9. Appendix 8: UK Resistance Guidelines (BHIVA) summary

5.16.4 Which method should be used?

Current HIV-1 genotyping assays mostly rely on dideoxynucleotide sequencing using the Sanger method that allows detection of mutations when present at a level of 20–25% of the virus quasispecies population.

There are assays available based on sequencing the pol (encoding the RT, PR, IN proteins) and envelope genes as well as genotyping tests predicting co-receptor usage.

A variety of methods is commercially available in the UK but in-house assays are also widely used. It is critical for laboratories performing resistance testing to be accredited, participate in a variety of quality assurance schemes, ensure tight quality-control programmes within the laboratory, and make the results of their performance available to users.

Detection of minority variants by deep sequencing is becoming available in many laboratories; however, reporting of those is not yet part of standard clinical algorithms. More laboratories will have validated next-generation sequencing methods in the near future and only tests validated by UKAS accredited laboratories should be used. In the absence of clinical cut-off data all minority variants should be interpreted with extreme caution and should be discussed with an expert before being acted upon (2B). Next-generation sequencing methods seem more sensitive in predicting CCR5 co- receptor antagonist failure (2B).

Phenotyping assays used to predict response to ART in cases of complex resistance profiles are not readily available (see section 5.16.5)

5.16.5 How should results be interpreted?

There is a variety of online tools available that aid interpretation of HIV-1 genotyping results. The Stanford database (http://hivdb.stanford.edu), ANRS (http://www.hivfrenchresistance.org) and REGA (http://rega.kuleuven.be/cev/regadb/download) are the most commonly used data-interpretation systems. Additionally, the International AIDS Society (IAS-USA) provides updates on significant drug resistant mutations.

• HIV physicians should be familiar with these online tools as well as with the basics of result interpretation and clinical significance of most common mutations.
  Specialist virology input by a clinical virologist is necessary for interpretation of complex patterns of mutations, for selection of alternative assays and tests in clinical dilemmas and ensuring quality control is clinically adequate in the diagnostic services responsible for resistance testing (1B).
• When a resistance test is performed by sequencing HIV-1 RNA, the subtype information is also automatically provided in the resistance report. Subtyping is of limited direct clinical use for managing the individual patient although it might aid interpretation of sequencing results. It provides, however, important epidemiological information.

Evidence

Testing at baseline: performing a genotypic resistance test (sequencing of pol gene that encodes RT and PR proteins) at time of diagnosis/entry to care is recommended for people diagnosed with primary HIV infection as well as those with chronic infection by all clinical guidelines, although implementation capacity differs in different parts of the world [1-3]. There is consensus on discriminating TDR mutations as opposed to naturally occurring polymorphisms [4] and the estimated prevalence of TDR mutations in the UK remains stable at 10%. Recent evidence suggests that it is self-sustained and sequences with TDR mutations are mainly derived from people who are treatment-naïve, the main group of people contributing to this phenomenon [5].

Testing needs to be performed on the earliest available sample close to time of infection, since TDR mutations can disappear (revert back to wildtype) over time [6]. However, specific TDR mutations can be detected years after transmission [6-9]. This is particularly true for thymidine analogue mutations (TAMs) like M41L and T215F/Y, as opposed to K65R and M184V and mutations conferring resistance to non- nucleoside reverse transcriptase inhibitors (NNRTIs) and protease inhibitors (PIs). Persistent revertants therefore, such as those containing M41L are observed more frequently [10,11]. Detection of TDR minority variants by more sensitive next-generation sequencing (NGS, detecting variants present in frequencies as low as 0.1–1%) has been shown to predict a higher risk of virological failure with low genetic barrier drugs such as NNRTIs [12,13]. Clinical thresholds have not yet been established, but an ART regimen with a high barrier to resistance (e.g. a boosted PI regimen) should probably be selected when low minority variants compromising susceptibility to NNRTIs are detected in a validated assay. In a recent large meta-analysis NNRTI-associated TDR was associated with cases of high-level resistance [14]. The impact of TDR to NRTI use in therapeutic schemes containing INSTIs is not known. High genetic barrier agents like boosted PIs seem unaffected by TDR. Finally, there is currently no evidence of circulating transmitted drug resistant INSTI mutations [15]. As the use of INSTIs becomes widespread, this position might need to be reviewed and sequencing of the integrase gene might be required.

