Palella, Frank J. Jr.; Chmiel, Joan S.; Moorman, Anne C.a; Holmberg, Scott D.*; and the HIV Outpatient Study Investigators
The benefits of highly active antiretroviral therapy (HAART) in the treatment of HIV infection have been well described , including viral suppression, CD4 lymphocyte repletion, and durable reductions in AIDS-related opportunistic diseases and death. However, the durability of the effectiveness of HAART remains to be delineated [2,3].
Factors that limit the success of HAART include poor therapy adherence, regimen complexity [4,5], viral resistance , pharmacodynamic interactions, drug tolerability and toxicity, therapy costs, and presence of comorbid conditions such as substance abuse and addiction [7,8]. The optimal selection of HAART salvage regimens is of increasing interest and importance; practitioners may formulate initial treatment that will maximize potential benefits of subsequent salvage regimens.
In this report we describe correlates of HAART efficacy over time among HIV-infected patients in the HIV Outpatient Study (HOPS).
HOPS is a prospective observational cohort into which patients are continuously recruited . HOPS has included almost 6000 patients seen in more than 88 000 visits to 10 HIV clinical care sites since 1992. Current analyses include data from patient-visits from 1 January 1994 through 30 June 2000. Study sites were 10 clinics specializing in HIV care (eight private and two public) in eight US cities (Atlanta; Chicago; Denver; Oakland, California; Philadelphia; Stony Brook, New York; Tampa, Florida; Washington, DC) that, in aggregate, provide care for about 2500 HIV-infected patients yearly. Participating physicians have extensive experience treating HIV-infected patients.
Information from outpatient visits is abstracted from the outpatient record and entered electronically by trained data abstracters, compiled centrally, reviewed, edited, and then added to the HOPS database. Symptoms, diagnoses, and treatments since the previous visit, including interim changes in treatment, are noted at each clinic visit. Categories of information abstracted include: demographic characteristics, risk factors for HIV infection; symptoms; diseases (both definitive and presumptive diagnoses); medications prescribed, including dose and duration; and laboratory values, including CD4 cell counts and measurements of plasma HIV RNA (viral load). The primary payment source for medical care is documented for each patient and categorized as private insurance (including fee-for-service care, health maintenance organizations, and preferred provider organizations), Medicare, Medicaid, self- payment, or prescription programs under the Ryan White Care Act.
We analyzed AIDS-related death and disease [opportunistic infection (OI)] rates among patients with CD4 nadir < 100 × 106/l seen between January 1994 and July 2000. As carried out in a previous analysis, data were stratified by calendar quarter and mortality and morbidity rates were calculated from the number of deaths (or events) that occurred in each quarter divided by the person-years of observation during that quarter. AIDS-related OIs were analyzed in the aggregate; in addition, separate analyses were performed for the three most common of these—Pneumocystis carinii pneumonia (PCP), Mycobacterium avium complex (MAC) infection, and cytomegalovirus (CMV) disease. Patients with a diagnosis of CMV or MAC before study entry or during the first 30 days of follow-up and patients with PCP at the beginning of follow-up were excluded from analyses of those OI.
Our definition of HAART was three or more drugs that included a single protease inhibitor (PI) or non-nucleoside reverse transcriptase inhibitor (NNRTI), or any regimen containing more than one PI. We evaluated the effectiveness of first and subsequent (salvage) HAART among HOPS patients seen between January 1996 and April 1999 with a CD4 nadir of < 500 × 106/l. Demographic data, type of health care funding, specific therapies (drugs) received, time spent on therapy, CD4 cell count and viral load before and after each HAART regimen were analyzed. Successful treatment response was defined as at least a 1.0 log10 decline in viral load, or achievement of undetectable viral load, as well as a stable or increasing CD4 cell count at least 30 days after HAART initiation but no later than 30 days after discontinuation when compared to baseline (pre-HAART) levels. A durable response was one that lasted for at least 12 months. Analysis of a subset of successful responders utilizing achievement of an undetectable plasma viral load as part of the definition of treatment success was also undertaken.
