AIDS-related morbidity and mortality have declined sharply with the use of potent antiretroviral regimens in both randomized clinical trials and observational, community-based studies [1–5]. However, it is estimated that 20–50% of the patients have viral breakthrough within the first year of triple drug therapy . Rates of virologic failure increase rapidly with successive therapeutic regimens [7–10]. Despite contemporary recommendations, given the limited number of antiretroviral drugs available, it is often not possible to change all drugs in a triple drug regimen when plasma viral load rebounds . Of note, at the time of inception of this cohort, resistance testing was not available and recycling of previously failed antiretroviral agents was generally regarded as futile.
The management of patients who have failed several previous courses of antiretroviral therapy represents a major challenge in clinical practice. Unable to offer a conventional regimen to these individuals, the British Columbia (BC) Centre for Excellence in HIV/AIDS agreed to provide multiple drug rescue therapy (MDRT) to such patients through a single, university-affiliated referral clinic. In the absence of a clear treatment option, MDRT was initially undertaken as a therapeutic trial with the intent of ruling out a fully suppressive virologic response before considering a partially suppressive treatment strategy as a palliative option. We report here on the results of our initial experience involving 106 consecutive patients treated with MDRT regimens.
The BC Centre for Excellence in HIV/AIDS distributes antiretroviral agents at no cost to eligible HIV-infected individuals in the province of British Columbia, Canada. Plasma viral load monitoring is also offered free of charge. A physician enrolling an HIV-positive individual into the Centre's Drug Treatment Program completes a drug request and enrolment form. Each request is reviewed by the Centre's consultants to ensure that it follows the Centre's therapeutic guidelines. Patients entering the Drug Treatment Program may sign consent for their data to be linked to other databases for research purposes. In addition, the St. Paul's Hospital's Ethics Committee has approved the use of the data generated from the program for research purposes.
This analysis was based on a prospective cohort study of 106 consecutive patients. Patients were referred to the Immunodeficiency Clinic at St. Paul's Hospital by their primary care physician for the purposes of discussing salvage therapy. Patients had failed at least two separate courses of antiretroviral therapy and started a MDRT regimen between August 1997 and June 1998. MDRT regimens included between five and nine drugs: up to four nucleoside reverse transcriptase inhibitors [NRTI; didanosine (ddI), lamivudine (3TC), stavudine (d4T), abacavir]; up to two protease inhibitors (PI; chosen from indinavir, nelfinavir, saquinavir and/or ritonavir); up to two non-nucleoside reverse transcriptase inhibitors (NNRTI; nevirapine, delavirdine); and hydroxyurea as an adjuvant, as long as the absolute neutrophil count was > 0.7 × 109 cells/l. The most common starting regimen was three NRTI plus two PI, which was used by 45 (42.5%) patients (Table 1). The two most common sets of initial antiretrovirals included in the MDRT regimen were d4T/ddI/3TC/ritonavir/saquinavir taken by 20 (18.9%) patients and d4T/ddI/3TC/indinavir/nelfinavir taken by 15 (14.2%) patients. Overall, 67 (63.2%) participants used hydroxyurea during the study period. The MDRT regimens were adapted according to the patients’ history of previous antiretroviral exposure, tolerability, and baseline laboratory profile. The principal investigator (J. S. G. M.) determined the specific MDRT regimen. Patients were followed monthly while on MDRT therapy for virologic and immunologic responses, and for the development of adverse effects, which were recorded using the World Health Organization grading system. All drugs were available through Expanded Access Programs or through the Drug Treatment Program of the BC Centre for Excellence in HIV/AIDS. Before initiation of MDRT, baseline plasma viral load, CD4 cell count, and safety laboratory tests were performed, including complete blood count with differential counts, uric acid, amylase, liver function tests, urea, and creatinine.
