DOI: 10.1111/hiv.12201 HIV Medicine (2014)
© 2014 British HIV Association
ORIGINAL RESEARCH
Time to initiation of antiretroviral therapy in HIV-infected patients diagnosed with an opportunistic disease: a cohort study L Deconinck,1,2,5 Y Yazdanpanah,2,3 RJ Gilson,1,4, H Melliez,5 N Viget,5 V Joly3 and CA Sabin1 UCL Research Department of Infection and Population Health, University College London, London, UK, 2Decision Sciences in Infectious Disease: Prevention, Control, and Care, IAME, UMR 1137, Paris Diderot University, Sorbonne Paris Cité, Paris, France, 3Department of Infectious Diseases, Bichat-Claude Bernard Hospital, Paris, France, 4The Mortimer Market Centre, Central and North West London NHS Foundation Trust, London, UK and 5Department of Infectious Diseases, Lille School of Medicine, Tourcoing Hospital, Tourcoing, France 1
Objectives
The aim of the study was to identify factors associated with the time between opportunistic disease (OD) diagnosis and antiretroviral therapy (ART) initiation in HIV-infected patients presenting for care with an OD, and to evaluate the outcomes associated with any delay. Methods
A multicentre cohort study was undertaken in London, Paris and Lille/Tourcoing. The medical records of patients diagnosed from 2002 to 2012 were reviewed. Results
A total of 437 patients were enrolled in the study: 70% were male, the median age was 40 years, 42% were from sub-Saharan Africa, 68% were heterosexual, the median CD4 count was 40 cells/μL, and the most common ODs were Pneumocystis pneumonia (37%), tuberculosis (24%), toxoplasmosis (12%) and Kaposi’s sarcoma (11%). Of these patients, 400 (92%) started ART within 24 weeks after HIV diagnosis, with a median time from OD diagnosis to ART initiation of 30 [interquartile range (IQR) 16–58] days. Patients diagnosed between 2009 and 2012 had a shorter time to ART initiation than those diagnosed in earlier years [hazard ratio (HR) 2.07; 95% confidence interval (CI) 1.58–2.72]. Factors associated with a longer time to ART initiation were a CD4 count ≥ 200 cells/μL (HR 0.30; 95% CI 0.20–0.44), tuberculosis (HR 0.40; 95% CI 0.30–0.55) and diagnosis in London (HR 0.62; 95% CI 0.48–0.80). Patients initiating ‘deferred’ ART (by ≥ 30 days) exhibited no difference in disease progression or immunovirological response compared with patients who had shorter times to ART initiation. Patients in the ‘deferred’ group were less likely to have ART modifications (HR 0.69; 95% CI 0.48–1.00) and had shorter in-patient stays (mean 14.2 days shorter; 95% CI 8.9–19.5 days) than patients in the group whose ART was not deferred. Conclusions
The time between OD diagnosis and ART initiation remains heterogeneous and relatively long, particularly in individuals with a high CD4 count or tuberculosis or those diagnosed in London. Deferring ART was associated with fewer ART modifications and shorter in-patient stays. Keywords: AIDS-related opportunistic infections, antiretroviral therapy, HIV, time to treatment Accepted 13 August 2014
Correspondence: Dr Laurène Deconinck, U 738 ATIP/Avenir Inserm Team, Université Paris 7, Faculté de Médecine – Site Bichat, 16 rue Henri Huchard, 75018 Paris, France. Tel: (33) 01 57 27 75 32; fax: (33) 01 57 27 75 21; e-mail: [email protected]
1
2 L Deconinck et al.
Introduction Across Europe, approximately 30% of HIV-infected patients present for care with advanced HIV disease [1,2]. These patients have a poor prognosis with a higher mortality, more AIDS- and non-AIDS-related events, a weaker immune response to antiretroviral therapy (ART), and more ART-related adverse events than those diagnosed earlier [3,4]. A crucial issue is the optimal time to initiate ART in HIV-infected patients diagnosed with an opportunistic disease (OD). ART is often delayed in these patients in order to prevent drug interactions, drug intolerance as a result of overlapping toxicities and the occurrence of immune reconstitution inflammatory syndrome (IRIS) [5]. Nevertheless, recent trials have provided evidence to support the earlier introduction of ART in patients with an OD [6–9]. Thus, guidelines now advise that ART should be started at an early stage among individuals with ODs, with the exception of cryptococcosis and tuberculous meningitis [10,11]. As ‘early’ is not clearly defined, however, medical practices remain heterogeneous. The main objective of this study was to identify factors associated with the time between OD diagnosis and ART initiation in HIV-infected patients presenting for care with an OD. A secondary objective was to determine the outcomes associated with the time to ART initiation in terms of immunovirological response, disease progression (mortality and AIDS and non-AIDS progression), ART modifications and in-patient stays.
Methods Study design and setting A multicentre cohort study was undertaken in three HIV reference centres in France and the UK: Bichat-Claude Bernard University Hospital in Paris; Dron Regional Hospital in Tourcoing; and the Mortimer Market Centre (MMC) in London. The study targeted all patients who entered care with an OD at the time of HIV diagnosis from 2002 to 2012.
Participants Eligible participants were aged > 18 years; had been diagnosed with HIV-1 infection at one of the participating centres between 1 June 2002 and 31 May 2012; and had a diagnosis of a stage C AIDS-defining event [12] within 45 days of HIV diagnosis. As this was a study of service delivery and all data were pseudonymized, ethics committee approval was not required.
