Secondary Logo

Journal Logo

Virological and immunological effects of treatment interruptions in HIV-1 infected patients with treatment failure

Miller, Veronicaa; Sabin, Carolineb; Hertogs, Kurtc; Bloor, Stuartd; Martinez-Picado, Javiere,*; D'Aquila, Richarde; Larder, Brendand; Lutz, Thomasa; Gute, Petera; Weidmann, Eckharta; Rabenau, Holgera; Phillips, Andrewb; Staszewski, Schlomoa

Clinical Science

Objective To analyse the immunological and virological effects of treatment interruptions in HIV-1-infected patients with treatment failure and multidrug-resistant virus.

Methods Drug susceptibility was assessed using Antivirogram and genotypic analysis was based on population and clonal sequencing for 48 patients who had interrupted treatment (≥ 2 months).

Results Treatment interruption resulted in viral load increases (mean 0.7 log10 copies/ml;P = 0.0001) and CD4 cell count decreases (mean 89 × 106 cells/l;P = 0.0001). A complete shift to wild-type virus at the phenotypic, genotypic and clonal level was observed in 28/45 patients. These patients differed from those that did not show a shift to wild type in baseline CD4 cell counts (192 versus 59 × 106 cells/l;P = 0.007) and in the relationship between baseline viral load and CD4 cell count (no correlation versus a significant negative correlation;P = 0.008). Response to re-initiation of treatment fell with increasing viral load [relative hazard (RH) 0.33;P = 0.001] and with increasing total number of drugs with reduced susceptibility (RH 0.51;P = 0.0003); it improved with the number of new drugs received (RH 2.12;P = 0.0002) and a shift to wild type (RH 5.22, P = 0.006).

Conclusions Changes in surrogate markers suggest that treatment provided benefit in spite of virological failure and resistant virus. Although patients with a shift to wild-type virus responded better in the short term to treatment re-initiation, the long-term effects are not known and the risk of immune deterioration needs to be carefully considered.

From the aKlinikum der Johann Wolfgang Goethe-Universität, Zentrum der Inneren Medizin, Infektionsambulanz Haus 68, Frankfurt, Germany, the bRoyal Free Centre for HIV Medicine and Department of Primary Care and Population Sciences, Royal Free and University College Medical School, London, UK, cVIRCO Laboratories, Mechelen, Belgium, dVIRCO Laboratories, Cambridge, UK and eMassachusetts General Hospital, Charlestown, Massachusetts, USA. *Present address: Retrovirology Laboratory, irsiCaixa Foundation, University Hospital Germans Trias i Pujol, Barcelona, Spain.

Requests for reprints to Dr V. Miller, Klinikum der Johann Wolfgang Goethe-Universität, Zentrum der Inneren Medizin, Infektionsambulanz Haus 68, Theodor-Stern-Kai 7, 60590 Frankfurt, Germany.

Received: 15 June 2000;

revised: 21 August 2000; accepted: 4 September 2000.

Sponsorship: This work was supported in part by a grant from the National Institute of Health, USA (NIH AI29193). Javier Martinez-Picado was supported by a postdoctoral fellowship from the Spanish Ministry of Education and a contract from the ‘‘Fundacio per a la Recerca Biomedica Germans Trias i Pujol.

Back to Top | Article Outline


Current guidelines for treatment of HIV disease recommend potent combination treatment on a continuous basis for maximal long-term suppression of virus replication [1,2] and patients are advised to adhere closely to prescribed dosing schedules. However, it is known that treatment interruptions may occur owing to drug toxicity, during management of specific opportunistic infections, or for reasons related to compliance and/or quality of life. In addition, patients with multidrug-resistant virus and their treating physicians may consider treatment discontinuation as an alternative to continuing treatment with apparently ineffective regimens after multiple previous treatment failures [2].

The safety of treatment interruption strategies in patients with uncontrolled virus replication is not known. The benefit of treatment in spite of viral rebound in terms of CD4 cell count [3,4] and clinical progression [5] has been described, but the duration of this effect is not known and CD4 cell decline will eventually set in after prolonged virological failure [6]. The potential benefit of continued treatment in the face of resistance needs to be weighed against the risk of further viral evolution and the risk of continued exposure to drug-related toxicities.

