Long-term outcomes of highly experienced people with HIV undergoing salvage therapy with raltegravir

Abstract

Raltegravir and other third-line drugs have shown promise in improving outcomes in treatment-experienced patients. However, the efficacy and tolerability of these agents vary. This study assessed real-life virologic success, long-term survival, and adverse events in patients receiving raltegravir or other third-line drugs as salvage regimens. This retrospective cohort study included adults who experienced treatment failure (human immunodeficiency syndrome-1 RNA plasma viral load >1000 copies/mL) and subsequently initiated raltegravir or other third-line drugs (darunavir/ritonavir, maraviroc, or etravirine). Propensity score matching methods were employed to account for differences at the time of switching from failing antiretroviral therapy regimens. The matched subset was analyzed using the Kaplan–Meier method and Generalized Wilcoxon tests to evaluate the probability of achieving virologic suppression (plasma viral load <50 copies/mL). Mortality rates, toxicity, treatment interruption, virologic failure, and loss to follow-up were determined using Poisson regression. One hundred and sixty-eight patients initiating salvage regimens were included, with 123 receiving raltegravir and 45 other third-line drugs. Propensity score matching resulted in a subset of 90 patients, 45 in each group. During the follow-up period, there were no significant differences observed between the groups in terms of virologic suppression (77.8% vs 82.2%, P = .73), mortality rates (4.04 vs 6.18 persons per 100 person-years [p-y]; P = .67), drug toxicity (0.00 vs 2.06 persons per 100 p-y; P = .49), treatment interruption (8.07 vs 0.00 persons per 100 p-y; P = .06), virologic failure (2.02 vs 4.12 persons per 100 p-y; P = .61), and loss of follow-up (6.05 vs 4.12 persons per 100 p-y; P = .70). Our findings indicate comparable survival and virological success rates between raltegravir and other drugs used in salvage regimens. Similar rates of drug toxicity, treatment interruption, virologic failure, and loss of follow-up were also observed. These results suggest that raltegravir may be a viable option for salvage therapy, demonstrating outcomes comparable to other third-line drugs in real life.

1. Introduction

Antiretroviral therapy (ART) is the established standard of care for human immunodeficiency syndrome (HIV) infection,^[1–3] offering improved survival and reduced risk of acquired immunodeficiency syndrome (AIDS)-related illnesses.^[4–7] However, a significant proportion of patients (approximately 15%) receiving ART fail to achieve virologic suppression,^[8–10] leading to treatment failure and the development of resistance,^[11] which limits further therapeutic options.^[12] Treatment success in heavily treatment-experienced patients depends on various factors, including comorbidities, tolerability, adherence, and cross-resistance within antiretrovirals (ARVs).^[13–19] The introduction of new classes of ARVs, such as fusion inhibitors, new boosted protease inhibitors, CCR5 antagonists, and integrase inhibitors, have considerably improved treatment outcomes in treatment-experienced patients.^[20–24]

However, these agents differ in tolerability, toxicity, and efficacy in maintaining virological suppression. Real-life data has helped to understand the complexity and challenges associated with the use of new ARVs classes to construct appropriate “salvage” regimens for heavily experienced patients.^[25–28] In Brazil, the Brazilian Unified Health System (Sistema Único de Saúde) and the National AIDS Program provide free access to ART and establish guidelines for its use.^[3] Raltegravir is an option for salvage regimens in cases of drug-resistant HIV infections.^[1,3] As the first commercially available integrase inhibitor, raltegravir became an option after controlled trials demonstrated its efficacy, tolerability, and safety profile in treatment-experienced patients under optimal conditions.^[23,29–32] However, raltegravir has a low genetic barrier to resistance,^[33] and treatment interruptions could potentially affect its efficacy, particularly in resource-limited settings.^[34,35]

The primary objective of this study was to evaluate real-life virologic success, long-term survival, and adverse events in treatment-experienced patients receiving either raltegravir or other third-line drugs as salvage regimens.

