The course of HIV disease varies considerably in infected individuals, and one of the greatest challenges currently facing clinical researchers working in this area is to improve the understanding of factors that impact this variability. The progression of HIV disease is typically measured by the time to occurrence of clinical outcomes that reflect development of AIDS. However, some individuals develop AIDS-defining disease within 5 years of infection, whereas others do not progress as rapidly [1,2]. The reasons for this interindividual variability are not completely understood, and probably include both host and viral factors. Recent investigations of the kinetics of viral turnover underscore the fact that variation between patients is not simply reflective of the clinical disease stage [3,4]. In addition, although CD4 cell count is recognized as being strongly predictive of the development of future opportunistic infections (OIs), patients with similar CD4 lymphocyte counts often progress at different rates even when given the same therapy.
Current availability of plasma HIV RNA assays has permitted new insights that are now utilized both in clinical trials and the treatment of patients . Recent studies have suggested that HIV-infected patients with high circulating plasma HIV RNA levels tend to develop AIDS more rapidly than those with low levels [6-14]. The majority of these studies have identified HIV-1 RNA level as a strong predictor of clinical outcome. After primary infection, the rate of HIV replication is thought to stabilize at a particular level, or ‚setpoint‚, and these setpoints have been suggested to be predictors of future clinical progression .
Although plasma HIV RNA levels are now used along with CD4 cell counts and clinical status to guide clinical management, emphasis has tended to focus on determining antiretroviral therapies for HIV infection. However, little is known about the association between HIV viral load levels and the risk of specific OIs. Despite recent advances in antiretroviral therapy, the prevention of serious OIs remains an important goal. OIs are associated with considerable morbidity and mortality, significantly influence patients‚ quality of life, often require hospitalization, and increase the cost of medical care [15,16]. Important advances have been made since the start of the AIDS epidemic in the prevention and delay of OIs. The survival of HIV-infected individuals has been increased both by prevention of exposure to pathogens and prevention of new or recurrent opportunistic diseases. For example, several studies have shown that prophylaxis against Pneumocystis carinii pneumonia (PCP) prolongs life, decreases morbidity, and reduces medical costs [17-24].
Guidelines for prophylaxis against OIs are still mostly determined by the incidence of infections [15,16,25]. Because prophylactic therapies often have some negative consequences, such as associated laboratory toxicities, side effects, and the potential development of drug resistance, incidence rates by CD4 cell level have been used as a tool for targeting prophylaxis. As measurements of viral load become incorporated into clinical practice, it will be important to assess whether viral load data can be utilized to identify patients at highest risk of OIs. Improved guidelines incorporating viral load data could extend survival, improve patient care, and are likely to be cost effective.
The primary objective of this research was to evaluate the association between baseline viral load and the risk of specific OIs through the use of a retrospective data analysis of four large clinical trials conducted through the AIDS Clinical Trials Group (ACTG) between 1991 and 1995. Baseline viral load was analysed as a risk factor in conjunction with baseline CD4 cell count for the development of three of the most common serious OIs: PCP, cytomegalovirus (CMV) disease and Mycobacterium avium complex (MAC). By including both HIV-1 RNA and CD4 cell values as baseline covariates, the additional effect of viral load in predicting the risk of an OI could be assessed after accounting for the CD4 cell count. The predictive value of early changes in viral load and CD4 cell count on the risk of OIs was also evaluated. These specific OIs were selected because they occur frequently and confer significant morbidity and mortality; consequently, chemoprophylaxis is often recommended for HIV-infected patients at high risk of these infections. The analyses we conducted were undertaken in order to improve the identification of HIV-infected patients at risk of OIs, and to help target high-risk populations for prophylaxis.
For the purposes of our cross-protocol retrospective analysis, patient data was combined from the virology substudies of four ACTG trials: ACTG 116A, 116B/117, 175, and 241. Each of these studies assessed the safety and efficacy of didanosine (ddI) or zidovudine (ZDV) monotherapy or ZDV-containing combination regimens. Informed consent was obtained for all patient participants, and study protocols were approved by the appropriate institutional review boards at each study site. This manuscript received a review through the ACTG Publications Review Process, and was endorsed by the chairs of the relevant protocols. The analysis relied on datasets previously created for the primary analyses associated with each study, including separate datasets containing information on reviewed endpoints, baseline risk factors, patient status, concomitant medications, and measurements of CD4 cell counts and HIV RNA levels.