Drug resistance-associated mutations may affect viral fitness [7–9] and where a loss of fitness compared with wild type in the absence of drugs is the result, a shift to predominantly wild type within the population would be predicted upon withdrawal of drug pressure. Treatment discontinuation-associated shifts from predominantly mutant populations to predominantly wild-type populations have been described for some antiretroviral agents including lamivudine and indinavir [10–12]. A rapid reappearance of virus with sensitive genotype was observed in patients interrupting reverse transcriptase inhibitor therapy [13] However, mutations associated with resistance to other drugs may persist long after discontinuation of that particular drug [14]. Preliminary studies based on small numbers of patients with multiple previous treatment failures and multidrug-resistant virus had indicated that in some patients, but not all, a shift to complete wild-type population may occur upon discontinuation of all treatments [15,16]. The effect of stopping individual drugs or the complete combination regimen on changes in resistance patterns will depend on the extent of linkage of mutations and the heterogeneity of the virus population [17].

In this observational study we have analysed the immunological and virological effects of treatment interruption in a larger set of patients from the Frankfurt HIV cohort with extensive antiretroviral treatment history. We describe changes in viral load, CD4 cell count and viral drug susceptibility status occurring as a result of treatment interruption in the total population as well as in two subgroups: those with virus population that shifts to wild type and those that have a virus population that retains reduced susceptibility and resistance-associated mutations in spite of lack of drug pressure. In addition, we describe the response to subsequent re-initiation of treatment.

Back to Top | Article Outline

Subjects and methods


The Frankfurt HIV cohort consists of all patients who have presented at least once since 1 January 1995 to the HIV Outpatient Clinic, Frankfurt, Germany. Demographics, antiretroviral treatment, CD4 and CD8 cell counts, viral load and clinical events are collected prospectively on a continuous basis [18,19]. Stored plasma samples allow for retrospective virological analysis.

For the purposes of this study, antiretroviral-treated patients were selected who had a documented treatment interruption of at least 2 months starting at a time of detectable viral load. Each had stored plasma samples for resistance testing available at baseline (start of treatment interruption) and at end of treatment interruption.

To date, information has been collected on 48 patients with at least one treatment interruption over follow-up. One patient took two distinct treatment interruptions; this patient has been included in the analysis twice as it was felt that the two interruption periods could be treated independently. Therefore, the analysis is based on 49 treatment interruption episodes. Exclusion of the second treatment interruption for this patient had little impact on the results.

Back to Top | Article Outline

Viral drug susceptibility status

Viral drug susceptibility status to each of the antiretroviral agents currently in clinical use was assessed at the start and at the end of the treatment interruption period, using a method based on a recombinant virus assay, Antivirogram (Virco Laboratories, Mechelen, Belgium) [20]. Resistance was defined as an increase of at least fourfold in IC50 (50% inhibitory concentration) relative to a wild-type control. Genotypic changes associated with resistance were assessed using population-based sequencing [21].

For clonal sequencing, 1 ml plasma was concentrated by ultracentrifugation at 21 000 ×g for 1 h and virion RNA was extracted (Viral RNA preparation kit; Qiagen, Chatsworth, California, USA). RNA was reverse transcribed (Superscript II, Gibco BRL, Gaithersburg, Maryland, USA) as described [22]. The cDNA was amplified in nested polymerase chain reaction (PCR) with XL rTth DNA polymerase (PE Biosystems, Foster City, California, USA) and products were cloned using uracil deglycosylase into either pJM20 GPRT [22] or pAMP1 [23]. Cycle sequencing used dye-labelled terminators (BigDye terminator cycle sequencing ready reaction kit and ABI 377 sequencer; PE Biosystems). Sequences of 12–16 molecular clones per PCR product were analysed from the 3′ end of the p7gag open reading frame, through to the gag p7/p1 cleavage site, p1gag reading frame, gag p1/p6 cleavage site, p6gag reading frame, the protease and the reverse transcriptase coding sequences to codon 350 of the reverse transcriptase (Sequencer, Gene Codes, Ann Arbor, Mississippi, USA).

Patients were grouped according to the changes in susceptibility status as a result of treatment interruption. Those with a virus population that shifted to wild type are designated ‘shift-yes’ and those with a virus population that did not shift to wild type as ‘shift-no'. Demographic, clinical, virological and immunological factors were compared in these two groups using either the Mann–Whitney U test, or chi square test (Fisher's exact test when expected numbers were small) as appropriate. Multivariate logistic regression methods were used to assess which of these factors, if any, was independently associated with a shift to wild-type virus over the treatment interruption period. Changes in the patients’ viral load and CD4 cell count over the treatment interruption period were tested for significance using the paired t-test.

