The 15th HIV Drug Resistance Workshop took place in Sitges, Spain, from June 13th to 17th. What were the key take-home messages?
New and old epidemics: the evolving HIV epidemiology
Take home message: There are two parallel epidemics in HIV. The old epidemic consists of patients historically infected and treated with sequential mono and dual therapies. The new epidemic reflects individuals newly diagnosed, 10% of whom are likely to have transmitted resistance and increasing numbers are infected with non-B subtype virus. Whilst the goals for therapy remain the same, the needs of these two population groups may not be directly comparable. These differences should influence messages delivered to drug developers, regulatory bodies, patients and their advocates.
One of the most fascinating discussions at the workshop was the definition of old versus new HIV epidemics. A distinction astutely observed by Dr Charles Boucher from Utrecht University Medical School and rapidly integrated into subsequent discussions. Two distinctions are notable here between the new and old epidemics and two reasons why it may have important implications for patients.
Firstly, the old epidemic is marked by sequential mono and dual-therapy. The residual resistance profile will likely reflect numerous mutations historically archived and replenishing dominant viral populations in circulation. Secondly, this could severely limit future options for switching strategies. Finally, without the constant supply of new drugs with novel mutational profiles, constructing effective rescue therapies will continue to challenge physicians and patients.
Why is it important to distinguish between these two epidemics? Essentially, the implications are important for two groups; drug developers and newly diagnosed patients. Firstly - and quite rightly, the threshold for licensing new drugs continues to rise. Doctors, patients and their advocates insist that pharmaceutical companies develop drugs that are safer, more tolerable, easier to store, simplified regimens with less pills, taken less often, free from food restrictions and as budgets tighten, relatively inexpensive. This demand will not change. The simultaneous pressure for drugs with novel resistance profiles whilst understandable and needed for patients with limited options may not continue to apply as the epidemic of new diagnoses evolves. With improved drugs and greater choice, regimens can now be constructed to sustain a durable response for a number of years.
Secondly, new infections in Europe, even in the US and the burgeoning epidemic in the developing world consist of patients infected with subtypes other than clade B. With increased migration, mobility of populations and scale-up of therapy in resource-poor countries, the need for drugs and diagnostics that are developed, tested and effective in non-B subtypes is paramount. Drug developers can no longer test their drugs in different groups as a postscript to their pivotal, registration studies. The new epidemic is increasingly heterosexual, non-B, poor and lacking treatment experience. The advocacy movement, scientific community and regulatory authorities need to reflect on these dichotomous demands and sharpen their messages so that the needs of patients from both epidemics are equitably reflected.
The (f)utility of resistance testing
Take home message: Resistance testing in patients failing therapy often serves to confirm rather than change decisions regarding switching treatments. Paradoxically, treatment naïve patients may benefit more from resistance testing. Testing at baseline to detect transmitted and archived resistance can improve selection of effective first-line regimens.
Andrew Zolopa from Stanford University was responsible for the most provocative review at the workshop. In an enlightening overview, he critically examined current dogmas asking:
- Do we still need resistance testing – and what’s its purpose?
- Why are we testing individual drugs when it’s the activity of the regimen that matters?
- Is resistance testing still useful for treatment-experienced patients?
The answers may be complex yet a tentative conclusion seems to be emerging, he argued. The utility or in many cases, the futility of resistance testing was described by Dr. Zolopa in the scenario of patients with prior treatment experience. Experienced physicians with patients on long-term treatment and knowledge of a range of HIV therapies may in fact use resistance tests simply to confirm rather than dictate their decisions for switching drugs. Yet as Bob Schafer also from Stanford noted, very few interpretation systems take into account patient treatment histories and advise accordingly. For patients in deep salvage, it could be argued that resistance testing provides little additional value when limited choices are available.
But instead of calling for a moratorium on resistance testing, what is being called for here are improved diagnostic techniques and better interpretation systems that can directly correlate with clinical scenarios. Precision in the tools, both diagnostically and interpretation is what is needed.
