Early Virological Failure in Naive Human Immunodeficiency Virus Patients Receiving Saquinavir (Soft Gel Capsule)-Stavudine-Zalcitabine (MIKADO Trial) Is Not Associated with Mutations Conferring Viral Resistance

Abstract

The MIKADO trial was designed to evaluate the efficacy of stavudine-zalcitabine-saquinavir (soft gel capsule) [d4T-ddC-SQV(SGC)] in 36 naive patients (−3.3 log₁₀ units at week 24 [W24]). Among the 29 patients remaining on d4T-ddC-SQV(SGC) until W24, 10 harbored a virological failure (viral load of >200 copies/ml at W24) (group 1). To determine the reasons for therapeutic failure, genotypic and phenotypic resistance test results and SQV concentrations in plasma were analyzed and compared to those in successfully treated patients (viral load of <200 copies/ml at W24) (group 2). Reverse transcriptase and protease genotypic analyses in group 1 revealed the acquisition of only one SQV-associated mutation (L90M) in only two patients. There was no significant increase in the 50 or 90% inhibitory concentration of SQV in patients with or without the L90M mutation. However, the fact that two patients developed an L90M mutation only 4 weeks after relapse points to the need for genotypic resistance testing in the context of an initial failure of the antiretroviral regimen. At W24, the median SQV concentration in group 1 (71 ng/ml) was significantly lower than in group 2 (475 ng/ml), and the plasma SQV concentration was correlated with the viral load at W24 (r = −0.5; P < 0.05) and with the drop in viral load between day 0 and W24 (r = −0.5; P < 0.01). These results and the fact that the plasma SQV concentrations in the two groups prior to relapse (W12) were not significantly different strongly suggest that the early failure of this combination is not due to viral resistance but to a lack of compliance, pharmacological variability, and drug interactions or a combination of these factors.

The reasons for early virological failure in patients on triple-drug therapy, including a protease inhibitor (PI) as the first line of defense, are currently being investigated, and many researchers are concerned about genotypic or phenotypic virological resistance. Several previous studies have analyzed early virological failure, especially in terms of viral resistance. Two of these studies, the ACTG 343 trial (7, 10a; D. Havlir, N. Hellmann, C. Petroupoulos, J. Whitcomb, A. Collier, M. Hirsch, P. Tebas, and D. Richman, Abstr. 6th Conf. Retrovir. Opport. Infect., abstr. 493, 1999; D. Havlir, C. J. Petropoulos, N. S. Hellmann, J. M. Whitcomb, D. D. Richman, and ACTG 343 Team, Abstr. 2nd Int. Work. HIV Drug Resist. Treat. Strat., abstr. 74, 1998) Trilege (ANRS 072) (4, 7a, 13; F. Brun-Vézinet, D. Descamps, G. Peytavin, V. Calvez, P. Flandre, V. Meiffredy, F. Raffi, G. Pialoux, J. P. Aboulker, and The Trilège Study Group, Abstr. Int. Conf. Discov. Clin. Dev. Antivir. Ther., abstr. 15, 1998; D. Descamps, V. Calvez, P. Flandre, G. Pialoux, F. Raffi, C. Delaugerre, G. Collin, V. Meiffrédy, G. Peytavin, J. P. Aboulker, F. Brun-Vezinet, and ANRS 072 Study Group, Abstr. 4th Int. Cong. Drug Ther. HIV Infect., abstr. 0P 3.3, 1998; D. Descamps, G. Peytavin, V. Calvez, P. Flandre, V. Meiffrédy, F. Raffi, G. Pialoux, J. P. Aboulker, F. Brun-Vezinet, and French Trilege (ANRS 072) Study Group, Abstr. 6th Conf. Retrovir. Opport. Infect., abstr. 493, 1999) analyzed the reasons for failure in the particular case of induction-maintenance trials. In the present study, MIKADO (Roche M61002), we were interested in the resistance profile of patients with a reemergence (>200 copies/ml) of viral load (VL) after 24 weeks (W24) of classical HAART triple therapy containing a new formulation of saquinavir (SQV), saquinavir soft gel capsule [SQV(SGC); Fortovase), a PI with a higher bioavailability (2a, 11, 12; Hoffman-La Roche, protocol NV15107, study report W-144981, data on SQV [RO 31-8959] on file).