The number of days of observation for each patient was calculated beginning with the time of HAART initiation. We sought to ensure that outcome measurements based upon receipt of therapy were such that changes or events that occurred while receiving such therapy could be considered attributable to that treatment. Hence, changes in CD4 cell count, plasma viral load, OI, or deaths that took place within the first 30 days after any initiation or change in antiretroviral drug therapy were considered to have occurred while receiving the previous antiretroviral regimen. Only events occurring within 90 days of a clinic visit were counted, unless there was a subsequent visit documenting events or receipt of therapy in the interim. Patients not seen since March 2000 contributed observation times for a maximum of 90 days after their last study visit. Because enrollment continued during the study period, data from new patients were included in each interval analysis. Patients were classified as continuously having a successful response between observed visits if they had documented successes at the adjacent visits. If a change in response was discerned at adjacent visits (i.e., success to non-success, or the reverse), the change point was interpolated as the mid-point between the adjacent visits.
Data were analyzed using SAS software (version 6.12; SAS Institute, Cary, North Carolina, USA). Subgroup characteristics of patients were compared using either chi-squared tests or two-sample t tests as appropriate. Kaplan–Meier survival curves were used to estimate the distribution of time spent on specific HAART regimens; log-rank tests were used to compare these curves. The null hypothesis was that the survival time distributions were equal. Univariate and multiple logistic regression methods were used to identify correlates of successful HAART therapy (for the 549 of the 1022 patients having sufficient laboratory data to be classified as treatment successes, or not) and of durable responses (for the 366 of the 549 patients having follow-up long enough to be classified as durable successes, or not.) See also footnotes to Tables 2 and 3 for definitions used in each analysis.
Characteristics of the patients
Mortality and OI rates were calculated for 1769 patients who ever had a CD4 cell count < 100 × 106/l, seen from January 1994 through June 2000. HAART's durability, effect, and patterns of utilization were analyzed using data from 1022 patients with at least one CD4 cell count < 500 × 106/l seen during January 1996 through March 1999. Both analyses evaluated patients who received one or more HAART regimens for at least 30 days. Patients were similar in each analysis: 14–15% were women; 59–62% were white, 26–29% were African–American, and 10% were Hispanic; mean age was 40 years; 45–50% received funding for health care from private sources such as insurance or health maintenance organizations (HMO) and 38–43% from public sources such as Medicare, Medicaid, and MediCal.
The mean number of CD4 cell count and viral load measurements per patient increased from 2.1 and 2.2 in 1996 to 2.6 and 2.8 in 1998, respectively.
Trends in mortality and morbidity
We observed sustained decreases in mortality and OI rates among ambulatory patients with CD4 cell counts < 100 × 106/l in association with HAART use (Fig. 1). Death rates declined from a quarterly (3-month time period) mean of 29.5 deaths/100 person-years (PY) of observation in 1995 – just prior to the widespread use of HAART – to a rate of 11.5 deaths/100 PY observed by the fourth quarter of 1996 when HAART utilization had a prevalence of 65.7% in our patient cohort. Deaths rates subsequently remained stably low with a mean quarterly mortality rate of 12.0 deaths/100 PY seen in 1997, 9.1 deaths/100 PY seen in 1998, 6.9 deaths 100 PY seen in 1999, and 4.4 deaths/100 PY for the first three quarters of 2000. Similar dramatic declines in rates of OI occurred over the same time period (Fig. 1b). Since the fourth quarter of 1997, the prevalence of HAART utilization among patients seen remained between 82.3% and 86.3%. Reductions in death and disease were seen regardless of the type of HAART received or baseline demographics of age, race, sex, HIV risk category, or source of medical payment (data not shown). Rates of chemoprophylaxis for PCP and MAC remained stable throughout the period of analysis (about 90% and 50%, respectively).