A total of 842 patients participated in the virology substudies of ACTG 116A, 116B/117, 175, or 241, and had baseline measurements available for both CD4 cell counts and HIV RNA levels. These 842 patients comprised the analysis set for all remaining evaluations. Six patients enrolled sequentially in two of the four studies; only the information from the earlier of the two trials was considered for these six patients. The proportion of patients with baseline RNA and CD4 cell data varied among the four ACTG studies, but overall represented 18% of the 4610 patients enrolled in the main studies (see Table 1). ACTG 116A and 116B/117 compared ddI and ZDV monotherapy among ZDV-naive and experienced patients, respectively [26,27]. ACTG 175 compared these two monotherapies with two ZDV-containing combination therapies , and ACTG 241 compared combination therapy with ZDV and ddI to ZDV, ddI, and nevirapine . The distribution of subjects by assigned treatment regimen and ACTG study is shown in Table 1. These studies were completed in October 1992 (ACTG 116A), March 1992 (ACTG 116B/117), November 1995 (ACTG 175), and November 1994 (ACTG 241).
For all four studies, the Roche Amplicor reverse transcriptase polymerase chain reaction (PCR) assay was utilized to quantify plasma HIV RNA levels. The lower limit of detection (LOD) for the Roche PCR assay was considered to be 200 copies RNA per millimeter, and all reported values below 200 copies were set to this LOD. The geometric mean of pre-entry measurements was used as the baseline measure for HIV RNA levels, whereas the arithmetic mean was used as the baseline for CD4 cell counts. For other variables, such as weight or Karnofsky performance score, the latest pre-entry measurement was selected as the baseline measure. Early changes in CD4 cell and RNA values were reflected by measurements taken at 8 and 24 weeks after randomization, with a maximum allowable window of 4 weeks on either side of the scheduled evaluation. For ACTG 175, RNA and CD4 cell measurements scheduled at week 20 were substituted for the week 24 measurements. Using the concomitant medication data for each patient, separate indicators were created for CMV prophylaxis with acyclovir, PCP prophylaxis with trimethoprim/ sulfamethoxazole (TMP/SMX), PCP prophylaxis with dapsone, and PCP prophylaxis with aerosolized pentamidine before or at study entry. Because MAC prophylaxis was relatively rare during the time period encompassed by the four studies, a single indicator was created to reflect the use of rifabutin, clarithromycin, or azithromycin before entry.
The primary analyses for this manuscript were based on ‚endpoint‚ datasets for each of the four studies, which contained all AIDS-defining events that had been reviewed and verified by the respective protocol chairs using standard ACTG endpoint definitions. From these endpoint files, only PCP, CMV (end-organ disease), and MAC events were selected for our analysis of ‚reviewed‚ OIs. However, two of the four studies only required review of the first OI for each patient, so that the use of the endpoint datasets may lead to underestimates of the true rates of OIs. Consequently, some supporting analyses also relied on including additional PCP, CMV, or MAC events from the full diagnoses dataset (all reported diagnoses, regardless of review as a study endpoint) in order to yield more accurate estimates of incidence rates for each of the OIs.
The joint effects of baseline CD4 cell counts and RNA viral load on the risk of each OI were assessed in several ways. First, crude and adjusted event rates for PCP, CMV, and MAC were calculated for the following four subgroups: (i) high CD4 cell count, low HIV-1 RNA level; (ii) high CD4 cell count, high HIV-1 RNA level; (iii) low CD4 cell count, low HIV-1 RNA level; and (iv) low CD4 cell count, high HIV-1 RNA level. The cutoffs for determining low versus high CD4 cell counts were set consistently with current prophylaxis guidelines : £50 cells/mm3 for the prediction of CMV and MAC endpoints and £200 cells/mm3 for PCP endpoints. The cutoffs for classifying high versus low HIV-1 RNA levels were set at ≥100000 copies/ml for CMV and MAC and ≥50000 copies/ml for PCP, under the parallel assumption that prophylaxis for PCP would be initiated at a lower viral burden than for MAC or CMV, as a result of its higher incidence rates and less costly, more effective prophylactic options.
