non-infectious comorbidities in people with HIV: HIV-1 positive individuals of CD4+ cell count and viral load an Italian cost impact study. AIDS Res Ther. 2017;14:1–7. response to antiretroviral therapy. AIDS. 2014;28:1193–1202. 16 Montaner JSG, Lima VD, Barrios R, et al. Association of highly 37 Engsig FN, Zangerle R, Raffi F, et al. Long-term mortality in HIV- active antiretroviral therapy coverage, population viral load, and yearly positive individuals virally suppressed for >3 years with incomplete new HIV diagnoses in British Columbia, Canada: a population-based CD4 recovery. Clin Infect Dis. 2014;58:1312–1321. study. Lancet. 2010;376:532–539. 38 Walmsley S, Baumgarten A, Berenguer J, et al. Dolutegravir plus 17 Quinn TC, Wawer MJ, Sewankambo N, et al. Viral load and abacavir/lamivudine for the treatment of HIV-1 infection in antire- heterosexual transmission of human immunodeficiency troviral therapy-naive patients: week 96 and week 144 results from virus type 1. Rakai project study group. N Engl J Med. the SINGLE randomized clinical trial. J Acquir Immune Defic Syndr. 2000;342:921–929. 2015;70:515–519. 18 Lapadula G, Chatenoud L, Gori A, et al. Risk of severe non AIDS 39 Rockstroh JK, Lennox JL, Dejesus E, et al. Long-term treatment events is increased among patients unable to increase their with raltegravir or efavirenz combined with tenofovir/emtricitabine CD4 + T-cell counts >200+/ml despite effective HAART. PLoS for treatment-naive human immunodeficiency virus-1-infected One. 2015;10:1–15. patients: 156-week results from STARTMRK. Clin Infect Dis. 19 Loutfy MR, Genebat M, Moore D, et al. A CD4+ cell count <200 2011;53:807–816. cells per cubic millimeter at 2 years after initiation of combination 40 Neesgaard B, Mocroft A, Zangerle R, et al. Virologic and antiretroviral therapy is associated with increased mortality in immunologic outcomes of treatment with integrase inhibitors HIV-infected individuals with viral suppression. J Acquir Immune in a real-world setting: the respond cohort consortium. PLoS Defic Syndr. 2010;55:451–459. One. 2020;15:1–19. 20 Mocroft A, Lundgren JD, Sabin ML, et al. Risk factors and outcomes 41 Molina JM, Clotet B, van Lunzen J, et al. Once-daily dolutegra- for late presentation for HIV-positive persons in Europe: results from vir versus darunavir plus ritonavir for treatment-naive adults the collaboration of observational HIV epidemiological research with HIV-1 infection (FLAMINGO): 96 week results from a Europe study (COHERE). PLoS Med. 2013;10:e1001510. randomised, open-label, phase 3b study. Lancet HIV. 2015;2: 21 Sobrino-Vegas P, Moreno S, Rubio R, et al. Impact of late presenta- e127–e136. tion of HIV infection on short-, mid- and long-term mortality and 42 Podzamczer D, Mican R, Tiraboschi J, et al. Darunavir/cobicistat/ causes of death in a multicenter national cohort: 2004-2013. J emtricitabine/tenofovir alafenamide versus dolutegravir/abacavir/ Infect. 2016;72:587–596. lamivudine in antiretroviral-na€ıve adults (SYMTRI): a multicenter 22 Serrano-Villar S, Martınez-Sanz J, Ron R, et al. Effects of first-line randomized open-label study (PReEC/RIS-57). Open Forum Infect antiretroviral therapy on the CD4/CD8 ratio and CD8 cell counts Dis. 2021;9:ofab595. www.thelancet.com Vol 52 Month , 2022 13