| |
Normalisation of CD4 counts in patients with HIV-1 infection and maximum virological suppression (<50 copies/ml) who are taking combination antiretroviral therapy: an observational cohort study
|
| |
| |
Combo HIV Drug Therapy May Restore Healthy Immune System
The Lancet Early Online Publication, 19 July 2007
Dr A Mocroft PhD a , Prof AN Phillips PhD a, Prof J Gatell MD b, B Ledergerber PhD c, M Fisher MBBS d, Prof N Clumeck MD e, M Losso MD f, Prof A Lazzarin MD g, Prof G Fatkenheuer MD h and Prof JD Lundgren MD i, for the EuroSIDA study group
Patients starting HAART with <200 CD4s and <50 copies/ml for 5 yrs or more had CD4 count of 500. See table 3 below. Note from Jules Levin: I think potency of the regimen is crucial in maximizing CD4 response and complete adherence is also important.
"....This study of CD4 count increases in antiretroviral-naive HIV-infected patients starting cART and who subsequently achieve maximum viral suppression has found little evidence of a plateau effect. Most patients continued to have significant rises in CD4 count, even at more than 5 years after cART initiation, whereas significant, but smaller, increases were seen in patients who started cART with low CD4 counts. Normalisation of CD4 counts in HIV-infected patients for all infected individuals was seen in patients starting cART with a CD4 count of more than 350 cells per μL and might be achieved in patients starting cART with lower CD4 counts if viral suppression can be maintained for a sufficiently long period of time....most patients with HIV who can maintain viral load at less than 50 copies per mL continue to have significant rises in CD4 counts, even after protracted exposure to combination therapy."
Summary
Background
Combination antiretroviral therapy (cART) has been shown to reduce mortality and morbidity in patients with HIV. As viral replication falls, the CD4 count increases, but whether the CD4 count returns to the level seen in HIV-negative people is unknown. We aimed to assess whether the CD4 count for patients with maximum virological suppression (viral load <50 copies per mL) continues to increase with long-term cART to reach levels seen in HIV-negative populations.
Methods
We compared increases in CD4 counts in 1835 antiretroviral-naive patients who started cART from EuroSIDA, a pan-European observational cohort study. Rate of increase in CD4 count (per year) occurring between pairs of consecutive viral loads below 50 copies per mL was estimated using generalised linear models, accounting for multiple measurements for individual patients.
Findings
The median CD4 count at starting cART was 204 cells per _L (IQR 85-330). The greatest mean yearly increase in CD4 count of 100 cells per _L was seen in the year after starting cART. Significant, but lower, yearly increases in CD4 count, around 50 cells per _L, were seen even at 5 years after starting cART in patients whose current CD4 count was less than 500 cells per _L. The only groups without significant increases in CD4 count were those where cART had been taken for more than 5 years with a current CD4 count of more than 500 cells per _L, (current mean CD4 count 774 cells per _L; 95% CI 764-783). Patients starting cART with low CD4 counts (<200 cells per _L) had significant rises in CD4 counts even after 5 years of cART.
Interpretation
Normalisation of CD4 counts in HIV-infected patients for all infected individuals might be achievable if viral suppression with cART can be maintained for a sufficiently long period of time.
Introduction
Combination antiretroviral therapy (cART) has been shown to reduce mortality and morbidity in patients with HIV.1 The goal of cART, according to current treatment guidelines, is to reduce HIV viral replication to below the limit of detection.2 As viral replication falls, the CD4 count increases.3,4 The initial increase is rapid and usually lasts 3-6 months, followed by a phase of slower CD4 count increases.5 The factors that determine CD4 count responses are only partly known and are thought to depend on both the host and the virus, and there is substantial variation in CD4 count recovery.6 In patients with virological suppression (HIV-RNA viral load <1000 copies per mL), older age, a longer duration of HIV infection, and lower CD4 counts at starting cART were predictors for maintaining lower CD4 counts.7,8 In the Swiss HIV Cohort Study,7 a third of patients were incomplete responders, and only half continued to have CD4 count increases. The remainder were described as reaching a CD4 plateau, with no further increases in CD4 count. This finding led to the conclusion that not all patients might eventually respond to cART by achieving a CD4 count in the normal range.