Testing at virological failure: the understanding that viruses and in particular RNA viruses exist in the infected host as quasispecies (from the Latin quasi – as if; almost) provides an important conceptual background in explaining resistance in virological failure [16]. HIV-1 replication is associated with a high mutation rate as the RT lacks proofreading capacity, leading to errors during replication and generation of a swarm of genetically distinct viruses. Genetic recombination when viruses infect the same cell, and proviral variants accumulated over time, also contribute [17,18]. When adherence to ART is suboptimal there is ongoing viral replication under drug pressure; selection of fit minor variants bearing drug resistance mutations is then possible and these can become dominant, leading to virological failure. There are several studies exploring the clinical utility of genotyping [19,20] and resistance testing at failure is a well- established practice in resource rich settings [1,2]. A public health approach, however, has been adopted in parts of the world with high prevalence [3,21] and new modelling approaches have been explored as an alternative to genotyping testing [22]. Modern ART is highly effective and new compounds are becoming available allowing more choice at failure. New patterns of resistance are however noted in clinical trials and in vitro. Therefore, resistance surveillance and database schemes need to be continuously updated.

Boosted protease inhibitors (bPIs) represent a potent class of ART. Mutants resistant to bPIs even in people failing therapy are rather rare. Emerging evidence suggests that sequencing of genes not currently part of standard testing algorithms like the gag and envelope (in particular the part of the genome encoding the cytoplasmic tail of gp41) might be required and mutations detected in cleavage sites (CS) in particular might be detrimental in virological failure experienced by people on ritonavir-boosted protease inhibitors [23,24]. After release from the host cell membrane, the viral protease (PR) cleaves the gag and gagpol precursor proteins to form the mature, infectious virion and the structural proteins are produced: matrix (MA), capsid (CA), nucleocapsid (NC), p6, and spacer peptides p1 and p2 [25,26]. HIV-1’s PR also processes the gagpol precursor polypeptide, releasing the PR, integrase (IN), and reverse transcriptase (RT) enzymes. Indeed, multiple mutations within all gag and spacer proteins have been linked to PI exposure, reduced susceptibility, and resistance [27]. The high genetic variability observed in the gag, however, translates to the need for comparison of individual patient gag sequences between baseline and failure.

Tropism assays can be performed on peripheral blood mononuclear cells (PBMCs) if the viral load is less than 500 copies/mL when maraviroc switch is considered. Most laboratories prefer to perform it on the last stored plasma (VL >500 copies/mL) because of technical difficulties of working with PBMCs and some concerns about the sensitivity of PBMCs in picking up minority X4 virus [28]. NGS genotypic tropism tests have been shown to have better sensitivity at predicting maraviroc virological failure due to minority X4 tropic virus and it would be the preferred test should a validated and accredited test be available [29]. Phenotypic tropism assays are not readily available the UK. Upon virological failure around two-thirds of people will have a tropism switch from R5 to X4 and therefore the tropism test needs to be repeated [30].

A resistance test should be performed at the time of virological failure and preferably within 4 weeks of stopping ART. Most laboratories will attempt resistance testing on low viral loads (down to 200 copies/mL, some even lower), but the risk of amplification bias is increased at viral loads below 1000 copies/mL. Not detecting drug resistant mutations does not exclude their existence. In the UK, laboratories test using population (Sanger sequencing), however, next-generation sequencing methodologies are becoming available. NGS emerges as a powerful new tool in managing people with HIV [34] as well as providing unique insights in the pathogenesis and epidemiology of the infection [35]. NGS does not increase the sensitivity for detecting mutations at low viral loads and suffers the same amplification bias problem. NGS however has the potential to detect minority drug resistant variants where people have stopped their medication and wildtype virus has become the dominant virus with viral rebound, i.e. high viral loads. The multiple technical considerations, cost efficiency issues and more importantly clinical utility considerations will probably be overcome as these technologies advance [36], laboratories are centralised and our knowledge about the impact of minority variants is enhanced [37,38].

Phenotyping assays measure the ‘fold resistance’: this is the ratio of the IC50 (drug concentration that inhibits 50% of patient virus population) to IC50 of reference strains. The cost, turnaround time and limitations in clinical interpretation explain why these assays are not widely available.

Result interpretation of genotyping testing relies on tools available online that are based on large databases of sequences. Quality control systems and ongoing evaluation of these data interpretation tools especially as larger more complicated NGS data become available is of paramount importance [38].