Time courses of the percentage of plasma human immunodeficiency virus–1 RNA levels below 400 copies/mL (A), CD4 T-lymphocyte count (B), and CD8 T-lymphocyte count (C) in the entire cohort (N = 2235), in the group continuously receiving highly active antiretroviral therapy (HAART) (CONT; n = 985), and in the group that intermittently discontinued HAART (DISCONT; n = 1250). For B and C, data are median and interquartile range (25th and 75th percentiles). D, Percentage of subjects in the CONT group reaching clinically relevant thresholds of CD4 T-lymphocyte counts (≥200/µL, ≥350/µL, and ≥500/µL) during the observation period.
Baseline CD4 T-lymphocyte count, virologic response, and CD4 recovery. A, Time course of CD4 T lymphocyte count (medians and interquartile ranges) in groups with continuous highly active antiretroviral therapy (HAART) (CONT; n = 985) stratified by baseline CD4 T-lymphocyte count. Of note, CD4 trajectories followed almost parallel time courses. Absolute increases in CD4 T-lymphocyte count were therefore in a similar range for all subgroups. B, However, fold-increase of CD4 T-lymphocyte count from baseline values was larger in individuals commencing HAART with lower CD4 T-lymphocyte counts. Dots represent absolute increases in CD4 T-lymphocyte count and open squares indicate the fold increase from baseline CD4 T-lymphocyte count. C, Time course of median CD4 T-lymphocyte count in the CONT group stratified by the reduction of plasma human immunodeficiency virus (HIV)-1 viremia.
Predictors of CD4 T-lymphocyte counts of 500/µL or greater at 4 years. Percentages of individuals reaching a CD4 T-lymphocyte count of 500/µL or greater in groups with continuous highly active antiretroviral therapy (n = 985) stratified by baseline CD4 T-lymphocyte count (A) and the percentage of viral loads below 400 copies/mL (B). P<.001 for all comparisons.
Kaufmann GR, Perrin L, Pantaleo G, Opravil M, Furrer H, Telenti A, Hirschel B, Ledergerber B, Vernazza P, Bernasconi E, Rickenbach M, Egger M, Battegay M, for the Swiss HIV Cohort Study Group. CD4 T-Lymphocyte Recovery in Individuals With Advanced HIV-1 Infection Receiving Potent Antiretroviral Therapy for 4 YearsThe Swiss HIV Cohort Study. Arch Intern Med. 2003;163(18):2187-2195. doi:10.1001/archinte.163.18.2187
Highly active antiretroviral therapy (HAART) for human immunodeficiency virus (HIV)-1 infection allows recovery of CD4 T lymphocytes. Few studies have explored the long-term T-lymphocyte responses to HAART.
Plasma HIV-1 RNA levels and CD4 and CD8 T-lymphocyte counts were longitudinally analyzed over 4 years in 2235 participants of the Swiss HIV Cohort, commencing HAART between 1996 and 1997. The CD4 T-lymphocyte count increase, the percentage of individuals with a CD4 T-lymphocyte count of 500/µL or greater and less than 200/µL, and the determinants of CD4 T-lymphocyte recovery were evaluated in individuals treated with continuous (CONT; n = 985) and discontinuous (DISCONT; n = 1250) HAART.
At 4 years, 69.5% of subjects (CONT, 84.5%; DISCONT, 53.6%; P<.001) showed HIV-1 RNA levels below 400 copies/mL, while the median CD4 T-lymphocyte count increased from 190/µL to 423/µL (CONT, 486/µL; DISCONT, 343/µL; P<.001). Of the 2235 participants, 38.8% (CONT, 47.7%; DISCONT, 29.4%; P<.001) reached a CD4 T-lymphocyte count of 500/µL or greater, but in 15.6%, CD4 T-lymphocyte count remained below 200/µL (CONT, 5.9%; DISCONT, 25.9%; P<.001). Larger increases in CD4 T-lymphocyte count were associated with higher baseline HIV-1 RNA, a larger percentage of undetectable HIV-1 RNA levels, lower baseline CD8 T-lymphocyte count, and younger age. Individuals reaching a CD4 T-lymphocyte count of 500/µL or greater at 4 years were characterized by higher nadir and baseline CD4 T-lymphocyte counts and a more sustained reduction of HIV-1 RNA levels.