Several factors have been investigated to establish the relation with CD4 count recovery, including viral pathogenicity, host factors, or co-infection with hepatitis B or C.6,9,10 The most consistent finding, however, is that patients who start cART with lower CD4 counts need longer treatment to achieve CD4 counts in the normal range.3,7 There has been little research to date on CD4 count increases in analyses restricted to patients with maximum virological suppression (viral load <50 copies per mL). Previous work from the EuroSIDA study11 assessed the increases in CD4 count in patients with maximum virological suppression, and found some differences according to cART regimen in use, but this previous study did not specifically address long term CD4 count increases and when or at what level CD4 counts were no longer increasing. The objectives of our study were therefore to describe the relation between duration of treatment, CD4 count at the start of cART, current CD4 count, and CD4 count increases in antiretroviral-naive patients starting cART who achieve maximum virological suppression.
Methods
Patients
EuroSIDA is a prospective, European study of 14 262 patients with HIV-1 infection in 92 centres across Europe (including Israel and Argentina as non-European representatives). Details of the study have been published previously.12 Seven cohorts have been recruited to date, the first in May, 1994, of 3116 patients, and the latest, of 2337 patients, was recruited from November, 2005. At recruitment, in addition to demographic and clinical information, a complete antiretroviral history was obtained, together with the eight most recent CD4 counts and viral load measurements. Viral load was measured within each clinic according to local guidelines; a lower limit of detection of 50 copies per mL was introduced from December, 1997. This analysis includes data to a median date of February, 2006, and includes details on all CD4 lymphocyte counts and viral load measured since last data collection and the date of starting and stopping each antiretroviral drug. Data for hepatitis B and C antibody status were recorded in 1997 at the inception of cohort III and for patients still under follow-up from cohorts I-II, similarly at the recruitment of cohort IV, and every year thereafter. Hepatitis B and C serological markers were assessed with commercial ELISAs.
Members of the coordinating office visited all centres to ensure correct patient selection, and that accurate data were provided, by checking the information against case notes for all reported clinical events and a random sample of 10% of all other patients.
Statistical methods
All antiretroviral naive patients from EuroSIDA with at least two consecutive viral loads of less than 50 copies per mL were eligible for inclusion. Patients were required to have a CD4 count measured in the 6 months before starting cART and distinct CD4 counts measured within at most 4 weeks (either side) of each viral load measure (95% of CD4 counts were measured on the same date as viral load measurements). cART was defined as exactly two nucleosides or nucleotides plus a single protease inhibitor, ritonavir boosted protease inhibitor, a non-nucleoside reverse transcriptase inhibitor (NNRTI), or abacavir.
Baseline (for descriptive purposes) was defined as the first of two consecutive viral loads of less than 50 copies per mL after starting cART. The first viral load measurement was used because that was the time point from which patients were included, conditional on there being a consecutive viral load of less than 50 copies per mL. The change in CD4 count occurring between each pair of consecutive viral loads of less than 50 copies per mL was calculated and standardised for the time between measurements to give the yearly change in CD4 count. Thus a patient with four consecutive viral loads of less than 50 copies per mL would contribute data from three viral-load pairs: change in CD4 between first and second viral load, second and third, and third and fourth. Pairs of viral-load measurements were excluded if any change (start or stop) in antiretrovirals had been made between the measurements.
Generalised linear models, using a normal distribution and an identity link function, with adjustments for repeated measures (since each patient could be included any number of times depending on their viral load, CD4, and cART history) were used to describe CD4 count changes stratified by time since starting cART, CD4 count at starting cART, and current CD4 count. These stratifications were decided a priori in order to investigate the relations between key variables. Where patients had multiple viral loads recorded within a 4-week period, the maximum viral load was used for analyses; similarly, patients with multiple CD4 counts measured within a 4-week period had the median of these values used.
In multivariable models, adjustments were made for factors previously shown to be associated with change in CD4 count in patients with a viral load of less than 50 copies per mL:11 age, time since cART initiation, change in CD4 since cART initiation, nucleoside pair (zidovudine-lamivudine, lamivudine-stavudine, stavudine-didanosine, tenofovir plus one nucleoside, abacavir plus one nucleoside, or any other combination of two nucleosides), and third drug (single protease inhibitor, ritonavir boosted protease inhibitor, NNRTI, or abacavir). The CD4 count previous to the pair used to calculate changes in CD4 counts was included in multivariable models to avoid regression to the mean and problems of overfitting. An additional adjustment was made for time between starting cART and initial virological suppression (<50 copies per mL).