Second, the risk of each OI was estimated on the basis of fitting a separate Cox proportional hazards model , including both baseline CD4 cell count and HIV-1 RNA level as joint predictors. A log10 transformation was applied to all HIV RNA measurements and CD4 cell count was transformed as -CD4 cells/50 before model fitting, so that the resulting relative risks can be interpreted as the risk ratio for every 1 log10 increment in RNA copies/ml or 50 cell reduction in CD4 cell count, respectively. Some models used the actual numerical values of transformed baseline CD4 cell count and RNA level, whereas others used indicators of whether the original (i.e. untransformed) CD4 cell and RNA values were above or below the previously described thresholds. Cox proportional hazards models were also used to assess the effect of changes in CD4 cell count or viral load after 8 or 24 weeks on study on the risk of the OIs of interest. Cumulative incidence functions were constructed from the predicted models to illustrate graphically the joint effects of CD4 cell count and RNA level on event rates for PCP, CMV, and MAC.
Other potentially confounding factors, such as specific protocol (ACTG 116A, 116B/117, 175, or 241), study treatment, sex, baseline risk factors (including hemophilia, intravenous drug use, and homosexual status), Karnofsky performance score, and indicators of previous chemoprophylaxis, were also evaluated and included in final models where appropriate. The effect of previous occurrence of PCP, CMV, or MAC on the future occurrence of one of these OIs was evaluated using proportional hazards models with time-dependent covariates as described in Finkelstein et al. .
The baseline characteristics of the 842 patients were typical for AIDS clinical trials conducted during this time period, with 85% men, 13% current or previous intravenous drug users, and a median age of 36 years. Only 7% (65/842) of the patients had a Karnofsky score of 80 or less. A comparison of the baseline characteristics specific to each ACTG study indicated that patients participating in the virology substudies and included in our analysis were generally representative of patients in the main study for each of the four trials. The median duration of follow up was 612 days (20 months) overall, with study-specific medians ranging from 417 days (14 months) for ACTG 116B/117 to 984 days (32 months) for ACTG 175.
The joint distribution of baseline CD4 cell counts and HIV RNA values by quartiles is shown in Table 2.Table 2 provides the numbers and percentages of the 842 patients falling into each cross-classification of CD4 cell quartile and RNA quartile, with the overall CD4 cell and RNA quartiles indicated in the margins. For example, 25% of patients had CD4 cell counts of 100 or less, whereas 25% had HIV-1 RNA levels over 150000 copies/ml. As expected, the highest counts in Table 2 are found on the diagonal (from bottom left to upper right), indicating that patients with low CD4 cell counts tended to have higher RNA values, and those with high CD4 cell counts were more likely to have low RNA values. If all patients fell into the same respective RNA quartile (from highest to lowest) as their corresponding CD4 cell quartile (from lowest to highest), then the use of baseline HIV RNA quartile would be unlikely to add much predictive value. However, note that only 39% of the patients fell into the same respective RNA quartile as their CD4 cell quartile, whereas 61% of patients either had both lower CD4 cell counts and lower RNA levels or higher values of both baseline measures.
The distribution of baseline CD4 cell counts and HIV-1 RNA levels showed fairly substantial shifts across the four trials considered in our analysis (see Table 1). Patients participating in the virology substudy of ACTG 175 tended to be less immunosuppressed than those participating in the virology substudies of the other three trials, with a median CD4 cell count of 333 compared with counts of 144, 89, and 125 for ACTG 116A, 116B/117, and 241, respectively. Similarly, the proportion of patients with more than 150000 copies HIV-1 RNA was only 7% for ACTG 175 patients, whereas 30-50% of patients in the other three studies had baseline viral loads in this higher range. These differences in initial status were obviously important in predicting the risk of OIs, and were accounted for by initially including study-specific covariates for ACTG 116A, 116B/117, and 241 in the proportional hazards models to reflect differences from ACTG 175. However, because these models also included covariates for baseline CD4 cell counts and RNA levels, much of the difference between patients in these studies was already accounted for, and these study-specific covariates were rarely significant.
