INSTI and weight gain: reports from CROI 2019
Polly Clayden, HIV i-Base
Several presentations at CROI 2019 showed data from analyses of INSTI-associated weight gain from mostly US cohorts (and one prevention study).
Although topical and a potential cause for concern – particularly with the accelerating roll out of dolutegravir worldwide – no clear conclusions emerged from what was presented.
ACTG A5001 and A5322
Annual within-person weight gain increased following switch to INSTI-based ART among AIDS Clinical Trials Group (ACTG) participants in protocols A5001 and A5322. The increase was particularly significant for women, black people and people aged 60 and above. 
This assessment included A5001 and A5322 participants, in follow-up from 1997–2017, who switched to INSTI. Participants were their own controls for estimation of background/age-related weight gain (before vs after switch).
A total of 972 adults switched to INSTI at a median of 7.8 years after enrolling in the parent trial. The evaluation included 691 with undetectable viral load at switch: 82% men, 45% non-white, median age 51 years, CD4 610 cells/mm3 and BMI 26 kg/m2; 289 switched to raltegravir, 204 to elvitegravir and 198 to dolutegravir.
Median follow up was 1.8 years – although this was up to 10 years for people receiving raltegravir. Approximately two thirds switched from a PI and the remainder from an NNRTI.
In adjusted models (for age, sex, race/ethnicity, parent study baseline BMI and their interactions, nadir CD4, smoking, diabetes and percent follow-up time with suppressed viral load <200 copies/mL), white or black race, age >60 and BMI >30 kg/m2 were associated with greater annual weight gain following switch for women, and age >60 was the greatest risk factor for men.
Dolutegravir was associated with the greatest annual weight gain with pre-post difference of 1 kg more per year p=0.0009. Pre-post differences in kg/year for elvitegravir and raltegravir were not statistically significant in this cohort.
Although subset analyses of NRTI back bone in this cohort were limited by sample size, switch to any INSTI with abacavir and switch to elvitegravir with tenofovir alafenamide were statistically significant, p<0.05.
Treatment-naive adults starting INSTI, especially dolutegravir and raltegravir, were at higher risk of weight gain compared to those starting older NNRTI-based regimens in the North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD). 
Participants in NA-ACCORD starting INSTI, protease PI, and NNRTI-based ART regimens after 1 January 2007 were included and followed through 31 December 2016.
Predicted weights by ART class were reported at years 2 and 5. Because of shorter follow-up for newer INSTI, predicted weights for raltegravir, elvitegravir, and dolutegravir were reported at years 1 and 2.
Of 24,001 participants, 4,720 started INSTI-based regimens: 1681 (35%) raltegravir, 2124 (45%) elvitegravir and 936 (20%) dolutegravir. A further 11,825 started NNRTI-based and 7436 started PI-based ART. The vast majority, 87% were men, and 41% were white.
At baseline, median age was approximately 42 years, BMI was 25 kg/m2 and CD4 approximately 300 cells/mm3.
Predicted weight gain for participants receiving INSTI at 2 and 5 years, was 4.9 and 6.0 kg, respectively, compared to 3.3 and 4.3 kg for NNRTI and 4.4 and 5.1 kg for PI.
Of those starting INSTI, predicted weight gain at year 2 was 6.0 kg for dolutegravir, 4.9 kg for raltegravir, and 3.8 kg for elvitegravir.
Weight gain associated with INSTI-based regimens did not vary by sex in this cohort (although this comparison is limited by few women vs men) or race (white vs non-white).
US ART-experienced cohort
Only psychiatric disorders were associated with weight gain in a treatment-experienced US cohort including 3468 virologically suppressed people with HIV – according to a retrospective analysis. 
Hypogonadism, obese or overweight at baseline and PI use were negatively associated with weight gain.
Although the association between INSTI-based ART and weight gain reached significance in bivariate analyses, this did not remain significant in multivariate analysis.
Participants were from 21 US states and Washington DC. The majority, 81% were men; 14% were women and the remaining 5% unspecified; 61% were white, 28% African-American and 11% unknown. Of these, 64% had received an INSTI, 28% an NNRTI and 20% a PI.