Baseline HIV drug resistance phenotype was determined (VIRCO Antivirogram)  using aliquots of the stored samples obtained immediately prior to the initiation of the MDRT, while patients were receiving the previous antiretroviral regimen and had detectable plasma viral load. Baseline plasma viral loads were available from 95 out of 106 individuals. Isolates with greater than fourfold increases in 50% inhibitory concentration compared to a laboratory strain were deemed ‘resistant'.
All patients who initiated MDRT between August 1997 and June 1998 were included in the analysis. The proportions of patients with plasma viral load levels < 400 copies/ml and < 50 copies/ml following initiation of MDRT were determined in each of five consecutive periods of 11 weeks. The analysis was performed both on an intention-to-treat (ITT) and on an as-treated (AT) basis. ITT included data for all patients at all time points, whether or not they were taking antiretroviral therapy. Patients with missing measurements during that period were considered failures unless the plasma viral load measurements from the visit prior to, and the visit after the visit in question were both < 400 copies/ml. In these cases patients were assumed to have plasma viral load < 400 copies/ml. Similar methods were applied when the threshold was < 50 copies/ml. The AT analysis included data from patients only while they were taking antiretroviral therapy, either the original MDRT regimen or a modified regimen.
Logistic regression models were used to identify independent predictors of achieving plasma viral load < 400 copies/ml on two consecutive visits. Covariates entered into the model included baseline CD4 cell count, baseline plasma viral load and prior use of NNRTI.
Generalized estimating equation logistic regression models were used to examine the relationship between achieving plasma viral load < 400 copies/ml at any visit while on MDRT and the level of phenotypic drug susceptibility to different antiretroviral agents at baseline. All plasma viral load measurements for all patients were included in this analysis and classified as binary outcomes, either above or below 400 copies/ml. The generalized estimating equation model allows each plasma viral load measurement to contribute to the analysis but takes into account that plasma viral load measurements from the same individual are not independent . A block diagonal correlation matrix is used to specify the nature of the correlation among repeat observations within individuals and to ensure that measurements from different individuals are treated as independent observations. As with ordinary logistic regression models, the odds ratio associated with a particular variable is calculated as the exponential of the estimated regression coefficient for that variable. A covariate was created for the level of decreased susceptibility to each drug. The covariate was equal to zero if a patient was not receiving the drug, equal to one if the patient was on the drug and had less than a fourfold decrease in susceptibility to it, and equal to the inverse log10 of the fold resistance to the drug if the decrease in susceptibility was equal to or greater than fourfold. The covariate for hydroxyurea was coded as 1 if the patient ever took hydroxyurea during the study period and 0 otherwise, because resistance testing is not relevant for this drug. Covariates for each antiretroviral agent were entered in the model simultaneously. The model also controlled for the time on MDRT since baseline.
A total of 106 consecutive patients were studied. The median follow-up time was 15 months [interquartile range (IQR), 14 to 18 months]. There were 46 participants (43%) who had a diagnosis of AIDS at the time of initiation of the MDRT regimen. The median baseline plasma viral load was 62 000 copies/ml (IQR, 15 000 to 220 000) and median baseline CD4 cell count was 180 × 106/l (IQR, 90 × 106/l to 260 × 106/l). The median number of previous antiretrovirals used was seven (IQR, six to eight) and the median time of prior exposure to antiretroviral therapy was 43 months (IQR, 27 to 55 months). In addition to NRTI, all patients had been exposed to PI with a median exposure time of 15 months (IQR, 12 to 18 months). Among the 40 participants (38%) who had also used NNRTI, the median time of exposure was 7 months (IQR, 3 to 9 months).
Baseline phenotypes (VIRCO Antivirogram) were available for 95 out of 106 patients (90%). Forty-seven of the 95 patients (49.5%) had ≥ fourfold decreased susceptibility to all four protease inhibitors, 58 (61%) to zidovudine, 82 (87%) to 3TC, 24 (26%) to one NNRTI alone and a further 43 (46%) to both NNRTI. Fifty-six patients (59%) had ≥ fourfold decreased susceptibility to seven or more antiretroviral drugs (Fig. 1).