Paradoxically, noted Dr Zolopa, resistance testing may actually prove more useful in treatment naïve patients. Sensitive testing can alert to transmitted mutations that may be archived and persist as minority populations. Resistance testing at baseline can also help to improve public health surveillance and inform treatment decisions for effective and durable therapy.
Why does low-level drug resistance matter?
Take home message: Low level viremia is persistent and perhaps prognostic. It can represent pre-therapy resistance and endure for as long as 7 years. Even on effective suppressive therapy, minority variants continue to escape and repopulate the majority viral population and potentially comprise response to therapy.
Does low level drug resistance matter? Well, there are two schools of thought; both agree that viremia persists at low levels even on effective ‘suppressive’ therapy however its clinical significance remains inconclusive. Persistent viremia is the result of a biphasic phenomenon; viral populations that manifest from the pre-therapy era are distinct from resistance escape that occurs whilst on therapy.
The first phase was described by Sarah Palmer (see abstract 55) at the US National Cancers Institute (NCI) as a long-term event. Their research confirmed the presence of long-lived mutations that originate pre-therapy and persist for as long as seven years even for patients who subsequently start HAART.
Francois Clavel from Claude-Bichat Hospital in Paris defined the secondary population as quasi-species that are constantly released into plasma repopulating dominant populations, even under potent drug pressure. Minority variants with drug resistance mutations are significant, he added, since they can serve to lower the genetic barrier allowing for rapid evolution of high-level resistance. Low-level resistance is currently missed by commercially available assays since the threshold for amplification can be unreliable at lower viral loads - the answer, noted John Mellors from Pittsburgh University, is to improve the performance of existing tests, making it possible to measure accurately at lower viral thresholds.
When is resistance not resistance or what’s the point of cut-off values?
Take home message: Cut-off values in phenotypic resistance tests are important, they confirm resistance or susceptibility - but still the numbers themselves remain relatively arbitrary. There are three caveats for determining the significance of cut-off values: (1) variations exist in reporting fold-change (FC) between different interpretation systems for the same drugs, (2) adjustments are needed for other drugs in the regimen and (3) current cut-offs are based on responses expected from treatment-naïve patients, so for treatment-experienced individuals consider lower FC values as more relevant.
What is a relevant cut-off to measure resistance? Cut-off values are expressed as fold-changes in susceptibility (e.g. a five-fold reduction in susceptibility may be considered the boundary between activity and intermediate resistance for a drug).
Cut-off values are derived from a correlation of phenotypic resistance observed in clinical isolates when compared to wild-type viruses, but how useful are these numbers for clinical decision-making?
Dogmatic faith in the numbers proffered has persisted despite a number of caveats. These include the reproducibility and technical limitations of assays and the assumption that values reliably reflect responses that can be expected in treatment-naïve patients.
But the phenotypic IC50 can be meaningless without an understanding of the patient’s treatment history, some level of familiarity with the available assays, an adjustment for other drugs in the regimen and good knowledge of the drugs under consideration.
However, the real conundrum lies in the subtle region between sensitivity and intermediate resistance. What for example, is the significance of a 2.1 fold-change (FC) compared to 2.4 FC? What should be done with results that lie tantalisingly close to break-points? Should such a result be seen to confer resistance, is it a forewarning of emerging resistance or an indication to wait and see?
Debates over the past few years had led to the alignment of cut-off thresholds for assays so that similar measures are used to define resistance or susceptibility for almost all drugs. Where do these numbers come from anyway, asked Dan Kuritzkes from Harvard University. It would appear that technical cut-offs are largely a product of FDA and EMEA regulations and reference is dutifully made to these in drug product labels.
On the other hand, clinical trial data have done much to enrich the utility of clinically meaningful cut-offs, such as data from ANRS cohorts and the recent RESIST study, although it must be said that endpoints, methodological differences and measures of adherence differ between studies making it difficult to establish any consensus across trials. Dr Kuritzkes reasonably suggested “let’s all agree on the values and stick to them”. He added “let’s assume that anything outside of the treatment-naïve range is likely to have a decreased response and define a lower cut-off beyond this threshold”.