2. Methods

2.1. Patients and setting

The Complexo Hospitalar Prof Edgard Santos (COM-HUPES), located at the federal university of Bahia in Salvador, Brazil, is the second-largest referral center specializing in the treatment of HIV. This retrospective study focused on patients who received third-line drugs between March 2008 and December 2016 within an open cohort of HIV-1 patients who had prior treatment experience and were under the care of the COM-HUPES. The inclusion criteria were adult patients (>18 years old) who exhibited virologic failure, defined as 2 consecutive HIV RNA tests with a viral load exceeding 1000 copies/mL after at least 6 months of continuous therapy. These individuals initiated salvage regimens following genotypic antiretroviral resistance testing (GRT), indicating the need for such treatment. Salvage regimens were defined as those containing drugs classified as third-line ARV according to Brazilian guidelines, which include raltegravir, darunavir/ritonavir, maraviroc, etravirine, and enfuvirtide.^[3] Prescriptions for salvage regimens were determined based on various factors, including genotypic resistance testing, treatment history, toxicity or allergy to ARVs, and national recommendations.^[3] Subsequently, infectious disease specialists conducted regular patient assessments every 3 to 6 months and recorded relevant clinical data.

2.2. Variables

Most of the clinical information used in this study was collected from medical records. Demographic data encompassed enrollment characteristics such as sex, age, and race (categorized as black and non-black). Clinical variables included alcohol use, cigarette smoking, illicit drug use, hemoglobin level, body mass index, duration of ART, number of treatment failures, previous use of ARVs, and diagnosed comorbidities or coinfections. Information concerning GRT, including HIV-1 RNA plasma viral load (PVL), CD4+/CD8 + cell count, and patient ART management details, was manually extracted from 2 national databases maintained by the Ministry of Health: the system for control of laboratory tests (Sistema de Controle de Exames Laboratoriais, which monitors surrogate markers of HIV in laboratory settings, and the System for Control of Drug Logistics (Sistema de Controle Logístico de Medicamentos - SICLOM), which ensures traceability of all prescribed ARVs, including initiation dates, refill dates, and ART adjustments. For PVL quantification, plasma specimens were analyzed using the Abbott RealTime HIV-1 Assay (Abbott Laboratories, Chicago) (m2000 system) with automated RNA extraction and reverse transcription Polymerase Chain Reaction (RT-PCR) targeting a sequence within the HIV-1 pol integrase region, performed on the real-time amplification platform m2000rt. Counting CD4+/CD8 + lymphocytes was obtained by flow cytometry using FACSVia flow cytometer (BD Biosciences, San Jose). GRT was performed using the Sanger sequencing techniques and the Applied Biosystems ViroSeq HIV-1 Genotyping System (Foster City, Calif.). The genotypic sensitivity score (GSS) was obtained using the Stanford HIVdb genotypic resistance interpretation system.^[36] It was calculated as the sum of individual scores (1.0 for fully active, 0.5 for partial, and 0.0 for inactive) for each drug employed in the new ART regimen. General mortality information, including the dates and causes of death, was manually extracted from the national mortality information system (Sistema de Informações de Mortalidade). This high-quality database has centrally registered and stored death certificates from across the country since 1979 in compliance with Brazilian legislation.^[37]

This study evaluated various endpoints during the follow-up period. Virological suppression, defined explicitly as the first occurrence of a PVL test indicating <50 copies/mL, was assessed 1 month after the initiation of the salvage regimen and subsequently monitored every 3 to 6 months. Information about adverse events resulting from ART toxicity, as documented by the infectious disease specialist at the time of the event, was extracted from medical records and classified according to the Table for Grading the Severity of Adult and Pediatric Adverse Events.^[38] Loss of follow-up was defined as any patient enrolled in the study who was not reported deceased in the national mortality information system, did not seek medical attention at the COM-HUPES clinic, had no contact with the healthcare service, and did not regularly refill their ART prescription over the past 6 months, as indicated by SICLOM reports. Treatment interruption was defined as a situation in which a patient ceased filling their ART prescription for >3 months without a known reason, based on information obtained from SICLOM reports. Virologic failure was determined by 2 consecutive HIV RNA tests with viral loads exceeding 1000 copies/mL observed after a minimum of 6 months of continuous therapy. Patients whose viral loads ranged from 51 to 1000 copies/mL were not categorized as experiencing virologic failure, and it was not feasible to conduct a resistance analysis on them in accordance with the existing clinical guidelines in Brazil.^[39]