The MIKADO study evaluated a triple antiretroviral (ARV) therapy with stavudine-zalcitabine-SQV(SGC) [d4T-ddC-SQV(SGC)] in naive patients (median CD4, 369/mm³; median VL, 4.98 log copies/ml) as a first-line therapy. The magnitude of the human immunodeficiency virus (HIV) RNA drop was −3.3 log copies/ml at W24. At W24, 63.3% of patients had HIV RNA levels below 200 copies/ml and 36.7% of the patients had HIV RNA levels below 20 copies/ml (Roche Ultrasensitive assay) and the median CD4 increase was 106/mm³ at W24 (C. Katlama, V. Calvez, J. L. Pellegrin, D. Lacoste, M. A. Valantin, E. Dohin, J. F. Delfraissy, and The Mikado Study Group, Abstr. 12th World AIDS Conf., abstr. 12244, 1998).

At W24, two groups of patients were identified according to whether their VL was above or below 200 copies/ml. This study examines the mechanisms underlying early failure (VL > 200 copies/ml at W24) of patients treated with d4T-ddC-SQV(SGC) during 24 weeks. We analyzed genotypic and phenotypic resistance, pharmacology at day 0 (D0), week 12 (W12) and W24, and other parameters, such as VL and CD4.

(This work was presented in part at the Fourth International Congress on Drug Therapy in HIV infection, in Glasgow, United Kingdom, in November 1998 [abstr. P13] and at the International Conference on the Discovery and Clinical Development of Antiviral Therapies, St. Thomas, West Indies, U.S. Virgin Islands, in December 1998 [abstr. 19].)

MATERIALS AND METHODS

Patients.

Of the 36 patients enrolled in the MIKADO trial, 29 received the d4T-ddC-SQV(SGC) regimen up to W24 (Katlama et al., Abstr. 12th World AIDS Conf.). At W24, patients fell into one of three categories: 19 of the 29 patients were “complete responders” with a VL always well controlled and below 200 copies/ml; 2 of the 29 patients were “incomplete responders” with a significant VL reduction but always >200 copies/ml; and 8 of 29 patients were “relapsers” with a controlled VL below 200 copies/ml followed by a reemergence of VL increasing to >200 copies/ml at W24.

Two groups of patients were also defined and investigated as follows. Group 1 consisted of 10 patients, the 2 incomplete responders and the 8 relapsers (Table 1). These patients underwent reverse transcriptase (RT) and protease gene sequencing, phenotypic analysis (recombinant virus assay [RVA]) (1, 17) at D0 and W24, and plasma SQV concentration measurements at W12 and W24. Group 2 consisted of 10 of the 19 complete responders. These were investigated only for plasma SQV measurements as a control (Table 2). The low VLs of these patients made it impossible to perform genotypic and phenotypic resistance testing.

TABLE 1.

Virological characteristics and results of genotypic and phenotypic resistance testing for group 1 patients

Group 1 patienta	Virological characteristics of patientsb			Genotypic studyd		Phenotypic study (fold change in SQV IC90 from D0 to W24)e
	VL at D0	Time of relapsec (VL)	VL at W24	RT at W24	PROT at W24
A	461,329	W20 (475)	970	WT	L90M	1
B	403,042	W24 (3,197)	3,197	WT	WT	ND
C	4,245	W20 (270)	227	IMPOSS	IMPOSS	ND
D	19,397	W20 (119,266)	2,097	WT	WT	ND
E	66,215	W24 (225)	225	IMPOSS	WT	ND
F	267,937	W20 (223)	239	IMPOSS	IMPOSS	ND
G	34,315	W20 (445)	818	WT	WT	1.6
H	67,151	W20 (505)	590	WT	L90M	2
I	302,469	NA	8,326	WT	WT	2.8
J	538,689	NA	3,400	WT	WT	ND

^aGroup 1 patients A, B, C, D, E, F, G, and H are relapsers, and patients I and J are incomplete responders. 

^bVL expressed in number of copies per milliliter. 

^cThe time of relapse is the first time with reemergence of VL of up to 200 copies/ml. NA, not applicable (VL always >200 copies/ml). 

^dIn the genotypic study, RT and protease (PROT) genes of relapsers and incomplete responders to the HIV-1 clade B consensus sequence at D0 and W24 were compared. No d4T- or ddC-associated resistance mutations were found in the RT genes, and only one amino acid substitution associated with an SQV resistance mutation was found at W24 in only two patients. Abbreviations: WT, wild-type protease gene; IMPOSS, PCR amplification failure on plasma samples with the lowest VL; L90M, L90M mutation in the protease gene. 

^eBetween D0 and W24, there was no significant increase in the SQV IC₉₀ in patients with and without the L90M mutation (fold changes of L3). ND, not done. 

TABLE 2.

