| |
CD4 Cell Counts of 800 Cells/mm3 or Greater After 7 Years of Highly Active Antiretroviral Therapy Are Feasible in Most Patients Starting With 350 Cells/mm3 or Greater
|
| |
| |
JAIDS Journal of Acquired Immune Deficiency Syndromes:Volume 45(2)1 June 2007pp 183-192
Gras, Luuk MSc*; Kesselring, Anouk M MSc, MD*; Griffin, James T PhD; van Sighem, Ard I PhD*; Fraser, Christophe PhD; Ghani, Azra C PhD; Miedema, Frank PhD; Reiss, Peter MD, PhD_; Lange, Joep M A MD, PhD_; de Wolf, Frank MD, PhD* ; on Behalf of the ATHENA, Netherlands National Observational Cohort Study
From the *HIV Monitoring Foundation, Amsterdam, The Netherlands; Department of Infectious Diseases Epidemiology, Imperial College School of Medicine, London, United Kingdom; Department of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London, United Kingdom; Department of Immunology, University Medical Centre, Utrecht, The Netherlands; and _Center for Infection and Immunity, Department of Infectious Diseases, Tropical Medicine and AIDS, Academic Medical Centre of the University of Amsterdam, Amsterdam, The Netherlands.
"Adjusted hazard ratios compared with those of patients with a pre-HAART CD4 count of 200 to 350 cells/mm3 were 0.26 and 0.46, respectively, for those with <50 and 50 to 200 cells/mm3 and were 2.84 and 6.79, respectively, for those with 350 to 500 and ≥500 cells/mm3".....during periods of viremia (HIV RNA >500 copies/mL after the initial 6 months of HAART), the slope of CD4 cell count was estimated to be lower than when HIV RNA levels were less than 500 copies/mL (P < 0.0001)." (from Jules: patients with less than 100,000 c/ml viral load at baseline appeared to have better chance of achieving >800 CD4 count; viremia between 5 and 7 years at least once [at let 1 HIV RNA measurement >1000 c/ml] carries a 4.69 OR risk for plateauing CD4 count)....."The median CD4 counts at 7 years were 410, 548, 660, 780, and 870 cells/mm3 for those with pre-HAART CD4 counts of <50, 50 to 200, 200 to 350, 350 to 500, and ≥500 cells/mm3, respectively"....restoration to CD4 cell levels ≥800 cells/mm3 is a lengthy process and may not be feasible for patients who start HAART with <200 CD4 cells/mm3......The association between periods of HIV production despite HAART and reaching a CD4 cell plateau earlier and (thus) at a lower level confirms the importance of monitoring of HIV RNA and keeping plasma levels to less than 500 copies/mL..."
"HAART restoration of CD4 cell counts in HIV-infected individuals to levels normally seen in uninfected individuals takes a long time and is not feasible within 7 years in most patients who initiate HAART with CD4 cell counts <350 cells/mm3. Patients ≥50 years of age when starting HAART and patients with periods of viremia (HIV RNA level >500 copies/mL) experience smaller increases and are more likely to reach a CD4 cell plateau earlier and at a lower level. Given the better toxicity profiles of the currently used antiretroviral combinations, particularly in patients older than 50 years of age, it may be beneficial to start HAART earlier than current guidelines recommend."
Abstract
Objective: CD4 cell count changes in therapy-naive patients were investigated during 7 years of highly active antiretroviral therapy (HAART) in an observational cohort.
Methods: Three endpoints were studied: (1) time to ≥800 CD4 cells/mm3 in 5299 therapy-naive patients starting HAART, (2) CD4 cell count changes during 7 years of uninterrupted HAART in a subset of 544 patients, and (3) reaching a plateau in CD4 cell restoration after 5 years of HAART in 366 virologically suppressed patients.
Results: Among patients with <50, 50 to 200, 200 to 350, 350 to 500, and ≥500 CD4 cells/mm3 at baseline, respectively, 20%, 26%, 46%, 73%, and 87% reached ≥800 CD4 cells/mm3 within 7 years of starting HAART. Periods with HIV RNA levels >500 copies/mL and age ≥50 years were associated with lesser increases in CD4 cell counts between 6 months and 7 years. Having reached ≥800 CD4 cells/mm3 at 5 years, age ≥50 years, and ≥1 HIV RNA measurement >1000 copies/mL between 5 and 7 years were associated with a plateau in CD4 cell restoration.