At 4 years, only 39% of individuals treated with HAART reached a CD4 T-lymphocyte count of 500/µL or greater, and 16% with CD4 T-lymphocyte counts less than 200/µL remained susceptible to opportunistic infections. Treatment interruptions, a poor virologic response, and older age were the major factors negatively affecting the recovery of CD4 T lymphocytes.
HIGHLY ACTIVE antiretroviral therapy (HAART) has significantly improved the prognosis of individuals infected with human immunodeficiency virus (HIV)-1 and extended life expectancy.1- 5 The recovery of CD4 T lymphocytes in treated persons is usually accompanied by enhanced T-lymphocyte responses to antigens and mitogens,6- 8 providing adequate protection against opportunistic infections.3,9- 11 However, the immunological recovery shows a large variability from patient to patient.12- 16 A significant proportion of treated subjects experience only small increases in CD4 T-lymphocyte numbers despite optimum suppression of plasma HIV-1 viremia by HAART. In others, CD4 T-lymphocyte count rapidly rises despite modest reductions of plasma HIV-1 RNA levels.17 The reasons for these different responses remain unknown. It has been suggested that differences in residual thymic function may play a role,18 but additional factors may potentially affect the recovery of CD4 T lymphocytes, including age,19 the degree of immunodeficiency before initiation of HAART,20 residual viral activity,21 viral coinfections such as hepatitis C or G,22- 24 and the susceptibility of CD4 T cells to HIV-1 infection, which depends in part on the expression of the viral coreceptor CCR5.25- 27
Although the long-term recovery of CD4 T lymphocytes deserves particular attention because of its major clinical significance, few studies have addressed this issue. In this investigation, we have evaluated the long-term T-lymphocyte dynamics over 4 years in 2235 subjects enrolled in the Swiss HIV Cohort Study, who were infected with HIV-1, and commenced HAART between 1996 and 1997. In addition, potential predictors of CD4 T-lymphocyte recovery above the clinically relevant levels of 200/µL and 500/µL were analyzed, which represent the threshold of the occurrence of opportunistic infections and the lower limit of the physiological range.
Individuals of the Swiss HIV Cohort Study were eligible if they had commenced HAART between January 1996 and December 1997. Highly active antiretroviral therapy was defined as (1) a combination of 2 nucleoside analogue reverse transcriptase inhibitors (NRTIs) in combination with a protease inhibitor or a nonnucleoside analogue reverse transcriptase inhibitor (NNRTI), (2) 2 protease inhibitors in combination with at least 1 NRTI, (3) a combination of a protease inhibitor and an NNRTI with at least 1 NRTI, or (4) a combination of 3 NRTIs. Data of 2235 individuals (73.9%) were used in the final analysis after excluding 786 individuals (26.0%) owing to missing baseline CD4 and CD8 T-lymphocyte counts or plasma HIV-1 RNA levels. The baseline value was defined as the last laboratory determination within a period of 3 months prior to the initiation of HAART.
Most individuals were male (72.8%) and white (86.8%). The mean ± SD values for age and estimated duration of HIV-1 infection (based on the first positive HIV-1 test result) were 38 ± 9 and 7.2 ± 4.1 years, respectively. The cohort was in an advanced stage of HIV-1 infection. The median baseline CD4 T-lymphocyte count was 190/µL, and a third of patients met the criteria of Centers for Disease Control and Prevention (CDC, Atlanta, Ga) stage C. A large percentage of participants (65%) were pretreated with nucleoside analogue combinations. Additional baseline characteristics are given in Table 1.
