| |
HIV Drug resistance Predicts AIDS & Death
|
| |
| |
"Multiple drug class-wide resistance associated with poorer survival after treatment failure in a cohort of HIV-infected patients"
AIDS: Volume 19(10) 1 July 2005
Zaccarelli, Mauroa; Tozzi, Valerioa; Lorenzini, Patriziaa; Trotta, Maria Pa; Forbici, Federicab; Visco-Comandini, Ubaldoa; Gori, Caterinab; Narciso, Pasqualea; Perno, Carlo Fb; Antinori, Andreaa; for the Collaborative Group for Clinical Use of HIV Genotype Resistance Test (GRT) at National Institute for Infectious Diseases 'Lazzaro Spallanzani'
From the aClinical Department
bLaboratory of Antiviral Drug Monitoring, National Institute of Infectious Diseases 'Lazzaro Spallanzani', IRCCS, Rome, Italy.
".....The results described here, if confirmed in other sets of patients and in evaluations of different patterns of drug class resistance, may have implications in the management of HIV-infected patients treated with antiretroviral drugs. Therefore, GRT may also have a role in the evaluation of a possible increased clinical progression risk, particularly in heavily pretreated patients. Though the interval between the occurrence of drug resistance and death may be long, both clinicians and patients should be aware that the onset of a wide pattern of resistance could represent a strong negative prognostic factor for survival. Finally, our results may have implications in transmission of drug-resistant viruses. It should become a priority to prevent the development of resistance by the correct use of resistance tests, by drug sequencing in regimens and by patient education regarding the risks of developing resistance if they have poor adherence to therapy...."
ABSTRACT
Objective: To evaluate the effect of drug class-wide resistance (CWR) on survival in HIV-infected individuals who underwent genotypic resistance test after antiretroviral failure.
Design: Observational, longitudinal cohort study.
Methods: HIV-infected individuals experiencing treatment failure were enrolled at first genotypic resistance test. End-points were death for any cause, AIDS-related death and AIDS-defining event/death. CWR was defined according to the International AIDS Society consensus. Survival analysis was performed with Cox's model.
Results: Among 623 patients enrolled and followed for a median of 19 months (interquartile range, 12-29), Kaplan-Meier analyses for end-points at 48 months in patients with no CWR, one CWR, two CWR or three CWR were 8.9, 11.7, 13.4 and 27.1%, respectively, for death; 6.1, 9.9, 13.4 and 21.5%, respectively, for AIDS-related death; and 16.0, 17.7, 19.3 and 35.9%, respectively, for new AIDS event/death.
In a multivariate Cox's model, higher HIV RNA level, previous AIDS and detection of three CWR (hazard ratio, 5.34; 95% confidence interval, 1.76-16.24) were all significantly associated with increased risk of death, while higher CD4 cell count and use of a new boosted protease inhibitor drug after identifying genotypic resistance were associated with reduced risk.
Detection of three CWR was also significantly associated with higher risk of AIDS-related death and new AIDS event/death.
Conclusions: Even in the late era of highly effective antiretroviral treatments, detection of CWR, particularly if extended to all three drug classes is related to poorer clinical outcome and represents a risk-marker of disease progression and death.
INTRODUCTION
In the second half of 1990s, the availability of combination treatments [highly active antiretroviral therapy (HAART)] and new antiretroviral drug classes had a dramatic effect on survival and risk of AIDS-related illnesses [1-3]. Both the mortality rate and the number of AIDS-defining events progressively fell according to the use of HAART. The natural history of HIV infection changed to an 'unnatural' history, conditioned by success and/or failure of treatment.
Several clinical-based cohort studies have shown that 30-60% of patients have evidence of virological failure during their antiretroviral treatment [4-7]. In patients starting HAART, both current CD4 cell count and viral load independently predicted disease progression [8]. An increase in CD4 cell count can be observed even during virological failure [9], which suggests that immunological decline in such patients may be slower than in patients who discontinue treatment [10]. Therefore, despite achievements in increasing the interval to clinical progression, virological failure remains associated with a greater risk of disease progression and death [8,11]. In patients who have experienced a virological failure to all three antiretroviral drug classes, latest CD4 cell count remains the main death predictor, even though viral load appeared to be a major determinant of the change in CD4 cell counts [12].
