C11 Managing treatment failure


Although the very first antiretroviral regimen for a treatment-naive patient is designed for the long term, changes may be necessary with time as indicated by clinical needs or simply patient preference. With the current generation of drugs, virologic suppression is now almost taken for granted. This is followed by immunologic and clinical recovery. As a result, most switches of therapy in clinical practice are now based on toxicity profile, drug interaction, or treatment simplification.

Nevertheless, treatment failure does occur for a variety of reasons. When active viral replication is allowed to occur in the presence of drug exposure, drug resistance will emerge and become irreversible as ‘memory’ retained in the form of archived mutations in the host chromosome of latently infected cells. All treatment failure has to be taken very seriously and evaluated by experienced HIV physicians.

Reasons of treatment failure

Despite an expanding portfolio of antiretroviral drugs, the basic tenet of treatment remains one of continuously halting viral replication. Other than preventing continued damage to the immune system, this will forestall the development of resistance as well. In this environment, immune activation also decreases and recovery follows. A most convenient marker is a rise of CD4 cell count. The CD4:CD8 ratio will also improve but is rarely restored to >1.0 unless treatment has been started early and continued for long periods of time.[1]

Yet, highly active antiretroviral therapy (HAART) as we know it may not be that potent. With very sensitive assays, residual viraemia is still detectable in most patients. Even if control of viral replication is absolute, release of virus from latently infected cells continues, ready to rekindle viral replication when given the opportunity. Inadequate HAART will therefore quickly lead to failure. The reasons for inadequate treatment may include:

  • Suboptimal antiretroviral combination – antiretrovirals are not identical. They differ in potency, genetic barrier for resistance and tolerance.
  • antagonism between some drug combinations.
  • inadequate delivery – non-adherence, malabsorption, vomiting, pharmacokinetic interaction.
  • excessive metabolism – individual variation, pharmacokinetic interaction.
  • pre-existing, archived resistance that fails to be detected in testing.

At the clinical level, it is known that response to HAART is often poorer in those patients with advanced disease, treatment experience high baseline viral load (VL) or suboptimal drug adherence. In the early HAART era, treatment response was usually lower in the real world than in clinical trials. However, this gap has largely been closed with the recent generation of antiretrovirals.

Defining treatment failure

Strictly speaking, treatment failure may be virologic (failure of viral suppression or rebound of plasma VL), immunologic (low or falling CD4) or clinical (e.g. progression to new AIDS defining illness). And they typically occur in this order if alternative, effective antiretroviral therapy is not quickly given Algorithm 11 summarises the approach to treatment failure following HAART.

Virologic failure – Today, VL assays typically have a limit of detection at 50/mL or lower. Nevertheless, it is generally agreed that viraemia has to exceed 200/mL to lead to the development of resistance during treatment.[2] As such, the following scenarios may be described:

  • Viral suppression – VL below the limit of detection, usually <50/mL
  • Incomplete viral suppression – VL ≥200/mL at 6 months into initiation of HAART
  • Virologic failure – failure to maintain suppression of VL to <200/mL
  • Viral rebound – VL ≥200/mL after initial virologic suppression
  • Low level viraemia – detectable viraemia <200/mL
  • Viral blip – isolated, detectable viraemia <500-1000/mL

These descriptions are not stringently defined and may vary across studies and different international guidelines.

Immunologic failure – Immunologic failure is broadly understood as an unsatisfactory CD4 count. It may be defined as a failure to rise above a significant level, say 200/μL, or a drop toward the pre-therapy baseline. Of note, patients may respond immunologically even if viral suppression is incomplete. However, mutations will accumulate, eventually causing a rise in viral fitness and replication, and ultimately loss of CD4 cell. Until then, a temporary discord exists between virologic failure and immunologic response.

Clinical failure – It is the occurrence or recurrence of AIDS-defining events after at least three months of treatment, with the exception of immune reconstitution inflammatory syndrome (IRIS).[Chapter C17] This definition therefore excludes residual immune deficiency that persists in the early period of antiretroviral therapy. Clinical failure usually follows virologic and immunologic failure in that order. In fact, even successfully suppressed HIV infection continues to have a negative impact beyond traditional HIV associated diseases, such as in cardiovascular diseases [Chapter C15] and non-AIDS malignancies.[Chapter E32] However, this is not regarded as clinical failure.

