Higher Rates of Neuropsychiatric Adverse Events Leading to Dolutegravir Discontinuation in Women and Older Patients

C Hoffmann; T Welz; M Sabranski; M Kolb; E Wolf; H-J Stellbrink; C Wyen


HIV Medicine. 2017;18(1):56-63. 

In This Article


In this large cohort of 1704 HIV-infected patients treated with at least one INSTI, the estimated overall discontinuation rate because of any AEs was around 6% within the first year of initiation. By far the commonest reason for discontinuation of dolutegravir was neuropsychiatric AEs (6%), which occurred less frequently in patients on raltegravir or elvitegravir (as Stribild©). These discontinuation rates are markedly lower than those recently reported in a smaller Dutch cohort where dolutegravir treatment was discontinued in 55 of 387 (14.2%) patients after a median of 78 days because of side effects, in particular sleeping, gastrointestinal and psychiatric problems.[14] Unfortunately, the authors did not provide details of ethnicity (only about one-third of the patients were of Dutch origin) or of other potential risk factors.

Our rates of AEs leading to dolutegravir discontinuation were, however, higher than those reported in clinical trials, particularly with regard to neuropsychiatric AEs. In four RCTs in treatment-naïve patients (SPRING-1, SPRING-2, SINGLE and FLAMINGO) and in one RCT in treatment-experienced patients (SAILING), all AEs ascribed to dolutegravir led to drug cessation in 1.2–2.5% of patients within the first year, comparable to rates of AEs leading to discontinuation of raltegravir and lower than those leading to discontinuation of efavirenz and darunavir/ritonavir.[1,3,10,11,15] However, although discontinuation rates were low, significant numbers of neuropsychiatric AEs, although usually mild, were observed in all the RCTs. For example, in SPRING-1, SPRING-2 and SINGLE, the rates for dizziness of any grade were 3, 6 and 9%, respectively. The corresponding rates for sleep disturbances were 2, 5 and 23%.[4] It has been speculated that the higher prevalence of insomnia observed in the SINGLE study was attributable to the specific study questionnaire that had not been used in other trials.[15,16] In a meta-analysis funded by ViiV Healthcare, the odds of discontinuation because of AEs were significantly lower with dolutegravir than with all treatments except raltegravir and rilpivirine.[9] In the phase III STRIIVING study, 551 experienced patients were randomized to switch from a variety of ART regimens to the STR Triumeq© or remain on their prior regimen. Ten patients (4%) in the switch arm discontinued because of AEs vs. none in the continuation arm.[12]

Clinical trials are, however, designed to provide evidence of efficacy and safety under ideal conditions. Although a large amount deal of information on a product's safety and efficacy is gathered during clinical development, it is not possible to fully describe the safety profile of a product in pre-marketing clinical trials. Post-marketing studies are the only sources of information that allow the assessment of the real-life effectiveness and safety of a new drug.[17]

It is well known that patients participating in RCTs are highly selected and may differ substantially from broader populations treated in different clinical settings, as a consequence of explicit exclusion criteria and subtle recruitment biases. Predefined stopping criteria for study drugs may discourage premature discontinuations in cases of mild to moderate AEs. Moreover, limitations in health care resources or limited treatment options may motivate patients to continue an antiretroviral regimen, even in the presence of side effects. For example, in SAILING, 49% of patients came from centres outside of Europe and North America.[1] Given the limited treatment options for patients in these settings, it is plausible that continuation of INSTI was a higher priority for them than for our patients. This may be illustrated by the high numbers of mild or moderate neuropsychiatric symptoms reported compared with the low discontinuation rates seen in the RCTs.

The two RCTs comparing dolutegravir with efavirenz, a nonnucleoside reverse transcriptase inhibitor (NNRTI) with well-recognized neuropsychiatric side effects, suggest a possible signal for the frequency of neuropsychiatric AEs seen with dolutegravir in 'real-life'. In SINGLE, there were significantly more discontinuations of Atripla® than of DTG with ABC + lamivudine because of psychiatric and nervous system disorders but the incidence of insomnia was significantly greater in the Triumeq®group and the incidences of headache, fatigue and depression were similar. Dizziness, abnormal dreams, anxiety and somnolence were more common with Atripla®.[15] In the SPRING-1 dose-finding study, headache occurred more frequently in the dolutegravir 50 mg/day group than in the efavirenz group (10 vs. 4%, respectively), with dizziness (6 vs. 18%, respectively) and insomnia (6 vs. 10%, respectively) less common with dolutegravir but still occurring at similar frequencies to those found in our study.[11]

More recently, preliminary data have been reported from the phase IIIb randomized, open-label ARIA study designed to demonstrate noninferior antiviral activity of the STR Triumeq® compared with atazanavir boosted with ritonavir plus TDF and emtricitabine in 495 treatment-naïve adult women.[8] ARIA, conducted by ViiV Healthcare, enrolled women (41% of the study participants were of African heritage) in 12 countries, with the USA and South Africa contributing the two largest groups of participants. There were fewer dolutegravir AEs leading to discontinuation compared with those in the atazanavir arm (4 vs. 7%, respectively) and none of the AEs seen with dolutegravir was neuropsychiatric. The reasons for the conflicting results between this trial and our data remain unclear, but the facts that most women were enrolled in settings with restricted ART access and that the RCT was open-label may have influenced the outcome.