The emergence of drug-resistant HIV variants is one of the major causes of treatment failure [13-16]. Particularly, cross-resistance (defined as resistance to drugs to which a virus has never been exposed) within a given class of antiretroviral agents has been found to affect all three classes of drug currently available [17,18]. Compared with single-drug resistance, this cross-resistance may result in class-wide resistance (CWR), which could substantially reduce the clinical utility of antiretroviral drugs of the same class and reduce future treatment options [19].
Resistance testing has been show to be clinically usefulness in controlled studies [20-23] and has become the standard of care to guide response to treatment failure [24-28]. However, the potential role of results from genotypic resistance testing (GRT), and particularly of HIV-1 variants harbouring CWR to different antiretroviral classes, has not yet been evaluated as a prognostic factor in clinical progression.
For these reasons, the present study examined disease progression and survival of a cohort of HIV-1-infected patients who underwent GRT at the time of antiretroviral treatment failure. The main objective was to assess the extent to which CWR to different antiretroviral drug classes might affect clinical progression among HIV-infected persons failing antiretroviral therapy.
RESULTS
Overall, 623 patients failing under HAART and undergoing GRT for the first time were included in the study. Patient characteristics are shown in Table 1 stratified by the number of CWR detected. Patients had generally had substantial previous treatment, as confirmed by the high number of pre-GRT therapies (considered as all previous treatment regimens, including changes for intolerance) and of pre-GRT failures (considered as changes for virological failure) and by the long duration of pre-GRT HAART. Approximately half the patients (50.9%) had at least one CWR at GRT, and 24 patients (3.9%) were resistant to all three classes. The clinical status (AIDS diagnosis, CD4 cell count, HIV RNA) of patients was progressively advanced in line with the number of CWR.
The overall median duration of follow-up after GRT was 19 months [interquartile range (IQR), 12-29]; in patients with none, one, two or three CWR it was 19 months (IQR, 12-28), 19 months (IQR, 13-30) and 21 months (IQR, 7-37), respectively. Over 48 months of observation from GRT, 38 deaths, of which 29 were AIDS related, and 28 new AIDS events occurred.
The overall incidences of death by any cause, AIDS-related death and new AIDS event/death were 3.66, 2.79 and 6.59 per 100 persons/years, respectively. The incidence of death in patients with none, one, two or three CWR was 3.50, 2.79, 3.97 and 11.90 per 100 persons/years follow-up, respectively. Similarly, the incidence of AIDS-related death was 2.47, 1.95, 3.98 and 9.40 per 100 persons/years, respectively. At Kaplan-Meier estimate, the overall 48-month proportion of patients progressing to death, AIDS-related death or new AIDS event/death was 11.1%, 8.9% and 17.5%, respectively.
Among patients with a single CWR, the 48-month proportion of patients progressing to death was 18.1%, 12.1% and 9.7% in patients with the CWR for NRTI, NNRTI or PI drugs, respectively. Only a CWR for NRTI drugs was significantly correlated to higher risk of death (P = 0.019, log rank test).
The 48-month percentages for patients with none, one, two or three CWR were 8.9, 11.7, 13.4 and 27.1%, respectively, for progressing to death (P = 0.0286, log rank test); 6.1, 9.9, 13.4 and 21.5%, respectively, for progressing to AIDS-related death (P = 0.0299, log rank test); and 16.0, 17.7, 19.3 and 35.9%, respectively, for progressing to a new AIDS event/death (P = 0.0115, log rank test).
At Cox's model, the detection of three CWR was significantly associated with increased risk of death, AIDS-related death and AIDS event/death at univariate analysis. Other factors associated with increased crude hazard ratios for all end-points were current higher HIV RNA level and pre-GRT AIDS diagnosis; higher CD4 cell count at GRT significantly reduced the risk. Moreover, the use of a new boosted PI after GRT, though not significantly associated at univariate analysis, showed a protective role over survival if included in a bivariate model together with three CWR. Consequently, all these factors were included in a multivariate Cox's proportional hazard model together to the correction variables, as indicated in the Methods.