Conclusions: Restoration to CD4 cell counts ≥800 cells/mm3 is feasible within 7 years of HAART in most HIV-infected patients starting with ≥350 cells/mm3 and achieving sufficient suppression of viral replication. Particularly in patients ≥50 years of age, it may be beneficial to start earlier than current guidelines recommend.
RESULTS
Notice baseline viral load predicts CD4 increase, viral load below 100,000 c/ml increases chance for achieving 800 CD4.
Time to CD4 Cell Counts ≥800 Cells/mm3
The characteristics of the 5299 patients at the start of HAART are shown in Table 1. Most (76%) were male, 50% were MSM, and 63% originated from WCE/NA/A. The median HIV RNA concentration in plasma was 5.0 log10 copies/mL. A pre-HAART CD4 count of <200 cells/mm3 was found in 2703 patients (51%).
The time required to restore CD4 counts to ≥800 cells/mm3 was associated with a higher pre-HAART CD4 cell count. After 7 years of HAART, Kaplan-Meier estimates of the percentage of patients reaching ≥800 CD4 cells/mm3 were 20%, 26%, 46%, 73%, and 87% for those with a pre-HAART CD4 count of <50, 50 to 200, 200 to 350, 350 to 500, and ≥500 cells/mm3, respectively. Adjusted hazard ratios compared with those of patients with a pre-HAART CD4 count of 200 to 350 cells/mm3 were 0.26 and 0.46, respectively, for those with <50 and 50 to 200 cells/mm3 and were 2.84 and 6.79, respectively, for those with 350 to 500 and ≥500 cells/mm3 (Table 2). Female gender and higher pre-HAART HIV RNA levels were associated with a shorter time to CD4 cell counts ≥800 cells/mm3. Older age, Southeast Asian or sub-Saharan African origin, and HIV infection through intravenous drug use were associated with a longer time to this endpoint. There were no significant differences according to different pre-HAART CD8 cell count strata in adjusted models (P = 0.58).
Long-Term CD4 Cell Response in Patients on Uninterrupted HAART
Also in Table 1 are the demographic and clinical characteristics of the subset of 554 patients on uninterrupted HAART for 7 years. These patients started HAART between July 1, 1996, and June 30, 1998, and were among those included in the first analysis. Because the inclusion criteria for the second analysis implied starting HAART in earlier calendar years, there was a higher proportion of men, MSM, and patients originating from WCE/NA/A.
The median CD4 count increased from 221 (interquartile range [IQR]: 80-340) cells/mm3 at the start to 607 (IQR: 440-800) cells/mm3 after 7 years of HAART. The median CD4 counts at 7 years were 410, 548, 660, 780, and 870 cells/mm3 for those with pre-HAART CD4 counts of <50, 50 to 200, 200 to 350, 350 to 500, and ≥500 cells/mm3, respectively (Fig. 1A). Overall, increases were a median of 136 cells/mm3 during the first 24 weeks and leveled off over time to 40 cells/mm3 in weeks 96 through 144 and to 0 cells/mm3 in weeks 312 through 360. Median increases in CD4 cell counts after 7 years of HAART were between 367 and 410 cells/mm3 for the 4 pre-HAART CD4 count strata <500 cells/mm3, whereas increases were 287 cells/mm3 for patients in the ≥500 cells/mm3 stratum (see Fig. 1B; Wilcoxon test, P = 0.007).
Of 554 patients, 344 (62.1%) had HIV RNA plasma concentrations <500 copies/mL at all measurements taken between 6 months and 7 years after starting HAART. The remaining 210 patients had at least 1 HIV RNA result ≥500 copies/mL. In 80 of them, plasma concentrations were between 500 and 1000 copies/mL, which, in the majority (54 patients), occurred between 6 months and 3 years after the start of HAART. Periods of HIV viremia occurring after initial virologic success were found in 27.6% of the patients from WCE/NA/A and 42.2% of the patients (P = 0.006) with a non-WCE/NA/A origin.
