6.7. When to Switch

– Christian Hoffmann –

Antiretroviral therapy has to be modified frequently, even though the rates of modification and interruptions have declined during recent years. In EuroSIDA, among almost 1200 patients who began ART after 1999, at one year after initiation, only 70% of patients remained on their original regimen. 24% had changed, and 6% were off all treatment (Mocroft 2005). In an evaluation of the Swiss cohort, 42% of 1318 patients beginning with ART between 2005 and 2008 had modified therapy after one year, 22% of them due to side effects (Elzi 2010). In general, ART is changed for three main reasons (interruptions will be discussed separately):

  1. Acute side effects
  2. Long-term toxicity (or concerns regarding them)
  3. Virologic treatment failure

Switching due to acute side effects

Not every acute side effect requires immediate modification. Mild nausea or diarrhea at the beginning can and should be tolerated. Gastrointestinal side effects that occur during the first weeks are not dangerous and often improve spontaneously or can be treated symptomatically. The same is true for some allergic reactions and for relatively mild CNS disorders. Talking with the patient, suggestions on how to tolerate or palliate certain problems with the idea that these will not continue indefinitely will help. However, certain adverse drug events almost always require discontinuation or changing of ART (see box).

Side effects that almost always require discontinuation/change of ART

  • Severe diarrhea, which persists despite loperamide even after several weeks (usually with nelfinavir, lopinavir/r, fosamprenavir/r, saquinavir/r)
  • Severe nausea, which persists despite metoclopramide, which requires continuous treatment or leads to significant weight loss (usually AZT, ddI)
  • Persistent sleeping disorder (efavirenz)
  • Polyneuropathy (d4T, ddI, possibly also 3TC) often resolves very slowly
  • Severe anaemia (AZT)
  • Severe, progressive muscular weakness (d4T, ddI)
  • Pancreatitis (ddI, ddI+TDF, d4T+ddI, in rare cases lopinavir/r)
  • Lactic acidosis (most often d4T+ddI, but also all other NRTIs)
  • Severe allergies with involvement of mucous membranes, fever (typically aba-cavir, all NNRTIs, more rarely fosamprenavir or darunavir)
  • Renal failure (tenofovir, indinavir), nephrolithiasis (indinavir)
  • Hepatotoxicity with transaminases >5 x normal values (nevirapine, tipranavir)
  • Jaundice (nevirapine, atazanavir, indinavir, tipranavir)
  • Rhabdomyolysis (raltegravir)
  • Severe repetitive onychitis (indinavir, possibly also 3TC)
  • Depression, psychosis (efavirenz, possibly also AZT)

Switching due to concerns over long-term toxicity

In the last few years, many clinicians have started to change virologically successful combinations out of concern for cumulative long-term toxicities, especially in cases of lipodystrophy and dyslipidemia. The switch strategy is based on the assumption that not all antiretroviral agents have similar toxicities. The most important switch studies are discussed below.

PI replacement with other agentsPIs may cause side effects in the long-term. Among these are lipodystrophy with abdominal and fat accumulation at the back of the neck, but also gastrointestinal side effects and dyslipidemia. The data on replacement of a successful PI with other classes, such as NNRTI, NRTI or more recently integrase inhibitors shows the following picture: replacement is virologically safe in most cases, provided viral load is constantly suppressed and no evidence of resistance exists (see Table 7.1).

Table 7.1. Randomized studies on switching from PIs to other drugs.



VL Effect

Effect of switch on lipids (L) or lipodystrophy (LD)
PI to  NVP
Barreiro 2000




L unchanged, LD better
Ruiz 2001




L possibly better, LD unchanged
Arranz-Caso 2005




L better, LD better
Becker 2001




L unchanged
Molina 2005




L/LD n.a., side effects similar
Clumeck 2001




L better, LD subjectively better
Opravil 2002




L better, LD unchanged
Katlama 2003




L better, LD better
Keiser 2002




L better
Negredo 2002




L only better on NVP, LD unchanged
Calza 2005




L actually worse if the PI arm contained lipid reducer
Martinez 2003



Trend against ABC

L only better on ABC, LD probably unchanged
Eron 2010Martinez 2010







L betterL better
In all studies (except Martinez 2003), randomization was against continuing PIs. All had an open-label design and all patients had been on PIs for several months at the time of the switch, with undetectable viral load. VL=viral load in the switch arm versus the continuing arm. Wk=weeks, LD=lipodystrophy, L=lipids, n.a.=not available, n.s.=not significant. *Here only 62% of patients were taking a PI, the rest were on NNRTIs or a triple nuke regimen.