In 1996 and 1997, potent antiretroviral therapy predominantly consisted of indinavir sulfate, ritonavir, or nelfinavir mesylate in combination with 2 nucleoside analogues (Table 2). During the study period, the proportion of subjects receiving double protease inhibitor therapy, NNRTI therapy, and combined protease inhibitor and NNRTI therapy increased.
Data were extracted from the January 2002 update of the Swiss HIV-1 Cohort database and included patient characteristics, treatment history, CD4 and CD8 T-lymphocyte counts, and plasma HIV-1 RNA levels. Laboratory values were determined irregularly at 1- to 6-month intervals, of which the 6-month values were calculated by linear interpolation. These calculations were based on a mean ± SD number of 12.6 ± 5.4 laboratory determinations per patient during the observation period. Although recently more sensitive assays have become available, undetectable plasma HIV-1 RNA levels were defined as values below 400 copies/mL.
Longitudinal T-lymphocyte counts and HIV-1 RNA levels were separately analyzed in individuals who had continuously received HAART (CONT; n = 985) and in subjects who discontinued HAART for at least 1 month (DISCONT; n = 1250). In the CONT group, 725 individuals (73.6%) changed the HAART regimen, but only 51 (5.2%) were lost to follow-up, 10 of whom had died. In the DISCONT group, 458 (36.6%) individuals were lost to follow-up during the observation period, including 179 (14.3%) subjects who had died, 39 (3.1%) who had moved to another country, 28 (2.2%) who had discontinued participation in the Swiss Cohort study, and 212 (17.0%) who were lost to follow-up for unknown reasons. Of the 1250 subjects in the DISCONT group, 950 (76.0%) stopped antiretroviral therapy at least once for a median time of 3 (interquartile range, 1-10) months. The remaining 300 individuals (24.0%) were intermittently switched to non-HAART regimens for a median time of 4 (interquartile range, 1-12) months.
Three end points were evaluated and included absolute increases in CD4 T-lymphocyte count over 4 years, the percentage of subjects reaching a CD4 T-lymphocyte count of 500/µL or greater, and the percentage of individuals with a CD4 T-lymphocyte count less than 200/µL. Immunological end points and potential predictors of the immunological response were analyzed in individuals who had continuously received HAART (CONT; n = 985). The following factors that could potentially affect the recovery of CD4 T lymphocytes were evaluated: sex, age, hepatitis C virus (HCV) serology (data available for 810 individuals), the duration of HIV-1 infection, the CDC stage, antiretroviral pretreatment before initiation of HAART, baseline HIV-1 RNA levels, baseline CD4 and CD8 T-lymphocyte counts, the number of therapy changes, and the percentage of HIV-1 RNA levels below 400 copies/mL.
The statistical analysis was performed using a general linear model and logistic regression. In the multivariate models, variables were entered stepwise in the order of strength of relationship found in the univariate model. A 2-sided P value less than .05 was considered statistically significant. All statistical calculations were performed using SPSS release 10.0 (SPSS Inc, Chicago, Ill).
In the entire cohort (N = 2235), median plasma HIV-1 RNA declined from 4.6 log10 copies/mL to values below the limit of detection. At 48 months, 69.5% of treated individuals had undetectable plasma HIV-1 RNA levels (Figure 1A). During the observation period, the median CD4 T-lymphocyte count increased from 190/µL to 423/µL (Figure 1B). Annual increases in CD4 T-lymphocyte count became gradually smaller over time. The percentage of individuals reaching CD4 T-lymphocyte counts of 500/µL or greater increased from 9.1% at baseline to 38.8% at 48 months. Of the 2235 subjects, 15.6% showed poor immunological responses and had persistently low CD4 T-lymphocyte counts below 200/µL.
In the CONT group, 84.5% showed HIV-1 RNA levels below 400 copies/mL at 48 months, whereas the median CD4 T-lymphocyte count increased from 203/µL to 486/µL (Figure 1B). In 50.3%, CD4 T-lymphocyte count increased to 500/µL or greater, whereas in 26.8%, the number of CD4 T lymphocytes remained below 350/µL (Figure 1D). In this group, 264 individuals (6.1%) were classified as poor immunological responders (CD4 T-lymphocyte count <200/µL).