Evaluation of virologic failure

Clinical evaluation

Virologic failure precedes clinical failure. It should therefore be taken seriously. Expeditious evaluation should be made of possible non-adherence and factors that may have contributed to it: adverse effects of drugs, conflict with a patient’s lifestyle, pill burden, use of recreational drugs, etc. History taking should be thorough and yet non-accusatory, as a compassionate attitude is a pre-requisite to obtain truthful information. Instead of asking ‘did you miss any dose?’ one may tactfully ask ‘have there been times that you could not take your medicines as prescribed?’ There is no standard measure of adherence.[Chapter C12] A patient’s self-report is as good as any.[3]

Concomitant medications such as tuberculosis treatment have to be reviewed for change. Chinese herbs and health foods may be important and should be avoided if there is doubt. The antiretroviral combination in use should be reassessed for adequacy. Past experience with antiretrovirals should also be examined for possible resistance. This was exemplified in a clinical trial in which patients with undetectable viral load were switched from Kaletra® to raltegravir (RAL). Excess failure occurred, especially in those whose Kaletra® regimen was not their first treatment.[4] Last but not least, evaluation has to be made of new development of HIV-related complications.

Laboratory investigations

Viral load – Repeated VL measurements are necessary during HIV treatment. The first one should be done 4-6 weeks after treatment initiation. The interval between measurements can be gradually lengthened to every 6 months, but only if the patient maintains suppression and shows good adherence and tolerance to the regimen. Virologic failure has occurred if the VL rebounds to >200/mL. Viral blips can only be diagnosed by hindsight after it has spontaneously resolved. Blips are more common in those with high pre-treatment VL. However, those that are frequent or ≥500/mL, and persistent low level viraemia may predate virologic failure and should be evaluated as such.

Although viral suppression is expected to occur by six months, a high pre-treatment viral load >100,000-500,000/mL may preclude this from happening despite perfect adherence to a potent regimen. In this case, the slope of decrease could be reassuring as the viral load should have decreased by ≥1 log at 4 weeks. Viral decay will be quicker with integrase strand transfer inhibitor (INSTI).

CD4 count – For patients without viable treatment options, both the trend and the current level of CD4 count may be important in the timing of changing to a new treatment regimen. CD4 count also indicates the need to start or re-start prophylaxis against opportunistic infections for all patients. However, CD4 count generally drops only after virologic failure. Therefore, it does not have to be measured more frequently than annually if the last CD4 count was higher than 250/μL and viral suppression has been well maintained, unless intercurrent conditions or treatment is expected to affect its level.

Resistance test [Chapter C13] – Not all failures are due to drug resistance. Yet, on its occurrence, the very approach to virologic failure will immediately become one of aggressive change of therapy. Resistance testing by the genotypic methodology has now become the accepted standard of care. Phenotypic tests are expensive and have a long turnaround time. Their use is now limited to complex cases of failure with limited options. Resistance testing should be done when the patient is still on the failing treatment or at least within 4 weeks of stopping it. A viral load of ≥500/mL is generally required for successful testing, although lower levels may still work. Identified mutations will help exclude certain drugs to be used, but absence of mutations does not guarantee success.

Tropism test [Chapter C13] – This should be done when maraviroc (MVC) is contemplated as an option in the next treatment regimen. As a CCR5 coreceptor antagonist, MVC will not be effective against viruses that use CXCR4 for entry. The current gold standard of tropism testing is the phenotypic Enhanced Trofile® test which is validated to detect dual/mixed X4 variants as low as 0.3% of the viral population. Genotypic testing of the V3 region is also acceptable. It is specific but not as sensitive as Trofile®. It is thus reasonable to reserve Trofile® for those with only R5 viruses detected in the genotypic test.

For those who are failing on a MVC-containing regimen, the emergence of X4 variants rules out the continued use of MVC. For those having only R5 viruses, resistance to MVC may still exist as it can be mediated by mechanisms other than tropism shift.

Therapeutic drug monitoring (TDM) [Chapter C13] – Where drug interaction is suspected to be a cause of treatment failure, TDM can be useful. Non-adherence may also be captured by TDM. Especially for protease inhibitors (PI), resistance may be relative and can be overcome by a higher level of plasma concentrations. By relating the plasma level to the IC50 of the resistant virus (the inhibitory quotient), simply increasing the dosage may enhance control of the resistant virus.[5] On the other hand, considerable inter-individual variations in the pharmacokinetics of antiretrovirals occur. An unusually high Cmax may explain some of the intolerable adverse effects and inform the physician how to adjust the antiretroviral regimen to better adherence. However, recommendation cannot be made to lower drug dosage to below standard because of high drug levels.