Virological response to re-initiation of antiretroviral treatment was considered by studying the time for the viral load to fall to < 500 copies/ml. Patient follow-up was considered from the time of starting therapy until the time of the first viral load < 500 copies/ml. Follow-up was right-censored at the last visit. Factor-associated responses were studied using Kaplan–Meier analyses and Cox proportional hazards regression models.

Back to Top | Article Outline


Patient population

A total of 49 treatment interruption episodes (ocurring in 48 patients) were included in this analysis; 92% of the patients were male, with a median age of 37.7 years (range 21.7–57.6) at baseline (start of treatment interruption). The median duration of treatment interruption was 121 days (range 54–322).

Previous antiretroviral treatment experience was extensive; the median number of drugs to which patients had already been exposed was nine (range 4–13) and 45 patients had been exposed to all three classes of drug. Prior to the treatment interruption episode, all patients had received zidovudine and lamivudine. Forty-eight patients (98%) had received stavudine, 42 (85.7%) didanosine, 41 (83.7%) saquinavir, 41 (83.7%) indinavir, 39 (79.6%) nevirapine, 35 (71.4%) ritonavir, 32 (65.3%) zalcitabine, nine (18.4%) abacavir and seven (14.3%) loviride. Patients had first started taking antiretroviral therapy a median of 3.9 years (range 0.3–9.2) prior to the start of treatment interruption.

Back to Top | Article Outline

Baseline virological and immunological characteristics

The median viral load at baseline was 5.07 log10 copies/ml (range 2.70–6.70) and the median CD4 cell count was 155 × 106/l (range 2–777) (Table 1).

Table 1

Table 1

Resistance was analysed primarily at the phenotypic level. It was possible to obtain phentoypes for 45 pre- and postsample pairs. Prior to interrupting treatment, viral resistance to all three drug classes was extensive (Table 2). The individual patient's viruses were resistant to a median number of eight (range 2–11) drugs. Among the nucleoside reverse transcriptase inhibitors, the drugs with most frequent resistance were lamivudine (80.5% of patients) and zidovudine (78.0% of patients). Of note, 471% of the patients had viruses resistant to abacavir, whereas only 18.4% had received abacavir prior to treatment interruption. The frequency of resistance to non-nucleoside reverse transcriptase inhibitors and protease inhibitors was high in this patient population (57.7–77.8% for the former and 50–75% for the latter). If resistance was defined as an increase of 10-fold or greater in IC50, 66% of samples displayed resistance towards at least one nucloside reverse transcriptase inhibitor (excluding lamivudine), 70% towards at least one non-nucloside reverse transcriptase inhibitor and 73% to at least on protease inhibitor. Twenty seven patients had viruses with resistance to all three drug classes at the fourfold level and 23 patients at the 10-fold level. The extent of baseline resistance is aslo reflected by the increase in IC50 for the individual drugs. For example, the mean (median; range) increase was 199-fold (18; 1–2067) for zidovudine, 43-fold (44; 1–102) for lamivudine, 232-fold (63; 1–1845) for nevirapine, 14-fold (6; 1–102) for indinavir, 24-fold (10; 1–127) for ritonavir, 230-fold (5; 1–8696) for saquinavir and 54-fold (17; 11–727) for nelfinavir.

Table 2

Table 2

Mutations associated with reduced drug susceptibility in protease and reverse transcriptase that were observed in the patient samples are shown in Fig. 1.

Fig. 1.

Fig. 1.

Back to Top | Article Outline

Virological and immunological changes during treatment interruption

At the end of the treatment interruption period, the median viral load was 5.87 log10 copies/ml (range 4.11–6.70) (Table 1). The viral load had increased in 36 (75%) patients, decreased in eight (16.7%) and remained stable in four (8.3%). This translates into a significant intra-individual change in viral load, with a median value of 0.71 log10 copies/ml (range −1.73 to 2.37;P = 0.0001, paired t-test).

At the end of the treatment interruption period, the median CD4 cell count was 49 × 106 cells/l (range 1–439) (Table 1). CD4 cell counts had decreased in 44 patients, resulting in a significant intra-individual decrease in CD4 cell count, with a mean value of −89 × 106 cells/l (range −410 to 16;P = 0.0001, paired t-test).