© 2014 British HIV Association
In the French centres, the patients were selected from the medical database Nadis [13] using the International Classification of Diseases, 10th revision (ICD-10) coding system. At the MMC, patients were selected through data submitted to the UK Collaborative HIV Cohort Study [14]. These databases contain routinely collected information on all HIV-positive individuals who have attended any of the collaborating centres in France and the UK. Data collection took place from December 2012 to March 2013 and was performed by the same investigator in all centres. A standardized questionnaire was used to review the patient’s electronic medical record and when necessary nonelectronic records. For the analysis, follow-up of patients started on the date of HIV diagnosis and ended at ART initiation. Follow-up of patients who did not start ART was censored at 24 weeks after HIV diagnosis. The patients who started ART within 24 weeks after HIV diagnosis were then followed until 24 weeks after ART initiation. Patient follow-up was censored at death, or at date of last visit for those lost to follow-up (LFU).
Statistical analysis Kaplan−Meier analysis was performed to determine the median time between OD diagnosis and ART initiation. When several ODs occurred in a single patient, only the first OD diagnosed was considered in this analysis. When ART was introduced before OD diagnosis, the time between ART and OD diagnosis was set to 0 days. Factors associated with the time between OD diagnosis and ART initiation were first identified using an unadjusted Cox proportional hazards regression model. In order to create mutually exclusive groups for the initial OD in patients who presented with multiple ODs, the primary OD was defined for each patient as the one that was the most likely to lead to a deferral of ART initiation [by decreasing order of likelihood: tuberculosis (TB), toxoplasmosis, Pneumocystis pneumonia (PCP), Kaposi’s sarcoma (KS) and other OD]. Any variables that remained associated with the time to ART (P < 0.2) after adjustment for the type of OD were considered for inclusion in the fully adjusted Cox model, with the final covariates selected using a backwards stepwise approach (P < 0.05). Patients who started ART were then classified into one of two groups: ‘immediate initiation of ART’ (< 30 days after OD diagnosis) or ‘deferred initiation of ART’ (≥ 30 days after OD diagnosis). The threshold of 30 days was used to permit comparisons with other published studies. Logistic regression was performed to evaluate the impact of the time to ART on the odds of having an undetectable (≤ 50 HIV-1 RNA copies/mL) viral load at 24 weeks after ART initiation. Linear regression was performed to evaluate the
HIV Medicine (2014)
Determinants and impact of time to ART 3
outcome considered in the univariate analysis (P < 0.2). Multivariable analyses were developed using a backwards stepwise approach to select the final covariates for each model (P < 0.05). All analyses were performed using SAS software, version 9.3 (SAS Institute, Cary, NC).
impact of the time to ART on the CD4 count increase seen between ART initiation and 24 weeks after ART initiation and on the total duration of any in-patient stays within the first 24 weeks after ART initiation. Cox models were used to evaluate the impact on: time to death, time to a new AIDS event, time to IRIS (according to the Shelburne et al. definition [15]), time to a non-AIDS-related event (cardiovascular, renal, hepatic or non-HIV-related cancer) or serious biological adverse events (grade 3 or 4 [16]), time to any clinical event (death, AIDS-related event/IRIS or nonAIDS-related/adverse event), and time to ART modification, where these occurred within the first 24 weeks after ART initiation. In order to avoid misclassifying any preexisting conditions diagnosed after ART initiation as new events, we only considered events that were diagnosed more than 45 days after HIV diagnosis. Sensitivity analysis was performed taking into account all events diagnosed after ART initiation regardless of their time after HIV diagnosis. Potential confounders were defined as factors that (1) had previously been identified as significantly associated with the time to ART in our study (P < 0.05 in the multivariable model) and (2) were associated with the
Results Participants Overall, 651 patients were screened, of whom 437 were included in the cohort (Fig. 1): 206 (47.1%) from Bichat, 132 (30.2%) from MMC and 99 (22.7%) from Tourcoing. Of these, 11 patients (2.5%) died and four patients (0.9%) were LFU before ART initiation.
Baseline characteristics Patient characteristics at the time of HIV diagnosis are presented in Table 1, both overall and stratified by centre. The patients were mainly male (307; 70.3%), with a median age of 40 years [interquartile range (IQR) 34–48 years].
