Integrase inhibitors and resistance

Simon Collins, HIV i-Base

The workshop provided the first forum with numerous early studies addressing the issue of resistance to integrase inhibitors (INIs):

  • Which key mutations develop and impact on drug sensitivity?
  • How quickly do they develop?
  • What is the role of polymorphisms in naive and treated patients?
  • What is the impact of cross class resistance?
  • The relationship between integrase mutations and those in RT and protease genome;
  • and the activity in HIV-1 subtypes and HIV-2.

With raltegravir already available on named patient programmes, these presentations probably had the greatest clinical significance. As with every other class, the risk of cross-resistance to other INIs should be expected until proved otherwise.

Ten abstracts at the resistance workshop addressed this subject, with the two most important studies providing the first in vivo data from Merck and Gilead compounds.

While these are the first important pool of studies, more are sure to follow at the IAS meeting in Sydney later in July.

Raltegravir resistance in experienced patients

One caution from the otherwise impressive results available on raltegravir, is that it appears to have a low genetic barrier to resistance. Michael Miller from Merck presented results from their Phase 2 dose-finding study that randomised patients to either 200 mg, 400 mg or 600 mg raltegravir or placebo, all in addition to optimised background therapy. [1]

By week 16, approximately 70% of these three-class experienced patients achieved viral suppression to <400 copies/mL. But 38 non-responders either failed to reduce viral load by > 1 log or reach and maintain levels <400 copies/mL. Mutations in integrase were found in 35/38 of these patients.

Two distinct pathways were also identified in in vitro studies –– either via N155H (n=14) or Q148H/R/K (n=20) –– which reduced raltegravir susceptibility by 10 and 25-fold respectively. One patient developed Y143R.

Secondary mutations found with N155H included L74M, E92Q, T97A, Y143H, V515I, G163R and D232N; and with Q148H/R/K included L74M, E138A/K, and G140S/A. All secondary mutations led to increased resistance. Fold changes in IC50 were greatest in people whose virus developed the 148 pathway with secondary mutations increasing from 10-fold (Q148H) to over 500-fold (Q148H+G140S and Q148K+E138A+G140A).

There was no dose-related relationship with the pattern of these mutations. A higher percentage of virological failures had baseline viral load >100,000 copies/mL (53% vs 47% <100,000), baseline CD4 <200 cells/mm3 (58%, vs 42% >200), and genotypic sensitivity score of 0 (68%, vs 32% when GSS>/=1).

Time to developing resistance was not clear, although no resistance was seen in the 10 day monotherapy study, but some patients failed by week 8 and accumulated multiple mutations by week 16. Longitudinal analysis presented on four patients showed detection of the primary N155H mutation from 50 to 300 days on study.

Impact on fold-change of other integrase inhibitors indicated cross-resistance to the Gilead GS-9137 compound. Merck has a pipeline compound MK-2048 which retained sensitivity to at least some of these mutations.

Vincent Calvez and colleagues from Hopital Pitie-Salpetriere reported resistance patterns from four highly treatment experienced patients who received raltegravir as part of a rescue therapy. [2]

Baseline viral load ranged from 4.3-5.5 logs copies/mL. Two patients had a genotypic sensitivity score (GSS) of 0 and two patients had a GSS of 1.

Two patients achieved suppression to <50 copies/mL and then rebounded, one suppressed to 2.8 log and one only had a -0.5 log reduction (this patient had a GSS of 1 at baseline). Resistance mutations at failure included Q140S+Q148H, which occurred at the same time; N155H; E92Q and E157Q with viral load returning to pre-treatment levels by week 8-24 as these mutations were detected, with a loss of 7-14 fold drug sensitivity.

The researchers concluded that selection of only one mutation that led to failure in these patients emphasised the low genetic barrier to resistance, and highlights that this drug should only be used with other active drugs in the regimen.