The multivariate longitudinal analyses included 13,528 CD4 cell count measurements for the 554 patients. The median number of measurements per patient was 25 (minimum to maximum: 6-58). The model estimates can be interpreted as the slope or annual rate of change in CD4 cell count (on a square root scale). During the first 6 months, the slope of CD4 cell count was higher in patients with a pre-HAART HIV RNA measurement ≥4.5 log10 copies/mL than in those with <4.5 log10 copies/mL (P = 0.009). Furthermore, the slopes of CD4 cell count during the first 6 months were higher in women than in men (P = 0.003), in patients originating from regions other than WCE/NA/A (P = 0.04), and in patients with a pre-HAART CD8 count <1300 cells/mm3 as compared with ≥1300 cells/mm3 (P = 0.0007). The effect of body weight at the start of HAART on the slope of CD4 cell count was not significant in univariate or multivariate models (data not shown). Between 6 months and 7 years after starting HAART, the slopes did not differ significantly between men and women, between patients from various origins, or between patients with various levels of pre-HAART CD8 cells or pre-HAART HIV RNA. The slopes of CD4 cell count between 6 months and 7 years were significantly higher in patients <50 years of age as compared with those ≥50 years of age at the start of HAART (P < 0.0001). There were no significant differences in CD4 cell count increases between 0 and 6 months according to age. Finally, during periods of viremia (HIV RNA >500 copies/mL after the initial 6 months of HAART), the slope of CD4 cell count was estimated to be lower than when HIV RNA levels were less than 500 copies/mL (P < 0.0001).
To facilitate interpretation, estimates from the longitudinal mixed-effect model were back-transformed from the square root scale to the usual absolute CD4 cell count scale. The estimated median CD4 count after 6 months of HAART was 448 (95% confidence interval [CI]: 419 to 478) cells/mm3 for a male reference patient of Western origin aged <50 years who started HAART with <1300 CD8 cells/mm3, 310 CD4 cells/mm3, and HIV RNA <4.5 log10 copies/mL. This is compared with 504 (459-551) CD4 cells/mm3 for a female patient, 483 (457-509) cells/mm3 for a patient with a pre-HAART HIV RNA level ≥4.5 log10 copies/mL, 403 (370-437) cells/mm3 for a patient with ≥1300 CD8 cells at the start of HAART, and 418 (381-456) cells/mm3 for a patient not born in WCE/NA/A. The estimated median CD4 count after 7 years of uninterrupted HAART for the reference patient was 704 (656-754) cells/mm3 compared with 585 (522-652) cells/mm3 for a patient aged ≥50 years at the start of HAART and 648 (598-701) cells/mm3 for a patient with viremia >500 copies/mL at all HIV RNA measurements between 5 and 7 years. The estimated median CD4 cell values and values at other time points are graphically depicted in Figure 2.
FIGURE 2. A-F, Estimated long-term median CD4 cell response according to the 5 pre-HAART CD4 cell count strata of <50, 50 to 200, 200 to 350, 350 to 500, and ≥500 cells/mm3. The solid lines in each graph denote the median CD4 cell response for the reference patient: a male patient originating from WCE/NA/A who is <50 years of age at the start of HAART with a pre-HAART HIV RNA level <4.5 log10 copies/mL and a pre-HAART CD8 count <1300 cells/mm3 and in whom all HIV RNA measurements between 6 months and 7 years after starting HAART were <500 copies/mL. The dashed lines in A display the median CD4 cell response for patients with all HIV RNA measurements ≥500 copes/mL between 6 months and 3 years, between 3 and 5 years, and between 5 and 7 years after starting HAART. The dashed lines in the other parts of this figure display the median CD4 cell response for patients ≥50 years of age at the start of HAART (B), female patients (C), patients with a pre-HAART CD8 cell count ≥1300 cells/mm3 (D), patients with a pre-HAART HIV RNA level ≥4.5 log10 copies/mL (E), and patients not from WCE/NA/A origin (F).
Decreases in CD4 Cell Count in Virologically Suppressed Patients
Finally, we selected 366 patients who took 7 years of uninterrupted HAART and in whom all measured HIV RNA levels between 6 months and 5 years after starting HAART were <500 copies/mL. Their distribution over the 5 pre-HAART CD4 cell strata and the other baseline variables was similar to that in the previous longitudinal analysis. The estimated median CD4 count at 5 years after the start of HAART for patients aged <50 years was 631 (IQR: 459-812) cells/mm3 and 489 (IQR: 412-725) cells/mm3 for those aged ≥50 years (Wilcoxon test, P = 0.03). In total, 150 patients had negative CD4 cell slopes between 5 and 7 years of uninterrupted HAART use. Variables independently associated with this outcome, as identified in a multivariate logistic regression analysis, are shown in Table 3 and were age ≥50 years at the start of HAART (odds ratio [OR] for a negative slope between 5 and 7 years = 3.01 [95% CI: 1.60 to 5.67] compared with age <50 years); at least 1 HIV RNA measurement ≥1000 copies/mL compared with all HIV RNA measurements <1000 copies/mL between 5 and 7 years after starting HAART (OR = 6.10 [95% CI: 2.12 to 17.51]); and, finally, a higher CD4 cell count at 5 years. Compared with patients with ≥800 cells/mm3 at 5 years, the OR for patients with <400 cells/mm3 was 2.23 (IQR: 1.10-4.51). The OR for patients with 400 to 600 CD4 cells/mm3 at 5 years was 1.07 (IQR: 0.54-2.11; P = 0.85), and the OR for patients with 600 to 800 cells/mm3 at 5 years was 1.08 (IQR: 0.54-2.15; P = 0.82); neither was significantly different from those for patients with <400 cells/mm3 at 5 years. When age was included in the model as a continuous variable, the OR was 1.027 (95% CI: 1.003 to 1.052, P = 0.02) for each year by which the starting age was increased. In multivariate analysis, region of origin (P = 0.81), pre-HAART HIV RNA level (P = 0.55), or gender (P = 0.55) was not associated with negative CD4 cell slopes between 5 and 7 years, nor was the pre-HAART CD4 cell count (P = 0.66) after controlling for the CD4 cell count at 5 years.