Taken together, these studies show that lipid levels are most likely to improve after switching to other agents, in particular abacavir and raltegravir, and to a lesser extent, efavirenz. In cases of lipodystrophy the effects are clearly poorer. Quality of life and treatment satisfaction improved significantly in the switch arms of most studies, probably due to the reduced pill burden. A large study focused on investigating quality of life showed a clear improvement after switching from PIs to efavirenz (Campo 2010).

Switching from a PI to other drugs poses an increased risk of virologic failure, particularly with prior NRTI treatment and the associated resistance mutations. One example of what could happen when the drug is changed for strategic reasons is shown in Table 7.3. This case demonstrates how careful one must be when switching drugs, if there is a past history of inadequate treatment (i.e., dual therapy).

There is a risk of a higher virological failure when switching from PI based regimens to triple nuke, especially in patients with prior NRTI pretreatment. The SWITCHMARK trials showed similar results for the integrase inhibitor raltegravir (Eron 2010). In these two large-scale phase II studies, a total of 702 patients on a stable and functioning lopinavir-containing regimen were randomized to change to raltegravir or to continue with lopinavir. Lipids improved with the switch, however after 24 weeks a non-inferiority of raltegravir compared to lopinavir/r in efficacy was seen. In the ITT analysis, only 82% of patients on raltegravir compared to 88% on the continued PI maintained viral load below the limit of detection after 24 weeks. The viral load breakthrough applied especially for pre-treated patients with previous therapy failure (66%). A smaller, open-label randomized study in Spain, did not make the same observations, however. Patients were below detection for a longer period (Martinez 2010).

Potential side effects also need to be considered with every switch. Although less frequently than with treatment naïve patients, a rash or hepatotoxicity can be expected with nevirapine, and efavirenz may be associated with adverse CNS events. There is the risk of a hypersensitivity reaction with abacavir. There is still no data on a change or a PI substitution with maraviroc yet, but it is being investigated.

Switching to atazanavir

Possibly the PI does not always have to be replaced with another drug class. In cases of dyslipidemia switching to atazanavir could make sense as it is associated with a comparably good lipid profile (Gatell 2007, Soriano 2008, Mallolas 2009). Lipodystrophy and glucose metabolism may also be improved (Stanley 2009), although not endothelial function (Flammer 2010, Murphy 2010). There may be additional favorable effects on the lipids when atazanavir is unboosted and ritonavir is omitted, which seems to work well with pretreated patients with a viral load below detection (Sension 2009, Elion 2010, Ghosn 2010). Patients must be informed about the risk of jaundice, which is typical for atazanavir.

Replacement of thymidine analogs with other NRTIs

The thymidine analog d4T, which plays a leading role in mitochondrial toxicity, is frequently replaced with other nucleoside analogs. Despite their heterogeneity, most studies show that lipoatrophy improves if d4T, and probably also AZT, is replaced (recent review: Curran 2011). In particular, the subcutaneous fat of the limbs increases, although at first the improvement is often unrecognizable clinically and can only be detected in DEXA scans (Martin 2004). Histological investigations have shown that the elevated rate of apoptosis in adipocytes normalizes when d4T has been replaced (Cherry 2005, McComsey 2005).

Based on the available data, it seems advisable to replace d4T with another nucleoside analog. According to a warning letter by the company BMS of March 2011, d4T should only be used, if other antiretroviral substances can not be applied and duration of treatment should be as short as possible and patients should change to a more suitable therapy alternative whenever possible. unfortunately, it may still need to play a role in resource-poor regions for the time being. A dose reduction may be able to reduce adverse events (McComsey 2008). With regard to AZT, a replacement should be considered when lipoatrophy becomes manifest.

To avoid a hypersensitivity reaction the patient’s HLA status should be known before switching to abacavir (Carr 2002).

Table 7.2. Controlled clinical studies on switching from d4T or AZT to other drugs (all randomized, except McComsey 2004).