Elvitigravir resistance in experienced patients

Damian McColl presented the first analysis fo resistance found in the Gilead Phase 2 dose-ranging (20 mg, 50 mg, 125 mg) study of elvitegravir (EVG) in treatment-experienced patients. [3]

The control arm in the Gilead study was investigator chosen boosted-PI. All patients had optimised nucleoside background regimens (OBR), T-20 could also be used (but not NNRTIs), and a protocol amendment allowed use of darunavir and tipranavir in the EVG/r arms after 8 weeks.

30/73 patients in the 125mg EVG/100mg ritonavir arm had virological failure (with higher failure rates In the lower dose arms which were stopped early) and results were presented for 28/30 patients with integrase genotype results.

Although mean viral load responses of approximately -2.0 logs from baseline were seen in the 125mg arm at week 2, this rebounded in patients with no active drugs in their OBR to -0.7 log by week 24, compared to maintaining -2.1 log reductions when one or more active drugs were included in the OBT or who were using T-20 for the first time.

The first observed mutations included T66I/A/K, E92Q, E138K, S147G, Q148R/H/K and N155H.

The most common mutation patterns by week 24 as detailed in Table 1.

Table 1: Most common EVG-related mutations

Integrase resistance mutations N (%)
Any E92Q 11 (39%)
E92Q + N155H (+/- others)

E92Q + T66A (+/- others, no N155H)

E92Q (+/- others, no T66A or N155H)

4 (14%)

3 (11%)

4 (14%)

Any N155H 11 (39%)
N155H + other mutations (No E92Q)

N155H + T66I (+/- others, no E92Q)

N155H + E138K (+/- others, no T66I, no E92Q)

N155H + others (no E138K, T66I, or E92Q)

E138K + S147G + Q148R (+/- others)

G140C/S + Q148R/H/K (+/- others inc T66I)

7 (25%)

1 (4%)

3 (11%)

3 (11%)

6 (21%)

3 (11%)

Again, resistance not reported in the 10-day monotherapy studies but developed by week 4 in some patients who had no other active drugs. Analysis of longitudinal data is ongoing.

Mean phenotypic changes in sensitivity to elvitegravir were >150-fold at viral failure (median 152, range 1.02-301-fold). Cross resistance to raltegravir was indicated by mean changes in phenotypic sensitivity of 28-fold (median 10-fold, range 0.78->256-fold).

There was a close correlation between reduced susceptibility to both EVG/r and raltegravir in isolates with most common mutation patterns (R2=0.66).

Polymorphisms in integrase

Several studies also analysed the importance of pre-exiting integrase mutations in integrase-naive patients. While the range of natural polymorphisms seems extensive, they do not seem to be related to reduced susceptibility to diketo acid compounds or to a higher risk of treatment failure, in either treatment-naive or -experienced patients. [4, 5, 6]

However, early data was presented that tentatively suggested that the relationship between changes in integrase may be linked to shifts in other parts of the genome –– reverse transcriptase, protease and perhaps other regions –– and that this might have an impact in treatment sequencing for experienced patients. [6]

Richard Myers from the Health Protection Agency in the UK and Deenan Pillay from UCL presented an analysis from 1,250 INI-naive isolates from the Los Alamos database, where they identified 41 mutations at 30 positions in integrase. [4]

The most prevalent variants were V201I (in 80%) and V72I (46%). 15 mutations occurred at >5% frequency in different HIV subtypes. Other common mutations were identified at codons 74, 97, 125, 154, 163 and 206. However, some of the between-clade differences led the researchers to conclude that susceptible to integrase inhibitors may vary by sub-types and that this should be examined in future in vitro models.

Kurt van Baelan and colleagues from Virco also presented data looking at the prevalence of naturally occurring integrase polymorphisms amplified from the RT-RNase-H-IN region in 47 patient samples and 89 clones. [5]

The most common polymorphisms were V72I (in 41 patients and 77 isolates) and V201I (in 33 patients and 48 clones), and included L74M/I, T97A, V151I, K156N, 165I, I203M, T206S and S230R/N, though none were associated with significant fold-changes. They also reported no differences between HIV sub-type, or between RTI sensitive and resistant isolates.