The different scenarios

A synthesis of information gained from the clinical evaluation and laboratory investigations allows one to reasonably explain the failure and hopefully devise an appropriate change of therapy:

Non-adherence to an otherwise adequate HAART without resistance – In the ideal world, this would have been preventable if a patient’s preference and lifestyle were taken into account in designing the initial regimen. But non-adherence continues to occur in spite of the best intentions. Not uncommonly, the dosing frequency and pill burden are a factor and need to be addressed. Adverse effects such as diarrhoea, nausea, or jaundice may also be important and should be proactively managed with symptomatic treatment and adequate warning. Certain patients may be habitual users of recreational drugs and alcohol, which will make it difficult for them to even remember taking the drugs. Referral to expert help should be considered. Although resistance may not be present in testing, the physician should strive to reasonably conclude that this is the case by ensuring that the resistance testing is done in a timely manner, and repeating the test on archived or new samples if in doubt. It is important that the patient’s adherence be improved as far as possible. This can be achieved by appropriately addressing the reasons behind non-adherence and continuing the same regimen. Alternatively, one may substitute components of therapy that are hindering adherence.

Non-adherence with development of resistance – Change of therapy in this circumstance should be prompt, addressing resistance as well as non-adherence. Procrastination will risk extension of resistance to other agents of the same class. In first treatment failure, adequate treatment options are usually available. However, the patient will have to understand that second line therapy and beyond is usually not as tolerable. If the CD4 count is high, a temporary drug holiday before change to the new therapy is acceptable. This will allow the patient and his care provider to work out issues hindering adherence.

Common patterns of resistance – In Hong Kong, antiretroviral use in the past was characterised by a combination of lamivudine (3TC), thymidine analogues and first generation non-nucleoside reverse transcriptase inhibitors (NNRTI) or PI. Later, the thymidine analogues, zidovudine (ZDV) and stavudine (d4T), were replaced by abacavir (ABC) and tenofovir with a different resistance profile. Against this background, most resistance issues are related to TAMs (thymidine-associated mutations), M184V, K65R, and mutations related to NNRTI at 100, 103, and 181. PI-associated mutations are relatively uncommon and usually few when they do occur. In recent years, INSTI-based treatment has gained popularity. Primary transmitted INSTI resistance is still rare, but resistance by way of pathways including Q148H or N155H have arisen where adherence was poor with RAL and elvitegravir (EVG), the two INSTI with a low genetic barrier.

The longer a patient has remained on a failing regimen, the more likely further resistance-associated mutations will occur and confer a higher and broader level of resistance. This is particularly true with NNRTI-based regimen.

Drug interactions with or without resistance – In Hong Kong, consumption of Chinese medicine is common and not often regarded as drugs by patients. History taking should be specific and TDM performed if necessary. Pharmacokinetic properties of most Chinese herbs are not known and, if suspicious, patients should be advised to stop them. Multiple medicines also interact with antiretrovirals. Anti-tuberculosis treatment, methadone, and macrolides are some of the common medications that patients with HIV infection may take concomitantly. Standard recommendations on dosage adjustment should be supplemented by TDM whenever there is doubt, such as when the patient is over or under-weight, treatment response has been unsatisfactory, or multiple concomitant medications are being taken.

Changing treatment in light of resistance

Resistance tests should be done for all VL >500/mL while on the failing therapy. Archived blood samples will have to be retrieved if pre-treatment resistance testing has not been performed. Ideally, treatment should be changed to three new drugs to which current therapy has no cross resistance. If not, a new regimen should have at least two active agents according to resistance test (i.e. a genotypic or phenotypic sensitivity score of ≥2), and with different mechanisms of action. Do not sequentially add one active drug to a failing regimen.

Cover suspected as well as known resistance. Accordingly, a detailed antiretroviral history should be taken to infer mutations that may have back-mutated or become minor species that failed detection. In particular, previous use of pre- and post-exposure prophylaxis should be noted.[Chapter B8] An accurate history of treatment in overseas countries which use generic compounds can be challenging but is important to obtain.