Effect of switch
Carr 2002
Martin 2004


ABC instead of d4T or AZT


LA better, lipids unchanged
John 2003


AZT instead of d4T and ABC instead of PI


LA of limbs slightly better, lipids and abdominal fat unchanged
Moyle 2003


ABC instead of d4T or PI/NNRTI, or AZT+ABC instead of d4T+PI


LA better (when replacing d4T)Lipids better (when replacing PI)
McComsey 2004


AZT or ABC instead of d4T


LA better, lactate better
Moyle 2005


TDF or ABC instead of d4T or AZT


LA better, lipids better on TDF
Valantin 2010


TDF+FTC instead of 2 NRTIs


Lipids better
Fischer 2009


TDF+FTC instead of AZT+3TC


LA better, lipids better
Ribera 2008


TDF instead of D4T


Lipids better, lactate better, LA slowly better
Tebas 2009


ABC or nuke sparing instead of d4T or AZT


LA better
Milinkovic 2007


TDF or d4T reduction (30 mg) instead of d4T


LA, lipids better (TDF effects better than d4T reduction)
No study showed any difference with respect to virologic failure. Wk=weeks, LA=lipoatrophy. In McComsey 2004 and Moyle 2005, only patients with LA were investigated.

Switching to tenofovir

Studies on therapy-naïve patients have shown that the short-term toxicity of tenofovir is lower than of d4T or AZT (Gallant 2004+2006). In the 903 Study, lipids improved in patients that were switched from d4T to tenofovir. There was also an increase of the mean limb fat after three years (Madruga 2007). Several studies, some of them randomized trials, point in the same direction. Lipids, lipoatrophy and mitochondrial toxicity and patient’s satisfaction improve on tenofovir (Milinkovic 2007, DeJesus 2008, Ribera 2008, Fischer 2010).

Recently, a double-blind, placebo-controlled, randomized study showed unexpected results. In ACTG A5206, the addition of tenofovir alone to existing virologically-suppressed ART regimens improved lipid parameters compared to placebo (Tungsiripat 2010). However, the mechanism of the lipid-lowering effect warrants further study. In a retrospective study, replacing d4T with tenofovir improved both lipids and liver enzymes (Schewe 2006).

It must be noted that negative effects may arise when switching too quickly to tenofovir. Randomized studies have observed a stronger reduction of bone density on tenofovir, compared to other NRTIs (Martin 2009, Stellbrink 2010). The potential nephrotoxicity of tenofovir is another point. Switching to tenofovir-containing triple nuke combinations must be avoided, as several studies have shown a high risk for an increase in viral load when switching to this combination (Hoogewerf 2003, Perez-Elias 2005), even after many years on successful ART. The resistance barrier is too low, as the following example shows.

Table 7.3. Example of what could happen on switching drugs (n.k.=not known).
Date ART

CD4 cells

Viral load

1996-98 AZT+ddC



Since 1998 AZT+3TC+NFV (always under the limit of detection)



Nov. 2002 Findings: significant lipoatrophy. Decision to switch



Feb. 2003 ABC+3TC+NFV



Apr. 2003 ABC+TDF+NVP (= targeted regimen,  notes below)



May 2003 Severe rash, ALT/AST > 500 U/l



Jun. 2003 ABC+TDF+3TC
Aug. 2003 Resistance: M41L+D67N+M184V+L210W+T215Y



Sep. 2003 AZT+3TC+NFV



Oct. 2003



Oct. 2004



Notes: On account of possible allergies to both ABC and NVP, ART was changed in February 2003 in two steps. Rash with hepatic involvement occurred on NVP, so in July 2003 NVP was replaced by 3TC – a triple nuke. The resistance mutations then detected were acquired almost certainly on the earlier treatment with AZT+ddC, but sufficiently suppressed while on PI therapy.

In practice, changes are often made that go further than PI and d4T/AZT, such as replacement of ddI, simply due to concerns over long-term toxicity. Such switching is based on laboratory studies showing a certain hierarchy with respect to mitochondrial toxicity (see chapter on Mitochondrial Toxicity).

A lot of attention is being drawn to simplification of therapy, in which mono- or nuke-sparing strategies are being used (see below). So far, there is no clear clinical evidence to show that this procedure has any benefit for the patient. If the patient has no complaints, a switch to monotherapy or nuke-sparing can not be justified and subjects the patient to unnecessary risks. Below, current data on this topic will be discussed.

Switching to nuke-sparing

Nuke-sparing is the attempt to completely avoid NRTIs in antiretroviral therapy. In treatment-naïve patients nuke-sparing proved virologically effective (see previous chapter) and is presently being investigated in treatment experienced patients with well suppressed viral load (Table 7.4). Data however is still very limited.