Between August 2013 and August 2018, 30% of participants had annualised weight gain >3%, 16% had weight loss >3%, and 54% had weight change <3%.
In multivariate analysis, psychiatric disorders were associated with weight gain >3%: OR 1.28 (95% CI: 1.0 to 1.6), p=0.020. Factors negatively associated with >3% weight gain were: hypogonadism OR 0.81 (95% CI 0.6 to 1.0), p=0.050; overweight at baseline OR 0.69 (95% CI 0.6 to 0.8); obese at baseline OR 0.62 (95% CI 0.5 to 0.8), and PI use OR 0.58 (95% CI 0.4 to 0.7), all p<0.001.
The proportion with >3% weight gain was significantly lower among participants receiving a PI vs no PI. And the proportion with >3% weight gain was higher among those receiving INSTI vs no INSTI. There was no statistically significant difference between the NNRTI and no NNRTI groups.
Analysis was by drug class, analysis by individual drug was not presented.
In the Women’s Interagency Health Study (WIHS) a switch to INSTI was associated with significant increases in body weight and measurements, body fat, and blood pressure vs no INSTI. 
Women who switched to or added an INSTI to ART were compared to women who remained on non-INSTI ART. The study examined changes in body weight; body mass index (BMI); percentage body fat; circumference of waist, hip, arm, and thigh; blood pressure; and incident diabetes mellitus, between 2008 and 2017.
There were 1118 WIHS participants included, of which 234 switched to or added an INSTI. The mean follow up was 2 years and there were no differences in baseline characteristics between groups: mean age 48.8 years; 61% African American; and mean CD4 669 cells/mm3. Baseline weight and BMI were 80.8 kg and 31 respectively.
Women receiving an INSTI had 2.14 kg greater increase in weight, 0.78 kg/m2 greater increase in BMI, 1.35% greater increase in percentage body fat, and 2.05, 1.87, 0.58, and 0.98 cm greater increases in waist, hip, arm, and thigh circumference, respectively.
Greater BMI increases among people who switched to INSTI-based ART than those who switched to non-INSTI-based ART found in the HIV Outpatient Study (HOPS). The association was most pronounced in women and people of Latino/a ethnicity. 
This study was an analysis medical record data of participants from nine US clinics who were INSTI-naive and virally suppressed for >1 year on non-INSTI ART, and who switched to INSTI-based ART and remained suppressed. They received INSTI-based ART for >6 months, had >2 weights recorded in the year before switch and >1 after.
Of 653 participants, 368 (56.4%) switched to an INSTI-based regimen and 285 (43.6%) switched to a non-INSTI- based regimen.
Of the participants receiving INSTI, median age was 51 years and 17.7% were women. INSTI regimens included raltegravir (48.6%), elvitegravir (21.7%), or dolutegravir (29.6%). Mean duration of INSTI use after switch was 2.4 years
Mean change in weight on INSTI-based regimens was greater than that on non-INSTI-based regimen: 1.2 kg vs 0.3 kg, p=0.05.
In multivariate analysis women and people of Latino/a ethnicity had greater BMI, and people who inject drugs had lower BMI than MSM, all p=<0.01.
BMI trajectory slopes post-ART switch were greater for dolutegravir-based ART than for regimens with other INSTIs, p=0.03 vs raltegravir and p=0.003 vs elvitegravir.
Only elvitegravir-based ART was not associated with increases in BMI, p=0.67.
Parkland Health and Hospital System in Dallas, Texas
Elvitegravir was associated with greater BMI gains overall; dolutegravir and raltegravir were associated with greater BMI gains in women; and dolutegravir with greater gains in black and Latino/a people in a large urban clinic in Dallas. 
All patients starting ART at the Parkland Health and Hospital System in Dallas, Texas from 2009 to 2017 were included in this analysis.
Of 4,048 participants, 69% were men, 53% black, 28% Latino/a, and 16% white. Mean age was 46.3 years (SD +/– 11.9). Mean baseline BMI was 27.0 kg/m2 (SD +/– 6.4). Median duration of ART was 6.7 years (IQR: 2.8 to 11.2).