The median plasma viral load at baseline and at each follow-up period is shown in Fig. 2. A patient's change in plasma viral load was calculated by comparing the last available plasma viral load with his/her baseline value. The median change in plasma viral load was −1.04 log10 HIV RNA copies/ml (IQR, −2.63 to −0.03 log10) over a median time of 15 months and the median changes in absolute and percentage CD4 cell count were −10 × 106/l (IQR, −70 to +40 × 106/l) and 0% (IQR, −2% to +4%), respectively over a median time of 14 months. The median changes in CD4 cell counts among patients who did not and did take hydroxyurea during the study period were 0 cells/l (IQR, −40 × 106/l to 50 × 106/l) and −10 × 106/l (IQR, −70 × 106/l to 35 × 106/l) respectively (P = 0.31). The median change in helper/suppressor ratio was 0.02 (IQR, −0.02 to 0.10).
The proportions of patients with plasma viral load < 400 or < 50 copies/ml in each of five consecutive periods of 11 weeks are shown in Fig. 3. AT (ITT) analysis showed that 48% (40%) of 88 subjects had plasma viral load < 400 copies/ml during weeks 47–57.
Among the subset of patients who achieved plasma viral load < 400 copies/ml at least twice during the study period (n = 52), the probability of sustaining virologic suppression to < 400 copies/ml was calculated out to 40 weeks of follow-up using Kaplan–Meier methods to characterize the durability of the virologic response (Fig. 4). Time zero was defined as the patient's first plasma viral load test date in a series of at least two consecutive plasma viral loads of < 400 copies/ml. The time of virologic failure was defined as the time plasma viral load first rebounded to ≥ 400 copies/ml. The probability of remaining suppressed at 40 weeks of follow-up was 59% (± 8%).
Among the 40 triple class experienced (and virologically failed) patients, 37.5% responded to MDRT. For this purpose, a response is defined using an ITT principle as the ability of the patients to sustain a plasma viral load < 400 copies/ml at two consecutive visits at least 1 month apart after starting MDRT. One-third of the 40 patients in this subset had a plasma viral load < 400 copies/ml at weeks 47–57 using an ITT principle.
Predictors of virologic response
Fifty-two patients (49%) achieved at least two consecutive plasma viral loads < 400 copies/ml during the study period (Table 2). The comparison of baseline characteristics showed that median plasma viral load (P < 0.01), median CD4 cell counts (P = 0.01), and the proportion of patients naive to NNRTI (P = 0.06) were different between patients who achieved the outcome and those who did not. Responders and non-responders were comparable with respect to age, sex, proportion of injecting drug users, proportion diagnosed with AIDS at baseline and the overall number and time of previous exposure to antiretroviral agents.
In a multivariate logistic regression model controlling for baseline CD4 cell count and follow-up time, higher baseline plasma viral load decreased the likelihood of responding to MDRT by a factor of 0.45 per log10 copies/ml (95%CI, 0.25–0.83;P = 0.01). In addition, being NNRTI-naive increased the likelihood of response to MDRT by 2.83 times (95% CI, 1.14–7.00;P = 0.025). When the same analysis was done for the subgroup of 92 participants who were prescribed NNRTI, higher baseline plasma viral load (in log10 copies/ml) decreased the likelihood of response by a factor of 0.39 (95%CI, 0.20–0.76;P = 0.006), whereas being naive to NNRTI increased the likelihood of response 3.06 times (95%CI, 1.15–8.12;P = 0.025). If response was defined as having a plasma viral load < 50 copies/ml on at least two consecutive measurements, only baseline viral load differed between responders and non-responders (Table 3). In a logistic regression model, baseline plasma viral load was associated with decreased likelihood of response by a factor of 0.47 per log10 copies/ml (95% CI, 0.29–0.77;P = 0.002).