Retrospective SQV concentrations in plasma samples from group 1 and 2 patients^a

Time and parameterb	SQV concn in plasmac								Statistical significanced
	Group 1 (n = 9)				Group 2 (n = 10)
	Median	Mean	Min	Max	Median	Mean	Min	Max
D0
VL	267,937	233,174	4,245	538,689	91,491	136,807	57,997	475,682	NS
CD4	353	375	230	726	382	464	289	899	NS
W12
VL	223	412	56	875	88	100	<20	200	NS
CD4	352	412	324	558	398	424	288	679	NS
SQV	673	692	103	1,520	734	754	112	1,574	NS
W24
VL	970	2,207	227	8,326	21.5	48	<20	165	0.0002
CD4	475	495	247	907	507	512	228	938	NS
SQV	71	100	<10	248	475	988	70	4,120	0.025

^aAt W24, the VLs for group 1 patients were >200 copies/ml, while the VLs for group 2 patients were <200 copies/ml. At D0, the two groups showed no differences in term of VL and CD4. At W12, prior to virological relapse, the two groups were statistically identical in terms of VL and CD4. Plasma SQV concentration measurement did not reveal any statistical difference between these two virologically well-controlled groups. At W24, the median SQV concentration of relapsers and incomplete responders (group 1) was dramatically lower than that of responders (group 2). 

^bVL is expressed in number of copies per milliliter, CD4 is expressed in number per millimeter³, and SQV is concentration in nanograms per milliliter. 

^cMedian, mean, minimum (Min) and maximum (Max) values are given. 

^dStatistical significance measured by the Mann-Whitney nonparametric test. Numbers are P values. NS, not significant. 

VL determinations.

HIV type 1 (HIV-1) plasma RNA was quantified with Amplicor Monitor Assay (Roche Diagnostic Systems, Inc., Branchburg, N.J.) following the manufacturer's recommendations. Aliquots of plasma with values below 200 copies/ml were assayed with the Roche Ultrasensitive Assay (detection limit, 20 RNA copies/ml).

Genotypic study.

We looked for mutations in group 1 patients by sequencing the RT and protease genes after RT-nested PCR amplifications on plasma HIV RNA collected at D0 and W24 as previously described. HIV-1 RNA was extracted from 200 μl of plasma using the HCV Roche extraction kit (Roche Diagnostic Systems, Inc.), reverse transcribed to cDNA using Titan One Tube RT-PCR System (Boehringer Mannheim), and directly amplified by nested PCR. For RT gene amplification, first PCR was performed with primers MJ3 (5′-AGTAGGACCTACACCTGTCAAC-3′) and MJ4 (5′-CTGTTAGTGCTTTGGTTCCTCT-3′) and with the following cycle conditions: 30 min at 50°C and then 40 cycles, with 1 cycle consisting of 2 min at 94°C, 30 s at 55°C, and 3 min at 68°C. Nested PCR used primers A 35 and NE1 35 and cycling conditions, both already described (3, 17). For protease gene amplification, primers and cycle conditions have been described previously (3, 17). Nested-PCR products were purified before sequencing, using Microcon (Microconcentrators; Amicon) and directly sequenced, using ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction Kit (Perkin-Elmer, Paris, France), on an automated DNA sequencer (ABI 377; Perkin-Elmer). Following electrophoresis, sample files were processed by use of Factura Software. Sample files were aligned by use of autoassembler software Sequence Navigator to generate a deduced nucleotide sequence. The resulting nucleotide sequence was compared with HIV-1 consensus clade B (10a) and then translated to yield the amino acid sequence.

Phenotypic study.

Phenotypic studies for SQV resistance were performed using an RVA on D0 and W24 plasma samples. Protease genes were amplified by RT-nested PCR and used to create a recombinant HIV-1 virus after cotransfection with HIV-1 plasmid in which the protease gene had been deleted (plasmid kindly provided by E. D. Blair [9]). Phenotypic assays were performed as previously described for the MT2 cell line (1, 8). The results are expressed as fold resistance ratios and compared to the sensitivity at D0.

Plasma SQV concentration measurement.

A retrospective SQV(SGC) plasma quantification was performed on samples collected at W24 for group 1 and group 2 patients. A sensitive high-performance liquid chromatographic method was used to measure plasma SQV concentrations on frozen plasma samples. The limit of quantification was 10 ng/ml. We considered a concentration to be in the therapeutic range if it was above 216 ng/ml [the minimum concentration of SQV(SGC)] (2a; Hoffman-La Roche data on file). We also compared the median plasma SQV concentrations of group 1 and 2 patients at W12 and W24.