On the basis of reviewed endpoints, patients who developed confirmed PCP, CMV or MAC during follow up were identified from among our virology substudy data. A total of 39 patients (4.6%) developed PCP, 33 (3.9%) developed CMV, and 21 (2.5%) developed MAC. These counts yielded adjusted rates of 2.6, 2.2, and 1.4 events per 100 patient years of follow up for PCP, CMV, and MAC, respectively. Although these rates of OIs may seem lower than those reported in the literature, it should be noted that two of the studies only required review of the first OI as an endpoint. When all reported diagnoses of these OIs were included, the total number of events increased to 64 patients (7.6%) with PCP, 55 (6.5%) with CMV, and 44 (5.2%) with MAC, yielding adjusted rates of 4.3, 3.7, and 2.9 events per 100 patient years follow up for PCP, CMV, and MAC, respectively. On the basis of reviewed endpoints only, the median values for baseline CD4 cell counts and HIV-1 RNA copies for patients with and without each type of infection are shown in Table 3. Not surprisingly, those who developed infections had substantially lower median CD4 cell counts at baseline than those who did not (CD4 cell count £80 versus ≥220), and higher median RNA copies (5.1-5.4 versus 4.6 log10 copies).
The crude and adjusted event rates by CD4 cell count and RNA level cutoffs based on all reported diagnoses are shown in Table 4. Although it may at first appear that the rates of CMV and MAC were higher than PCP in the low CD4 cell count strata, the allocation of patients into the four subgroups is defined differently for PCP than for CMV and MAC, so that a larger number of patients in the low CD4 cell count strata (£200) for PCP resulted in a higher predicted number of events. It is clear from Table 4 that higher RNA levels at baseline were associated with higher event rates within both the low CD4 cell count strata and the high CD4 cell count strata. For CMV in particular, the adjusted event rate for those with high RNA levels (≥100000) was 7.4 events per 100 patient years even within the upper CD4 cell count stratum. For PCP and MAC, the adjusted rates within the high CD4 cell count strata were less than five events per 100 patient years. Within the low CD4 cell count strata, the adjusted event rates for CMV were substantially higher among patients with high versus low RNA levels (25.2 versus 11.3 events).
The results of fitting a Cox proportional hazards model with baseline CD4 cell counts and mean log HIV-1 RNA levels for predicting the risk of each OI are shown in Table 5. For each of the OIs, the baseline covariates for both HIV RNA level and CD4 cell count were highly significant, which confirms that both baseline CD4 cell count and HIV RNA level were independent predictors for the risk of development of PCP, CMV or MAC. For every 50 cell reduction in CD4 cell count, the risk of developing one of the specific OIs considered in our analysis was 40-60% higher. In other words, a subject with a baseline CD4 cell count of 50 was estimated to have approximately 1.5 times the risk of one of these OIs as a subject with a baseline CD4 cell count of 100. The effect of CD4 cell count on the risk of these OIs was fairly consistent across the three infection types. On the other hand, the relative risk associated with baseline log10 RNA levels varied across the three OIs, ranging from 1.74 for PCP to 3.13 for MAC. Thus, a patient with baseline log10 RNA equal to 5 (i.e. 100000 copies/ml) had over three times the risk of MAC as a patient with log10 RNA equal to 4 (i.e. 10000 copies/ml) at entry.
When the baseline predictors were dichotomized on the basis of the previously described cutoffs, a similar picture emerged. The relative risk of infection for high versus low RNA levels was 2.99 for PCP (P=0.009), 6.22 for CMV (P=0.0001), and 4.88 for MAC (P=0.001), whereas the relative risk for low versus high CD4 cell counts was 5.75 for PCP (P=0.0001), 4.52 for CMV (P=0.0001) and 10.75 for MAC (P=0.0001). Possible interactions between these baseline covariates were also explored. There was no evidence of interaction between low CD4 cell counts and high RNA levels on MAC development, which implies that patients with high RNA levels are estimated to have 4.88 times the risk of MAC than those with low RNA levels, regardless of the CD4 cell count. However, there was some evidence of interaction between CD4 cell count and RNA level for the development of both PCP and CMV. In both cases, the relative risk associated with high compared with low RNA levels was estimated to be larger within the high than the low CD4 cell count category. That is, taking into account RNA levels had less of an effect for patients who were already at higher risk on the basis of having low CD4 cell counts at baseline.