Generalized estimating equation regression models were used to examine the relationship between achieving plasma viral load < 400 copies/ml while on MDRT, the level of phenotypic resistance to different drugs at baseline and whether or not patients actually took various drugs during the study period (Table 4). Patients who took 3TC, d4T, ddI, and saquinavir and who were susceptible to the drugs at baseline were more likely to achieve plasma viral load < 400 copies/ml at P-values < 0.05 (OR, 3.41, P = 0.009; OR, 2.88, P = 0.01, OR, 3.13, P = 0.008 and OR, 3.82, P = 0.005 respectively). Patients who took indinavir and nevirapine and who were susceptible to these drugs as baseline were also more likely to achieve plasma viral load < 400 copies/ml (OR, 2.11, P = 0.05 and OR, 1.94, P = 0.08 respectively). Interestingly, patients taking abacavir who were susceptible at baseline were less likely to achieve plasma viral load < 400 copies/ml (OR, 0.52;P = 0.04). These results remained substantially unchanged when a plasma viral load < 50 copies/ml was used as the outcome of interest.
Overall, 91 subjects (86%) had either laboratory abnormalities or subjective adverse drug effects (ADE). The frequency distribution of these various abnormalities/ADE are shown in Table 5. The most common type of mild/moderate laboratory abnormalities was changes in liver function tests found in 67 (63%) patients. Alterations in lipid profile (increased triglycerides and/or cholesterol) found in six (6%) patients were the most common severe laboratory abnormalities. Gastrointestinal complaints were the most common type of subjective ADE and these included nausea, vomiting, diarrhea, and abdominal pain. Neurological/psychological disorders include circumoral paresthesias, peripheral neuropathy, irritability, and mood swings. Laboratory abnormalities and subjective ADE of varying severity led to regimen modifications in 28% of MDRT patients.
A total of 16 patients (15%) discontinued therapy during the study. Among them, six discontinued due to ADE, three were lost to follow-up and seven patients died due to AIDS-related illnesses. All deceased patients had an AIDS diagnosis at the beginning of the study. Altogether, 26 patients (25%) had severe clinical or laboratory ADE. One patient with AIDS diagnosis at baseline developed cryptosporidiosis by week 20 on MDRT. One patient developed Pneumocystis carinii pneumonia as primary AIDS-defining illness by week 9 on MDRT.
Multiple drug rescue therapy regimens consisting of five to nine drugs were prescribed to a group of 106 consecutive, heavily pretreated patients resulting in a median plasma viral load decrease of 1.04 log10 copies/ml over a median of 15 months. According to an ITT analysis, 40% of the patients had plasma viral load < 400 copies/ml between weeks 47 and 57 of follow-up. Overall, 52 (49%) and 35 (33%) of the patients had at least two consecutive plasma viral load determinations < 400 copies/ml and < 50 copies/ml, respectively.
Patients taking any of 3TC, d4T, ddI, saquinavir, indinavir and nevirapine who were susceptible to these drugs at baseline were more likely to achieve plasma viral load < 400 copies/ml. Patients taking abacavir who were susceptible to abacavir at baseline were less likely to achieve plasma viral load < 400 copies/ml (OR, 0.52, P = 0.04). This unexpected result arose because of the fact that abacavir was not available to most patients when they started their MDRT therapy and only became widely available through expanded access part-way through the study. Thus, most of the patients were not offered abacavir as part of their initial MDRT regimen but were only offered it later as further salvage when they were failing their assigned MDRT regimen. The negative association between abacavir and virologic success is probably related more to the illness level of patients offered abacavir than to the effectiveness of the drug.
Our results are in general agreement with those presented recently by Staszewski et al., who reported on 37 heavily pretreated HIV-1 positive patients who were followed for a median of 7 months and received treatment with six or more antiretrovirals (with 65% of patients taking three PI). Fifteen patients (41%) exhibited plasma viral load < 20 copies/ml in one determination after initiation of therapy . More recently, Youle et al. reported data on 63 subjects who previously failed PI and were offered multi-drug combinations with a backbone of efavirenz, ddI, ritonavir, indinavir and hydroxyurea, showing 3.08 log10 copies/ml median decrease in plasma viral load levels after 28 weeks of therapy .