Profound changes in resistance were noticed as a result of discontinuation of all antiretroviral treatments. As seen in Table 2, significant reduction in the proportion of patients with viruses resistant to 11 out of 13 drugs was observed. In some patients (28), the virus population appeared to have shifted completely to wild type; in others (17) it remained drug resistant. The two groups (those with complete shift to wild type and those who retained resistant virus populations) did not appear to differ in the resistance pattern prior to starting the treatment interruption. In both groups, patients had viruses that were resistant to a median of eight (range 2–11) drugs. Of the patients with virus resistant (10-fold increase in IC50) to at least one drug from each class, 68% experienced a shift to wild-type virus and 32% did not. The phenotypic changes were also reflected at the genotypic level, with a shift to wild-type amino acid residues at resistance-associated positions in both the protease and reverse transcriptase (Fig. 1), with the exception of positions known for extensive polymorphism, such as position 63 of protease (shift from 100% pre-interruption to 90% post-interruption) and, to a lesser extent, positions 10 and 77 (96 or 41%, respectively, pre-interruption to 15% post-interruption).

In patients with multidrug-resistant viruses owing to sequential suboptimal treatment regimens, mutations need not be genetically linked on a single genome. In order to address this, genomic sequences were analysed at the clonal level for samples from one patient (Table 3). The samples originated from 14 days prior to treatment interruption [ER (–14)] and days 30, 60 and 90 after stopping all medications [ER (+30), ER (+60) and ER (+90), respectively, where ER is the patient identification tag]. In general, the data indicate that mutations were linked on the same genome for both protease and reverse transcriptase, with some indication for viral heterogeneity. Prior to treatment interruptions, both bulk and clonal analysis revealed mutations associated with resistance at protease positions 10, 36, 54, 63, 71 and 82 as well as a p7/p1 cleavage site mutation. In addition, mutations at positions 46 (3 out of 16 clones) and 84 (3 of 16 clones) were evident only by clonal analysis, indicating viral heterogeneity. Reverse transcriptase bulk and clonal analysis revealed mutations at positions 67, 70, 181, 184 and 219. One variant was found with the additional mutation K103N, associated in this case with Y181I rather than Y181C, which was found in 10/12 sequenced clones. A frame-shifting single-base insertion at position 67 of reverse transcriptase was found in 4/12 clones originating from the pre-treatment interruption sample. Therefore, although relatively homogenous, evidence for heterogeneity was found in both protease and reverse transcriptase. Thirty days after stopping all medications, protease positions associated with resistance were wild type at the population sequencing level, and predominantly wild type at the clonal level, with isolated incidences of mutations at positions 10, 24, 46, 54, 71, 82, 84 and 90. Consistent with the loss of mutations conferring resistance to protease inhibitors at this time point, the p7/p1 cleavage site mutation was also lost at 30 days. However, mutations that were present in the gene for the reverse transcriptase prior to treatment interruption largely remained, with evidence of wild-type/mutated mixtures at positions 67 and 181; interestingly, wild-type/mutated mixture was also observed at position 98, which had been fully wild type prior to treatment interruption. Position 184, for which rapid switches to wild type have previously been described [24], was consistently mutated at the bulk and clonal level. After 60 and 90 days without treatment, the virus population shifted to a more homogenous wild type; however proline was retained at protease position 63 and arginine at reverse transcriptase position 70. Reverse transcriptase position 98 evolved via a wild-type/serine/glycine mixture to serine (11/14 clones).

Table 3

Table 3

Back to Top | Article Outline

Factors associated with a shift to wild type

Changes to wild type at resistance-associated positions after cessation of specific drug pressure has been observed previously for mutations associated with individual drugs [10,11]. However, this is not always the case, and for positions associated with resistance to non-nucleoside reverse transcriptase inhibitors, for example, resistance-associated mutations may remain in the predominant virus population long after stopping that drug treatment [14]. The complete reversal to wild type – including positions associated with resistance to the non-nucleoside reverse transcriptase inhibitors – was surprising. There were no differences in terms of demographic characteristics amongst patients with virus that shifted to wild type (n = 28) and those whose viruses retained drug resistance (n = 17) (Table 1). The duration of treatment interruption did not differ among patients in the shift-yes and shift-no groups (P = 0.59, Mann–Whitney U test). Factors measured at the start of the treatment interruption (CD4 cell count, viral load, demographic and prior treatment factors, viral resistance phenotype) were assessed for association with the complete shift to wild type. Based on logistic regression analyses, only the CD4 cell count at the time of starting the treatment interruption [odds ratio (OR) 2.32 for 100 × 106/l higher; 95% confidence interval (CI) 1.14–4.71;P = 0.02] and the time since first starting antiretroviral therapy (OR 0.66 per additional year; 95% CI 0.47–0.95;P = 0.02) were associated with a shift to wild type over the treatment interruption period. The median CD4 cell count was 192 × 106 cells/l (range 2–777) for the patients with shift to wild type and 58 x 106 cells/l (range 1–367) for those with no shift (Table 1). As for the time on drugs, patients not experiencing a shift to wild type were treated for a median of 5.5 years (range 1.8–9.2) and those with a shift to wild type were treated for a median of 3.5 years (range 0.7–8.1) (P = 0.01, Mann Whitney U test). No other demographic factors, previous treatment history (number or type of drug received) or phenotypic resistance to any of the drugs were associated with a shift to wild type. In a multivariate model, both the CD4 count and the time since starting therapy were associated with shift to wild type at a marginally significant level (P = 0.07).

