Risk Factors for Selection of the L74I Reverse Transcriptase Mutation in Human Immunodeficiency Virus Type 1-Infected Patients

Abstract

We analyzed 3,475 human immunodeficiency virus sequences and 241 therapeutic histories. The L74I mutation was carried by 7% of viruses. L74I was strongly associated with T215F, K70R, and V75M/S/T/A mutations and increased with the number of thymidine analog mutations. It seemed to be linked to the use of abacavir or efavirenz.

Among resistance mutations to nucleoside reverse transcriptase inhibitors (NRTI), L74V is well documented (1-3, 7, 9-11, 13, 14). Conversely, very little is known about L74I, although it is also associated with human immunodeficiency virus treatment failure and is currently found in virologic practice (5, 6). We analyzed 3,475 RT sequences and 241 therapeutic histories. For each non-NRTI (NNRTI) and NRTI resistance mutation of the International AIDS Society-USA list, we studied the risk of an increase in the 74I/74V ratio. For each NNRTI or NRTI use, we also assessed the risk of L74I and L74V selection. Odds ratios (ORs) were calculated in univariate analysis and by multivariate logistic regression analysis.

The L74I and L74V mutations were present in 7% and 14% of viruses, respectively. Of the cases in which a mutation was present, the virus was of the B subtype in more than 93%. Analysis of the triplet of nucleotides at codon 74 showed that both mutations differed from the wild-type L74L by a single nucleotide change, with the thymidine in the first position replaced by an adenine for 74I and a guanine for 74V. Thus, L74I is not an intermediate that evolves to L74V. Furthermore, in patients with subsequent genotypic tests, the L74I mutation was replaced by L74V in only 5% of cases. Finally, for patients harboring 74I for more than 1 year (48%), the mutation persisted for a median of at least 3 years.

Table 1 summarizes the ORs of the mutations associated with resistance to NRTI and NNRTI that significantly increased the 74I/74V ratio. Almost all thymidine analog mutations (TAMs) increased this risk in univariate analysis. Actually, the number of TAMs was a strong risk factor for carriage of the 74I mutation (Fig. 1). The number of L74I and L74V mutations increased with the number of TAMs, but the rate of increase was higher for L74I. In a multivariate analysis, five mutations were statistically significant (Table 1). The highest risk was associated with the T215F, V75M/S/T/A, and K70R mutations (Fig. 2). The ORs for each of these mutations remained significant following adjustment for the number of TAMs. In patients who had undergone previous resistance tests, T215F and K70R were present before the emergence of L74I/V in 88% and 100% of cases, respectively. Finally, the presence of the L100I mutation (n = 153) significantly reduced the L74I/L74V ratio, and none of the sequences harboring the F77L (n = 55) or the F116Y (n = 50) mutation carried a concomitant L74I or V change.

TABLE 1.

ORs and P values for an increase in the L74I/L74V ratio in univariate analysis and multivariate logistic regression tests according to the presence or absence of the reverse transcriptase inhibitor mutation

Resistance mutationa	% of sequences in the following group		ORs and P values for an increase in the 74I/74V ratio according to the presence or absence of the mutationb
	L74I (n = 242)	L74V (n = 500)	Crude OR (95% CI)	P value	Adjusted OR (95% CI)	P value
T215F	33.9	9.6	11.7 (5.9- 23.1)	<0.0001	7.0 (3.0-16.2)	<0.0001
V75M/S/T/A	34.7	4.2	8.0 (5.0-12.9)	<0.0001	8.1 (4.3-15.1)	<0.0001
K70R	38.0	15.8	3.3 (2.3-4.6)	<0.0001	4.0 (2.0-8.0)	<0.0001
T215Y	57.0	67.4	2.8 (1.5-5.2)	0.001	3.1 (1.5-6.4)	0.0031
K219N	22.3	12.4	4.4 (2.8-7.0)	<0.0001	2.6 (1.5-4.7)	0.0012
K219E	10.7	5.2	5.0 (2.7-9.3)	<0.0001		Not significant
K219Q	33.9	17.4	4.7 (3.1-7.2)	<0.0001		Not significant
D67N	70.7	50.4	2.8 (1.9-3.9)	<0.0001		Not significant
M41L	86.8	76.2	2.1 (1.4-3.3)	0.0008
V108I	31.8	19.0	2.0 (1.4-2.9)	0.0001
K101E	13.6	8.8	1.7 (1.1-2.9)	0.0241
M184V/I	89.3	84.0	1.7 (1.1-2.9)	0.0271
G190A/S	34.7	25.8	1.5 (1.1-2.1)	0.0155
L100I	5.0	17.4	0.2 (0.1-0.5)	<0.0001

^aFor the percentage of sequences with the wild-type codon (not shown), see Fig. 2.

^bOnly the results with P values of <0.05 are shown.

FIG. 1.

Proportion of RT sequences carrying L74V or L74I mutations according to the number of TAMs and increase in 74I/74V ratio (OR of 1.60 per TAM carried with a range 1.41 to 1.81; P < 0.0001).

FIG. 2.

Proportion of RT sequences carrying L74V or L74I mutations according to the presence of mutations statistically linked to a high risk for an increase in the 74I/74V ratio.