DISCUSSION
We performed 3 types of analyses. First, we evaluated determinants of CD4 cell count recovery up to 800 cells/mm3 in a cohort of HAART-naive patients. Second, we evaluated changes in CD4 cell count in patients on uninterrupted HAART for 7 years to determine the maximum capacity of the immune system to restore CD4 cell numbers. Finally, we determined predictors for decreases in CD4 cell count after 5 years of virologically successful uninterrupted HAART. The first 2 analyses showed that 7 years after starting HAART, patients starting with lower pre-HAART CD4 counts experienced less restoration of CD4 cell counts than patients starting with higher pre-HAART CD4 cell counts. The third analysis showed that the plateau effect found after long-term CD4 cell restoration is associated with achievement of CD4 levels in the normal range. Plateauing of CD4 cell counts at a less than normal range is associated with insufficient suppression of HIV replication and with older age at the start of HAART. The strength of this study is the long follow-up (7 years) in a large number of naive patients with a variety of pre-HAART CD4 cell counts. We did not look at differences in CD4 cell response between individual drugs or drug classes because that is beyond the scope of this report and is the topic of a future analysis.
The largest gains in the number of CD4 cells occurred in the first 6 months after starting HAART, presumably because of redistribution of CD4 cells from lymphoid tissue.35 Thereafter, the rate of increase in CD4 cell counts gradually slowed. Between 5 and 7 years of uninterrupted HAART, CD4 cells still continued to increase in patients with a pre-HAART CD4 count less than 350 cells/mm3. Because of the slow rate of increase, however, restoration to CD4 cell levels ≥800 cells/mm3 is a lengthy process and may not be feasible for patients who start HAART with <200 CD4 cells/mm3.
The association between periods of HIV production despite HAART and reaching a CD4 cell plateau earlier and (thus) at a lower level confirms the importance of monitoring of HIV RNA and keeping plasma levels to less than 500 copies/mL. The single study31 that did not find a plateau effect, in contrast to our study and others,29,30 might reflect different levels of ongoing viremia or different age distributions among studies. The association of a lower CD4 cell count plateau with older age (≥50 years of age at the start of HAART) could reflect the lower normal CD4 cell range reported in older healthy individuals.36-38 Larger CD4 cell gain in treated patients has previously been associated with younger age,16,28 and less gain has been attributed to lower thymic function with older age.14,15 Because patients with low CD4 cell counts remain at risk of developing new AIDS events after starting HAART, it may be appropriate to start antiretroviral therapy in older (ie, ≥50 years of age) patients earlier than in younger patients.
Factors associated with differences in CD4 cell response during the first 6 months of HAART include gender, region of origin, pre-HAART HIV RNA plasma levels, and the number of pre-HAART CD8 cells. These differences persisted over the study period. The long-term CD4 cell response is, however, largely determined by age and by the degree of HIV RNA suppression. The findings that patients from sub-Saharan Africa, and possibly from Southeast Asia, have a slower recovery to 800 CD4 cells/mm3 may indicate geographic variation in normal CD4 ranges but also differences in adherence. The latter may be confirmed by our finding of a higher proportion of patients experiencing periods of HIV viremia while on uninterrupted HAART. Normal CD4 cell counts in HIV-seronegative Dutch individuals are reported to be higher than in HIV-seronegative individuals from Tanzania, Ethiopia, Kenya, and China39-44 but lower than in such individuals in Cameroon and Uganda.45,46 Seropositive patients from Ethiopia experience a slower decline in CD4 cells than seen in Dutch patients, but time to AIDS is not significantly different between these Dutch and Ethiopian patients.47 This suggests that immune restoration in patients originating from regions with low normal CD4 cell numbers might be slower than in patients with high normal CD4 cell numbers even if they are fully adherent.