These results are illustrated via cumulative incidence functions for each of the three OIs (Figs. 1-3). Pairwise comparisons of the cumulative incidence distributions supported the previous claim: the comparison of high versus low RNA levels was significant within the higher CD4 cell level (lowest two lines, P<0.001 for PCP, P<0.001 for CMV, P=0.0011 for MAC), but non-significant within the lower CD4 cell level (upper two lines, P=0.27 for PCP, P=0.30 for CMV, and P=0.13 for MAC). This is partly attributable to the fact that subjects with high viral loads had lower CD4 cell counts at baseline than those with low viral loads, even within the high CD4 cell count group (median CD4 cell count=173 versus 189), whereas the difference in baseline CD4 cell count is much smaller within the low CD4 cell count group. Yet it also remained clear that viral load added information after accounting for whether a patient met the guidelines for initiating prophylaxis. It is also worth noting that, although patients with low CD4 cell counts and low RNA levels had a consistently higher risk of infection than those with high CD4 cell counts and high RNA levels, these comparisons lacked significance for all three OIs (middle two curves, P=0.19 for PCP, P=0.42 for CMV, P=0.12 for MAC).
Before accounting for baseline CD4 cell count and HIV RNA values, several other factors were explored as potential predictors of risk for OI. Study treatment (combination therapy versus monotherapy) and baseline risk factors (hemophilia, intravenous drug use, and homosexual status) were not identified as significant predictors for any of the three OIs, but study indicators and separate indicators of previous PCP prophylaxis with dapsone, aerosolized pentamidine (AP), and TMP/SMX were often significant. However, after taking into account baseline CD4 cell counts and RNA levels, most of these baseline predictors were no longer significant. Patients enrolled in ACTG 241 continued to have an elevated risk of CMV compared with those in ACTG 175 [relative risk (RR)=2.55], previous dapsone or AP led to significantly higher risks of PCP (RR=3.96 for dapsone, RR=2.22 for AP), and Karnofsky £80 was associated with a significantly higher risk of MAC (RR=2.64) even after accounting for CD4 cell counts and HIV RNA levels. Although it may seem counterintuitive that the use of dapsone or AP before entry would lead to a higher risk of PCP, these indicators were strongly confounded with health status at baseline. For example, 41% of patients with low CD4 cell counts (£200) received AP, compared with only 8% of those with high CD4 cell counts. In addition, they may reflect treatment for previous episodes of PCP, which in turn puts patients at a higher risk of recurrence.
The effect of early changes in HIV RNA levels and CD4 cell values on the risk of OIs was examined in 690 patients with CD4 cell counts and HIV-1 RNA measurements taken 8 weeks after initiating antiretroviral therapy. Our analyses indicated that the 291 patients with a decrease of at least 0.5 log10 RNA were at a significantly reduced risk of PCP and CMV (RR=0.41 for PCP; RR=0.33 for CMV), whereas the 501 patients with any decrease in log10 RNA were at a significantly reduced risk of MAC (RR=0.32). In contrast, the 204 patients with an increase of at least 50 CD4 cells/mm3had no significant protection against the risk of PCP or CMV, and only a marginally significant reduction in the risk of MAC (RR=0.28, P=0.093). Similarly, the 402 patients identified as exhibiting any increase in CD4 cell count at the week 8 visit compared with baseline showed no benefit in risk reduction for any of the three OIs.
This general pattern persisted even after controlling for baseline levels of CD4 cell count and RNA level, which typically remained highly significant. After accounting for high RNA level and low CD4 cell count baseline indicators, patients with any decrease in viral load by week 8 were at a significantly reduced risk of both CMV (RR=0.44, P=0.031) and MAC (RR=0.23, P=0.004); the results were similar when the actual baseline values of CD4 cell count and RNA level were considered rather than the high/low indicators. In contrast, an increase of 50 CD4 cells was not a significant predictor of any of the three OIs after controlling for the baseline indicators, nor was an indicator of any CD4 cell count increase by 8 weeks. In some cases, the use of the actual week 8 values for CD4 cell counts and RNA levels increased the predictive power of these early changes over the use of simple indicators. For example, both the CD4 cell counts and RNA values at week 8 were highly significant predictors of the risk of all three of the OIs before controlling for the baseline levels. In addition, after controlling for high RNA levels and low CD4 cell counts at baseline, the actual RNA value at week 8 was significantly associated with the risk of PCP (RR=1.78, P=0.045) and MAC (RR=2.61, P=0.019) and showed a marginally significant association with CMV (RR=1.79,P=0.076). In the same model, the actual CD4 cell count at week 8 also exhibited a significant association with CMV risk (P=0.027), but no association with PCP or MAC. However, when the actual baseline RNA and CD4 cell values were included along with the actual values at week 8, neither week 8 predictor was significant.