Duration and patterns of use of HAART regimens
Comparisons of first ever to second and third HAART regimens received by the 1022 patients initiating therapy from January 1996 to March 1999 revealed that fewer salvage regimens containing a single PI and more with four or more drugs or two PIs (especially ritonavir plus saquinavir) were used. NNRTI, the combination of didanosine and hydroxyurea, and abacavir were more often prescribed as salvage therapy for those failing first or second HAART regimens (Table 1).
Rates of HAART receipt in those whose CD4 nadir was < 500 × 106/l did not differ by age, race, or HIV risk factor, but those with private insurance received HAART more promptly than those with public sources of funding (Medicaid or Medicare). For example, by the first quarter of 1997, 80.2% of those with private insurance were receiving HAART compared to 57.1% of those with Medicaid. By the first quarter of 1999, these differences had diminished, with 91.4% versus 84.5% of those with private insurance versus Medicaid receiving HAART, respectively. In the first quarter of 1999 91.2% of men and 83.9% of women were receiving HAART, and this sex disparity persisted when controlling for type health insurance (data not shown).
Over the 39-month period of analysis, the mean number of HAART regimens used was 1.8. As indicated in Fig. 2a, progressively less time was spent on third versus second versus first HAART regimen. Median time spent on first HAART (n = 1022 patients) was 11.8 months; on second HAART (n = 424) it was 7.4 months; and on third HAART (n = 213) 7.2 months (overall P < 0.0001).
Analysis of time spent receiving first HAART, stratified by HAART regimen type revealed that a longer time was spent on first HAART regimens containing nelfinavir (n = 262; 15.6 months) or indinavir (n = 336; 12.5 months) than on regimens containing ritonavir or saquinavir (n = 23; 9.7 months; and n = 51; 6.1 months, respectively), or regimens containing both ritonavir and saquinavir (n = 93; 6.7 months) (Fig. 2b;P < 0.0001.) Pair-wise comparisons of indinavir- or nelfinavir-based regimens versus ritonavir-, saquinavir-, or ritonavir + saquinavir-containing regimens confirmed these relationships (data not shown). These regimen-related differences were not seen when analyzing time on second HAART, although overall time spent on treatment was shorter (Fig. 2c).
In assessing factors predictive of time spent taking first HAART, there were no significant differences when controlling for pre-HAART viral load or CD4 cell count in those receiving regimens containing a single PI (data not shown). For those whose first HAART did not contain a PI, higher baseline viral loads (> 100 000 plasma HIV RNA copies/ml) were associated with shorter time receiving first HAART than persons whose baseline viral loads were < 100 000 plasma HIV RNA copies/ml (median, 5.6 versus 16.3 months, respectively;P = 0.02). A similar, but not significant, relationship was seen for those with baseline CD4 cell counts < 200 × 106/l versus ≥ 200 × 106/l (median, 7.9 months versus 10.6 or more months;P = 0.11).
Patients were also analyzed by baseline demographic characteristics and major source of health care reimbursement (Medicare versus Medicaid versus private). No significant differences in duration of time treated within first HAART treatment categories were seen (data not shown).
Durability of treatment effects
Most of the 549 persons with sufficient viral load and CD4 cell count data who demonstrated treatment benefit did so while receiving first HAART. Fifty-one percent of patients on their first HAART regimen achieved a successful response within 6 months; only an additional 5% achieved success while remaining on first HAART for a time longer than 6 months (Table 2).
We evaluated additional benefit derived from salvage HAART regimens (those prescribed subsequent to first HAART) (Fig. 3a). Six months after starting HAART, 40% of those still receiving first HAART and 42% of those receiving any HAART (first or salvage) were classified as successful responders. When stratifying first HAART recipients by type of therapy received, single PI-based regimens resulted in the most frequent and longest successful response; and dual PI-based therapy the least (Fig. 3b).