Back to Top | Article Outline

Virological and immunological effects of treatment interruption with and without viral population shift to wild type

Virus load increases and CD4 cell count decreases were compared in the two groups of patients: with and without a shift in their viral population to wild type (shift-yes and shift-no, respectively) (Table 1). Patients with virus populations that subsequently shifted to wild type appeared to have lower viral load values at the beginning of treatment interruption than patients in the shift-no group, but this difference was not significant. However, exact quantification for samples with high viral load was not possible since prior dilutions had not been carried out. A significant difference in terms of CD4 cell count at the time of interrupting treatment was noted. Patients in the shift-yes group interrupted treatment with a higher CD4 cell count (median 192 × 106 cell/l) than patients in the shift-no group (median of 59 × 106 cell/l) and this difference was highly significant (P = 0.007). This is in agreement with the findings described above, that a higher CD4 cell count was associated with a shift to wild type in a logistic regression model. At the end of the treatment interruption period, median CD4 cell counts were 49 and 58 × 106 cell/l for the shift-yes and the shift-no populations, respectively (P = 0.21), representing a median fall of 122 × 106 cells/l cells for the shift-yes group and of 25 × 106 cells/l for the shift-no group (P = 0.0006). This latter difference reflects the difference in the potential for CD4 cell count fall in the two groups.

It was of interest that the CD4 cell count in the shift-yes population was significantly higher than in the shift-no population, whereas the viral load levels were not significantly different. For the shift-yes group, there was no correlation between viral load and CD4 cell count at baseline (r = 0.00;P = 0.97) or at the end of treatment interruption (r = 0.05;P = 0.82), whereas baseline virus load and CD4 cell count were negatively correlated for the shift-no group at baseline (r = −0.75;P = 0.008) and at the end of treatment interruption (r = −0.46;P = 0.06)

Back to Top | Article Outline

Response to re-initiation of treatment

Patients re-initiated treatment with a median of five drugs (range one to nine). The most common drugs received were lamivudine (73.5%), didanosine (61.2%), nelfinavir (59.2%), ritonavir (53.1%), stavudine (49%), indinavir (46.9%), abacavir (44.9%) and zidovudine (36.%). Patients were followed for a median of 298 days (range 63–629) after re-starting antiretroviral therapy. The Kaplan–Meier estimate was 67.3% of patients reaching < 500 copies/ml by day 629 (not shown). Factors associated with response (univariate model) were viral load and CD4 cell count at time of interrupting treatment, number of new drugs started (i.e. drugs the patient had not been previously been exposed to), whether the patient experienced a shift to wild type, and the total number of drugs to which the patient's virus population exhibited reduced susceptibility at the time of re-initiating treatment. Factors that were not associated with response were CD4 cell and viral load at time of restarting therapy and previous treatment history. Two multivariate models were used, adjusting for either shift to wild type (model 1) or number of drugs with reduced susceptibility (model 2). As shown in Table 4, higher viral load at time of interrupting treatment and a greater number of new drugs received when restarting treatment remained significantly associated with virologic response in both models, with similar RH values. Patients in the shift-yes group were significantly more likely to respond (RH 5.22;P = 0.006). Those with a greater number of drugs to which the virus displayed reduced susceptibility had a significantly lower chance of responding (RH 0.51;P = 0.01).

Table 4

Table 4

Back to Top | Article Outline


This report provides the first characterization of the consequences of treatment interruptions in extensively pre-treated patients with uncontrolled virus replication and drug-resistant viruses in terms of changes in immunological, virological and resistance parameters as well as response to re-initiation of treatment. The results are based on an observational cohort study and on patients who decided on treatment interruptions for a variety of reasons. Nevertheless, our findings raise important questions regarding treatment strategies for heavily pre-treated patients with drug-resistant HIV. Although resistance was extensive, patients nevertheless appeared to be benefiting from the treatment regimen they were on, as shown by a rise in viral load and fall in CD4 cell count once treatment was interrupted. Therefore, on the one hand, treatment interruptions may pose a significant clinical risk to the patients. On the other hand, in some cases treatment interruptions resulted in a shift to wild-type virus populations. Although these shifts to wild type were associated with improved virologic response to re-initiation of treatment in the short term, the long-term implications are not clear. It is to be expected that the resistant viruses will persist as a minority and may be rapidly re-selected upon re-exposure to treatment. Clearly the assessment of long-term benefit or risk of treatment interruption versus treatment discontinuation in the face of multidrug-resistant viruses needs to be confirmed by randomized trials.