All analyses were done using SAS (Statistical Analysis Software, version 8.2, Cary, NC, USA).
Role of the funding source
The sponsors of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had full responsibility for the decision to submit for publication.
Results
There were 3365 antiretroviral-naive patients who started cART in the EuroSIDA study; of these, 2598 had at least one viral load measured with a lower limit of detection of 50 copies per mL, and 2000 had a pair of consecutive viral loads of less than 50 copies per mL. Of these 2000 patients, 1835 had a CD4 count measured before starting cART and were therefore included in analyses (table 1). The median CD4 at the arbitrarily defined baseline was 404 cells per _L (IQR 251-580); 742 patients (40¥4%) had a CD4 cell count of 350 cells per _L or less at baseline. The median change in CD4 between starting cART and baseline was 178 cells per _L (IQR 76-310). The median date of starting cART was May, 1999 (IQR September, 1997-September, 2001), and the median CD4 at starting cART was 204 cells per _L (IQR 85-330). At baseline, most patients were taking an NNRTI-based regimen (n=961, 52¥4%) or a protease inhibitor-based regimen (n=716, 39%). The most commonly used nucleoside pair was zidovudine-lamivudine (n=1014, 55¥3%), followed by stavudine-lamivudine (n=333, 18¥2%).
The median time between viral load measurements was 3 months (IQR 2-4), and a median of six pairs of viral load measurements of less than 50 copies per mL per patient were included in these analyses (IQR 3-10). There were 13 809 pairs with a viral load of less than 50 copies per mL included in analyses, representing 4095¥8 patient-years of follow-up. 6503 (47¥1%) viral load pairs were in patients taking a protease inhibitor-based regimen, 5660 (41¥0%) in patients taking a NNRTI-based regimen, and 1646 (11¥9%) in patients taking a triple-nucleoside regimen containing abacavir. Zidovudine-lamivudine was the most commonly used nucleoside pair (61¥5%), followed by stavudine-lamivudine (15¥3%).
Table 2 shows the unadjusted rate of increase in CD4 count (per year) stratified by current count (ie, the CD4 count measured before each viral load pair). For example, there were 1353 viral load pairs where the current CD4 count was 200 cells per _L or less. In these patients, the mean rate of increase in CD4 (per year) with viral load below 50 copies per mL was 74 cells per _L (95% CI 58-91), whereas the mean current CD4 count was 137 cells per _L (134-139). Notably, the only group that did not have significant increases in CD4 count were those patients with a current count of more than 700 cells per _L, where the mean yearly change in CD4 was 18 cells per _L (-6 to 43). The current mean CD4 count in these patients was 922 cells per _L (914-930).
Figure 1. Adjusted rate of CD4 count increases (per year) with viral load less than 50 copies per mL
Adjusted for nucleoside pair, cART regimen, age, change in CD4 count since cART initiation, time since starting cART, and time since starting cART to initial virological suppression (<500 copies per mL).
A multivariable model adjusted for nucleoside pair, cART regimen, age, change in CD4 count since starting cART, time since starting cART, and time to initial virological suppression, showed little evidence of no further increases in CD4 count at higher current CD4 count levels (figure 1). For current CD4 counts of less than 700 cells per μL, there was about a 40-60 cells per μL CD4 count rise per year with a viral load of less than 50 copies per mL. There was a lower, but still significantly increasing, yearly change in CD4 counts in patients with a current CD4 count of more than 700 cells per μL (mean 24 cells per μL; 95% CI 1-46, p=0¥047). Adjusting for other variables, such as hepatitis B or C co-infection, exposure group, ethnic origin, or time spent with a viral load of less than 50 copies per mL did not alter these findings and were not related to rate of CD4 count change. Additionally, in the multivariable model, there was no interaction between the rate of increase in CD4 count (per year) and age (p=0¥61), change in CD4 since starting cART (p=0¥88), peak viral load (p=0¥22), CD4 count at starting cART (p=0¥38), or between current CD4 count and duration of cART (p=0¥60).