The percentage of patients reaching HIV-1 RNA levels below 400 copies/mL and the median CD4 T-lymphocyte count at 48 months was smaller in the DISCONT group than in the CONT group (53.6% vs 84.5%; 343/µL vs 486/µL, respectively; P<.001 for both comparisons). In addition, the percentage of individuals reaching CD4 T-lymphocyte counts of 500/µL or greater was smaller (29.4% vs 47.7%; P<.001) and the percentage of poor immunological responders was larger (25.9% vs 5.9%; P<.001). CD8 T lymphocytes showed smaller changes over time compared with CD4 T lymphocytes, increasing from 740/µL to 869/µL in the CONT group and from 697/µL to 872/µL in the DISCONT group (Figure 1C).
The potential factors associated with the recovery of CD4 T lymphocytes are given in Table 3. In the multivariate model, sex did not predict the recovery of CD4 T-lymphocyte count, although larger increases of CD4 T lymphocytes were observed in female study participants (323/µL vs 279/µL; Table 3). In contrast, younger age represented an independent predictor of larger rises in CD4 T-lymphocyte count (P = .003). The difference between men and women regarding the increase in CD4 T-lymphocyte count may therefore be the result of the significantly younger age of female study participants (33 vs 37 years; P<.001).
The recovery of CD4 T lymphocytes showed no association with the mode of HIV transmission and was comparable in HCV-seropositive and -seronegative individuals. Nevertheless, significantly smaller rises in CD4 T-lymphocyte count were observed in HCV antibody–positive subjects in the first year of therapy (126/µL vs 151/µL at 1 year [P = .005] and 281/µL vs 294/µL at 48 months [P = .34]).
In the CONT group, 60.1% of individuals had been pretreated with antiretroviral agents for a median duration of 19 months. They showed smaller median increases in CD4 T-lymphocyte count than treatment-naive individuals (256/µL vs 335/µL at 48 months; P<.001). Nevertheless, in the multivariate model, pretreatment was not independently associated with the recovery of CD4 T lymphocytes (P = .59).
Importantly, baseline and nadir CD4 T-lymphocyte counts before initiation of HAART were not correlated with the recovery of CD4 T lymphocytes. Hence, the longitudinal time course of CD4 T lymphocytes in groups stratified by baseline CD4 T-lymphocyte count followed roughly parallel time courses (Figure 2A). As a consequence, the proportional increase of CD4 T-lymphocyte count from baseline was larger in individuals commencing HAART with lower baseline CD4 T-lymphocyte counts (Figure 2B). In contrast, lower baseline CD8 T-lymphocyte counts were independently associated with significantly larger rises in CD4 T-lymphocyte count (P = .008).
Higher baseline viral load was significantly associated with larger increases in CD4 T-lymphocyte count, both short term (3-6 months; r = 0.235; P<.001) and long term (48 months; r = 0.218; P<.001). In addition, more sustained suppression of HIV-1 viremia as measured by the percentage of plasma HIV-1 RNA values below 400 copies/mL during the observation period improved the recovery of CD4 T lymphocytes (P<.001; Figure 2C).
Higher nadir CD4 T-lymphocyte count indicating the maximum extent of HIV-1–related immunodeficiency increased the likelihood of reaching CD4 T-lymphocyte counts of 500/µL or greater within 48 months (adjusted odds ratio [OR], 1.38 per 100-cell increase; Table 4). Similarly, the likelihood of reaching this CD4 T-lymphocyte count was higher in individuals with a larger number of CD4 T lymphocytes at baseline (adjusted OR, 2.04 per 100-cell increase). A baseline CD4 T-lymphocyte count ranging from 300/µL to 400/µL appeared thereby to represent a critical threshold. Almost 70% of subjects commencing HAART with a CD4 T-lymphocyte count of 300/µL to 400/µL reached the end point of a CD4 T-lymphocyte count of 500/µL or greater, whereas only 51% initiating HAART with a CD4 T-lymphocyte count of 200/µL to 300/µL and 40% with a count of 100/µL to 200/µL reached this target level (Figure 3A).