Category of exposure for HIV infection, number of pre-GRT drugs used, number of pre-GRT treatments and duration of pre-GRT treatments (considering both HAART and mono/dual pre-HAART therapies) did not show statistically significant association with death in univariate or bivariate analysis and were not included in the Cox model. Moreover, all new drugs used during follow-up after GRT were tested to assess their effect on survival. No drug added after GRT, other than a new boosted PI added after GRT, showed a significant effect in either the crude analysis or the bivariate analysis on survival with three CWR (data not shown), so these were not included in the Cox model. No patient was treated with pentafuside (T20) before or after GRT in this set.
After adjusting in the multivariate model, detection of three CWR, higher HIV RNA and pre-GRT AIDS diagnosis adversely affected survival when considering death for any causes or AIDS related death, while higher CD4 cell count and use of new boosted PI after GRT were associated with a reduced risk of death (Table 2). Comparable results were observed if a new AIDS event/death was the end-point, but detection of three CWR was associated with an even greater risk. Higher number of pre-GRT treatment failures was associated with better survival, while the protective effect of adding a new boosted PI was not observed at this end-point.
Among those with only a single CWR, only resistance to NRTI was found to be significantly associated with higher risk of death at multivariate Cox's model adjusted by the same set of covariates (data not shown in tables).
Author Discussion
The detection of any CWR was relatively frequent in our sample and present in at least half of the included patients. In contrast, the presence of all three CWR was rare and decreased proportionally over time, probably because there was always a higher proportion of patients at the point of early failure by GRT. Despite the infrequency of three CWR, the study showed that its presence was a strong predictor of death in multivariate analysis controlling for CD4 cell count, plasma HIV RNA level, clinical stage, gender, age, calendar year of GRT and drug exposure.
Drug-resistant HIV-1 has become a major concern in antiretroviral therapy because of its clinical and public-health impact. Among one large population sample of HIV-infected adults experiencing virological failure, prevalence of any genotypic resistance to three drug classes was estimated as 13%, with resistance to two classes as 48% and any resistance as 76% [13]. In another observation, prevalence of extended resistance to three drug classes was estimated as ranging from 4 to 15%, with any single CWR ranging from 19 to 77%, according to the method of analysis [28]. This intrinsic variability may be further influenced by high variation in the prediction of virological profiles using different systems to interpretation detected genotypic resistance [29]. To minimize this effect, the current study used a consensus-based standardized definition for a high degree of resistance rather than a rules-based algorithm or any other expert system.
Assuming that most drug failures within any class were the result of developing resistance, our observations estimated the risk of disease progression related to increased acquired resistance at drug failure. This occurred independently of CD4 cell count changes during therapy, although the latter is considered the most powerful predictor of death in treated patients so far [8,12]. As viral load remains a major risk factor for maintaining CD4 cell count in patients with multiple failed regimens [12], and assuming that, at present, achieving viral suppression after acquiring several CWR is generally unlikely, our data give a possible explanation of the clinical effects of virological failure among patients at advanced stages of treatment [30].
Two recent studies failed to demonstrate a correlation of drug-resistant HIV-1 with disease progression or death [31,32]. In the cohort in British Columbia, three CWR was observed in 11% of living individuals but only in 5% of those who had died. Based on this observation and on a previous report [33], the authors concluded that non-adherence and lack of access to care rather than drug resistance were the main driving factors leading to death [31]. In the Johns Hopkins HIV clinical database, the extent of antiretroviral resistance in patients treated with GRT-guided therapy was not related to disease progression or death [32]. Two explanations were put forward for these results: very low adherence levels in patients who had therapy failure with little or no drug resistance [34] and reduced 'fitness' of HIV-1-resistant isolates compared with the wild-type virus. In contrast, in our cohort, the specific design based on CWR as the predictive baseline factor for disease progression could have allowed a more sensitive analysis than achievable with clinical end-points. To our knowledge, an association between high degree of CWR and poorer survival has never been reported in the literature. This finding reinforces the concept that ongoing viral replications and few residual treatment options are associated with an increased probability of disease progression [35].