2.3. Propensity score analysis

To mitigate potential indication bias and ensure comparability between the 2 nonrandomized treatment groups, we employed propensity score (PS) matching.^[40,41] PS was calculated using a multivariate logistic regression model, which estimated the likelihood of patients being prescribed raltegravir based on their clinical characteristics. Our final PS model incorporated the following binary variables: HIV-1 RNA PVL, CD4 cell count below 100 cells/mL, and GSS <3. Certain baseline variables, such as age, episodes of previous treatment failure, malnutrition, and anemia, were not included in the PS model because they did not contribute to the model fit. Patients with similar PS values were matched 1:1 using the nearest-neighbor algorithm, pairing those receiving raltegravir with those receiving other drugs. Non-matched patients were excluded from further analysis.

2.4. Statistical analysis

The characteristics of the participants are presented as proportions for categorical variables and means (standard deviations) for continuous variables. A comparison was made between participants receiving raltegravir and those receiving “other drugs” at the time of treatment initiation, both before and after matching, using Student t test, Pearson chi-square test, or Fisher exact test, as indicated in Table 1. “Other drugs” encompassed therapies involving darunavir/ritonavir, maraviroc, or etravirine. Notably, none of the patients received enfuvirtide or dolutegravir as a salvage therapy during the study period.

Table 1.

Baseline characteristics of people with HIV receiving salvage regimen.

Characteristics	Eligible population (n = 168)			Matched subset (n = 90)
	Raltegravir (n = 123, 73.2%)	Other drugs (n = 45, 26.8%)	P	Raltegravir (n = 45, 50.0%)	Other drugs (n = 45, 50.0%)	P
Age	44.2 (9.7)	46.0 (10.6)	.30‡	45.5 (10.1)	46.0 (10.6)	.82‡
Male sex	78 (63.4)	34 (75.6)	.14*	31 (68.9)	34 (75.6)	.48*
Black race	108 (87.8)	37 (82.2)	.35*	38 (84.4)	37 (82.2)	.78*
Yr receiving ART	10.6 (4.2)	9.7 (3.4)	.16‡	11.1 (4.0)	9.7 (3.4)	.07‡
Previous treatment failures ≥3	37 (30.1)	9 (20.0)	.19*	9 (20.0)	9 (20.0)	1.00*
Alcohol use	58 (47.2)	16 (35.6)	.18*	18 (40.0)	16 (35.6)	.66*
Cigarette smoking	41 (33.3)	13 (28.9)	.59*	16 (35.6)	13 (28.9)	.50*
Illicit drug use	15 (12.2)	4 (8.9)	.55*	4 (8.9)	4 (8.9)	1.00†
PVL (log10 copies/mL)	4.5 (0.8)	4.2 (0.6)	.01‡	4.2 (0.6)	4.2 (0.6)	.57‡
CD4 count (cells/mL)	230.7 (210.3)	342.7 (232.3)	.00‡	286.8 (178.3)	342.7 (232.3)	.20‡
Hemoglobin <9 g/dL	8 (6.5)	0 (0.0)	.11†	1 (2.2)	0 (0.0)	1.00†
Body mass index <18.5 kg/m2	15 (12.2)	2 (4.4)	.25†	2 (4.4)	2 (4.4)	1.00†
Existing comorbidities	58 (47.2)	23 (51.1)	.65*	27 (60.0)	23 (51.1)	.40*
Diabetes mellitus	10 (8.1)	3 (6.7)	.75†	8 (17.8)	3 (6.7)	.11*
Blood hypertension	20 (16.3)	13 (28.9)	.07*	13 (28.9)	13 (28.9)	1.00†
Hyperlipidemia	21 (17.1)	11 (24.4)	.28*	9 (20.0)	11 (24.4)	.61*
Past coinfections	69 (56.1)	27 (60.0)	.65*	20 (44.4)	27 (60.0)	.14*
Pulmonary TB	27 (22.0)	10 (22.2)	.97*	7 (15.6)	10 (22.2)	.42*
CNS toxoplasmosis	26 (21.1)	5 (11.1)	.14*	7 (15.6)	5 (11.1)	.54*
Syphilis	12 (9.8)	5 (11.1)	.78†	4 (8.9)	5 (11.1)	1.00†
Genotypic sensitivity score <3	62 (50.4)	31 (68.9)	.03*	30 (66.7)	31 (68.9)	.82*