This analysis was then stratified both by current CD4 count and years since starting cART (figure 2). The greatest increases in CD4 count were seen in the year after starting cART (around 100 cells per μL), irrespective of the CD4 count at starting cART. Significant, but lower, increases in CD4 count (about 50 cells per μL per year) were seen up to 5 years after starting cART in patients whose current CD4 count was below 500 cells per μL. The only group without significant increases in yearly CD4 count was patients who had taken cART for more than 5 years with a current count of more than 500 cells per μL. The current mean CD4 count in this patient group was 774 cells per μL (95% CI 764-783), whereas in the first year of cART the current mean CD4 count was 658 cells per μL (640-676), in years 1-3 was 710 cells per μL (700-720), and in years 3-5 was 747 cells per μL (737-756). Similar results were seen after adjustment for nucleoside backbone, cART regimen started, age, and change in CD4 count since starting cART.
Additional analyses were done to investigate both the sensitivity and potential sources of bias in these findings. Analyses were repeated with stratification by CD4 count at starting cART, rather than current CD4 count, and the adjusted mean changes are shown in table 3. Of the patients with maximum virological suppression, those who started cART with CD4 counts of less than 200 cells per μL continued to have significant increases in CD4 count 5 years after starting cART. In this analysis, the only group without significant increases in CD4 count were those who started cART with a CD4 count of more than 350 cells per μL and had been taking cART for more than 3 years. A further sensitivity analysis censored patients on stopping cART, because patients with treatment interruptions and subsequent virological suppression might have higher rates of CD4 count change. Only a small proportion of the patients included in this analysis underwent a treatment interruption (less than 5%) and hence the results were entirely consistent with those shown above (data not shown). Similarly, ten viral load pairs were included from patients treated with interleukin-2 (0¥07%), and exclusion of these data did not alter our findings. Finally, various transformations of the change in CD4 count were investigated to assess the sensitivity of the generalised linear model to variability in the data. We also repeated analyses using alternative methods, including mixed models and binomial regression, modelling whether there was an increase in CD4 in different strata, rather than modelling the absolute change in CD4. All sensitivity analyses gave consistent results to those reported above.
Discussion
This study of CD4 count increases in antiretroviral-naive HIV-infected patients starting cART and who subsequently achieve maximum viral suppression has found little evidence of a plateau effect. Most patients continued to have significant rises in CD4 count, even at more than 5 years after cART initiation, whereas significant, but smaller, increases were seen in patients who started cART with low CD4 counts. Normalisation of CD4 counts in HIV-infected patients for all infected individuals was seen in patients starting cART with a CD4 count of more than 350 cells per _L and might be achieved in patients starting cART with lower CD4 counts if viral suppression can be maintained for a sufficiently long period of time.
Pre-infection CD4 counts were not known and we cannot to say whether the counts returned to this level. With sufficient duration of virological suppression, CD4 levels in patients starting cART with CD4 counts in excess of 350 cells per _L were approaching levels seen in HIV negative patients, of around 800 cells per _L.13,14 The rates of CD4 count increases seen after the first year of cART were consistent with those seen in HIV-negative patients after chemotherapy.15 Although patients who started cART with low CD4 counts continued to have significant rises in cell count, the absolute current level of CD4 count, roughly 500 cells per _L, was consistently lower than patients who started cART with higher counts. The rate of CD4 count increase diminished with increasing time since starting cART in patients starting cART with lower CD4 counts. This finding is consistent with an asymptotic rather than a plateau effect, whereby patients continue to have significant but progressively smaller increases in CD4 counts. Longer follow-up in this patient group is required to investigate whether CD4 counts in this group can reach the levels seen in patients starting cART with higher CD4 counts. Additionally, there is some evidence to suggest that functional immune reconstitution is incomplete in patients starting cART with lower CD4 counts.16
The increase in CD4 cell count will result in a corresponding reduction on the risk of opportunistic diseases or death associated with HIV, as the CD4 count, rather than viral load, remains one of the strongest markers of clinical disease progression.17-19 The risk has been shown to decrease as CD4 count increases,20,21 although there remain a small number of patients who develop opportunistic infections at higher than expected CD4 counts.22 Whether or not there will be a residual increased risk of clinical progression in patients after CD4 counts return to above 500 cells per _L or the levels seen in uninfected patients is unknown. Research from the SMART study group would suggest that the lowest rates of HIV-disease progression, liver-related events, or non-AIDS defining malignancies would be seen in patients who continue to take cART with virological suppression.23