The ‘treatment benefit in spite of failure to suppress viral replication’ phenomenon may be associated with residual antiviral activity or with changes in viral characteristics owing to the developed resistance that are of benefit to the host. The latter is supported by the rapid shift to wild-type phenotype and genotype in a subpopulation (the shift-yes group). The shift to wild type may be associated with changes in viral fitness [7–9]; depending on the combination of mutations present, viruses may have increased or decreased fitness in the presence or absence of drug. In this regard, it will be of interest to determine whether mutations in both protease and reverse transcriptase genes are necessary to drive the shift-to-wild-type effect. Changes in viral characteristics may also be responsible for the differences that we observed with regards to the CD4 cell/HIV-1 RNA correlation in patients in whom the rapid shift to wild type was observed and those in whom it was not observed. Our data are consistent with the notion that the presence of resistance may lead to changes in the HIV-1/CD4 cell interaction in some patients that are of benefit to the host (less CD4 cell destruction); however, the benefit may disappear after continued viral evolution. This is supported by data from Deeks et al., who demonstrated that the ‘shift to wild type’ in patients interrupting treatment was temporarily associated with a fall in CD4 cell count, increase in viral RNA and a change towards better replicative capacity in a small group of patients [25]. Our data may help to explain the ‘disconnect’ that has been described in patients experiencing treatment failure with highly active antiretroviral treatment [3–5,7–9] in that the mechanism responsible for this phenomenon may be associated with differences in the viral characteristics or virus populations that were observed in the two groups of patients we describe. However, the fact that virus load increased and CD4 cell count decreased in both the shift-yes and the shift-no patient groups would speak for an additional role being played by residual antiviral activity. In addition, for a given CD4 cell count and viral load level, there appears to be a lower risk of progression for patients on potent combination treatment than for patients who are not on treatment [26].

Clonal genotypic analysis for virus samples from one patient from four different time points indicated that resistance to protease inhibitors is selected against first upon cessation of treatment and was evident within 30 days of cessation of treatment. Resistance to reverse transcriptase inhibitors was present longer but was lost within 60 days in this individual patient. This implies heterogeneity within the virus population: wild-type virus, viruses resistant to both drug classes and viruses resistant to reverse transcriptase only were each likely to be present at the time of treatment interruption. Mutations appeared to be linked on one genome within each of the resistant populations. The composition may reflect the treatment history: the patient described in Table 3 received reverse transcriptase inhibitors and developed resistance to this drug class prior to protease inhibitor exposure. It is possible that the more rapid disappearance of protease inhibitor resistance compared with reverse transcriptase resistance is a result of increased loss of fitness for the protease inhibitor resistance viruses; however, this speculation requires further investigation.

We identified two variables based on patient characteristics – CD4 cell count and duration of antiretroviral treatment – as predictors for shift to wild type. Although the association was of marginal significance, these findings are in keeping with the notion that a shift to wild type is more likely to occur in patients who have a higher proportion of wild type in their virus population (shorter duration of treatment) and less disease progression (higher CD4 cell count). However, there did not appear to be an association between previous CD4 cell count nadir and shift to wild type (data not shown). Although eradication of the complete ‘wild-type’ population would seem improbable, this possibility has not been tested. Our findings, if substantiated in larger cohorts, may have important clinical implications in terms of how long patients stay on failing regimens.

We have identified a potential benefit of treatment interruptions, in that patients from the shift-yes group appeared to respond better in the short term to re-initiation of treatment. But we stress that until these data are reproduced by randomized studies, the results should be considered as of preliminary nature, and any benefit the patient may experience needs to be weighed against the risk of drastic reductions in CD4 cell count. It was not possible to evaluate consequences of treatment interruptions in terms of clinical progression in this cohort because of its non-randomized nature, and this aspect needs to be evaluated in randomized clinical trials. Similarly, it will be of great importance to assess the effect of treatment interruptions on drug-related toxicities and quality-of-life parameters.