The higher increase in CD4 cell count in women than in men follows most probably from the higher CD4 cell counts in uninfected women than in uninfected men33,38 but might also reflect prescription of different antiretroviral drugs between men and women. As in other studies,48-50 our patients with a lower pre-HAART CD8 cell count experienced higher rates of CD4 cell increase during the first 6 months. The reason for this relation is unclear, but it might be related to the level of CD4 and CD8 cell activation.17 CD4 and CD8 cell activation was not measured in this cohort, however.
It is well established that patients with pre-HAART CD4 cell counts <200 cells/mm3 are much more likely to progress to AIDS or death.10 Combined observational cohort data suggest that the long-term prognosis might be better for patients starting HAART when having ≥350 CD4 cells/mm3 as compared with 200 to 350 cells/mm3,51 although the absolute risk difference is small. In our subset of patients who used HAART uninterrupted for 7 years, the restoration of CD4 cell counts was sufficient to minimize the risk for development of AIDS, even for those starting HAART with CD4 cell counts less than 200 cells/mm3. These patients were likely to be adherent, and those who died were excluded from the analysis. Therefore, the results of the longitudinal model and the analysis of the decrease in CD4 cell response after 5 years of HAART use cannot be generalized to all patients using HAART but do give an estimate of the immune system's maximum capacity for CD4 cell restoration during 7 years of therapy.
CONCLUSIONS
HAART restoration of CD4 cell counts in HIV-infected individuals to levels normally seen in uninfected individuals takes a long time and is not feasible within 7 years in most patients who initiate HAART with CD4 cell counts <350 cells/mm3. Patients ≥50 years of age when starting HAART and patients with periods of viremia (HIV RNA level >500 copies/mL) experience smaller increases and are more likely to reach a CD4 cell plateau earlier and at a lower level. Given the better toxicity profiles of the currently used antiretroviral combinations, particularly in patients older than 50 years of age, it may be beneficial to start HAART earlier than current guidelines recommend.
MATERIALS AND METHODS
Study Population
Patients were selected from the AIDS Therapy Evaluation Project, Netherlands (ATHENA) national observational HIV cohort.32 HAART was defined as a combination of 3 or more drugs from at least 2 drug classes or a combination of 3 or more nucleoside reverse transcriptase inhibitors, including abacavir or tenofovir. All patients were 16 years of age or older and had a recorded pre-HAART CD4 cell count. We performed 3 analyses.
Time to CD4 Cell Counts ≥800 Cells/mm3
In the first analysis, we used a longitudinally followed cohort of 5299 antiretroviral therapy-naive patients who initiated HAART between July 1, 1996 and December 31, 2004 to analyze the probability of reaching CD4 cell counts ≥800 cells/mm3 in relation to pre-HAART values and other baseline characteristics. Most studies25,28,30 have used a threshold of 500 CD4 cells/mm3, the lower limit of the normal range in uninfected individuals. The mean observed CD4 counts in HIV-negative adults are 1050, 840, and 800 cells/mm3 for women, heterosexual men, and men who have sex with men (MSM), respectively.33 Because our cohort is largely MSM, we chose ≥800 CD4 cells/mm3 as an endpoint. Potential predictors tested for association with time taken to reach the endpoint included pre-HAART CD4 count (<50, 50-200, 200-350, 350-500, and ≥500 cells/mm3), pre-HAART CD8 count (<600, 600-1300, and ≥1300 cells/mm3), pre-HAART HIV RNA plasma levels (<3, 3-4, and ≥4 log10 copies/mL), Centers for Disease Control and Prevention category C (CDC-C) events before starting HAART, age at starting HAART, gender, and region of birth (Western or Central Europe, North America, or Australia combined [WCE/NA/A] and sub-Saharan Africa, Caribbean, Latin America, Southeast Asia, and all other regions combined were tested for association with time taken to reach the endpoint). The CD4 cell count measured closest to starting HAART from 6 months before to 7 days after starting was selected as the pre-HAART CD4 cell count. Patients were allowed to change or interrupt regimens and were retained in the analysis regardless of the level of HIV RNA.