Previous studies considering the occurence of a plateau in CD4 counts have addressed a range of questions using a various methods in both antiretroviral-naive and experienced patients with variable viral suppression.3,7,24-28 In 2001, Tarwater and colleagues26 found no further increases in CD4 count after 2-3¥5 years of cART in 314 HIV-infected patients from the MACS study, most whom were treatment-experienced. In 20 patients with moderately advanced disease enrolled in the ACTG 375 study, most changes in CD4 count took place within the first year of treatment, and there were no significant increases in CD4 count in the second or third year of therapy.27 Data from a larger patient group from the Swiss HIV Cohort Study3,7 describe CD4 count increases up to 4 years after starting cART. There was some evidence of a plateau, and around 40% of patients achieved a CD4 count of more than 500 cells per _L after starting cART. However, the subgroup of patients who remained on cART continued to have CD4 count increases at 4 years after starting treatment. In two further studies of patients with virological suppression,24,25 there was evidence of significant increases in CD4 counts at 3-4 years after starting cART, but there were limited data beyond this time to show whether further CD4 count increases would be seen. A more recent paper reported a return to near-normal levels in CD4 count among patients starting cART with a CD4 count less than 350 cells per _L, consistent with our findings, but unlike our results, found no significant increases in CD4 counts after 4 years of cART, irrespective of the level of immunodeficiency at starting cART.28
Crucially, this study differs in several ways from those previously published. We have used different methods to estimate CD4 count changes, taking advantage of the serial measurements in CD4 count and viral load recorded over time. This method means that all changes in CD4 count while a patient had maximum viral suppression were included in analyses, greatly increasing power of this study. Alternative statistical methods and sensitivity analyses showed consistent results. Patients in this study had extensive follow-up while viral load was below 50 copies per mL, had maximum virological suppression, and the analysis focused on their current CD4 count. This method indicates the current level of immunodeficiency, taking into account changes that have occurred to date. Other studies have had shorter follow-up, or have used higher limits of detection for viraemia.3,7,24-28
There are several points which should be considered. We excluded around half the antiretroviral-naive patients starting cART because of either incomplete virological suppression or because they did not satisfy the precise inclusion criteria, which were chosen a priori to determine CD4 count increases under optimum antiretroviral treatment. This means that our findings are only generalisable to patients who have an optimum response to cART and should therefore be regarded as a best-case scenario. Up to 30% of patients starting cART will not achieve viral suppression and the CD4 counts in these patients will not increase to the same extent.24,29,30 Many cART regimens are associated with both long term and short term toxicities,2,31,32 treatment interruptions are common,33-35 and viral load might not always be maintained at such a low level. This study includes patients with long term follow-up after starting cART, and half of the patients started cART before 2000. Since that time, there has been a change in the antiretrovirals available and the way they are combined,36,37 which has led to an improvement over time in the initial response to cART.38-40 This improvement in response would suggest that a higher proportion of patients starting cART using contemporary regimens would be more likely to have sustained maximum suppression of viraemia, and therefore the potential to achieve a CD4 count in the range seen in HIV-negative individuals. We chose an arbitrary baseline for our analyses, and this is clearly different to the baseline as defined in clinical trials. CD4 count changes most rapidly during the first few months of treatment with cART.5 As a consequence, in this study the observed change in CD4 count after some years of cART might be more similar to the changes observed in earlier years when compared with clinical trials. Finally, the patients included in our study were predominantly of white ethnic origin and the extent to which the results will be generalisable to patients from different ethnic origins is unclear.
In conclusion, we have shown that most patients with HIV who can maintain viral load at less than 50 copies per mL continue to have significant rises in CD4 counts, even after protracted exposure to combination therapy. Patients who started cART with a CD4 cell count of more than 350 cells per _L had CD4 cell counts approaching the level seen in HIV-negative individuals after more than 3 years of cART and had no further significant rises in CD4 counts.
|
|
| |
| |
|
|
|