If choices exist with new treatment options that circumvent existing resistance, those with a high genetic barrier are preferred. Attempts should also be made to select drugs that address the root cause of the culpable non-adherence, be it pill burden, intolerance, or food requirements that are difficult to follow.

Failing an NNRTI-based regimen – considerable overlap exists between efavirenz (EFV) and nevirapine (NVP) in terms of resistance. Although cross resistance is less extensive with etravirine (ETR) and doravirine (DOR), they cannot be counted on to replace the failing NNRTI. Even if absent from resistance tests, resistance to lamivudine should be suspected especially if it has been given together with the NNRTI for a long time.

A change to either a boosted PI-based or INSTI-based regimen is possible, such as that based on boosted darunavir (DRV), dolutegravir (DTG) or bictegravir (BIC). The backbone of two nucleoside reverse transcriptase inhibitors (NRTI) should contain at least one fully active component. Using a boosted PI in combination with an INSTI may be considered if extensive resistance to NRTI exists, but adherence has to be carefully supported.

Failing a PI-based regimen – resistance is generally slow to develop, especially to the PI component as long as it is boosted. Adherence has to be re-examined and encouraged in this circumstance. To this end, it may require a change of PI for simpler dosing, e.g. from twice a day Kaletra® to once-daily DRV, or for adverse effects, e.g. from atazanavir (ATV) to DRV to avoid jaundice. Alternatively, a change to the better tolerated DTG can be considered to encourage adherence. This becomes advisable if resistance mutations to PI begin to accumulate. In the meantime, it is important to maintain an NRTI backbone with at least one active agent.

Failing an INSTI-based regimen – RAL and EVG may result in resistance when adherence is suboptimal. It is a natural decision to switch to a PI-based regimen with at least one active NRTI in its backbone. An alternative is to switch to a new regimen based on twice-daily DTG if the resistance pattern suggests effectiveness.

Multi-drug resistance – The advent of first generation HAART in 1997 was followed by development of triple-class resistance in some patients. Non-adherence, contributed by the relative toxicity of those drugs and their large pill burden, was the major culprit. It used to be referred as a salvage situation. In this event, it was very difficult, if not impossible, to construct a regimen that could reliably suppress viral load. That was until the appearance of new classes of drugs. In one clinical trial, 90% of patients with triple class resistance were able to achieve virologic suppression by a combination of RAL, boosted DRV, and ETR.[6]

Despite the potency of new generation antiretrovirals, true salvage situations do rarely occur where resistance has extended to INSTI. Almost all such cases are due to inappropriate sequencing and prolonged non-adherence of antiretrovirals. With these patients, management is the domain of experienced HIV physicians. Very often, complicated regimens using a combination of four or more drugs will be required. Phenotypic resistance tests and TDM will be leveraged to make the maximal use of partially active drugs by increased dosing. One may also have to resort to clinical trials of investigational drugs or compassionate access for more novel drugs such as ibalizumab, a CD4 receptor antagonist, and albuvirtide, a fusion inhibitor, both of which are licensed overseas but not available in Hong Kong.

In a deep salvage situation where there is no effective HAART, complete virologic suppression is no longer realistic. Therefore, treatment goals should be realigned to preservation of CD4 count and prevention of clinical deterioration. In this vein, partial viral suppression becomes the immediate objective of therapy. Its rationale is based on impaired viral fitness with resistance mutations. This ‘cost’ to the virus is particularly relevant with NRTI mutations. For instance, M184V, the signature mutation to 3TC and emtricitabine (FTC), impairs RT processivity and hence viral fitness. Continuing an NRTI backbone is therefore desirable. Boosted PIs in general have a high genetic barrier and corresponding mutations undermine viral fitness. They should be continued. On the other hand, mutations to NNRTI and INSTI do not significantly impair fitness. Their continuation may actually allow mutations to accumulate and affect susceptibility to future drugs of the same class. Similarly, if X4 virus emerges with the use of MVC, it should be discontinued. It is unclear if X4 will revert to R5 as a result.

Immunologic and clinical failures

Most immunologic failure follows virologic failure. However, isolated immunologic failure sometimes happens in which VL is suppressed but the CD4 count fails to rise. It is important to rule out laboratory and clerical error in processing the blood samples for viral load. Coinfections, some concurrent treatment, old age, or the development of new HIV-related and unrelated clinical disease may also blunt or reverse the immunologic response to antiretroviral treatment. Tenofovir+didanosine (ddI) as a backbone of HAART is also known to have a negative impact on immunologic recovery.