In ACTG 5116, the largest study so far with 236 patients on successful ART, the switch to lopinavir/r plus efavirenz compared to efavirenz plus NRTIs led to higher discontinuation rates due to increased virologic failure and other side effects (Fishl 2007). The results of this study are contrary to those of some other studies and to the results of lopinavir/r plus efavirenz with ART-naïve patients (Riddler 2008). At this point in time, it seems too early to recommend nuke-sparing as a transition strategy simply on the basis of theory. This not only concerns the new drug classes of integrase inhibitors and CCR5 antagonists (no data yet), but also monotherapies with boosted PIs (see below). Several studies are ongoing with raltegravir in combination with atazanavir/r (BATAR), darunavir/r (RALDAR, SPARE) and lopinavir/r (KITE). Results are expected in one to two years.

Table 7.4. Studies on switching to nuke-sparing regimens.


Switch to


Main effects of the switch




Lopez-Corles 2003




Virologically effective
Boyd 2005(HIVNAT 009)




Virologically effective, but many side effects due to IDV, LA probably slightly better
Negredo 2009 (MULTINEKA)




Virologically effective, lipids and mitochondrial DNA better
Tebas 2009 (ACTG 5110)




Virologically effective, lipodystrophy better
Tebas 2007 (ACTG 5125)




Many metabolic disturbances and LA better
Fischl 2007 (ACTG 5116)




Trend towards more virologic failure, more side effects
Ruane 2009




Virologically safe, but in 26% blips
Allavena 2009




Virologically safe
Ripamonti 2009 (CARDS)




Virologically safe, good PK data also for ATV (unboosted)
LA = lipoatrophy, * in Switch Arm (these studies were randomized) ** 7 patients received other PIs, among them 4 atazanavir

Switching due to virologic failure

Any change in treatment due to virologic failure requires experience, a certain degree of finesse and decisiveness. There are many possibilities for mistakes here. On the one hand, there is a threat of acquiring more resistance (if they have not already developed), but on the other hand, young physicians often want to quickly change treatment, which is not always necessarily the right solution. In many cases a frequent change of therapy confuses the patient and causes anxiety. If the problem is adherence, switching the regimen without talking about adherence may not be the solution. A switch only brings up more misunderstandings and, consequently, may generate later resistance. It is always important to explain to patients, who often tend to be skeptical (“should I save the other drugs for later?”) when and why treatment changes must be made.

As a rule of thumb, ART should be changed quickly with insufficient viral suppression and/or a rise in plasma viremia, as otherwise future options could be limited. One speaks of insufficient viral suppression or virologic failure if the viral load is repeatedly above the level of detection. A switch is not recommended with temporary viremia (blips – more on this topic in the chapter Principles of Therapy). Even single point mutations can be a problem. Abacavir, 3TC, FTC and ddI lose their efficacy in the presence of the K65R mutation, which is often selected by tenofovir-containing triple nuke therapies. Viral replication with insufficient plasma levels are the best breeding ground for resistance. Therefore, it is recommended to act fast if a clear virologic failure should occur. The longer one waits, the more complicated it becomes. An insufficient viral suppression means, as stated before, a repeated viral load above 50 copies/ml. Some clinicians tolerate levels of up to 500 or even 1000 copies/ml for months. We believe such hesitation is not justified in most cases when patients have good options and good adherence. A patient’s frequent assertions of not having symptoms should not count too much, either. Obviously, such thoughts do not always play a role in clinical reality. In an analysis in Great Britain 34% out of 694 patients remained on a virologically unsuccessful combination for over 6 months. Factors associated with an early switch were low CD4 T cells, a high viral load and older age (Lee 2008).

Arguments for a rapid switch in the case of virologic failure Arguments for a later switch in the case of virologic failure
The virus becomes incapable of generating more resistance New therapies bear the risk of new toxicities/ intolerance, which can lead to a termination of therapy
Options are maintained Most patients are immunologically stable for a long time with low viremia (clinically)
The switch is more successful with less resistance Replication fitness is reduced on failing treatment
The lower the viral load at time of switch, the better the response to the new therapy Resistance testing is often not possible with low viral load, even though they are there, so you may switch “blindly”
The following regimens do not have to be as complex as the present one – some things can be simplified (QD, no more d4T/ddI, etc.) It is sometimes difficult to explain to the patient why change of a well-tolerated and simple regimen is necessary

To date, only a few randomized trials have investigated two randomized strategies in patients in whom several ART combinations have failed: either the patients change immediately or when the viral load reaches a certain level (early versus deferred switch). The preliminary results of some small randomized studies indicate that even in such cases one can wait a short time (Nasta 2006, Tenorio 2009). However, these trials were small. It seems difficult to recruit physicians and patients to participate in such strategy trials.