An evaluation of longer term changes 24 weeks after randomization based on 580 patients with both CD4 cell counts and RNA measurements available suggested that the 197 patients with a decrease of at least 0.5 log10 RNA were at a significantly reduced risk of CMV (RR=0.25, P=0.024) and a marginally reduced risk of MAC (RR=0.14, P=0.067), but had no significant difference in the risk of PCP. The 147 patients with an increase of at least 50 CD4 cells after 24 weeks had a marginally reduced risk of CMV (RR=0.27, P=0.077), but no change in the risk of PCP or MAC. Again, an indicator for any decrease in RNA levels seemed to be more predictive of the risk of MAC than a decrease of 0.5 log10 or more, with an associated relative risk of 0.18 (P=0.006). Although the smaller number of patients with these measurements at 8 or 24 weeks after study initiation reduced the power of such analyses, early changes in RNA were clearly associated with a greater reduction of risk for the OIs we considered than were early changes in CD4 cell count.
Although much attention has focused on the association between HIV-1 RNA levels and future clinical progression, little research has addressed the relationship of these laboratory markers and specific OIs. Our evaluations of four completed ACTG trials indicated that baseline RNA values were significant independent predictors for the risk of PCP, CMV and MAC, even after accounting for CD4 cell count. The difference in median baseline RNA levels for those with and without each type of opportunistic infection was greater than 0.5, which is considered to reflect the upper range of intra-patient variability .
The predictive value of high versus low RNA levels appeared to be greatest for those patients with higher baseline CD4 cell counts, indicating that prophylactic strategies based on RNA levels in combination with CD4 cell levels may have the potential to identify patients at high risk of OIs, who were previously overlooked using CD4 cell count alone. For example, the adjusted CMV event rate for those with high RNA levels and high CD4 cell counts was 7.4 events per 100 patient years, which may be sufficiently high to warrant prophylaxis. In addition, even within the low CD4 cell count strata, the use of RNA levels would help target patients at highest risk of OIs. The relative risks associated with baseline RNA levels tended to be higher for CMV and MAC than for PCP, suggesting that it may be especially important to account for viral load in evaluating the risk of these later stage OIs.
Early decreases in HIV-1 RNA levels were associated with significant reductions in risk for the OIs studied, and appeared to be more important in predicting risk than were increases in CD4 cell count. The significance of a decrease in RNA level by 8 weeks after initiating treatment persisted even after controlling for baseline CD4 cell counts and RNA levels. The actual values of RNA levels and CD4 cell counts at 8 weeks (as opposed to increase/decrease indicators) were highly significant independent predictors of the risk of all three OIs, but failed to maintain their significance when controlling for the actual baseline values. This is probably due to the high correlation between baseline and week 8 values, making it difficult to evaluate the independent contributions of these repeated measures, given the relatively small numbers of events.
Other analyses  have indicated that the risk of clinical progression tended to increase in a fairly linear manner as a function of log10 RNA, when quantified in terms of the log hazard ratio. Thus, we are not suggesting that a biological threshold of HIV viral load exists, i.e. that patients above a certain RNA level experience a sudden jump in risk of these OIs. However, this research suggests that it may be possible to identify levels of HIV RNA below which the yearly risk is considered acceptable and above which there is a consensus that prophylaxis is desirable. Such considerations would require taking into account both: (i) the increase in risk associated with certain RNA levels; and (ii) the overall incidence rates of infections predicted within categories defined by both HIV RNA level and CD4 cell count.
Previous research has shown that patients who develop the OIs considered in this analysis are at higher risk of the future development of PCP, CMV, and MAC. For example, Finkelstein et al.  found that the occurrence of PCP significantly increased the risk of MAC and CMV, and that the occurrence of MAC led to a higher risk of CMV and vice versa. In our cross-protocol analysis, two of the OIs of interest were confirmed for each of nine subjects. One patient had both PCP and CMV reported on the same date. Three patients first developed PCP and then later developed CMV. MAC occurred in two patients with previous PCP, and three patients with previous CMV. Despite the small numbers of events, a Cox proportional hazards model for the development of MAC identified a marginally significant effect of a time-dependent indicator of previous CMV or PCP (RR=2.61, P=0.076) after adjustment for baseline CD4 cell count, RNA level, and Karnofsky score.