RESULTS

Patients.

At D0 and W12, the differences in the VLs of group 1 and 2 patients were not statistically significant (Table 2). At W24, as expected because of the group definitions, the mean VL of group 1 patients was significantly greater than that of group 2 patients (P < 0.001) (Table 2).

At D0, CD4 counts were 353/mm³ in group 1 patients and 382/mm³ in group 2 patients (range, 300 to 500/mm³) with no statistically significant difference between them (Table 2). At W12 and W24, neither the median numbers of CD4 (Table 2) nor the median increases in group 1 and group 2 were statistically different.

Genotypic study.

In group 1 (10 patients) (virological failure), PCR amplifications at W24 were successful in eight patients for protease and seven patients for RT (Table 1).

Protease genotypic analysis at D0 and W24 revealed an acquisition of an SQV-associated mutation (L90M) in viruses from only two patients (Table 1). No other SQV-associated primary mutations (G48V or I84V) were seen. None of the patients developed secondary SQV-associated mutations (I54V, L63P, A71V/T, G73S, or V82A/F/T/S) (16).

RT gene analysis at D0 and W24 revealed no mutation in the gene associated with dideoxynucleoside resistance (Table 1).

In group 2 (no virological failure), PCR amplifications at W24 were unsuccessful due to the low VLs (<200 copies/ml). At D0, no mutations were present in the RT gene.

Phenotypic study.

RVA was performed on four patients of group 1: two patients with a wild-type (WT) protease gene at D0 and acquisition of L90M at W24 and two patients with a WT protease gene at D0 and W24. We did not observe any significant increase in the 50% inhibitory concentration (IC₅₀) or 90% inhibitory concentration (IC₉₀) of SQV in patients with or without the L90M mutation (fold changes between D0 and W24 of ≤3) (Table 1).

Pharmacological studies.

Retrospective pharmacological studies were performed on 9 of the 10 patients of group 1 (virological failure) and all 10 patients of group 2 (no virological failure). SQV could be detected in the plasma samples of all patients in group 2 and in all but one patient in group 1. In group 2, 3 of the 10 (30%) patients had an SQV concentration below the therapeutic range. By contrast, in group 1, eight of nine (89%) patients had SQV concentrations below the therapeutic range. At W24, the median concentration of SQV in plasma for group 1 patients was significantly lower (P = 0.025 by the Mann-Whitney test) than those of group 2 patients (Table 2). Moreover, there were correlations between the plasma SQV concentration and the VL at W24 (r = −0.5; P < 0.05) and between plasma SQV concentration and the VL decrease between D0 and W24 (r = −0.6; P < 0.01).

However, the SQV concentrations in the two groups were not significantly different at W12, prior to relapse. This is in accordance with the fact that both groups harbored similar VLs at this time (Table 2).

DISCUSSION

As part of the MIKADO trial evaluating a triple ARV combination [d4T-ddC-SQV(SGC)], we have investigated the differences between patients with an incomplete response (a significant reduction in VL but always >200 copies/ml) or a reemergence of VL (>200 copies/ml) at W24 (group 1) and patients with a complete response (<200 copies/ml) at W24 (group 2). Many factors may be involved in these early virological failures. The first factor, the most obvious one, is viral resistance. The second factor which could be evoked are HIV VL and CD4 counts, as some degree of risk factor has been ascribed to their baseline values and rate of response. We report an additional factor, the PI concentration in plasma, which shows statistically significant differences between patients with and without virological failure.

The first factor which may be attributed to virological failure is the genotypic resistance. In the case of the protease gene, only two patients had one SQV-associated mutation (L90M). There was no acquisition of other primary SQV-associated mutations such as G48V (16). This is in accordance with the work of Schapiro et al. (15), who found that the most frequent mutation in the protease gene following treatment with the zidovudine (AZT)-SQV or AZT-didanosine (ddI)-SQV combination is at codon 90, and mutation at codon 48 is very rare. No acquisition of secondary mutations was observed in this study. This last fact suggests that resistance had not developed in the majority of failing patients. For those who developed an L90M mutation, it has been clearly demonstrated (3) that acquisition of numerous secondary mutations in association with the L90M mutation is needed to affect sensitivity to SQV. In our study, phenotypic assays showed no elevation of IC₉₀ for SQV, even in patients with an L90M mutation. However, the fact that two patients developed an L90M mutation only 4 weeks after relapse points to the need for genotypic resistance testing in the context of an initial failure of the ARV regimen. No mutation was found in the RT gene. This is in accordance with the virological studies of two induction-maintenance trials, in Trilège (ANRS 072) (4, 7a; Brun-Vézinet et al., Abstr. Int. Conf. Discov. Clin. Dev. Antivir. Ther.; Descamps et al., Abstr. 4th Int. Cong. Drug Ther. HIV Infect.; Descamps et al., Abstr. 6th Conf. Retrovir. Opport. Infect.) and ACTG 343 (10a; Havlir et al., Abstr. 6th Conf. Retrovir. Opport. Infect.; Havlir et al., Abstr. 2nd Int. Work. HIV Drug Resist. Treat. Strat.). In these trials, following induction with zidovudine-lamivudine-indinavir (AZT-3TC-IDV) regimen, patients were randomly assigned to three types of regimens: (i) AZT-3TC-IDV, (ii) AZT-3TC, and (iii) AZT-IDV (Trilège) or IDV (ACTG 343). Viral resistance analysis of early virological failure revealed the absence of mutations in the RT gene in the AZT-IDV or AZT-3TC-IDV arms of the study (except for M184V).