651 patients screened 119 note consultations
214 patients excluded : • • • • • • •
81 no stage C disease 59 initial care elsewhere a 24 time between HIV and OD diagnosis >45 days 28 HIV previously diagnosed 12 OD after HIV diagnosis b 4 age 5 log10 copies/mL [n (%)] Number of ODs per patient [n (%)] 1 2 ≥ 3† OD at diagnosis [n (%)] PCP Tuberculosis Toxoplasmosis Kaposi’s sarcoma Candidiasis HIV encephalopathy/wasting syndrome CMV infection Cryptococcosis Other OD‡ Time from OD to HIV diagnosis (days) [median (IQR)]
Total
Bichat
MMC
Tourcoing
437 (100)
206 (47.1)
132 (30.2)
99 (22.7)
307 (70.3) 40 (34–48)
136 (66.0) 40 (33–51)
96 (72.7) 39 (34–45)
75 (75.8) 42 (36–48)
183 (41.9) 166 (38.0) 87 (19.9) 1 (0.2)
112 (54.4) 39 (18.9) 54 (26.2) 1 (0.5)
49 (37.1) 56 (42.4) 27 (20.5) 0 (0.0)
22 (22.2) 71 (71.7) 6 (6.1) 0 (0.0)
297 (68.0) 126 (28.8) 8 (1.8) 6 (1.4)
168 (81.6) 34 (16.5) 4 (1.9) 0 (0.0)
70 (53.0) 60 (45.5) 2 (1.5) 0 (0.0)
59 (59.6) 32 (32.3) 2 (2.0) 6 (6.1)
181 (41.4) 131 (30.0) 125 (28.6) 355 (81.2) 79 (18.1)
89 (43.2) 62 (30.1) 55 (26.7) 176 (85.4) 53 (25.7)
59 (44.7) 46 (34.8) 27 (20.5) 102 (77.3) 9 (6.8)
33 (33.3) 23 (23.2) 43 (43.5) 77 (77.8) 17 (17.2)
25 (5.7) 8 (1.8)
16 (7.8) 5 (2.4)
8 (6.1) 3 (2.3)
1 (1.0) 0 (0.0)
40 (12–111) 244 (55.8) 142 (32.5) 35 (8.0) 15 (3.4) 5.4 (4.9–5.8) 305 (69.8)
30 (11–97) 127 (61.7) 54 (26.2) 17 (8.3) 7 (3.4) 5.5 (4.9–5.8) 150 (72.8)
50 (20–130) 61 (46.2) 52 (39.4) 12 (9.1) 7 (5.3) 5.3 (4.7–5.7) 81 (61.4)
39 (11–100) 56 (56.6) 36 (36.4) 6 (6.1) 1 (1.0) 5.5 (5.0–5.8) 74 (74.7)
342 (78.3) 70 (16.0) 25 (5.7)
171 (83.0) 25 (12.1) 10 (4.9)
111 (84.1) 17 (12.9) 4 (3.0)
60 (60.6) 28 (28.3) 11 (11.1)
64 (48.5) 18 (13.6) 4 (3.0) 25 (18.9) 10 (7.6) 9 (6.8) 4 (3.0) 6 (4.5) 17 (12.9) 0 (−5–2)
40 (40.4) 19 (19.2) 13 (13.1) 9 (9.1) 17 (17.2) 19 (19.2) 12 (12.1) 4 (4.0) 17 (17.2) −3 (−12–0)
161 (36.8) 107 (24.5) 52 (11.9) 48 (11.0) 42 (9.6) 39 (8.9) 33 (7.6) 24 (5.5) 54 (12.4) −1 (−7–0)
57 (27.7) 70 (34.0) 35 (17.0) 14 (6.8) 15 (7.3) 11 (5.3) 17 (8.3) 14 (6.8) 20 (9.7) −2 (−8–0)
CMV, cytomegalovirus; HBV, hepatitis B virus; HCV, hepatitis C virus; HBsAg, HBV surface antigen; IQR, interquartile range; MMC, Mortimer Market Centre; MSM, men who have sex with men; ICU, intensive care unit; OD, opportunistic disease; PCP, Pneumocystis pneumonia. *Injecting drug use, five (1.1%); blood products, two (0.5%); mother-to-child transmission, one (0.2%). † Three individuals had four ODs (0.7%). ‡ Lymphoma, 15 (3.4%); cryptosporidiosis, 13 (3%); herpes simplex virus infections, six (1.4%); histoplasmosis, six (1.4%); progressive multifocal leukoencephalopathy, five (1.1%); mycobacteriosis, four (0.9%); isosporiasis, three (0.7%); recurrent Salmonella septicaemia, one (0.2%); invasive cervical cancer, one (0.2%).
They were mostly from sub-Saharan Africa (183; 41.9%), France (110; 25.2%) and the UK (56; 12.8%). The main routes of HIV acquisition were heterosexual transmission (297; 68%) and homo/bisexual transmission (126; 28.8%). The median CD4 count at HIV diagnosis was 40 cells/μL
© 2014 British HIV Association
(IQR 12–111 cells/μL); the median HIV viral load was 5.4 log10 copies/mL (IQR 4.9–5.8 log10 copies/mL). The proportion of patients hospitalized at the time of diagnosis did not differ significantly between MMC and the French hospitals (77.9% versus 83.2%, respectively; P = 0.22), but the dura-
HIV Medicine (2014)
Determinants and impact of time to ART 5
tion of the first hospitalization was significantly shorter at MMC than in the French hospitals median 15 (IQR 11–25) versus 27 (19–42) days, respectively; P < 0.01). Three hundred and forty-two patients (78.3%) presented with a single OD, and 95 (21.7%) presented with multiple ODs. The first OD was diagnosed at a median of 1 day (IQR 0–7 days) after HIV diagnosis. The most frequent ODs at presentation were PCP (161; 36.8%), TB (107; 24.5%), cerebral toxoplasmosis (52; 11.9%), and KS (48; 11%).
There was a trend towards a shorter time to ART initiation in patients with KS versus PCP (median 31 versus 26 days, respectively; HR 1.41; 95% CI 0.96–2.08), and towards a longer time to ART initiation in those with toxoplasmosis versus PCP (median 32 versus 26 days, respectively; HR 0.71; 95% CI 0.50–1.01) (Kaplan−Meier plots of the univariate analyses are shown in Supporting information Fig. S1).