We acknowledge that the generalizability of our analyses is limited by the retrospective nature of the study, the combination of different antiretroviral trials, and the time period in which these trials were conducted. Recently, much interest has centered around the question of what happens to viral load at the time of an OI; however, none of the trials included in this analysis collected this information in a deliberate way, and the differences in timing of CD4 cell counts and RNA level measurements does not allow us to address this question in the context of this meta-analysis. Protease inhibitor therapy was unavailable during the conduct of these trials, and many participants received monotherapy with nucleoside analogs. Chemoprophylaxis for MAC has subsequently improved, although prophylaxis for PCP has changed little, and prophylaxis for CMV remains controversial. Within the era of combination therapy with protease inhibitors, a smaller percentage of HIV-infected individuals would be expected to develop HIV RNA values greater than 100000 copies/ml. However, our analyses suggest that it may still be possible to identify clinically meaningful cutoff values for targeting prophylaxis for these specific OIs.
In the era of highly active antiretroviral therapy, questions have been raised regarding the need for chemoprophylaxis in patients whose CD4 cell counts have risen above previously defined thresholds. Although it is still unknown how much value viral load adds above CD4 cell counts for predicting the risk of OIs among those receiving the current standard of care, the above analyses suggest that consideration of viral load is likely to be important. Prospective studies in the era of combination antiretroviral therapy including protease inhibitors are urgently needed to address the relationship between CD4 cell count, HIV RNA level and the risk of OIs, and to help formulate improved guidelines for chemoprophylaxis.
We thank the study teams of the associated clinical trials that allowed us to use these data: ACTG 116A: Raphael Dolin, MD, Margaret Fischl, MD; ACTG 116B/117: James Kahn, MD; ACTG 175: David Katzenstein, MD; ACTG 241: Richard D‚Aquila, MD. In addition, we thank the patient volunteers and the staff who made these trials a success.
1.Phair J, Jacobson L, Detels R, et al
. Acquired immune deficiency syndrome occurring within 5 years of infection with human immunodeficiency virus type 1: the Multicenter AIDS Cohort Study.J Acquired Immune Defic Syndr
2.Sheppard HW, Lang W, Ascher MS, Vittinghoff E, Winkelstein W. The characterization of non-progressors: long term HIV-1 infection with stable CD4+ T cell levels.AIDS
3.Wei X, Ghosh SK, Taylor ME, et al. Viral dynamics in human immunodeficiency virus type 1 infection.Nature
4.Ho DD, Neumann AU, Perelson AS, Chen W, Leonard JM, Markowitz M. Rapid turnover of plasma virions and CD4 lymphocytes in HIV-1 infection. Nature
5.Carpenter CCJ, Fischl MA, Hammer SM, et al. Antiretroviral therapy for HIV infection in 1996.JAMA
6.Henrard DR, Phillips JF, Muenz LR, et al
. Natural history of HIV-1 cell-free viremia.JAMA
7.Mellors JW, Kingsley LA, Rinaldo CR, et al
. Quantitation of HIV-1 RNA in plasma predicts outcome after seroconversion.Ann Intern Med
8.Mellors JW, Rinaldo CR, Gupta P. White RM, Todd JA, Kingsley LA. Prognosis of HIV-1 infection predicted by the quantity of virus in plasma. Science
9.Welles SL, Jackson JB, Yen-Lieberman B, et al
. Prognostic value of plasma human immunodeficiency virus type 1 (HIV-1) RNA levels in patients with advanced HIV-1 disease and with little or no prior zidovudine therapy.J Infect Dis
10.O‚Brien WA, Hartigan PM, Martin D, et al
. Changes in plasma HIV-1 RNA and CD4+ lymphocyte counts and the risk of progression to AIDS.N Engl J Med
11.O‚Brien TR, Blattner WA, Walters DW, et al
. Serum HIV-1 RNA levels and time to development of AIDS in the Multicenter Hemophilia Cohort Study.JAMA
12.Coombs RW, Welles SL, Hooper C, et al
. Association of plasma human immunodeficiency virus type 1 RNA level with risk of clinical progression in patients with advanced infection. J Infect Dis
13.Katzenstein DA, Hammer SM, Hughes MD, et al
. The relation of virologic and immunologic markers to clinical outcomes after nucleoside therapy in HIV-infected adults with 200 to 500 CD4 cells per cubic millimeter.N Engl J Med
14.Marschner IC, Collier AC, Coombs RW, et al
. The use of changes in plasma levels of human immunodeficiency virus type 1 RNA for assessing the clinical benefit of antiretroviral therapy.J Infect Dis
15.Kaplan JE, Masur H, Holmes KK, et al
. USPHS/IDSA guidelines for the prevention of opportunistic infections in persons infected with human immunodeficiency virus: introduction.Clin Infect Dis
1995, 21 (Suppl.)