Presented values as n (%) or mean (standard deviation).

ART = antiretroviral therapy, HIV = human immunodeficiency syndrome, PVL = plasma viral load.

^*Pearson χ² test.

^†Fisher exact test.

^‡Student’s t test; were used to compare groups.

For the subgroup of PS-matched participants, a time-dependent survival analysis was conducted using the Kaplan–Meier method, and the equality of survival curves was assessed using the Wilcoxon test. The analysis aimed to compare the time to virologic suppression from the date of drug administration to the patient, as reported in SICLOM, or until December 31, 2016, whichever occurred first. Patients who did not achieve virologic suppression at the last medical evaluation were excluded. Cases of loss of follow-up, death, toxicity, or treatment interruption at the time of the previous contact while on salvage regimens were also censored. Cox proportional hazard regression analysis could not be performed to examine the adjusted association between salvage regimens and virologic suppression because of violations of the proportional hazard assumptions.

Incidence rates per 100 person-years were used to describe outcomes such as drug toxicity, treatment interruption, virological failure, loss to follow-up, and death from any cause during the follow-up period. Poisson regression analysis was employed to compare these outcomes between raltegravir and “other drugs” regimens.

This clinical study was approved by the COM-HUPES Ethics Committee (Ethics Committee Resolution No. 2340107; Salvador, October 20th, 2017). Financial support was partially provided by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, Brasil (CAPES), Finance Code 001.

3. Results

Of the 367 people with HIV who initiated salvage regimens during the study period, 168 met the inclusion criteria. Among these patients, 123 received raltegravir, and 45 received other third-line drugs, including darunavir/ritonavir (40 patients), maraviroc (3 patients), and etravirine (2 patients). After PS matching, 90 patients (45 in each group) were included in the analysis. The allocation process for patients is illustrated in Figure 1.

Figure 1.

Flow diagram for study participants.

In the overall study population (n = 168), the 2 groups differed in PVL, CD4 cell count, and drug sensitivity (GSS < 3). However, in the PS-matched subgroups, the patients were comparable (Table 1). In the PS-matched analysis, the rate of virologic suppression was similar between the 2 groups (P = .73; Kaplan–Meier analysis; Generalized Wilcoxon test). Among the patients prescribed raltegravir-based salvage regimens, 77.8% (35 of 45) achieved virologic suppression, compared to 82.2% (37 of 45) of patients prescribed other drugs (Fig. 2).

Figure 2.

Kaplan–Meier survival plot of time to virologic suppression in patients receiving raltegravir or other third-line drugs for salvage regimens.

The incidence of drug toxicity was similar between groups (incidence rate ratio [IRR] = 0.99, 95% confidence interval [CI]: 0.74–1.33). In the group receiving other salvage regimens, 1 case of grade 2 lipohypertrophy (associated with ART: tenofovir, lamivudine, and darunavir/ritonavir) and 1 case of grade 1 depression (not associated with ART) were identified. Among the non-matched raltegravir recipients, 1 case of grade 2 aminotransferase elevation (possibly unrelated to raltegravir) and 1 case of grade 2 pancreatitis (unrelated to raltegravir) were identified. The rates of treatment interruption were also similar between the groups (IRR = 1.05, 95%CI: 0.78–1.41), with 4 episodes identified in raltegravir recipients, none of which were associated with a known cause.