The virologic response was a further relevant predictor. Higher baseline plasma HIV-1 RNA levels (adjusted OR, 1.34 per 1-log increase) and a higher percentage of undetectable plasma HIV-1 RNA levels increased the likelihood of reaching a CD4 T-lymphocyte count of 500/µL at 48 months (adjusted OR, 1.27 per 10% increase; Figure 3B).
The risk of a poor recovery of CD4 T lymphocytes (<200/µL at 48 months) was significantly higher in subjects with lower baseline CD4 T-lymphocyte counts (adjusted OR, 0.26 per 100-cell increase; Table 5), lower baseline HIV-1 RNA levels (adjusted OR, 0.65 per 1-log increase), and a smaller percentage of undetectable plasma HIV-1 RNA levels during the observation period (adjusted OR, 0.75 per 10% increase).
In most HIV-1–infected individuals treated with HAART, CD4 T lymphocytes recover to levels above 200/µL, at which HIV-1–related clinical complications are exceedingly rare.11 Nevertheless, many clinicians as well as patients aim at higher, more physiological levels of CD4 T lymphocytes, preferably above 500/µL. The percentage of individuals who ultimately reach this goal and the factors affecting the recovery of CD4 T lymphocytes were the focus of the present study.
At 4 years, only 39% of 2235 participants of the Swiss HIV Cohort reached CD4 T-lymphocyte counts above 500/µL and 16% had counts less than 200/µL, remaining susceptible to opportunistic infections. Higher baseline and nadir CD4 T-lymphocyte counts increased the likelihood of reaching a CD4 T-lymphocyte count of 500/µL within 4 years. Poor immunological responders failing to reach a CD4 T-lymphocyte count of 200/µL commenced HAART with a smaller number of CD4 T lymphocytes and showed poorer virologic responses. Larger absolute increases in CD4 T-lymphocyte count were observed in individuals who were younger, had higher baseline plasma HIV-1 RNA levels, lower CD8 T-lymphocyte counts, and a more sustained reduction of plasma HIV-1 viremia.
At 4 years, almost 70% of individuals commencing HAART with a CD4 T-lymphocyte count of 300/µL to 400/µL had CD4 T-lymphocyte counts of at least 500/µL. The percentage of subjects reaching this goal declined to 51% in individuals commencing HAART with a CD4 T-lymphocyte count less than 300/µL, which suggests that HAART should preferably be initiated before CD4 T-lymphocyte count has declined below 300/µL. This is in good agreement with current treatment guidelines that propose to commence HAART at a CD4 T-lymphocyte count of 350/µL.28
CD4 T lymphocytes showed a rapid increase in the first 3 months, followed by an almost linear rise in the subsequent 2 years. However, annual changes in CD4 T-lymphocyte count thereafter became gradually smaller, suggesting that the number of CD4 T lymphocytes may reach a plateau level. Therefore, a normalization of CD4 T-lymphocyte count does not appear feasible in all individuals or will require at least very long-term therapy. This is further supported by the fact that the increase of CD4 T lymphocytes did not depend on baseline values. The comparison of CD4 T lymphocytes of groups stratified by baseline CD4 T-lymphocyte count revealed that CD4 trajectories followed almost parallel time courses. Hence, individuals commencing HAART with a small number of CD4 T lymphocytes will reach, after the same observation period, on average lower CD4 T-lymphocyte counts than subjects who initiate HAART at higher CD4 T-lymphocyte numbers.
The delayed reconstitution of the CD4 T-cell compartment in elderly patients has previously been observed in smaller studies.19,27 The thymus significantly contributes to the recovery of CD4 T lymphocytes,29 but loses partly its functionality after physiological involution during puberty. This may provide an explanation for the poorer recovery of CD4 T lymphocytes in older patients.