Finally, we draw attention to the implications for resistance testing in clinical practice. In order to capture the extent of resistance development fully for an individual patient, samples need to be drawn at a time when the patient is in fact exposed to drug.

Treatment interruptions occur for various reasons and will continue to occur in HIV-positive patients taking antiretroviral treatments. Although the data to date are not sufficient to make recommendations regarding the usefulness of treatment discontinuation as a treatment strategy, this possibility is being considered more and more frequently among patients with few remaining treatment options.

Salvage therapy for extensively pre-treated patients is a challenging area, and few studies are available to guide clinicians. That response to salvage therapy is strongly linked to viral drug susceptibility is clearly established [27–32]. However, although knowledge of drug susceptibility status may assist in making rational treatment regimen choices, it does not alleviate problems in management of toxicities and intolerances in patients treated over a long period nor will it lessen the challenge of treating patients with high-level resistance to all currently available drugs. Availability of new drug combinations with less within-class cross-resistance will no doubt continue to be a deciding factor in the successful management of these patients.

Back to Top | Article Outline


We gratefully acknowledge the expertise of Brenda Dauer in manuscript preparation and of Beverley Jennings in data management.

Back to Top | Article Outline


1. Guidelines for the use of antiretroviral agents in HIV-infected adults and adolescents.Ann Intern Med 1998, 128 :1079–1100.
2. Carpenter CCJ, Cooper DA, Fischl MA. et al. Antiretroviral therapy in adults: updated recommendations of the InternationalAIDSSociety–USA Panel. JAMA 2000, 283: 381 –390.
3. Kaufmann D, Pantaleo G, Sudre P for the Swiss HIV Cohort Study. CD4 cell count in HIV-1-infected individuals remaining viraemic with highly active antiretroviral therapy (HAART). [Letter] Lancet 1998, 351 :723–724.
4. Deeks S, Barbour J, Martin JN, Swanson MS, Grant RM. Sustained CD4 T cell response after virologic failure of protease inhibitor based regimens in patients with human immunodeficiency virus infection. J Infect Dis 2000, 181: 946 –953.
5. Miller V, Sabin C, Rottmann C. et al. Prognostic value of virus load levels in patients receiving HAART: the Frankfurt HIV-Cohort. AIDS 1998, 12 (Suppl. 4) : 14 –68.
6. Deeks SG, Barbour JD, Martin JN, Grant RM. Delayed immunologic deterioration among patients who virologically fail protease inhibitor-based therapy.Seventh Conference on Retroviruses and Opportunistic Infections. Chicago, January 2000 [abstract 236].
7. de la Carrière LC, Paulous S, Clavel F, Mammano F. Effects of human immunodeficiency virus type 1 resistance to protease inhibitors on reverse transcriptase processing, activity, and drug sensitivity. J Virol 1999, 73: 3455 –3459.
8. Martinez-Picado J, Savara AV, Sutton L, D'Aquila RT. Replicative fitness of protease inhibitor-resistant mutants of human immunodeficiency virus type 1. J Virol 1999, 73: 3744 –3752.
9. Nijhuis M, Schuurman R, de Jong D. et al. Increased fitness of drug resistant HIV-1 protease as a result of acquisition of compensatory mutations during suboptimal therapy. AIDS 1999, 13: 2349 –2359.
10. Rusconi S, de Pasquale MP, Milazzo L. et al. Loss of lamivudine resistance in a zidovudine and lamivudine dual-resistant HIV-1 isolate after discontinuation ofin vitrolamivudine drug pressure. Antiviral Ther 1998, 3: 203 –207.
11. Condra JH, Holder DJ, Graham DJ. et al. Genotypic or phenotypic susceptibility testing may not predict clinical responses to indinavir. Antiviral Ther 1997, 2: 31 –32.
12. Lange JMA, Balzarini J, Harrer T. et al.(eds.) The m184v Mutation in HIV Therapy: Impact on Clinical Care. New York: World Health Communications; 2000.
13. Verhofstede C, van Wanzeele F, van der Gucht B, de Cabooter N, Plum J. Interruption of reverse transcriptase inhibitors or a switch from reverse transcriptase to protease inhibitors resulted in a fast reappearance of virus strains with a reverse transcriptase inhibitor-sensitive genotype. AIDS 1999, 13: 2541 –2546.