Impact of time to ART initiation Time to ART initiation Overall, 400 patients (91.5%) started ART within 24 weeks after HIV diagnosis. The median time between OD diagnosis and ART initiation was 30 days (IQR 16–58 days). ART was introduced before OD diagnosis in 16 patients, who presented mainly with TB (six cases) and lymphoma (four cases). The proportion of individuals starting ART while an in-patient was significantly higher in the French hospitals than in MMC (47.7% versus 16.8%, respectively; p < 0.01). Two hundred and sixty-three patients (65.8%) started ART while still receiving OD treatment, 789 patients (22.3%) started ART after all OD treatments were completed, and 40 patients (10%) did not receive any OD treatment (20 with HIV encephalopathy/wasting syndrome, 16 with KS, three with progressive multifocal leucoencephalopathy, and one with isosporiasis). The first ART regimen was protease inhibitor (PI)-based in 268 cases (67%), nonnucleoside reverse transcriptase inhibitor (NNRTI)-based in 113 cases (28.3%), and other regimens [nucleoside reverse transcriptase inhibitor (NRTI)-based, PI monotherapy or raltegravir] in 18 cases (4.5%) (one was unknown), with similar distributions over time. Twenty-two patients did not start ART within 24 weeks after HIV diagnosis (14 with TB, three with oesophageal candidiasis, two with PCP, one with mycobacteriosis, one with toxoplasmosis, and one with TB and HIV encephalopathy).
Determinants of time to ART initiation In the fully adjusted multivariable analysis (Table 2), factors remaining significantly associated with a longer time to ART initiation (p < 0.05) were a CD4 count ≥ 200 cells/μL versus 50–199 cells/μL [median 98 versus 26 days, respectively; hazard ratio (HR) 0.30; 95% confidence interval (CI) 0.20–0.44], a diagnosis of TB versus PCP (median 64 versus 26 days, respectively; HR 0.40; 95% CI 0.30– 0.55) and diagnosis at MMC versus Bichat (median 39 versus 31 days, respectively; HR 0.62; 95% CI 0.48–0.80). Those diagnosed in 2009–2012 versus 2006–2008 had a significantly shorter time to ART initiation (median 18 versus 39 days, respectively; HR 2.07; 95% CI 1.58–2.72).
© 2014 British HIV Association
The main outcomes within 24 weeks after ART initiation are described in Table 3, overall and stratified by time to ART initiation (baseline characteristics of both groups are reported in Supporting information Table S1). Among the 400 patients who started ART, 207 (51.7%) started < 30 days after OD diagnosis, while 193 (48.3%) started ≥ 30 days after OD diagnosis. The median time from OD diagnosis to ART initiation was 15 days (IQR 7–21 days) in the ‘immediate’ group and 52 days (IQR 38–74 days) in the ‘deferred’ group. Patients in the ‘immediate’ group were more likely than those in the ‘deferred’ group to have a period of overlap between the OD treatment and ART (73.9% versus 57%, respectively; p < 0.01), and to start ART as in-patients (66.2% versus 15.6%, respectively; p < 0.01). The CD4 count at ART initiation was higher in the ‘deferred’ group than in the ‘immediate’ group (median 42 versus 31 cells/μL, respectively; p < 0.01). Patients in the ‘immediate’ group were more likely than those in the ‘deferred’ group to start a PI-based regimen (74.4% versus 59.1%, respectively), whereas patients in the ‘deferred’ group more often started an NNRTI-based regimen than those in the ‘immediate’ group (36.8% versus 20.3%, respectively) (p < 0.01). Among these 400 patients, 13 (3.3%) were LFU. Of the remaining patients, 240 (60%) had an undetectable HIV viral load at 24 weeks and the median CD4 count increase since ART initiation was 130 cells/μL (IQR 75–196 cells/ μL). Eleven patients (2.8%) died, with eight deaths (2%) occurring more than 45 days after HIV diagnosis. Seventysix patients (19%) experienced a new AIDS-related event, with 55 (13.8%) of these events occurring more than 45 days after HIV diagnosis; 33 (8.3%) of these events were secondary to IRIS, 19 (4.8%) of which occurred more than 45 days after HIV diagnosis. Eighty-five patients (21.3%) experienced a non-AIDS-related or adverse event, with 63 events (15.8%) occurring more than 45 days after HIV diagnosis. These events were mostly haematological (36 cases), including 22 cases of neutropenia. Overall, 142 patients (35.5%) experienced any clinical event, with 105 events (26.3%) occurring more than 45 days after HIV diagnosis. In addition, 126 patients (31.5%) required a
HIV Medicine (2014)