Recovery of CD4 T lymphocytes was significantly associated with more sustained reduction of plasma HIV-1 viremia. However, we did not find a significant difference regarding the increase in CD4 T-lymphocyte count between individuals with 75% to 90% and more than 90% undetectable viral loads during the observation period. In contrast, individuals with more than 25% detectable HIV-1 RNA levels showed a trend toward a poorer recovery of CD4 T lymphocytes. This observation supports the concept of a critical viral load threshold, which determines whether CD4 T lymphocytes recover or further decline.30
In the first year of HAART, HCV-seropositive individuals showed smaller increases in CD4 T-lymphocyte count than HCV-seronegative persons, which is in agreement with the results of a previous investigation of the Swiss HIV Cohort Study Group.22 The negative effect of HCV infection on the recovery of CD4 T lymphocytes disappeared during the observation period. However, the CONT and DISCONT groups were not balanced for HCV infection. Most HCV-seropositive individuals were observed in the DISCONT group (41% vs 31%), but the analysis of predictors of CD4 T-lymphocyte recovery was based on data of the CONT group.
The substantial percentage (26%) of poor immunological responders in the DISCONT group indicates that a sizable proportion of individuals remains at risk for opportunistic infections over a lengthy period despite the availability of potent antiretroviral regimens. The substantial percentage of individuals who intermittently discontinued HAART also reflects the problems associated with currently available antiretroviral drug regimens such as frequent adverse events and poor adherence to the drug regimen.31,32
One limitation of this prospective cohort study is the follow-up pattern, which could have introduced bias. For example, patients with slow disease progression may have avoided outpatient visits, ultimately reducing the amount of information. This "walking well" phenomenon may have resulted in an underestimation of the recovery of CD4 T lymphocytes. Conversely, patients in more advanced stages of HIV-1 infection were more likely to discontinue HAART or to leave the study, which may have introduced a bias in the opposite direction.
This prospective observational cohort study shows that the recovery of CD4 T lymphocytes in treated HIV-1–infected individuals is slower than initially anticipated. Indeed, CD4 T-lymphocyte count failed to recover significantly in 16% of individuals, and more than 50% did not reach CD4 T-lymphocyte counts of 500/µL or greater after 4 years of HAART. Of further concern is the observation that the number of CD4 T lymphocytes appeared to reach a plateau level after 2 to 3 years. This suggests that CD4 T-lymphocyte count remains in a critically low range in a significant number of individuals. Therefore, HAART should not be deferred to late stages of HIV-1 infection, although the advantages of an earlier initiation of HAART have to be individually counterbalanced against potential adverse events and the rapid emergence of drug resistance.
Corresponding author and reprints: Manuel Battegay, MD, Division of Infectious Diseases, Department of Internal Medicine, University Hospital Basel, Petersgraben 4, CH-4031 Basel, Switzerland (e-mail: email@example.com).
Accepted for publication December 10, 2002.
This study has been financed in the framework of the Swiss HIV Cohort Study and was supported by grant 3345-062041 (project 343) from the Swiss National Science Foundation.
The study has been presented in part at the Ninth Conference on Retroviruses and Opportunistic Infections (abstract LB8); February 24-28, 2002; Seattle, Wash.
M. Battegay, M.-C. Bernard, E. Bernasconi, H. Bucher, P. Bürgisser, M. Egger, P. Erb, W. Fierz, M. Flepp (Chairman of the Clinical and Laboratory Committee), P. Francioli (President of the Swiss HIV Cohort Study), H. J. Furrer, M. Gorgievski, H. Günthard, P. Grob, B. Hirschel, C. Kind, T. Klimkait, B. Ledergerber, U. Lauper, M. Opravil, F. Paccaud, G. Pantaleo, L. Perrin, J.-C. Piffaretti, M. Rickenbach (Head of Data Center), C. Rudin (Chairman of the Mother & Child Substudy), J. Schupbach, A. Telenti, P. Vernazza (Chairman of the Scientific Board), T. Wagels, R. Weber.