We evaluated factors potentially associated with the success of first HAART by 3 and 6 months. No significant differences were noted among persons receiving different types of HAART (Table 2). Patients who discontinued therapy prior to successful response were considered failures to treatment. However, again, being antiretroviral drug-treatment-naive (see also Fig. 3c), having a higher pre-HAART viral load and, to a lesser extent, having a lower CD4 cell count prior to starting HAART were predictive of treatment success. Other baseline pre-HAART characteristics including race, age, sex, HIV risk category, and insurance status were not predictive of treatment success.
Finally, we attempted to identify those clinical and demographic factors associated with durable (12 or more continuous months) treatment response (Table 3). There were 366 patients for whom adequate follow-up data (for at least 15 months) existed; 134 could be classified as durable successes, while 232 others were not. Type of HAART regimens received was again predictive of durable success, with those receiving dual PI-based HAART doing significantly worse than those taking single PI-based HAART regimens (the majority of these indinavir- or nelfinavir-based). Durable treatment success was, again, associated with being therapy naive prior to first HAART, and the likelihood of success decreased with each successive HAART regimen. Race, sex, age, HIV risk category, and type of health insurance were not predictive of treatment success (Table 3). After adjusting for pre-HAART antiretroviral experience (none versus any), the duration of pre-HAART antiretroviral experience (in days) among those with such experience did not emerge as a significant determinant of durable success (data not shown), nor did the effect of the other factors listed above change. Patients with 15 months of follow-up were more likely to have a durable success with first (49.0%) than with second (29.6%) or third (14.9%) HAART (Table 3).
We evaluated durably successful responders who achieved or maintained an undetectable viral load after HAART initiation (n = 116; 32%). In that analysis, additional factors that emerged as significant predictors of durable treatment success included white race (P = 0.0003) and having private insurance (versus Medicaid;P < 0.0001).
Earlier responders (mean time to first documented response, 3.6 months) were more likely than later responders (mean, 5.6 months) to have a durable response to first HAART (P = 0.002). Only 10 (44.4%) patients without a durable response who were followed for at least 15 months ever responded successfully.
Low HIV-related death and disease rates have been maintained among the patients in this cohort in the 33 months since our first report . This benefit has come at a cost to patients that includes the use of sequential HAART regimens of progressively shorter duration with less CD4 cell count and HIV viral load suppressive benefits in spite of sustained dramatic reductions in AIDS-related mortality and morbidity.
The likelihood of achieving either short-term or durable successful response to a HAART regimen was highest among those who were naive to antiretroviral therapy, those who responded most promptly to HAART, and those with high viral loads and low CD4 cell counts before starting HAART (presumably because treatment benefit was most readily demonstrable among those with the greatest ‘room for improvement'). Those achieving and maintaining undetectable HIV RNA levels were more likely to be of white race or had private insurance, but it is difficult to interpret these findings because the viral load assays used over the period of observation varied widely in sensitivity, and no such association with race or insurance status was found in the larger group demonstrating success by our broader definition. In no analysis, however, was durable treatment success associated with patient age, sex, or HIV transmission risk group.
Durable (> 12 months) treatment success was most likely among those whose first-ever HAART included a single PI, especially indinavir or nelfinavir. Reasons for this are unclear, yet it persisted even when controlling for pre-HAART CD4 cell counts, viral loads, and extent of prior antiretroviral therapy. Unmeasured factors such as patient adherence to therapy or treatment-related toxicities may explain this finding.
Caution is needed in interpreting these analyses of therapeutic outcomes up to 3 years after initiation of a given therapy. Patients who had discordant or equivocal responses to therapy, i.e., both their CD4 cell counts and their viral load went up, or both went down, following start of therapy, may still have enjoyed benefits of treatment not described here [10–12]. Though our ‘success’ rates (based on CD4 cell count improvement and HIV plasma viral load suppression) appear meager in comparison to clinical trial-derived data, that often include only treatment-naive patients receiving specific drugs over short periods, such differences may serve to further emphasize the value and distinctness of observational data, especially in populations with extensive antiretroviral treatment histories prior to HAART. In fact, lower than anticipated rates of virologic and/or immunologic treatment success make the continuing low morbidity and mortality rates observed even more impressive.