With failing PI therapies there is more time. In the prospective Johns Hopkins Cohort there was no association between a deferral of ART modification and mortality in the course of treatment in patients on a PI showing virologic failure (Petersen 2008). This is why in the TITAN study, the number of acquired PI mutations had no effect on the success of darunavir/r, although it did play a role for lopinavir/r (De Meyer 2008).

In cases of clinical treatment failure (disease progression) or immunological failure (stagnation or decrease in the level of CD4 T cells) where the viral load remains below 50 copies/ml, the value of a change in therapy is unclear. Some combinations such as TDF+ddI are clearly unfavorable for immunological reconstitution (Negredo 2004). This may also be the case for AZT-containing regimens; such combinations should be changed.

It is important that when virologic failure occurs, the individual situation of the patient is carefully analyzed. In particular, several questions need to be addressed:

What are the reasons for the measurable viral load? A viral load above 50 copies/ml does not necessarily mean that resistance mutations have developed. A frequent cause may be a blip (see section on Goals and Principles of ART). These transient and, almost always, small increases in viral load usually have no relevance. However, a measurable viral load may be due to treatment failure. It may indicate insufficient plasma drug levels (measure these if possible). This may be due to drug malabsorption, drug interactions or simply insufficient dosing (e.g., in larger, heavy patients).

How is the patient’s adherence? Adherence is critical. Any difficulties related to the regimen should be openly addressed. Is it the number of pills? Do restrictions in food intake cause problems? Would once-daily treatment be better? Are there other reasons, such as depression? Any misunderstanding on how to take the drugs? The risks of resistance development as a result of non-compliance should be reiterated. If plasma levels are sufficient and viral load remains detectable (monitor blips at short intervals – within a few weeks), treatment should be changed as soon as possible.

How vulnerable is the present combination? NNRTI regimens are extremely sensitive, and cross-resistance can develop particularly rapidly for the class. A prompt change in therapy is more vital than with the other drug classes. Delaying this by even a few days or weeks may be too long. Rapid development of resistance can also be expected with 3TC/FTC and probably with the integrase inhibitor raltegravir. A PI-containing regimen without an NNRTI may allow a little more time, but the credo still applies. The higher the viral load at the time of modification, the lower the chance of success. One should not wait too long.

What options does the patient have, and what are the consequences of the change in therapy? The more options that remain available the sooner they should be utilized. Therapy can often be intensified quite easily (e.g., adding abacavir plus an NNRTI). In such cases, the decision to change or intensify a regimen is less difficult. On the other hand, it may be advisable in certain circumstances to continue therapy in a patient, even if the plasma viremia is not completely suppressed. Often, the viral load does not rise above the baseline value, and the CD4 T cells remain stable or even increase. Some experts advocate waiting in these cases. Resistance to nucleoside analogs are to be expected, so NNRTIs and PIs can be saved by waiting.

Even when multiple resistance mutations are already present, one is probably able to wait (see above). Especially in patients with adherence problems, it does not make sense to run through new drug classes. Adherence will not automatically be better with newer regimens. One should talk with the patient, find out what needs to be made better, and clarify if they are really ready for intensification or modification of therapy.

Virologic failure: to be considered before changing therapy

  • How resistance-sensitive is the present therapy? NNRTIs, 3TC/FTC, raltegravir: rapid development of resistance, change quickly
  • The lower the viral load, the greater the prospect of success with a change
  • Are you sure it is virologic failure and not a temporary blip?
  • Are there other reasons for a detectable virus load? What about malabsorption/drug uptake?
  • Do you know all the other therapies the patient is taking? Ask. Whether a gastric stimulant prescribed by the family doctor (i.e., PPIs) or herbal agents prescribed by an alternative practioner, it should all be laid out
  • Has the patient been adherent to current ART or have there been misunderstandings? Was the therapy discontinued ad hoc?
  • What do the plasma levels say and what does the patient say?
  • What options are there and what does a change mean for the patient? Is the patient able to start a new therapy?
  • Does a reasonably up-to-date resistance test exist? (if not, do one)
  • If relevant mutations against the current agents have already developed, calmly wait and prepare the patient for a new regimen, possibly with more adherence counseling

Leave a Comment

Filed under 6. ART 2011, 6.7. When to Switch?, Part 2 - Antiretroviral Therapy

Leave a Reply

Your email address will not be published. Required fields are marked *