Similar to Youle et al., we included hydroxyurea as part of the MDRT regimen as long as the regimen included ddI and there was no contraindication for its use (i.e. history of pancreatitis or severe neutropenia). The ability of hydroxyurea to enhance the antiviral effect of ddI in vivo has been clearly documented among chronically ddI-treated patients  and antiretroviral therapy naive patients . However, the use of hydroxyurea has been associated with a blunting of the absolute CD4 cell response without detriment to the increase in the CD4 fraction. Of note, Staszewski et al. reported some degree of CD4 cell count increase with the use of multiple drug regimens, but this was not the case in either Youle's cohort or ours, both of which included hydroxyurea. The ultimate contribution of this agent to the MDRT regimen deserves to be examined in a randomized, controlled setting, particularly because the use of hydroxyurea was not associated with improved virologic response in our study. Overall, we did not see a change in CD4 cell counts in our patient population despite a substantial plasma viral load response. It should be noted that patients switched to MDRT without interruption of antiretroviral therapy and, therefore, this could have minimized the magnitude of changes in CD4 cell counts as a result of the subsequent antiviral response. Further work will be required, however, to understand better to what extent this is the case. Lack of CD4 cell count improvement could also represent a subtle toxicity of the multiple drugs used in the regimen and could eventually limit the viability of this approach. As currently put forward, this approach is intended as a bridging mechanism to prevent further viral evolution (with regard to resistance to antiretroviral agents) while newer therapeutic options become available for a given patient. In this context it is reassuring that even when hydroxyurea was included in the MDRT regimen we did not see a decrease in CD4 count.
Almost all patients in this analysis experienced at least some ADE, most were mild to moderate and did not interfere with continuation of treatment. Subjective ADE, although less frequent than laboratory abnormalities, were often more serious and most frequently involved the gastrointestinal tract. No unexpected toxicity was observed as a result of combining multiple agents. Of note, these data do not allow us to make any inferences regarding the potential for long-term toxicity associated with MDRT.
Responders were characterized by having a lower plasma viral load at the time of initiation of MDRT. Our results support the current recommendation of changing the antiretroviral therapy regimen soon after plasma viral load rebound is confirmed . This strategy could potentially increase the likelihood of achieving an antiviral response with a subsequent regimen by reducing viral replication under incomplete drug selective pressure, thereby limiting the development of resistance to the drugs involved, as shown recently by Havlir et al. .
As expected, NNRTI naive patients had a significantly greater rate of virologic response than triple class experienced patients. The logistic regression model demonstrated that being NNRTI naive increased the response rate by almost threefold. This is consistent with current guidelines, which encourage the use of as many new agents (or drug classes) as possible in this setting. Of note, the response rate was 37.5% among triple class experienced patients. It should be emphasized that even this relatively low rate of responsiveness was more than anticipated at the inception of the cohort. Also of interest is to note that approximately one-third of the NNRTI resistant patients demonstrated decreased susceptibility to only one of the NNRTI. Recent data suggest that these cases may represent intermediate levels of decreased susceptibility rather than a high level of resistance. Harrigan et al.  recently presented data from a cohort of antiretroviral therapy naive individuals who initiated triple drug therapy with one NNRTI and two NRTI. Moderate decreases in baseline susceptibility (in the range of four to 10-fold) did not appear to have a negative impact on virologic responses in this patient population. Whether these results apply to heavily pretreated individuals remains to be elucidated.