Carlo Perno and colleagues from Universities of Rome and Milan, looked at gene sequences for the whole sequence of RT and IN (1-320 residues) from 448 patient with HIV-1 sub-type B (134 naive and 314 experienced). [6]

Protein sequences in integrase were unaltered at 62% and 67% of codons in naive and experienced patients respectively. However, 24/37 integrase mutations were not found in either naive or RTI-treated patients. The eight position changes showing >5% variability were I72V, T125A/V, M154I, K156N, V165I, V201I, T206S and S230N.

Some mutations were present significantly more frequently in treated compared to naive patients: M154I (21% vs 6%, p<0.001), V165I (13% vs 6%, p =0.022), M185L (6% vs 0%, p=0.003). M185L was significantly positively associated with V165I in integrase and F227L and T215Y in RT.

I72V occurred more frequently in untreated patients, and was positively associated with the protective R83K mutation in RT, and negatively associated with D67G and M184V in RT (p<0.03).

The authors concluded that the association found between selected INI and RT mutations supports the hypothesis of a tight interaction of these two proteins, and suggest the importance of IN-sequencing for RT-experienced patients prior to starting integrase-based regimens.

Integrase inhibitors sensitivity to HIV-2

Dianne Descamps and colleagues from Bichat-Claude Bernard Hopital Paris presented a paper on the phenotypic sensitivity of raltegravir and elvitegravir in isolates from 19 integrase-naive patients infected with HIV-2 (9 with subtype A, 9 with subtype B, 1 with sub-type H). [7]

The HIV-2 integrase gene differed from HIV-1 at 133/288 codons, including some that have been described as integrase resistant sites for HIV-1. However, mean IC50 was similar between HIV-2 and HIV-1 reference strains, including two isolates with MDR Q151M in RT. Despite overall amino acid polymorphism of 31% this didn’t alter phenotypic susceptibility of HIV-2 to either inegrase inhibitor.


All references are to the Programme and Abstracts of the XVI International HIV Drug Resistance Workshop, 12-16 June 2007, Barbados. Published as part of Antiviral Therapy (Volume 12 Issue 5).

  1. Hazuda DJ, Miller MD, Hguyen BY et al. Resistance to the HIV-integrase inhibitor raltegravir: analysis of protocol 005, a Phase II study in patients with triple-class resistant HIV-1 infection. Antiviral Therapy. 2007;12:S10. Abstract 8.
  2. Malet I, Delelis O, Calvez V et al. Biochemical characterizations of the effect of mutations selected in HIV-1 integrase gene associated with failure to raltegravir (MK-0518). Antiviral Therapy. 2007;12:S9. Abstract 7.
  3. McColl DJ, Fransen S, Gupta S et al. Resistance and cross resistance to first generation integrase inhibitors: insights from a Phase II study of elvitegravir (GS-9137). Antiviral Therapy. 2007;12:S11. Abs 9.
  4. Myers RE, Pillay D. HIV-1 integrase sequence variation and covariation. Antiviral Therapy. 2007;12:S65. Abstract 3.
  5. Van Baelen K, Clynhens M, Rondelez E et al. Low level of baseline resistance to integrase inhibitors L731,988 and L870,810 in randomly selected subtype B and non-B HIV-1 strains. Antiviral Therapy. 2007;12:S7. Abstract 5.
  6. Ceecherini-Silberstein F, Malet I, Perno CF et al. Specific mutations related to HIV-1 integrase inhibitors are associated with reverse transcriptase mutations in HAART-treated patients. Antiviral Therapy. 2007;12:S6. Abstract 4.
  7. Roquebert B, Dmaond F, Descamps D et al. Polymorphism of HIV-2 integrase gene and in vitro phenotypic susceptibility of HIV-2 clinical isolates to integrase inhibitors: raltegravir and elvitegravir. Antiviral Therapy. 2007;12:S92. Abstract 83.

Links to other websites are current at date of posting but not maintained.