6 L Deconinck et al.
Table 2 Results from unadjusted and adjusted Cox regression analyses of the factors associated with the time to initiation of antiretroviral therapy (ART) following opportunistic disease (OD) diagnosis Time between OD diagnosis and ART initiation (n = 437) Median (IQR)
Sociodemographic characteristics Sex Male Female Age < 30 years 30–49 years ≥ 50 years Place of birth France/UK Sub-Saharan Africa Other place of birth Risk group MSM/bisexual Heterosexual Other risk group Year of first access to care 2009–2012 2006–2008 2002–2005 Centre Bichat MMC Tourcoing Initial in-patient stay No Yes Initial ICU stay No Yes Comorbidities HBV coinfection No Yes HCV coinfection No Yes Clinical and biological characteristics CD4 T-cell count < 50 cells/μL 50–199 cells/μL ≥ 200 cells/μL HIV RNA ≤ 5 log10 copies/mL > 5 log10 copies/mL Number of ODs 1 >1 Type of OD PCP Tuberculosis Kaposi’s sarcoma Toxoplasmosis Other OD
Unadjusted HR (95% CI)
P
Adjusted HR (95% CI)
P
29 (14–52) 36 (20–79)
1 0.73 (0.59–0.91)
–
© 2014 British HIV Association
ORIGINAL RESEARCH
Time to initiation of antiretroviral therapy in HIV-infected patients diagnosed with an opportunistic disease: a cohort study L Deconinck,1,2,5 Y Yazdanpanah,2,3 RJ Gilson,1,4, H Melliez,5 N Viget,5 V Joly3 and CA Sabin1 UCL Research Department of Infection and Population Health, University College London, London, UK, 2Decision Sciences in Infectious Disease: Prevention, Control, and Care, IAME, UMR 1137, Paris Diderot University, Sorbonne Paris Cité, Paris, France, 3Department of Infectious Diseases, Bichat-Claude Bernard Hospital, Paris, France, 4The Mortimer Market Centre, Central and North West London NHS Foundation Trust, London, UK and 5Department of Infectious Diseases, Lille School of Medicine, Tourcoing Hospital, Tourcoing, France 1
Objectives
The aim of the study was to identify factors associated with the time between opportunistic disease (OD) diagnosis and antiretroviral therapy (ART) initiation in HIV-infected patients presenting for care with an OD, and to evaluate the outcomes associated with any delay. Methods
A multicentre cohort study was undertaken in London, Paris and Lille/Tourcoing. The medical records of patients diagnosed from 2002 to 2012 were reviewed. Results
A total of 437 patients were enrolled in the study: 70% were male, the median age was 40 years, 42% were from sub-Saharan Africa, 68% were heterosexual, the median CD4 count was 40 cells/μL, and the most common ODs were Pneumocystis pneumonia (37%), tuberculosis (24%), toxoplasmosis (12%) and Kaposi’s sarcoma (11%). Of these patients, 400 (92%) started ART within 24 weeks after HIV diagnosis, with a median time from OD diagnosis to ART initiation of 30 [interquartile range (IQR) 16–58] days. Patients diagnosed between 2009 and 2012 had a shorter time to ART initiation than those diagnosed in earlier years [hazard ratio (HR) 2.07; 95% confidence interval (CI) 1.58–2.72]. Factors associated with a longer time to ART initiation were a CD4 count ≥ 200 cells/μL (HR 0.30; 95% CI 0.20–0.44), tuberculosis (HR 0.40; 95% CI 0.30–0.55) and diagnosis in London (HR 0.62; 95% CI 0.48–0.80). Patients initiating ‘deferred’ ART (by ≥ 30 days) exhibited no difference in disease progression or immunovirological response compared with patients who had shorter times to ART initiation. Patients in the ‘deferred’ group were less likely to have ART modifications (HR 0.69; 95% CI 0.48–1.00) and had shorter in-patient stays (mean 14.2 days shorter; 95% CI 8.9–19.5 days) than patients in the group whose ART was not deferred. Conclusions
The time between OD diagnosis and ART initiation remains heterogeneous and relatively long, particularly in individuals with a high CD4 count or tuberculosis or those diagnosed in London. Deferring ART was associated with fewer ART modifications and shorter in-patient stays. Keywords: AIDS-related opportunistic infections, antiretroviral therapy, HIV, time to treatment Accepted 13 August 2014
Correspondence: Dr Laurène Deconinck, U 738 ATIP/Avenir Inserm Team, Université Paris 7, Faculté de Médecine – Site Bichat, 16 rue Henri Huchard, 75018 Paris, France. Tel: (33) 01 57 27 75 32; fax: (33) 01 57 27 75 21; e-mail: [email protected]
1
2 L Deconinck et al.
Introduction Across Europe, approximately 30% of HIV-infected patients present for care with advanced HIV disease [1,2]. These patients have a poor prognosis with a higher mortality, more AIDS- and non-AIDS-related events, a weaker immune response to antiretroviral therapy (ART), and more ART-related adverse events than those diagnosed earlier [3,4]. A crucial issue is the optimal time to initiate ART in HIV-infected patients diagnosed with an opportunistic disease (OD). ART is often delayed in these patients in order to prevent drug interactions, drug intolerance as a result of overlapping toxicities and the occurrence of immune reconstitution inflammatory syndrome (IRIS) [5]. Nevertheless, recent trials have provided evidence to support the earlier introduction of ART in patients with an OD [6–9]. Thus, guidelines now advise that ART should be started at an early stage among individuals with ODs, with the exception of cryptococcosis and tuberculous meningitis [10,11]. As ‘early’ is not clearly defined, however, medical practices remain heterogeneous. The main objective of this study was to identify factors associated with the time between OD diagnosis and ART initiation in HIV-infected patients presenting for care with an OD. A secondary objective was to determine the outcomes associated with the time to ART initiation in terms of immunovirological response, disease progression (mortality and AIDS and non-AIDS progression), ART modifications and in-patient stays.
Methods Study design and setting A multicentre cohort study was undertaken in three HIV reference centres in France and the UK: Bichat-Claude Bernard University Hospital in Paris; Dron Regional Hospital in Tourcoing; and the Mortimer Market Centre (MMC) in London. The study targeted all patients who entered care with an OD at the time of HIV diagnosis from 2002 to 2012.
Participants Eligible participants were aged > 18 years; had been diagnosed with HIV-1 infection at one of the participating centres between 1 June 2002 and 31 May 2012; and had a diagnosis of a stage C AIDS-defining event [12] within 45 days of HIV diagnosis. As this was a study of service delivery and all data were pseudonymized, ethics committee approval was not required.