The rates of virological failure were similar in both groups (IRR = 0.99, 95%CI: 0.74–1.33). A total of 5 cases of virologic failure were identified during the follow-up period, with 3 occurring within the matched subset. Among these cases, 1 patient with a history of 3 treatment failures and treatment interruptions, receiving raltegravir, tenofovir, lamivudine, and zidovudine, developed resistance to lamivudine and raltegravir strains with the N155H mutation. Two additional occurrences of the N155H mutation were observed among non-matched raltegravir users who were on regimens involving lamivudine, tenofovir, and lamivudine, tenofovir, zidovudine, respectively, both of which also developed resistance to lamivudine. Furthermore, 2 cases of darunavir resistance were confirmed in patients on lamivudine, tenofovir, and zidovudine regimen; both of whom had experienced 3 treatment failures, and one had previously interrupted treatment. Mortality rates were similar between the groups (IRR = 0.99, 95%CI: 0.74–1.33). Two participants died while receiving raltegravir, with undetectable PVL and a CD4 cell count of >350 cells/mL before death (Table 2). Three other patients died while receiving other drugs due to non-AIDS-related infections; 2 had undetectable PVL and CD4 cell count >350 cells/mL. Four unmatched raltegravir recipients died, none of whom achieved undetectable PVL, and the causes of death were non-AIDS-related infections (2), myocardial infarction (1), and non-AIDS-related cancers (1).

Table 2.

Incidence rates of events during follow-up per 100 person-yr of patients prescribed raltegravir and other third-line drugs for salvage regimens.

Events during follow-up	Eligible population (n = 168; 178.05 p-y)		Matched subset (n = 90; 98.06 p-y)		IRR (95%CI)*; P†
	Raltegravir (n = 123, 129.54 p-y)	Other drugs (n = 45, 48.51 p-y)	Raltegravir (n = 45, 49.55 p-y)	Other drugs (n = 45, 48.51 p-y)
Drug toxicity	0.77	2.06	0.00	2.06	−, .49
Treatment interruption	12.35	0.00	8.07	0.00	−, .06
Virologic failure	2.32	4.12	2.02	4.12	.49 (.04–5.40), .61
Loss of follow-up	3.86	4.12	6.05	4.12	1.47 (.25–8.79), .70
Death	4.63	6.18	4.04	6.18	.65 (.11–3.91), .67

CI = confidence interval, p-y = person-year.

^*IRR (incidence rate ratio) and 95% CI for rates of each event by salvage regimen were calculated by Poisson regression, where other drugs were considered as comparators.

^†Wald test P value.

4. Discussion

Heavily experienced people with HIV who received raltegravir achieved and maintained high levels of virologic suppression after nearly 2 years of follow-up, consistent with findings from clinical trials and observational studies.^[29,32,42] The BENCHMRK trials, comparing raltegravir to placebo plus optimized background therapy in treatment-experienced patients, reported a virologic success rate of 77% in the raltegravir arm, maintained over 5 years of follow-up.^[29] Similarly, the SALIR-E study, which focused on treatment-experienced patients with detectable viremia starting a raltegravir-based regimen, found that 73% of participants had undetectable viral load at week 206.^[32] However, in our study involving heavily experienced patients treated in a non-clinical trial setting, survival analysis showed similar virologic suppression rates between patients receiving raltegravir and those receiving other salvage regimens based on darunavir/ritonavir, maraviroc, and etravirine. Similarly, when comparing patients receiving raltegravir-based and raltegravir-sparing regimens (with etravirine, maraviroc, enfuvirtide, or elvitegravir), Buchacz et al^[42] found comparable virological and clinical outcomes (76% and 63%, respectively, P = .51). These results contrast with the superiority of raltegravir over the optimized background therapy observed in the BENCHMRK randomized clinical trials.