It has been reported that mutations conferring resistance may also impair viral replication and fitness, by modifications of key viral proteins. The association between PI resistance and HIV replication capacity is still not understood, but there are reports of an association between resistance and reduced viral fitness in PI-exposed patients [36]. Nonetheless, continuing long-term viral replication during therapy can ultimately lead to progressive accumulation of mutations increasing viral fitness [37,38]. It is noteworthy that the reduced pathogenicity of drug-resistant HIV can reasonably explain the prolonged interval from virological failure to death as well as the lack of clinical progression among patients harbouring resistant virus in time-limited prospective observations [32]. In our cohort, an overall 4-year follow-up with a median of 19 months offers a reasonable time for observation of long-term effects that may not be detectable with shorter periods of observations.
In our study, different hazard ratios were found for resistances to different drug classes. The major contribution to the effect of drug resistance on disease progression was from NRTI, with some effect of NNRTI drugs and less-significant effects for PI drugs. This probably reflects the non-linear relationship between PI-associated resistance mutations and virological outcome, allowing remarkable chances of achieving virological success even in the presence of a high number of certain mutations [39]. Another possible explanation concerns the use of boosted PI as a component of the salvage regimen. Boosted PI regimens have been associated with higher efficacy against resistant HIV strains [40], lower rates of resistance at high-adherence level than with a single PI [41] and a lack of PI-associated resistance mutations after virological failure in treatment-naive patients [42]. A new boosted PI regimen of lopinavir/ritinovir was used in approximately half the patients with a PI CWR in our cohort; it showed a marginal independent protective effect on disease progression, which agrees with other reported data [43]. However, there is still controversy over how to define wide PI class resistance and different criteria have been used in different published studies [44], particularly in order to evaluate the efficacy of new PI drugs [45].
Our study may have some limitations. At first, data were obtained from a single centre and so, even if they could be representative of a large urban HIV-infected treated population, may not be simply generalized to other settings. The practice of prescription based on GRT in our centre, even if in agreement with recent guideline on the use of GRT, could have influenced our results, making them not easily extendable to other settings with different GRT practice. Moreover, patients underwent GRT at different failure points: some patients at the first one, other patients after failing multiple regimens. This potential bias was partially corrected using several correction variables in the multivariate analysis.
A second limitation was possibly related to the fact that only those with documented CWR at the first genotyping were included in the dataset. As the rate of acquiring new drug resistance mutations increases if a failing regimen is continued [35], it could be hypothesized that CWR would be more likely to develop under progressive drug pressure in patients who had experienced successive drug failures after the first GRT, and even in patients with failing therapy in whom a genotype test has not been requested. Therefore, the prevalence of CWR in our cohort could be an underestimation, and a potential bias for survival analysis cannot be excluded. However, development of new drug resistance mutations has been shown to be inversely proportional to the number of mutations in the first genotype sequence [32], and this should minimized risks related to a lack of evaluation of CWR as a time-dependent covariate.
A third limitation is that a validated measure of adherence was not included in our study. Poor adherence is associated with an enhanced probability of virological rebound [46], an increased risk of death [33] and a increased probability (not-linear) of selecting drug-resistant viral mutants [34,40]. It is also conceivable that CWR would enhance low adherence by its accompanying higher pill burden and the more severe toxicity of salvage regimens.
A final limitation may be that our analysis was based on cross-resistance to currently used PI, NRTI and NNRTI drugs; as new drugs and drug classes are in development [47], the clinical scenario could change with the availability of more potent antiretroviral dugs that are active against resistant strains and/or affect different stages of virus replication.
The results described here, if confirmed in other sets of patients and in evaluations of different patterns of drug class resistance, may have implications in the management of HIV-infected patients treated with antiretroviral drugs. Therefore, GRT may also have a role in the evaluation of a possible increased clinical progression risk, particularly in heavily pretreated patients. Though the interval between the occurrence of drug resistance and death may be long, both clinicians and patients should be aware that the onset of a wide pattern of resistance could represent a strong negative prognostic factor for survival. Finally, our results may have implications in transmission of drug-resistant viruses. It should become a priority to prevent the development of resistance by the correct use of resistance tests, by drug sequencing in regimens and by patient education regarding the risks of developing resistance if they have poor adherence to therapy.