© 2014 British HIV Association
In the French centres, the patients were selected from the medical database Nadis [13] using the International Classification of Diseases, 10th revision (ICD-10) coding system. At the MMC, patients were selected through data submitted to the UK Collaborative HIV Cohort Study [14]. These databases contain routinely collected information on all HIV-positive individuals who have attended any of the collaborating centres in France and the UK. Data collection took place from December 2012 to March 2013 and was performed by the same investigator in all centres. A standardized questionnaire was used to review the patient’s electronic medical record and when necessary nonelectronic records. For the analysis, follow-up of patients started on the date of HIV diagnosis and ended at ART initiation. Follow-up of patients who did not start ART was censored at 24 weeks after HIV diagnosis. The patients who started ART within 24 weeks after HIV diagnosis were then followed until 24 weeks after ART initiation. Patient follow-up was censored at death, or at date of last visit for those lost to follow-up (LFU).
Statistical analysis Kaplan−Meier analysis was performed to determine the median time between OD diagnosis and ART initiation. When several ODs occurred in a single patient, only the first OD diagnosed was considered in this analysis. When ART was introduced before OD diagnosis, the time between ART and OD diagnosis was set to 0 days. Factors associated with the time between OD diagnosis and ART initiation were first identified using an unadjusted Cox proportional hazards regression model. In order to create mutually exclusive groups for the initial OD in patients who presented with multiple ODs, the primary OD was defined for each patient as the one that was the most likely to lead to a deferral of ART initiation [by decreasing order of likelihood: tuberculosis (TB), toxoplasmosis, Pneumocystis pneumonia (PCP), Kaposi’s sarcoma (KS) and other OD]. Any variables that remained associated with the time to ART (P < 0.2) after adjustment for the type of OD were considered for inclusion in the fully adjusted Cox model, with the final covariates selected using a backwards stepwise approach (P < 0.05). Patients who started ART were then classified into one of two groups: ‘immediate initiation of ART’ (< 30 days after OD diagnosis) or ‘deferred initiation of ART’ (≥ 30 days after OD diagnosis). The threshold of 30 days was used to permit comparisons with other published studies. Logistic regression was performed to evaluate the impact of the time to ART on the odds of having an undetectable (≤ 50 HIV-1 RNA copies/mL) viral load at 24 weeks after ART initiation. Linear regression was performed to evaluate the
HIV Medicine (2014)
Determinants and impact of time to ART 3
outcome considered in the univariate analysis (P < 0.2). Multivariable analyses were developed using a backwards stepwise approach to select the final covariates for each model (P < 0.05). All analyses were performed using SAS software, version 9.3 (SAS Institute, Cary, NC).
impact of the time to ART on the CD4 count increase seen between ART initiation and 24 weeks after ART initiation and on the total duration of any in-patient stays within the first 24 weeks after ART initiation. Cox models were used to evaluate the impact on: time to death, time to a new AIDS event, time to IRIS (according to the Shelburne et al. definition [15]), time to a non-AIDS-related event (cardiovascular, renal, hepatic or non-HIV-related cancer) or serious biological adverse events (grade 3 or 4 [16]), time to any clinical event (death, AIDS-related event/IRIS or nonAIDS-related/adverse event), and time to ART modification, where these occurred within the first 24 weeks after ART initiation. In order to avoid misclassifying any preexisting conditions diagnosed after ART initiation as new events, we only considered events that were diagnosed more than 45 days after HIV diagnosis. Sensitivity analysis was performed taking into account all events diagnosed after ART initiation regardless of their time after HIV diagnosis. Potential confounders were defined as factors that (1) had previously been identified as significantly associated with the time to ART in our study (P < 0.05 in the multivariable model) and (2) were associated with the
Results Participants Overall, 651 patients were screened, of whom 437 were included in the cohort (Fig. 1): 206 (47.1%) from Bichat, 132 (30.2%) from MMC and 99 (22.7%) from Tourcoing. Of these, 11 patients (2.5%) died and four patients (0.9%) were LFU before ART initiation.
Baseline characteristics Patient characteristics at the time of HIV diagnosis are presented in Table 1, both overall and stratified by centre. The patients were mainly male (307; 70.3%), with a median age of 40 years [interquartile range (IQR) 34–48 years].