Long-Term CD4 Cell Response in Patients on Uninterrupted HAART for 7 Years
Second, we analyzed the immune system's maximum capacity to restore CD4 cell numbers in a longitudinally followed subcohort of 554 patients who started HAART between July 1, 1996, and June 30, 1998, and took HAART continuously for at least 7 years. These patients were also part of the first analysis; they were included regardless of whether they had HIV RNA measurements greater than detectable limits because we were interested in estimating the effect of periods of viremia on the rate of change in CD4 cell count.
We selected CD4 cell counts measured closest to weeks 24, 48, and 72 after initiating HAART (within a time frame of 3 months) and, subsequently, at 24-week intervals up to 360 weeks. Median increases in CD4 cell count at these time points were calculated and graphically summarized.
All CD4 cell counts measured between starting HAART and 7 years thereafter were longitudinally modeled. The same potential predictors as in the first analysis were tested for their association with slopes of CD4 cell counts over time. Given the smaller sample size for this analysis, the predictors were subdivided differently: region of origin (WCE/NA/A and all other regions combined), pre-HAART HIV RNA plasma concentration (<4.5 log10 copies/mL and ≥4.5 log10 copies/mL), and CD8 count (<1300 cells/mm3 and ≥1300 cells/mm3). To study the effect of viremia, we created a time-updated variable with values between 0% and 100% and denoting the percentage of time a patient's plasma HIV RNA concentration was ≥500 copies/mL after the initial 6 months of HAART. The value was 0% when a patient never had HIV RNA levels <500 copies/mL after the initial 6 months and 100% when a patient always had levels <500 copies/mL after the initial 6 months. We also tried to distinguish the effect of low-level viremia and high-level viremia on the slopes of CD4 cell count using different cutoffs (500-1000, 1000-10,000, and >10,000 copies/mL). Given our inclusion criteria, however, the number of HIV RNA measurements greater than 500 copies/mL was limited, and we lacked statistical power to detect significant differences in slopes of CD4 cell count during periods of low-level and high-level viremia.
Decreases in CD4 Cell Count Between 5 and 7 Years in Virologically Suppressed Patients
The third analysis determined predictors for reaching a plateau in CD4 cell restoration between years 5 and 7. To counter random fluctuations in CD4 cell measurements, we used the individual slopes between 5 and 7 years derived from a longitudinal model similar to that used in the second analysis (ie, with 5 intercepts for the 5 pre-HAART CD4 cell strata and 4 slopes for the 4 time intervals but without any other covariates) to determine whether a patient had reached a plateau in CD4 cell restoration. This plateau can be interpreted as an on-average decreasing CD4 cell count between years 5 and 7. Included in this model was a subset of 366 patients on uninterrupted HAART who were among those included in the second analysis. Additional inclusion criterion for this subset was that all HIV RNA measurements between 6 months and 5 years after starting HAART were <500 copies/mL. Variables included in the analysis were the same as in the second analysis, with the exception of HIV viremia, which was now defined as at least 1 HIV RNA measurement ≥1000 copies/mL between 5 and 7 years after starting HAART (yes/no). This excludes HIV RNA measurements between 500 and 1000 copies/mL from the definition because these are unlikely to cause a plateau in CD4 cell restoration.
Statistical Analysis
The Cox proportional hazards model and Kaplan-Meier estimates were used for the first analysis of time to ≥800 CD4 cells/mm3. Time was censored at the end of follow-up or time of death, whichever occurred first. The statistical model used for the longitudinal analyses was a mixed-effects model with a random intercept and 4 random slopes for each patient. A first-order autoregressive covariance structure was used to correlate intraindividual serial measurements. We divided the 7-year time period into 4 intervals: 0 to 6 months after starting HAART, 6 months to 3 years, 3 to 5 years, and 5 to 7 years; slopes were allowed to differ between them. The intervals were chosen by visual inspection of the graphs of median CD4 cell response. CD4 cell counts were square root transformed to comply with model assumptions. Slopes of CD4 cell count increase during each interval were estimated for the 5 pre-HAART CD4 cell count strata (<50, 50-200, 200-350, 350-500, and ≥500 cells/mm3). The other variables were allowed to have 1 effect on the slopes between the start of HAART and 7 years thereafter (ie, the whole period) or to have an effect in each of the 4 previously stated time intervals. Model fit was determined by the Akaike Information Criterion statistic.34 Logistic regression was used for the third analysis of decreasing CD4 cell counts between 5 and 7 years after first starting HAART. All calculations were performed using SAS 9.1.3 (SAS Institute, Cary, NC).
|
|
| |
| |
|
|
|