The pattern of neuropsychiatric AEs leading to dolutegravir discontinuation in our cohort was heterogeneous. The most frequent (but not universal) events were insomnia and sleep disturbances, but also dizziness and painful paraesthesia which could not be explained otherwise. Of note, in almost all patients, the symptoms occurred during the first few weeks on dolutegravir, were not life-threatening, did not lead to hospitalization and disappeared quickly after discontinuation of dolutegravir, which was usually the only action taken.

Of concern was the fact that the rate of dolutegravir discontinuation because of AEs was almost threefold higher in women and in patients older than 60 years. Further studies evaluating the effect of sex, age, body mass index and race on dolutegravir pharmacokinetics (PK) are warranted. RCTs to date have primarily enrolled white and normal weight men between the ages of 19 and 54 years, leaving important sections of the HIV-infected population underrepresented.[4] In the four trials evaluating dolutegravir as first-line therapy, the median age was 34–37 years and only 13–16% of patients were female. It is possible that different PK may have contributed to the higher rate of dolutegravir-related AEs in these groups. Although the pathways of metabolism and transport suggest that the drug should be relatively unaffected by such differences, no study has specifically assessed the effects of demographic diversity on dolutegravir PK parameters. This also applies to patients with hepatic comorbidity, in which the unbound fraction of dolutegravir in plasma is higher than in healthy volunteers. Finally, it remains to be investigated whether neuropsychiatric events represent a direct neurotoxic effect. Dolutegravir achieves high therapeutic concentrations in the central nervous system (CNS), mainly by passive diffusion with a low possibility of active transporter involvement.[7]

We did not find any impact of CD4 T-cell counts or ethnicity on AE rates. Correlations between tolerability or PK and race, sex and immune status are well described for NNRTIs such as nevirapine and efavirenz.[13] There was also no evidence for any cross-drug effect. Eighty-six per cent of patients who switched away from an INSTI because of AEs had no tolerability problems with the subsequent antiretroviral regimen which contained other INSTIs, protease inhibitors or NNRTIs.

Our study has several important limitations.

Firstly, because of the retrospective design, we were unable to account for confounding and bias resulting from patient selection. However, potential sources of bias and confounding related to patient selection for INSTI initiation are likely to be fewer in Germany than in most other settings as, unlike in most settings world-wide, there are no financial restrictions to ART (in particular INSTI) prescribing. While the treatment histories of our patients did indeed show some differences in the context in which the INSTI was initiated (i.e. salvage therapy, treatment simplification, or prior intolerability of efavirenz or other regimens), these reflect the availability of more INSTI alternatives over time. For example, the fact that more treatment-experienced patients switched to raltegravir can be explained by the fact that raltegravir had been the only INSTI available in 2007–2012.

Patients initiating raltegravir before 2013 were likely to have had fewer ART options and therefore a higher motivation to remain on therapy despite side effects. However, the discontinuation rates for raltegravir remained relatively stable over time. One might also hypothesize that previous neuropsychiatric AEs on efavirenz may have influenced the reporting of neuropsychiatric symptoms and/or patients' willingness to persevere with a subsequent regimen in the presence of such symptoms. We found no association between neuropsychiatric events on INSTIs and previous treatment history or reason for switch among experienced patients, and our findings on the tolerability of various INSTIs are not explained by differences in the reason for INSTI initiation.

The high rates of discontinuation of dolutegravir in 2016 are difficult to explain other than by an enhanced awareness of the treating physicians of the possibility of dolutegravir being the drug responsible for neuropsychiatric AEs. Of note, our results and conclusions remained consistent when patients initiating dolutegravir in 2016 were excluded.

It has been shown that patients' beliefs and satisfaction with therapy modulate the impact of mild AEs.[2] However, in most patients (including women and older patients) who discontinued dolutegravir, the subsequent ART regime was well tolerated. Neuropsychiatric AEs recurred in six patients who were re-exposed to dolutegravir after stopping it once.

It was not possible to determine whether the AEs were solely related to dolutegravir. Discontinuation rates were markedly higher in patients initiating abacavir at the same time, despite prior HLA testing. This was seen in both naïve and treatment-experienced patients, and it remains unclear whether some of these events were directly related to abacavir. We found no difference between dolutegravir neuropsychiatric AEs for patients on abacavir as STR and those on a two or three tablet regimen (data not shown). Data for the dolutegravir/abacavir combination remain relatively limited as, except for the SINGLE trial, the majority of patients did not receive abacavir in the RCTs.

As we only had data on AEs leading to discontinuation, the true rate of AEs related to dolutegravir may be higher, even in the setting of a broad range of antiretroviral options and a lower threshold to modify a regimen. Indeed, anecdotally several patients complained of neuropsychiatric AEs during the first few weeks on dolutegravir which disappeared thereafter despite continuing dolutegravir.

Finally, because this was a retrospective observational cohort study, it was not possible to characterize neuropsychiatric AEs in detail and we also did not have information on the timing of dolutegravir administration. As these AEs were also, to a lesser extent, seen with other INSTIs, it remains unclear whether they represent a class effect with variable incidence between different INSTIs.

In this retrospective analysis of more than 1700 patients treated with INSTIs, almost 6% of therapies with dolutegravir were discontinued because of neuropsychiatric side effects within a year of initiation. The discontinuation rate for dolutegravir because of these AEs was higher than that reported in clinical trials and higher than for raltegravir and elvitegravir. Neuropsychiatric AEs were seen especially in women, in older patients and in patients who initiated abacavir at the same time. As dolutegravir is likely to remain among the preferred antiretroviral options for HIV-infected patients, it is vital that post-marketing surveillance and further research be done on the safety of dolutegravir outside of RCTs and on mechanisms for potential neurotoxicity, especially in populations underrepresented in the RCTs.