651 patients screened 119 note consultations
214 patients excluded : • • • • • • •
81 no stage C disease 59 initial care elsewhere a 24 time between HIV and OD diagnosis >45 days 28 HIV previously diagnosed 12 OD after HIV diagnosis b 4 age 5 log10 copies/mL [n (%)] Number of ODs per patient [n (%)] 1 2 ≥ 3† OD at diagnosis [n (%)] PCP Tuberculosis Toxoplasmosis Kaposi’s sarcoma Candidiasis HIV encephalopathy/wasting syndrome CMV infection Cryptococcosis Other OD‡ Time from OD to HIV diagnosis (days) [median (IQR)]
Total
Bichat
MMC
Tourcoing
437 (100)
206 (47.1)
132 (30.2)
99 (22.7)
307 (70.3) 40 (34–48)
136 (66.0) 40 (33–51)
96 (72.7) 39 (34–45)
75 (75.8) 42 (36–48)
183 (41.9) 166 (38.0) 87 (19.9) 1 (0.2)
112 (54.4) 39 (18.9) 54 (26.2) 1 (0.5)
49 (37.1) 56 (42.4) 27 (20.5) 0 (0.0)
22 (22.2) 71 (71.7) 6 (6.1) 0 (0.0)
297 (68.0) 126 (28.8) 8 (1.8) 6 (1.4)
168 (81.6) 34 (16.5) 4 (1.9) 0 (0.0)
70 (53.0) 60 (45.5) 2 (1.5) 0 (0.0)
59 (59.6) 32 (32.3) 2 (2.0) 6 (6.1)
181 (41.4) 131 (30.0) 125 (28.6) 355 (81.2) 79 (18.1)
89 (43.2) 62 (30.1) 55 (26.7) 176 (85.4) 53 (25.7)
59 (44.7) 46 (34.8) 27 (20.5) 102 (77.3) 9 (6.8)
33 (33.3) 23 (23.2) 43 (43.5) 77 (77.8) 17 (17.2)
25 (5.7) 8 (1.8)
16 (7.8) 5 (2.4)
8 (6.1) 3 (2.3)
1 (1.0) 0 (0.0)
40 (12–111) 244 (55.8) 142 (32.5) 35 (8.0) 15 (3.4) 5.4 (4.9–5.8) 305 (69.8)
30 (11–97) 127 (61.7) 54 (26.2) 17 (8.3) 7 (3.4) 5.5 (4.9–5.8) 150 (72.8)
50 (20–130) 61 (46.2) 52 (39.4) 12 (9.1) 7 (5.3) 5.3 (4.7–5.7) 81 (61.4)
39 (11–100) 56 (56.6) 36 (36.4) 6 (6.1) 1 (1.0) 5.5 (5.0–5.8) 74 (74.7)
342 (78.3) 70 (16.0) 25 (5.7)
171 (83.0) 25 (12.1) 10 (4.9)
111 (84.1) 17 (12.9) 4 (3.0)
60 (60.6) 28 (28.3) 11 (11.1)
64 (48.5) 18 (13.6) 4 (3.0) 25 (18.9) 10 (7.6) 9 (6.8) 4 (3.0) 6 (4.5) 17 (12.9) 0 (−5–2)
40 (40.4) 19 (19.2) 13 (13.1) 9 (9.1) 17 (17.2) 19 (19.2) 12 (12.1) 4 (4.0) 17 (17.2) −3 (−12–0)
161 (36.8) 107 (24.5) 52 (11.9) 48 (11.0) 42 (9.6) 39 (8.9) 33 (7.6) 24 (5.5) 54 (12.4) −1 (−7–0)
57 (27.7) 70 (34.0) 35 (17.0) 14 (6.8) 15 (7.3) 11 (5.3) 17 (8.3) 14 (6.8) 20 (9.7) −2 (−8–0)
CMV, cytomegalovirus; HBV, hepatitis B virus; HCV, hepatitis C virus; HBsAg, HBV surface antigen; IQR, interquartile range; MMC, Mortimer Market Centre; MSM, men who have sex with men; ICU, intensive care unit; OD, opportunistic disease; PCP, Pneumocystis pneumonia. *Injecting drug use, five (1.1%); blood products, two (0.5%); mother-to-child transmission, one (0.2%). † Three individuals had four ODs (0.7%). ‡ Lymphoma, 15 (3.4%); cryptosporidiosis, 13 (3%); herpes simplex virus infections, six (1.4%); histoplasmosis, six (1.4%); progressive multifocal leukoencephalopathy, five (1.1%); mycobacteriosis, four (0.9%); isosporiasis, three (0.7%); recurrent Salmonella septicaemia, one (0.2%); invasive cervical cancer, one (0.2%).
They were mostly from sub-Saharan Africa (183; 41.9%), France (110; 25.2%) and the UK (56; 12.8%). The main routes of HIV acquisition were heterosexual transmission (297; 68%) and homo/bisexual transmission (126; 28.8%). The median CD4 count at HIV diagnosis was 40 cells/μL
© 2014 British HIV Association
(IQR 12–111 cells/μL); the median HIV viral load was 5.4 log10 copies/mL (IQR 4.9–5.8 log10 copies/mL). The proportion of patients hospitalized at the time of diagnosis did not differ significantly between MMC and the French hospitals (77.9% versus 83.2%, respectively; P = 0.22), but the dura-
HIV Medicine (2014)
Determinants and impact of time to ART 5
tion of the first hospitalization was significantly shorter at MMC than in the French hospitals median 15 (IQR 11–25) versus 27 (19–42) days, respectively; P < 0.01). Three hundred and forty-two patients (78.3%) presented with a single OD, and 95 (21.7%) presented with multiple ODs. The first OD was diagnosed at a median of 1 day (IQR 0–7 days) after HIV diagnosis. The most frequent ODs at presentation were PCP (161; 36.8%), TB (107; 24.5%), cerebral toxoplasmosis (52; 11.9%), and KS (48; 11%).
There was a trend towards a shorter time to ART initiation in patients with KS versus PCP (median 31 versus 26 days, respectively; HR 1.41; 95% CI 0.96–2.08), and towards a longer time to ART initiation in those with toxoplasmosis versus PCP (median 32 versus 26 days, respectively; HR 0.71; 95% CI 0.50–1.01) (Kaplan−Meier plots of the univariate analyses are shown in Supporting information Fig. S1).