The second factor was evoked by the results of studies of the predictors of durable response to ARV therapy. The first level of investigation is the measurement on D0. The VL at D0 as a predictive factor of treatment success is found in some clinical trials but not in other studies (5, 6). In our study, the baseline VL in group 1 and 2 patients did not show a statistically significant difference. Another factor which is said to influence the early reemergence of VL (10) is the baseline level of CD4. The VL in patients with a higher CD4 cell count was more likely to become and stay undetectable. In our study, we did not observe any difference in the CD4 cell count or any relationship between the CD4 cell count at D0 and the VL during the trial. The second level of investigation is follow-up studies. V. Miller et al. (V. Miller, S. Stasweski, A. Hill, A. Cozzi, I. Lepri, C. Sabin, and A. Phillips, Abstr. 6th Conf. Retrovir. Opport. Infect., abstr. 167, 1999) have shown the predictive value of the time needed to get the VL below the limit of detection (500 copies) and the time needed to observe a rebound in VL after initial suppression by therapy with three or four drugs. The best situation is a VL < limit of quantification before 12 weeks, which was always the case, for our group 1 and 2 patients. The time of relapse was close to W24 (Table 1). Moreover, it has been shown that suppression of the plasma VL to below 20 copies/ml is required to achieved a long-term response to therapy (14). Except for the incomplete responders, group 1 and 2 patients had at least two consecutive VLs under 20 copies/ml during the 2 months before the time of relapse for group 1 and during the 2 months before W24 for group 2. Moreover, the increase of CD4 was in the same range in group 1 and 2 patients. The last level of investigation is located at the moment of virological failure. V. Miller et al. (Abstr. 6th Conf. Retrovir. Opport. Infect.) have shown the predictive value of the latest CD4 cell count, one that is not observed in our study.

The last factor is evoked by many studies showing that the difference between patients who do and do not fail is signaled by PI concentrations in their plasma samples. In induction-maintenance therapy trials (Trilège, ANRS 072), failure was not associated with viral resistance but with pharmacological impairment (4, 7a). Burger et al. (2) showed that low concentrations of IDV in plasma are related to virological treatment failure in patients on IDV-containing triple therapy. In our study, the plasma SQV concentration levels differed between patients who had been therapeutic failures and therapeutic success; we also observed a correlation between plasma SQV concentration and VL. The differences could be related to many phenomena leading to pharmacological variability and including compliance, drug interactions, drug absorption, metabolism, and cytochrome P450 interactions. The fact that plasma SQV levels prior to relapse (at W12) were in the therapeutic range of and were not significantly different in the two groups gives additional support to the conclusion that therapeutic failure initiation is due to suboptimal concentrations of SQV in plasma.

These results suggest that early failure to respond to this ARV combination is not due to drug resistance, VL, or CD4 characteristics but to compliance with medication, poor drug absorption, and adverse drug interactions.

In our study, at W12, patients were comparable in terms of VL, CD4 counts, and plasma saquinavir concentration and became different between W12 and W24 in terms of VL and plasma SQV concentration. Most probably, adherence to a treatment regimen is essential. The time and degree of failure observed in the Trilège trial were associated with the degree of adherence. Other parameters, such as poor drug absorption and adverse drug interactions, were not investigated in our study. Our study provides supplementary elements to point out the urgent need for less-toxic, simpler, and more patient-friendly regimens. We need prospective studies to evaluate precisely the exact roles of patient noncompliance, poor drug absorption, and adverse drug interactions in the context of an initial failure of ARV therapy.