Methods
Study design and population
This prospective cohort study analysed the database of all patients who underwent GRT in the National Institute for Infectious Diseases 'Lazzaro Spallanzani' after HAART failure between June 1999 (when routine GRT test at HAART failure in our Institute was started) and June 2002. All subjects routinely underwent GRT for clinical purposes if they had a viral load rebound to >500 copies/ml or failed to achieve a viral load below the detection limit (< 80 copies/ml) on the current HAART regimen.
At the time of GRT, a complete retrospective clinical history was collected, including all laboratory tests, HIV/AIDS-related clinical events, treatment history as well as demographic and behavioural characteristics. After GRT, patients were prospectively evaluated in order to assess virological and immunological response to treatment, onset of new AIDS-related events and death. The follow-up was carried out up to December 2003.
Genotypic resistance testing
Direct sequencing of the genes for reverse transcriptase and protease was performed using the Viroseq HIV-1 genotyping kit, versions 1 and 2 and the 3100 ABI sequencer (Applied Biosystems, Foster City, California, USA). Briefly, RNA was extracted from plasma samples, reverse transcribed and amplified with specific primers using the polymerase chain reaction. The products covered the whole pol region, coding for all amino acids of the protease, and the first 320 amino acid residues of reverse transcriptase, which is the area where the majority (and relevant) of mutations conferring resistance to antiretroviral drugs are found.
Class-wide resistance assessment
The aim of the analysis was to evaluate if a high degree of resistance to antiretroviral drugs can influence survival and so a standardized assessment of extended class resistance was essential. The International AIDS Society-USA guidelines were adapted to identify single mutations or sets of mutations conferring high levels of resistance to each class (CWR) [27].
CWR to nucleoside reverse transcriptase inhibitors (NRTI) was considered if at least one of the following mutations or patterns of mutations was detected: Q151M mutation or at least three mutations belonging to the Q151M-complex (among A62V, V75I, F77L, F116Y); 69 insertion; at least four mutations from the nucleoside analogue mutation group (among M41L, E44D, D67N, K70R, V118I, L210W, T215Y/F, K219Q/E).
CWR to non-nucleoside reverse transcriptase inhibitors (NNRTI) was considered if at least one of the following mutations or patterns of mutations was detected: K103N; Y188L; or at least two mutations among L100I, V106A, Y181C/I, G190S/A, M230L.
CWR resistance to protease inhibitors (PI) was considered if the following pattern of mutations was detected: at least four mutations among L10F/I/R/V, V32I, M46I/L, I54L/M/V, V82A/F/T/S, I84V/A/C, L90M.
Antiretroviral treatment and clinical course
The GRT-guided therapy was decided by the reference physician for each patient on the basis of indications for active drugs. The most controversial GRT tests were discussed weekly in a setting including both physicians and virologists, and the best treatment for those patients was usually decided based also on behaviour and adherence characteristics of the particular patient. The subsequent clinical course, including the nature and timing of laboratory testing, was entirely under the responsibility of the treating physician.
Routine follow-up programme
After GRT, patients were routinely evaluated every 1-2 months with clinical examinations and blood tests including CD4 cell count and HIV-1 RNA determination. Changes of therapy during follow-up were recorded as well as clinical events.
Data analysis and statistical methods
Death for any cause was considered as the primary end-point of the study. AIDS related death, considered as death by AIDS-defining event, and occurrence of new AIDS events/death were also considered as end-points. The occurrence of AIDS events and deaths was checked at the Regional AIDS Registry. However, patients who did not contact any of the Institute reference points (i.e., hospital wards or outpatient clinics) in the last 6 months were considered lost in follow-up and censored at the last visit. To avoid bias, if patients died outside our clinic later than 6 months, they were still considered lost in follow-up.
The Kaplan-Meier method was used to estimate the probability of progression to end-points and the log rank test was used to assess statistical differences between two groups. A multivariate analysis was performed using Cox's proportional hazard model, where factors found significantly associated at univariate, or bivariate analysis in association with CWR were included. In addition, other variables, such as gender, age at GRT, calendar year at GRT and number of pre-GRT treatment failures (considering as any therapy owing to virological rebound) were included in multivariate model as correction variables. The effect of each new drug added after GRT was tested to evaluate its effect over survival. Statistical analyses were performed using SPSS version 11.0 (SPSS, Chicago, Illinois, USA).
|
|
| |
| |
|
|
|