Resistance testing has just recently become available for the management of HIV infected patients. Drug resistance information was not available when our patients initiated MDRT regimens, but at a later time using samples stored at baseline. Not surprisingly, baseline resistance to multiple drugs was highly prevalent in our study group. Of note, decreased susceptibility at baseline was inversely correlated with the likelihood of a favorable virological response. Recent studies have demonstrated that the development of resistance tends to be an evolutionary process [17,19]. In other words, a detectable HIV RNA while on a triple drug regimen does not necessarily indicate resistance to all three drugs in the regimen. As such, recycling agents based on drug exposure carries a meaning different from that based on the results of resistance testing; this is supported by our results in that recycling of drugs tended to be more successful as a function of residual phenotypic sensitivity to the drugs used. This suggests that in the future, it might be possible to modify therapeutic choices by taking resistance testing results into account, as has been suggested initially by several recent small studies [20–24]. Strategies designed to prospectively assess the utility of incorporating resistance testing into the management of heavily pretreated patients need to be evaluated urgently in larger randomized clinical trials. We identified individuals with susceptibility to 3TC, d4T, ddI and saquinavir as more likely to achieve a significant virologic response to MDRT. However, among the group of patients phenotypically resistant to all available antiretroviral drugs, one was a long term MDRT responder, once again highlighting the fact that historical and laboratory evidence are insufficient to rule out a therapeutic response to MDRT. Pharmacokinetics, drug metabolism, between-drug effects with regard to drug activation or even antiviral activities are only a few of the multiple remaining issues to be explored in this context.
It should be noted that the degree of resistance (fold-change in 50% inhibitory concentration) observed may not be directly comparable between the different drugs. For example, the fold-changes observed for the nucleoside analogue ‘di-deoxy’ drugs are generally small (< 10-fold), relative to those observed for the PI, NNRTI or other nucleoside analogues. Furthermore, in some cases the fold-change may be a minimum estimate, as concentration of inhibitor required to inhibit viral replication can be so high that it results in cell death.
In summary, the MDRT approach led to a virologic response in 30–50% of heavily pretreated patients at 1 year of follow up. Lower baseline plasma viral load, being NNRTI naive at the time of initiation of the regimen, and having virologic susceptibility to 3TC, ddI, d4T and saquinavir were associated with favorable virologic response. Despite the relatively high incidence of ADE, most patients were able to continue therapy. Our data support the need for the critical evaluation of MDRT regimens in prospective randomized clinical trials. Until then, we suggest that in the absence of other clear treatment options, a carefully monitored therapeutic trial of a MDRT regimen may be warranted for selected heavily pretreated patients.
The authors thank VIRCO (Belgium) for assistance with the virological susceptibility testing, C. Sherlock for assistance with HIV RNA levels determination, and S. Rae and J. Tilley for help with statistical analyses.
1. Hogg RS, Rhone SA, Yip B. et al. Antiviral effect of double and triple drug combinations amongst HIV-infected adults: lessons from the implementation of viral load-driven antiretroviral therapy.
AIDS 1998, 12: 279 –284.
2. Hogg RS, Heath KV, Yip B. et al. Improved survival among HIV-infected individuals following initiation of antiretroviral therapy.
JAMA 1998, 279: 450 –454.
3. Palella F, Delaney K, Moorman A. et al. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection.
N Engl J Med 1998, 338: 853 –860.
4. Hammer S, Squires K, Hughes M. et al. A controlled trial of two nucleoside analogues plus indinavir in persons with human immunodeficiency virus infection and CD4 cell counts of 200 per cubic millimeter or less.
N Engl J Med 1997, 337: 725 –733.
5. Hogg RS, Yip B, Kelly C. et al. Improved survival among HIV infected individuals following initiation of triple combination antiretroviral regimens.
Can Med Assoc J 1999, 160: 659 –665.
6. Montaner JSG, Hogg R, Raboud J, Harrigan PR, O'Shaughnessy MV. Antiretroviral treatment in 1998.
Lancet 1998, 352: 1919 –1922.
7. Carpenter CCJ, Fischl MA, Hammer SM. et al. Antiretroviral therapy for HIV infection in 1998.
:Updated recommendations of the International AIDS Society-USA Panel.
JAMA 1998, 280: 78 –86.
8. Leigh Brown AJ, Richman DD. HIV-1: gambling on the evolution of drug resistance?.
Nature Med 1997, 3: 268 –271.