16.Kaplan JE, Masur H, Holmes KK, et al
. USPHS/IDSA guidelines for the prevention of opportunistic infections in persons with human immunodeficiency virus: an overview.Clin Infect Dis
1995, 21 (Suppl.)
17.Rothenberg R, Woelfel M, Stoneburner R, Milberg J, Parker R, Truman B. Survival with the acquired immunodeficiency syndrome. N Engl J Med
18.Bacchetti P, Osmond D, Chaisson RE, et al
. Survival patterns of the first 500 patients with AIDS in San Francisco.J Infect Dis
19.Fischl MA, Dickinson GM, La Voie L. Safety and efficacy of sulfamethoxazole and trimethoprim chemoprophylaxis for Pneumocystis carinii pneumonia in AIDS.JAMA
20.Graham NMH, Zeger SL, Park LP, et al
. Effect of zidovudine and Pneumocystis carinii pneumonia prophylaxis on progression of HIV-1 infection to AIDS. The Multicenter AIDS Cohort Study.Lancet
21.Chaisson RE, Kreuly J, Richman DD, Moore RD, the Zidovudine Epidemiology Group. Pneumocystis prophylaxis and survival in patients with advanced human immunodeficiency virus infection treated with zidovudine.Arch Intern Med
22.McAvinue SM, Gallant JE, Stanton DL, Chaisson RE. The effect of prophylaxis on outcome of Pneumocystis carinii pneumonia (PCP). VIII International Conference on AIDS
. Amsterdam: Congrex Holland BV, 1992 [abstract PuB 7350].
23.Hardy WD, Feinberg J, Finkelstein DM, et al
. A controlled trial of trimethoprim-sulfamethoxazole or aerosolized pentamidine for secondary prophylaxis of Pneumocystis carinii pneumonia in patients with the acquired immunodeficiency syndrome. AIDS Clinical Trials Group Protocol 021.N Engl J Med
24.Schneider MM, Hoepelman AI, Eeftnick Schattenkerk JK, et al
. A controlled trial of aerosolized pentamidine or trimethoprim-sulfamethoxazole as primary prophylaxis against Pneumocystis carinii pneumonia in patients with human immunodeficiency virus infection. N Engl J Med
25.Gallant JE, Moore RD, Chaisson RE. Prophylaxis for opportunistic infections in patients with HIV infection.Ann Intern Med
26. Dolin R, Amato DA, Fischl MA, et al
. Zidovudine compared with didanosine in subjects with advanced HIV type 1 infection and little or no previous experience with zidovudine.Arch Intern Med
27.Kahn JO, Lagakos SW, Richman DD, et al
. A controlled trial comparing continued zidovudine with didanosine in human immunodeficiency virus infections. The NIAID AIDS Clinical Trials Group. N Engl J Med
28.Hammer SM, Katzenstein DA, Hughes MD, et al
. A trial comparing nucleoside monotherapy with combination therapy in HIV-infected adults with CD4 counts from 200 to 500 per cubic millimeter.N Engl J Med
29.D‚Aquila RT, Hughes MD, Johnson VA, et al
. A randomized double-blind placebo-controlled trial of nevirapine, zidovudine, and didanosine versus zidovudine and didanosine in patients with human immunodeficiency virus type 1 infection.Ann Intern Med
30.Cox DR. Regression models and life-tables [with discussion].J R Statist SocSeries B
31.Finkelstein DM, Williams PL, Molenberghs G, et al
. Patterns of opportunistic infections in patients with HIV infection. J Acquired Immune Defic Syndr
32.Hughes MD, Johnson VA, Hirsch MS, et al
. HIV-1 RNA levels in plasma in addition to CD4+ lymphocyte count improves assessment of antiretroviral therapeutic response.Ann Intern Med