Our results may differ partly because the subjects were heavily experienced with previous virologic failure, limiting their ART rescue options. In this study, raltegravir was preferred for sicker patients with advanced disease, as they had a higher viral load, a more significant proportion of CD4 cell count below 100 cells/mL, and a considerable number of patients with a GSS of <3 compared to those receiving other salvage regimens. Other more common characteristics in raltegravir recipients included a higher proportion of patients with 3 or more past treatment failures, alcohol use, anemia, and malnutrition. These characteristics indicate a potential risk of virologic failure, as observed in previous studies.^[43–46]

After matching, the outcomes of the raltegravir recipients did not appear to differ from those of patients treated with other third-line drugs. The rates of toxicity, treatment interruption, virologic failure, loss of follow-up, and mortality were comparable between the 2 groups. Raltegravir was generally well-tolerated, consistent with previous safety reports in treatment-experienced patients.^[23,29–31] A few patients in our study developed integrase mutations, with N155H being the primary pathway causing resistance to raltegravir. This may be particularly important when using raltegravir, as it seems sufficient to cause virologic failure.^[47,48] This mutation also reduces susceptibility to elvitegravir but does not affect dolutegravir or bictegravir alone.^[49,50] The slightly higher but non-significant proportion of patients with treatment interruption in the raltegravir group before and after matching raises concerns about factors that may affect patient adherence to ART. The possible reasons for treatment interruption remain unknown, as they were not systematically coded in the medical records. However, these data raise concerns about unmeasured patient characteristics, such as undiagnosed depression, unreported or unrecognized illicit drug use, and other social reasons that may affect adherence to ART.^[51,52] These factors are associated with increased mortality or treatment failure risk.^[47,51]

The overall mortality was very low, similar to controlled trials, such as the BENCHMRK and SALIR-E studies. However, the relatively small number of participants and deaths in our study limited our ability to detect a stronger association between the use of raltegravir-containing regimens and mortality, as well as associations with immunologic or virologic history.

Other limitations of our study include selection and information bias owing to reliance on medical records and national databases for data collection. PS matching allowed us to match the main covariates simultaneously. This helped mitigate these limitations by simulating a randomized controlled trial design using observational data from a real-life clinical population. However, it should be noted that matching all patients and basing conclusions on a subset of “good” matched individuals has its own limitations.

5. Conclusion

In summary, this real-world study provides valuable insights into the long-term outcomes of heavily treated patients, including mortality, treatment durability, and the safety and efficacy of different ARV. Our findings demonstrate that a virologic response is achievable in over 80% of heavily experienced patients initiating salvage therapy. Moreover, raltegravir-based regimens appear to be equally safe and effective as regimens based on other antiretroviral drug classes, despite the higher disease severity at the start of salvage therapy. These results are particularly significant, considering the limited availability of salvage therapy options.

Acknowledgments

We thank all the patients included in these analyses, the COM-HUPES facilities, and the staff of the Infectious Disease Research Laboratory (LAPI).

Author contributions

Conceptualization: Jesús Enrique Patiño Escarcina, Eduardo Martins Netto, Carlos Brites.

Data curation: Jesús Enrique Patiño Escarcina, Eduardo Martins Netto.

Formal analysis: Jesús Enrique Patiño Escarcina, Eduardo Martins Netto.

Investigation: Jesús Enrique Patiño Escarcina, Eduardo Martins Netto, Carlos Brites.

Methodology: Jesús Enrique Patiño Escarcina, Eduardo Martins Netto, Carlos Brites.

Project administration: Jesús Enrique Patiño Escarcina.

Software: Jesús Enrique Patiño Escarcina, Eduardo Martins Netto.

Supervision: Eduardo Martins Netto, Carlos Brites.

Validation: Jesús Enrique Patiño Escarcina, Eduardo Martins Netto, Carlos Brites.

Visualization: Eduardo M. Netto.

Writing – original draft: Jesús Enrique Patiño Escarcina, Eduardo Martins Netto, Carlos Brites.

Writing – review & editing: Jesús Enrique Patiño Escarcina, Eduardo Martins Netto, Carlos Brites.