9. Harrigan PR, Dong W, Weber AE, et al. Clinical failure is associated with highly mutated RT and protease. 38th Interscience Conference on Antimicrobial Agents and Chemotherapy.
San Diego, September 1998 [abstract I-115].
10. Harrigan PR, Hogg R, Yip B. et al. Baseline resistance profile predicts response to ritonavir/saquinavir therapy in a community setting.
AIDS 1999, 13: 1863 –1871.
11. Hertogs K, De Bethune M, Miller V. et al. A rapid method for simultaneous detection of phenotypic resistance to inhibitors of protease and reverse transcriptase in recombinant human immunodeficiency virus type 1 isolates from patients treated with antiretroviral drugs.
Antimicrobial Agents Chemother 1998, 42: 269 –276.
12. Liang K-Y, Zeger SL. Longitudinal data analyses using generalized linear models.
Biometrika 1986, 73: 13 –22.
13. Staszewski S, Gute P, Carlebach A. et al. Virological and immunological response to mega-HAART salvage therapy after failure of multiple antiretroviral regimens.
AIDS 1998, 12 (suppl 4): S40. S40.
14. Youle M, Mocroft A, Johnson M. et al. Surrogate marker responses to multidrug combinations comprising hydroxyurea, efavirenz, double protease inhibitors and nucleoside analogues in protease inhibitor failures.
Antiviral Ther 1999, 4 (suppl 1): 18 –19.
15. Montaner JSG, Zala C, Conway B. et al. A pilot study of hydroxyurea among patients with advanced human immunodeficiency virus (HIV) disease receiving chronic didanosine therapy: Canadian HIV Trials Network Protocol 080.
J Infect Dis 1997, 175: 801 –806.
16. Frank I, Boucher H, Fiscus S. Phase I/II dosing study of once-daily hydroxyurea (HU) alone vs didanosine (ddI) alone vs ddI+HU. Sixth Conference on Retroviruses and Opportunistic Infections
. Chicago, January–February1999 [abstract 402].
17. Havlir D, Hellmann N, Petropoulos C. et al. Drug susceptibility in HIV infection after viral rebound in patients receiving indinavir-containing regimens.
JAMA 2000, 283: 229 –234.
18. Harrigan PR, Verbiest W, Larder B. et al. Impact of moderate decreases in baseline NNRTI susceptibility on response to antiretroviral therapy.
Antiviral Ther 2000, 5 (suppl 3): 68. 68.
19. Descamps D, Flandre P, Calvez V. et al. Mechanisms of virologic failure in previously untreated hiv-infected patients from a trial of induction-maintenance therapy.
JAMA 2000, 283: 205 –211.
20. Baxter JD, Mayers DK, Wentworth DN, et al. A pilot study of the short-term effects of antiretroviral management based on plasma genotypic antiretroviral resistance testing (GART) in patients failing antiretroviral therapy. Sixth Conference on Retroviruses and Opportunistic Infections.
Chicago, January–February 1999 [abstract LB8].
21. Durant J, Clevenberg P, Halfon PH, et al. Prevalence of resistance-associated genotypic mutations in plasma HIV of patients failing triple therapy combination. XII International Conference on AIDS.
Geneva, June 1998 [abstract 32301].
22. Falloon J, Masur H, Brosgart C, et al. Salvage therapy with abacavir, amprenavir and efavirenz in subjects with plasma HIV-1 RNA >500 copies/mL despite protease inhibitor therapy. 36th IDSA Annual Meeting.
, Denver, Co, USA, November 1998 [abstract 441].
23. Deeks SG, Hellman S, Grant R. et al. Novel four-drugs treatment regimens after failure of a human inmmunodeficiency virus type 1 protease inhibitor-containing regimen: Antiviral activity and correlation of baseline phenotypic drug susceptibility with virologic outcome.
J Infect Dis 1999, 179: 1375 –1381.
24. Weidle PJ, Lichtenstein KA, Moorman AC. et al. Factors associated with the successful modification of antiretroviral therapy.
AIDS 2000, 31: 491 –497.