In most cases, the phenomenon cannot be explained. Watchful followup is therefore the only recourse while observing the usual prophylactic treatment according to CD4 thresholds. Most of these patients do well, but some fail to recover from an HIV-related complication as long as the CD4 count remains low. Specific treatment such as IL-2 has been tried. It elevated the CD4 count but otherwise had no impact on clinical outcome.[7] Certain antiretrovirals, such as PI, ABC and tenofovir, were associated with a better immunologic recovery, but in all likelihood the magnitude of effect is too small to be meaningful.

On the other hand, IRIS occasionally occurs in patients who have apparently responded to HAART, and yet deteriorates clinically. It is not considered clinical failure. Nonsteroidal anti-inflammatory drugs and systemic steroids may be required in addition to specific treatment. HAART is generally continued in IRIS unless life-threatening disease occurs. True clinical failure may develop despite virologic success and immunologic response. Certain HIV-related complications, such as lymphoma, are now occurring at higher CD4 strata than in the pre-HAART era. This is probably related to the fact that immunologic recovery from HIV-mediated damage is incomplete despite viral suppression and a good CD4 count.

Algorithm 11. Approach to treatment failure

Algorithm 11. Approach to treatment failure


  1. Lee SS, Wong NS, Wong BCK, Chan KCW. Combining CD4 recovery and CD4:CD8 ratio restoration as an indicator for evaluating the outcome of continued antiretroviral therapy: an observational cohort study. BMJ Open. 2017 Sep 15;7(9):e016886. link
  2. Esber A, Polyak C, Kiweewa F, Maswai J, Owuoth J, Maganga L, Adamu Y, Hickey P, Ake J, Crowell TA. Persistent low level viremia predicts subsequent virologic failure. Is it time to change the 3rd 90? Clin Infect Dis. 2018 Nov 20. link
  3. Almeida-Brasil CC, Moodie EEM, Cardoso TS, Nascimento ED, Ceccato MDGB. Comparison of the predictive performance of adherence measures for virologic failure detection in people living with HIV: a systematic review and pairwise meta-analysis. AIDS Care. 2018 Dec 5:1-13. link
  4. Eron JJ, Young B, Cooper DA, Youle M, Dejesus E, Andrade-Villanueva J, Workman C, Zajdenverg R, Fatkenheuer G, Berger DS, Kumar PN, Rodgers AJ, Shaughnessy MA, Walker ML, Barnard RJ, Miller MD, Dinubile MJ, Nguyen BY, Leavitt R, Xu X, Sklar P; SWITCHMRK 1 and 2 investigators. Switch to a raltegravir-based regimen versus continuation of a lopinavir-ritonavir-based regimen in stable HIV-infected patients with suppressed viraemia (SWITCHMRK 1 and 2): two multicentre, double-blind, randomised controlled trials. Lancet 2010;375(9712):396-407. link
  5. Morse GD, Catanzaro LM, Acosta EP. Clinical pharmacodynamics of HIV-1 protease inhibitors: use of inhibitory quotients to optimise pharmacotherapy. Lancet Infect Dis 2006;6(4):215-25. link
  6. Yazdanpanah Y, Fagard C, Descamps D, Taburet AM, Colin C, Roquebert B, Katlama C, Pialoux G, Jacomet C, Piketty C, Bollens D, Molina JM, Chêne G; ANRS 139 TRIO Trial Group. High rate of virologic suppression with raltegravir plus etravirine and darunavir/ritonavir among treatment-experienced patients infected with multidrug-resistant HIV: results of the ANRS 139 TRIO trial. Clin Infect Dis. 2009;49(9):1441-9. link
  7. Onwumeh J, Okwundu CI, Kredo T. Interleukin-2 as an adjunct to antiretroviral therapy for HIV-positive adults. Cochrane Database Syst Rev. 2017 May 25;5:CD009818. link

Further reading

  1. Panel on Antiretroviral Guidelines for Adults and Adolescents. Guidelines for the Use of Antiretroviral Agents in Adults and Adolescents Living with HIV. Department of Health and Human Services. Available from link
  2. European AIDS Clinical Society. Guidelines, version 9.1,2018 Available from link