Tentative phenotypic cut offs for dolutegravir: guidance for use with integrase resistance
Simon Collins, HIV i-Base
Several presentations presented new data relating to the investigational integrase inhibitors dolutegravir.
The most important of these included a correlation of baseline fold change (FC) in IC50 with change in plasma viral load following 8 days of monotherapy, plotting individual patient responses based on baseline integrase resistance. 
This analysis, presented by Cindy Vavro and colleagues from GlaxoSmithKline, was from the VIKING-3 study in treatment-experienced patients with integrase inhibitor resistance, using dolutegravir at a dose of 50 mg twice daily for 8 days as functional monotherapy (while on current failing treatment), before switching to optimised background therapy.
Mean reduction in viral load at day 8 in 183 patients was -1.43 log copies/mL (95% CI -1.52, +1.34; p<0.001) with 82% of patients dropping by greater than 1 log copies/mL.
Phenotypic cut offs for dolutegravir were derived based on viral response at day 8. These were defined in the study as full (>1 log), intermediate (0.5-1.0 log) and none (<0.5 log) with FC (95% CI) of <9.4 (5.98, 15.88) and ≥25.99 (15.92, – ), for full and intermediate responses respectively.
Table 1 summarises Day 8 observed responses by these FC sensitivity at baseline, which resulted in a negative predictive value of 67% for people with FC <9.45 who would still not have a full response and a positive predictive value of 87% (136/156).
|Est. FC cut-off||Full||Intermediate||None||Total||PPV||NPV|
|< 9.45||136 (77%)||16 (9%)||4 (2%)||156 (88%)||136/156 (87%)||10/15 (67%)|
|9.45-25.99||5 (2%)||2 (1%)||5 (3%)||12 (7%)||159/168 (95%)|
|> 25.99||0||0||3 (2%)||3 (2%)||3/3 (100%)|
|Total **||146 (82%)||19 (11%)||12 (7%)||177 (100%)|
** 6/177 patients had no FC at baseline
In this planned interim analysis of patients with 24 week results, 63% of those with baseline FC <9.45 (72/114) had viral load suppressed to <50 copies/mL. No patients with FC >9.45 at baseline achieved <50 copies/mL at week 24. Response rates were also closely related to integrase mutations at baseline, with Q148H/K/R present in 57 (31%) patients, usually with additional mutations. Pre-specified integrase mutations (with primary integrase mutations in bold) were: H51Y, T66A/I/K, L68V, L68I, L74I/M/R, E92Q/V, Q95K, T97A, G118R, E138A/K/T, G140A/C/S, Y143C/H/R, P145S, S147G, Q148H/K/R, V151I/L, S153F/Y, N155H, E157Q, G163R/K, G193E, R263K.
In multivariate analysis, three baseline mutations were most significantly associated with reduced day 8 response: Q148H/K/R (by -0.47 log); p=<0.001), L74I (by -0.27 log; p=0.037) and E138E/K/T (by -0.25 log; p=0.052). There was also a high correlation between Q148H/K/R and G140A/C/S (p<0.001).
When Q148 mutations were not present at baseline, 92% participants achieved >1 log reduction by day 8. This dropped to 71% when Q148 was present with one out of G140A/C/S, L74I or E138A/K/T and to 45% when two or more of these mutations were also present. The 24-week response rates for each of these baseline resistance categories was 79%, 45% and 11% respectively. The investigators noted that confidence intervals for FC values were either wide or not obtained and that this limited certainty of the results, but commented that this was partly related to high rates of viral efficacy and low numbers of patient with higher FC values in the current dataset.
In the Q&A after the presentation, Brendan Larder suggested that it would be more important to use the dataset to establish the cut-off that determined 100% response rates rather than the 69% presented. This improved sensitivity would clearly help management of patients who are unlucky enough to have already developed drug resistance to raltegravir or elvitegravir.
A poster from the VIKING-3 study presented an analysis of the resistance that developed in 35/183 (19%) based on interim results in 31/35 of these patients with paired baseline and week 24 samples. 
Treatment emergent resistance developed in 15/31 of these patients, with 13/15 having a Q148 mutation at baseline or in archived samples. Newly emergent Q148 mutations occurred in four patients, with 3/4 having archived Q148 and detection at day 1 occurring in the other. Dolutegravir fold-change increased from baseline to point at viral failure by >2 fold in 13/15 people with newly emergent resistance with >8-fold increases in 7/31 (23%) of patients.
An interim 24 week analysis from the phase-3 SAILING study, comparing raltegravir to dolutegravir in treatment-experienced but integrase-naïve patients, reported fewer protocol-defined failures in the dolutegravir arm (14/354 vs 34/361: 4% vs 9%). Additionally, that two patients on dolutegravir-based treatment had viral rebound with mutations at codon R263. 
These two cases sub-optimal ART and/or dolutegravir levels with one patient entering the study on a combination with only one additionally sensitive drug (efavirenz). Both patients experienced viral rebound prior to week 24, one with R263R/K at week 16 and one with V260I/R262K at week 24, but both osolates retained low IC50 fold changes to raltgravir an dolutegravir. A separate in vitro analysis suggested that dolutegravir binding remains prolonged (>50 hours) in the presence of with both these complexes and that this might explain the relatively high genetic barrier to resistance.
Finally, a poster looked at the evolution of integrase resistance mutations in the Monogram clinical isolate library with raltegravir resistance during two periods – from August 2008 to December 2009 (n=273) and from January 2010 to December 2011 (n=806). This was to give an indication of how resistance may be changing in the clinic setting. 
In general the percentage of both individual mutations and more complex combinations, was reduced in the second time period. They reported that 662 (93.9%) of raltegravir-resistant isolates had dolutegravir FC <10 and that only 13 isolates (1.8%) had dolutegravir FC >25. This was interpreted as reflecting prompt switching away from raltegravir in patients who developed early resistance on raltegravir-based combinations who did not achieve viral suppression to <50 copies/mL.
Most of these isolates retained sensitivity to dolutegravir with significantly reduced FC sensitivity to dolutegravir compared to raltegravir, but this was based on definitions for primary and secondary mutations that have since been updated based on the VIKING-3 results described above.  This study found a similar correlation to dolutegravir sensitivity based on presence of Q148 and contribution of additional changes at G140, E138 and L74.
In addition to identifying the role of key integrase mutations more accurately and the likely response to dolutegravir, these results further caution against continuing to use raltegravir or elvitegravir in combinations that are not maximally suppressing viral load. Continued use increases the risk or more complex integrase resistance that will limit the option to use dolutegravir.
All references are to the International Workshop on HIV and Hepatitis Virus Drug Resistance and Curative Strategies, 4-8 June 2013, Toronto.
- Vavro C et al. Integrase genotypic and phenotypic predictors of antiviral response to dolutegravir (DTG) in subjects with resistance to integrase inhibitors (INIs). Oral abstract 29.
http://www.informedhorizons.com/resistance2013/pdf/Presentations/Vavro.pdf (PDF slides)
- Vavro C et al. Evolution of integrase in virus at protocol-defined virologic failure from the VIKING-3 study. Poster abstract 81.
- Underwood M et al. Analysis and characterization of treatment-emergent resistance in ART-experienced, integrase inhibitor-naïve subjects with dolutegravir (DTG) vs raltegravir (RAL) in SAILING (ING111762). Oral abstract 21.
http://www.informedhorizons.com/resistance2013/pdf/Presentations/Underwood.pdf (PDF slides)
- Underwood M et al. Epidemiology of dolutegravir (DTG) resistance in ~700 RAL-resistant isolates. Poster abstract 85.