Use of successive and increasingly complex HAART regimens can result in the exhaustion of viable antiretroviral drug treatment options. In this study, almost half of the patients on a third course of HAART were taking four or more antiretroviral drugs with many receiving at least one agent from each of the three major classes (Table 1). Yet, our data indicate that the more extensively HAART pre-treated (i.e., the more deeply an individual is into salvage HAART), the less likely such therapy will be to provide durable viral suppression. Taken alone, this observation suggests an advantage to designing first HAART in order optimize the likelihood of successful response to salvage regimens, and may support a strategy of judicious sequencing of drugs to preserve subsequent treatment options .
Many clinicians have been likely to modify HAART when complete viral suppression has not been durable [13,14]. With experience, some clinicians are now more often willing to accept substantial if incomplete suppression of HIV replication in some situations [15,16]. A subset of HOPS patients exists for whom increased CD4 cell counts were seen despite a lack of observed viral load reduction, while remaining on initial HAART and doing no less well clinically than those who also demonstrated viral suppression .
Such observations challenge us to think critically about our notions of therapeutic success. Clearly, clinicians must weigh clinical, virologic, and immunologic criteria when deciding whether a given patient is successfully treated. Several pertinent questions arise. Should the goals of HAART therapy or the definition of treatment success vary for patients in different stages of clinical, immunologic, or virologic disease, or with different abilities to tolerate or adhere to complex therapies? Should our criteria for success differ for patients with more antiretroviral experience and, by implication, fewer remaining drug treatment options eligible for inclusion in a HAART salvage regimen to which viral isolates remain susceptible ? Are there circumstances in which incomplete viral suppression on therapy is, at least for a time, acceptable, such as when CD4 cell counts are stable or increasing [11,16]? Our observations suggest a possible ‘discordance’ between the progressively decreasing ability of sequential salvage HAART regimens to maintain optimal viral suppression, and such therapy's potential continued benefit in preventing HIV-related death and disease. They also raise the question of whether, among heavily treatment-experienced patients for whom novel antiretrovirals are not available, there may be clinical benefits consequent to continuing ‘failing’ salvage HAART therapy that are not linked per se to such treatment's ability to suppress plasma viremia (e.g., via attenuation of HIV fitness or direct immunological benefit of drug therapy).
If, however, our current indices of response to therapy are accurate, patients clearly derive progressively less benefit from successive and increasingly more complex, more expensive, and possibly more difficult to tolerate HAART regimens. If physician-prescribed regimens continue to exhaust remaining therapeutic options at a rate that outstrips the pace at which newer and more effective agents become available, will there come a time when we observe a recrudescence in HIV-associated death and disease rates?
The authors thank A. Greenberg, Centers for Disease Control and Prevention (CDC) and J. P. Phair, Northwestern University Medical School, for their ongoing encouragement and many helpful insights and suggestions. They also express their gratitude to T. Tong, CDC, for his assistance in the preparation of figures, and C. Chan, Northwestern University, for her assistance in data programming.
1. Palella FJ Jr, Delaney KM, Moorman AC. et al. Declining mortality and morbidity in HIV-infected ambulatory patients. N Engl J Med 1998, 328: 853–860.
2. Schooley RT. Longer-term immunologic effects and side effects of successful antiretroviral therapy. Clin Infect Dis 1999, 29: 12–18.
3. Forum for Collaborative HIV Research, George Washington University Medical Center. The Challenges of Clinical Trial Design in Assessing the Effects of Anti-HIV Therapy in Heavily Pre-treated Patients. A Report from a Workshop in Toronto, Canada, May 21 and 22, 1999.
4. Von Bargen JC, Moorman AC, Holmberg SD. How many pills do HIV infected patients take [letter]? JAMA 1998, 280: 29.29.
5. Wenger N, Gifford A, Liu H, et al. Patient characteristics and attitudes associated with antiretroviral adherenceSixth Conference on Retroviruses and Opportunistic Infections. Chicago, January–February 1999 [abstract 98].