Impact of time to ART initiation Time to ART initiation Overall, 400 patients (91.5%) started ART within 24 weeks after HIV diagnosis. The median time between OD diagnosis and ART initiation was 30 days (IQR 16–58 days). ART was introduced before OD diagnosis in 16 patients, who presented mainly with TB (six cases) and lymphoma (four cases). The proportion of individuals starting ART while an in-patient was significantly higher in the French hospitals than in MMC (47.7% versus 16.8%, respectively; p < 0.01). Two hundred and sixty-three patients (65.8%) started ART while still receiving OD treatment, 789 patients (22.3%) started ART after all OD treatments were completed, and 40 patients (10%) did not receive any OD treatment (20 with HIV encephalopathy/wasting syndrome, 16 with KS, three with progressive multifocal leucoencephalopathy, and one with isosporiasis). The first ART regimen was protease inhibitor (PI)-based in 268 cases (67%), nonnucleoside reverse transcriptase inhibitor (NNRTI)-based in 113 cases (28.3%), and other regimens [nucleoside reverse transcriptase inhibitor (NRTI)-based, PI monotherapy or raltegravir] in 18 cases (4.5%) (one was unknown), with similar distributions over time. Twenty-two patients did not start ART within 24 weeks after HIV diagnosis (14 with TB, three with oesophageal candidiasis, two with PCP, one with mycobacteriosis, one with toxoplasmosis, and one with TB and HIV encephalopathy).
Determinants of time to ART initiation In the fully adjusted multivariable analysis (Table 2), factors remaining significantly associated with a longer time to ART initiation (p < 0.05) were a CD4 count ≥ 200 cells/μL versus 50–199 cells/μL [median 98 versus 26 days, respectively; hazard ratio (HR) 0.30; 95% confidence interval (CI) 0.20–0.44], a diagnosis of TB versus PCP (median 64 versus 26 days, respectively; HR 0.40; 95% CI 0.30– 0.55) and diagnosis at MMC versus Bichat (median 39 versus 31 days, respectively; HR 0.62; 95% CI 0.48–0.80). Those diagnosed in 2009–2012 versus 2006–2008 had a significantly shorter time to ART initiation (median 18 versus 39 days, respectively; HR 2.07; 95% CI 1.58–2.72).
© 2014 British HIV Association
The main outcomes within 24 weeks after ART initiation are described in Table 3, overall and stratified by time to ART initiation (baseline characteristics of both groups are reported in Supporting information Table S1). Among the 400 patients who started ART, 207 (51.7%) started < 30 days after OD diagnosis, while 193 (48.3%) started ≥ 30 days after OD diagnosis. The median time from OD diagnosis to ART initiation was 15 days (IQR 7–21 days) in the ‘immediate’ group and 52 days (IQR 38–74 days) in the ‘deferred’ group. Patients in the ‘immediate’ group were more likely than those in the ‘deferred’ group to have a period of overlap between the OD treatment and ART (73.9% versus 57%, respectively; p < 0.01), and to start ART as in-patients (66.2% versus 15.6%, respectively; p < 0.01). The CD4 count at ART initiation was higher in the ‘deferred’ group than in the ‘immediate’ group (median 42 versus 31 cells/μL, respectively; p < 0.01). Patients in the ‘immediate’ group were more likely than those in the ‘deferred’ group to start a PI-based regimen (74.4% versus 59.1%, respectively), whereas patients in the ‘deferred’ group more often started an NNRTI-based regimen than those in the ‘immediate’ group (36.8% versus 20.3%, respectively) (p < 0.01). Among these 400 patients, 13 (3.3%) were LFU. Of the remaining patients, 240 (60%) had an undetectable HIV viral load at 24 weeks and the median CD4 count increase since ART initiation was 130 cells/μL (IQR 75–196 cells/ μL). Eleven patients (2.8%) died, with eight deaths (2%) occurring more than 45 days after HIV diagnosis. Seventysix patients (19%) experienced a new AIDS-related event, with 55 (13.8%) of these events occurring more than 45 days after HIV diagnosis; 33 (8.3%) of these events were secondary to IRIS, 19 (4.8%) of which occurred more than 45 days after HIV diagnosis. Eighty-five patients (21.3%) experienced a non-AIDS-related or adverse event, with 63 events (15.8%) occurring more than 45 days after HIV diagnosis. These events were mostly haematological (36 cases), including 22 cases of neutropenia. Overall, 142 patients (35.5%) experienced any clinical event, with 105 events (26.3%) occurring more than 45 days after HIV diagnosis. In addition, 126 patients (31.5%) required a
HIV Medicine (2014)
6 L Deconinck et al.
Table 2 Results from unadjusted and adjusted Cox regression analyses of the factors associated with the time to initiation of antiretroviral therapy (ART) following opportunistic disease (OD) diagnosis Time between OD diagnosis and ART initiation (n = 437) Median (IQR)
Sociodemographic characteristics Sex Male Female Age < 30 years 30–49 years ≥ 50 years Place of birth France/UK Sub-Saharan Africa Other place of birth Risk group MSM/bisexual Heterosexual Other risk group Year of first access to care 2009–2012 2006–2008 2002–2005 Centre Bichat MMC Tourcoing Initial in-patient stay No Yes Initial ICU stay No Yes Comorbidities HBV coinfection No Yes HCV coinfection No Yes Clinical and biological characteristics CD4 T-cell count < 50 cells/μL 50–199 cells/μL ≥ 200 cells/μL HIV RNA ≤ 5 log10 copies/mL > 5 log10 copies/mL Number of ODs 1 >1 Type of OD PCP Tuberculosis Kaposi’s sarcoma Toxoplasmosis Other OD
Unadjusted HR (95% CI)
P
Adjusted HR (95% CI)
P
29 (14–52) 36 (20–79)
1 0.73 (0.59–0.91)
–