14. Miller V, de-Bèthune M-P, Kober A. et al. Patterns of resistance and cross-resistance to HIV-1 reverse transcriptase inhibitors in patients treated with the non-nucleoside reverse transcriptase inhibitor loviride. Antimicrob Agents Chemother 1998, 42: 3123 –3129.
15. Miller V, Gute P, Hertogs K. et al. Baseline resistance and virological response to Mega-HAART salvage therapies. AIDS 1998, 12 (Suppl 4): 18. 18.
16. Devereux HL, Youle M, Johnson MA, Loveday C. Rapid decline in detectability of HIV-1 drug resistance mutations after stopping therapy. AIDS 1999, 13: F123 –F127.
17. Martìnez-Picado J, de Pasquale MP, Ruiz L. et al. Clonal sequencing for detection of resistance mutations and their linkage. Antiviral Ther 1999, 4 (Suppl 1): 50. 50.
18. Miller V, Staszewski S, Sabin C. et al. CD4 lymphocyte count as a predictor of the duration of HAART-induced suppression of HIV-1 virus load. J Infect Dis 1999, 180: 530 –533.
19. Staszewski S, Miller V, Sabin C. et al. Virological response to protease inhibitor therapy in an HIV clinical cohort. AIDS 1999, 13: 367 –373.
20. Hertogs K, de Béthune M-P, Miller V. et al. A rapid method for simultaneous detection of phenotpyic resistance to inhibitors of protease and reverse transcriptase in recombinant HIV-1 isolates of patients treated with antiretroviral drugs (PR-RT Antivirogram). Antimicrob Agents Chemoth 1998, 42: 269 –276.
21. Larder BA, Bloor S, Kemp SD. et al. A family of insertion mutations between codons 67 and 70 of human immunodeficiency virus type 1 reverse transcriptase confer multi-nucleoside analog resistance. Antimicrob Agents Chemother 1999, 43: 1961 –1967.
22. Martinez-Picado J, Sutton L, de Pasquala MP, Savara AV, D'Aquila RT. Human immunodeficiency virus type 1 cloning vectors for antiretroviral resistance testing. J Clin Microbiol 1999, 37: 2943 –2951.
23. Rashtchian A, Thorton CG, Heidecker G. A novel method for site-directed mutagenesis using PCR and uracil DANN glycosylase. PCR Meth Appl 1999, 2: 124. 124.
24. Harrigan PR, Stone C, Griffin P et al.Resistance profile of HIV-1 RT inhibitor abacavir (ABC, 1592U89) after monotherapy and combination therapy.J Infect Dis, 2000, in press.
25. Deeks SG, Wrin T, Hoh R, et al. Virologic and immunologic evaluation of structured treatment interruptions (STI) in patients experiencing long-term virologic failure.Seventh Conference on Retroviruses and Opportunistic Infections. Chicago, January 2000 [abstract LB10].
26. Miller V, Mocroft A, Clotet B for the EuroSIDA Cohort. Association of viral load, CD4 cell count, and treatment with clinical progression in HIV patients with very low CD4 cell counts: The EuroSIDA Cohort.Seventh Conference on Retroviruses and Opportunistic Infections. Chicago, January 2000 [abstract 454].
27. Piketty C, Race E, Castiel P. et al. Efficacy of a five-drug combination including ritonavir, saquinavir and efavirenz in patients who failed on a conventional triple-drug regimen: phenotypic resistance to protease inhibitors predicts outcome of therapy. AIDS 1999, 13: F71 –F77.
28. Deeks S, Hellman NS, Grant RM. et al. Novel four-drug salvage treatment regimens after failure of a human immunodeficiency 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.
29. Hammer S, Demeter L, de Gruttola V. et al. Relationship of phenotypic and genotypic resistance profiles to virological outcome in a trial of abacavir, nelfinavir, efavirenz and adefovir dipivoxil in patients with virological failure receiving indinavir (ACTG 372). Antiviral Ther 1999, 4 (Suppl 1) : 45. 45.
30. Harrigan PR, Hertogs K, Verbiest W. et al. Baseline HIV drug resistance profile predicts response to ritonavir–saquinavir protease inhibitor therapy in a community setting. AIDS 1999, 13: 1863 –1871.
31. Harrigan PR, Raboud J, Hertogs K. et al. Drug resistance and short term virological response in patients prescribed multidrug rescue therapy. Antiviral Ther 1999, 4 (Suppl 1) : 43. 43.
32. Miller V, Cozzi-Lepri A, Hertogs K. et al. HIV drug susceptibility and treatment response to mega-HAART regimen in patients from the Frankfurt HIV Cohort. Antiviral Ther 2000, 5: 49 –55.

Treatment interruption; resistance; shift to wild type; salvage therapy

© 2000 Lippincott Williams & Wilkins, Inc.