6. Wainberg MA, Friedland G. Public health implications of antiretroviral therapy and HIV drug resistance. JAMA 1998, 279: 1977–1983.
7. Lucas GM, Chaisson RE, Moore RD. Highly active antiretroviral therapy in a large urban clinic: Risk factors for virologic failure and adverse drug reactions. Ann Intern Med 1999, 131: 81–87.
8. Weidle PJ, Lichtenstein KA, Moorman AC. et al. Factors associated with the successful modification of antiretroviral therapy. AIDS 2000, 14: 491–497.
9. Moorman AC, Holmberg SD, Marlowe SI. et al. Changing conditions and therapies of ambulatory HIV-infected patients: the HIV Outpatient Study (HOPS). Ann Epidemiol 1999, 9: 349–357.
10. Kaufmann D, Pantaleo G, Sudre P, Telenti A, the Swiss HIV Cohort Study. CD4-cell count in HIV-1-infected individuals remaining viraemic with highly active antiretroviral therapy (HAART). Lancet 1998, 35: 723–724.
11. Levitz, S. Improvement in CD4+ cell counts despite persistently detectable viral loads [letter]. N Engl J Med 1998, 338: 1074–1075.
12. Deeks SG, Barbour JD, Martin JN, Swanson RM. Sustained CD4+ cell response after virologic failure of protease inhibitor-based regimens in patients with human immunodeficiency virus infection. J Infect Dis 2000, 181: 946–953.
13. Zackin R, Marschner I, Andersen J. et al. Perspective: human immunodeficiency virus type 1 (HIV-1) RNA end points in HIV clinical trials: issues in interim monitoring and early stopping. J Infect Dis 1998, 177: 761–765.
14. Clough LA, D'Agata E, Raffanti S, Haas DW. Factors that predict incomplete virologic response to protease inhibitor-based antiretroviral therapy. Clin Infect Dis 1999, 29: 75–81
15. Goetz MB. Editorial response: Discordance between virological, immunologic, and clinical outcomes of therapy with protease inhibitors among human immunodeficiency virus-infected patients. Clin Infect Dis 1999, 29: 1431–1434.
16. Deeks SG, Wrin T, Liegler T. et al. Virologic and immunologic consequences of discontinuing combination antiretroviral drug therapy in HIV-Infected patients with detectable viremia. N Engl J Med 2001, 344: 472–480.
17. Gilbert PB, Degruttola V, Hammer SM, Kuritzkes DR. Virologic and regimen terminal surrogate endpoints in AIDS clinical trials. JAMA 2001, 285: 777–784.
The HOPS Investigators
A. C. Moorman, T. Tong, S. D. Holmberg, Division of HIV/AIDS Prevention, National Center for HIV, STD, and TB Prevention, Centers for Disease Control and Prevention, Atlanta, GA; K. C. Wood, R. K. Baker, Health Research Network of APACHE Medical Systems, Inc., McLean, VA; F. J. Palella, J. S. Chmiel, M. Deloria Knoll, J. Arnold, Northwestern University Medical School, Chicago, IL; K. A. Lichtenstein, K. S. Greenberg, B. Young, B. Widick, C. Stewart, P. Zellner, Columbia Rose Medical Center, Denver, CO; B. G. Yangco, K. Halkias, C. Lapierre, Infectious Disease Research Institute, Tampa, FL; D. J. Ward, C. A. Owen, Dupont Circle Physicians Group, Washington, DC; J. Fuhrer, L. Ording-Bauer, R. Kelly, J. Esteves, State University of New York (SUNY), Stony Brook, NY; E. M. Tedaldi, L. Walker-Kornegay, Temple University Hospital, Philadelphia, PA; J. B. Marzouk, R. T. Phelps, M. Rachel, Adult Immunology Clinic, Oakland, CA; R. E. McCabe, M. Rachel, Fairmont Hospital, San Leandro, CA. Cited Here...
© 2002 Lippincott Williams & Wilkins, Inc.