We analyzed a total of 241 treatment histories, including similar numbers of L74L/V and I carriers (Table 2). The median numbers of prior regimens and of years on antiretroviral treatment were adjusted for the number of TAMs in multivariate analysis and had no significant impact on the selection of L74I or V or the L74I/L74V ratio. Conversely, in this adjusted test, the number of TAMs remained significantly linked to L74I selection (OR, 1.86; 95% confidence interval [CI], 1.42 to 2.42; P < 0.0001) and increase in L74I/L74V ratio (OR, 1.56 per TAM carried; 95% CI, 1.21 to 2.02; P = 0.0006). Table 2 summarizes the statistical results regarding the use of NRTI or NNRTI. In multivariate analysis, use of didanosine (ddI) or abacavir (ABC) was a risk factor for L74V selection, whereas use of ABC or efavirenz (EFZ) was a risk factor for L74I selection.

TABLE 2.

Crude and adjusted ORs for the risk of L74I and L74V selection according to the presence of each compound in the treatment history

Drug	% of patients who received the drug in the following group			L74I versus L74La		L74V versus L74L
	L74I (n = 70)	L74V (n = 96)	L74L (n = 75)	Crude OR (95% CI)	Adjusted OR (95% CI)	Crude OR (95% CI)	Adjusted OR (95% CI)
Zidovudine	97.1	87.5	92.0	Not significant		Not significant
Stavudine	97.1	96.9	94.7	Not significant		Not significant
Lamivudine	97.1	97.9	97.3	Not significant		Not significant
Didanosine	92.8	96.8	82.7	Not significant		6.5 (1.8-23.7) (1)	5.7 (1.5-21.1) (2)
Abacavir	67.1	59.4	40.0	3.1 (1.5-6.1) (1)	2.2 (1.1-4.4) (2)	2.2 (1.2-4.1) (2)	2.0 (1.1-3.7) (2)
Tenofovir	24.3	6.2	10.7	2.7 (1.1-6.7) (2)	Not significant	Not significant
Efavirenz	60.0	51.0	34.7	2.8 (1.4-5.5) (1)	2.4 (1.20-5.0) (2)	2.0 (1.1-3.7) (2)	Not significant
Nevirapine	67.1	59.4	50.7	2.0 (1.0-3.9) (2)	Not significant	Not significant

^a(1), P < 0.005; (2), P < 0.050.

Analysis of the 3,475 RT sequences showed that the L74I and L74V mutations were carried by 7% and 14% of viruses, respectively. Thus, L74I accounted for about one-third of changes at this position. According to the Stanford University website (http://hivdb.stanford.edu/), our prevalences were close to those reported for patients treated with at least four NRTIs (6% and 12%), and our L74I/l74V ratio (1/2) was close to that for patients previously treated with at least one NRTI. This may be because most of the genotypic tests carried out at our hospital were prescribed after at least two antiretroviral treatment failures and because our clinical department follows a large number of patients with large treatment histories.

The T215F and K70R mutations were the most strongly linked to a high risk of L74I selection. Both of these mutations are specific to TAM profile #2 (8). Multivariate analysis showed that the number of TAMs also had a significant effect on the likelihood of L74I selection independently of the TAM profile. This is why L74I was found together with many #1 profiles. As T215Y/F and K70R were detected before L74I or V in most cases, the resistance mechanism and replication constraints created by the number or profile of TAMs already present may affect the likelihood of selection of the 74I or 74V mutation. As it has already been suggested that L74V has an effect on the primer-unblocking activity conferred by TAMs, it is not surprising that the amino acids resulting from these mutations interact (2, 11). Interestingly, Miranda et al. showed that the negative effect of L74V on primer unblocking was blunted by the presence of a larger number of TAMs (11). Thus, L74I selection may be a better compromise for the virus in this context. Therefore, studies of biochemical interactions between TAMs and L74V should be repeated with the L74I mutation. The proximity of codons 74, 75, and 70 may account for interactions between these codons. Finally, simultaneous changes at codons 74 and 75 were also reported by Kleim et al. (6).

In multivariate analysis, the use of ABC or ddI was associated with the presence of L74V in RT sequences. These findings are consistent with published data, but interestingly, the presence of L74I was linked to treatment with slightly different compounds: ABC and EFZ. Kleim et al. also reported a possible association between L74I and NNRTI (6). In our study, the small number of patients naive for ddI may decrease the statistical power for analysis of the effects of this NRTI. The impact of this compound should therefore be studied in patients with a less extensive pretreatment histories. Despite these data, the impact of the L74I mutation on the antiretroviral activity of ABC and EFZ requires further clarifications in in vitro and in vivo studies. However, the low sensitivity of phenotypic testing, particularly in such a strong TAM context, and the large number of patients required for significant results to be obtained in clinical trials will probably make such studies difficult (4, 12).

Overall, this study shows that the baseline mutations present and the number of such mutations may affect the emergence of subsequent mutations. In addition to resistance effects, replicative capacity may favor the selection of one amino acid residue rather than another. Thus, for some compounds, such as abacavir, selection pressure seems to result in the selection of L74I as the best compromise between resistance and replication capacity effects in profile #2 or in a background containing a large number of TAMs.

This study shows that the L74I mutation is linked to antiretroviral treatment and independent of the L74V substitution. However, further studies are required to determine the real impact of this mutation on the antiretroviral activity of compounds such as abacavir or efavirenz. Analysis of this mutation should therefore be systematically included in all clinical trial analyses to clarify its impact on antiretroviral therapies. Similarly, further analyses are also required to determine the biochemical mechanisms favoring the selection of L74I rather than L74V in a TAM profile #2 background or in patients with a large number of TAMs.

(Part of this study was presented at the 13th Conference on Retroviruses and Opportunistic Infections, Denver, 2006, and at the 15th International HIV Drug Resistance Workshop, Sitges, Spain, 2006)