There was no significant interaction between sex and race/ethnicity on BMI gains. Proportion of overweight /obese participants (BMI ≥ 25) increased from 51% at ART initiation to 65% at year 3, p<0.001.
The BMI slope per year on NNRTI, PI and INSTI were 0.22, 0.24 and 0.32 kg/m2, respectively.
Among INSTIs, elvitegravir appeared to be associated with greater BMI gains overall (0.39/year), but the effect did not vary or by sex or race/ethnicity.
Dolutegravir and raltegravir were associated with greater BMI gains in women than men: 0.44 vs 0.12/year, p<0.01; and 0.3 vs 0.08/year, p=0.03, respectively. Dolutegravir was also associated with greater BMI gains in black and Latino/a people vs white.
There was no difference in weight gain among HIV negative people receiving cabotegravir compared with placebo in HPTN 077. 
This is a phase 2a randomised placebo-controlled prevention study of two dose/dose-interval regimens of cabotegravir, conducted at sites in the US, Brazil and sub-Saharan Africa. A total of 199 participants were enrolled and randomised 3:1 to cabotegravir or placebo.
Median weight change over 41 weeks was +1.1 kg (IQR: –0.9 to +3.0) in the cabotegravir arm and +1.0 kg (IQR: –1.2 to +3.2) in the placebo arm (p=0.66). There were no differences by sex, dose, age, race/ethnicity, smoking, or BMI.
Table 1: Weight changes on ART: studies at CROI 2019
|Abstract no.||Design||Details||Findings||Higher risk|
Lake JE et al.
|ACTG switch study (US).
Prospective cohort: 2007– 2017 – within <2 years of switch during follow-up. BUT weight changes were modelled, adjusted for age, sex, BMI, CD4, smoking, diabetes etc.
VL <200 c/mL
55% white, 26% black, 19% Latino/a
|Generated weight trajectories before and after switch.
Weight inc with DTG, only when switch from PI and with EVG and DTG from NNRTI.
Older >60 y
Bourgi K et al.
|NA-ACCORD cohorts (US). Naive (<45 d), 2007-2016. Modelled weight gain by class: NNRTI vs PI vs INSTI.||n-24,000 (n=4740 INSTI)
BMI 25 (IQR: 23 to 29)
| Predicted weight increase (kg) at 2 yr: +3.3 (NNRTI), +4.4 (PI), +4.9 (INSTI).
DTG (+6.0) > RAL (+4.9) > EVG (+3.9).
|No differences by gender or race.|
McComsey G et al.
|Retrospective US observational switch study 2013–2017. VL<200 c/mL and 2 x BMI. compared risk factors for weight gain (>3%) vs no weight gain.||n=3468
|Overall. 30% weight gain, 54% minimal change and 16% weight loss.
INSTI > no INSTI (32% vs 28%).
No PI > PI (32% vs 22%).
|No INSTI effect in multivariate analysis BUT not by ART. Lower BMI (NS) but high BMI significantly reduced risk.
No differences by gender or race.
Psychiatric illness increased risk.
AM et al.
|Retrospective analysis WIHS cohort (US) 2008-2017 switch analysis.||n=1118 (n=234 switch)
VL <1000 c/mL.
~2 yrs follow up.
Only difference between arms was higher PI use in switch group (69% vs 46%m p<0.0001).
|Significant increases in INSTI switch group for weight (difference +2.14 kg) BMI (0.78 kg/m²), percentage body fat (+1.35%), and waist, hip, arm, and thigh circumference (2.05, 1.87, 0.58, and 0.98 cm respectively).
Also greater change in systolic and diastolic BP (2.24 and 1.17 mmHg, p<0.05) and new-onset DM (4.5% and 2.2%) in STAY, p=0.11.
|No significant differences by INSTI.|
Palella FJ et al.
|Retrospective analysis HOPS cohort (US) 2007–2017. All switch analysis with >2 x BMI.||n=653 (n=363 INSTI switch vs 285 non-INSTI switch).
VL <200 c/mL. Approx 20% women. 60% white, 25% black, 12% Latino/a.
|Greater mean weight increase
on INSTI-based switch (1.2 kg vs 0.3 kg, p=0.05).Increases with both DTG and RAL but not with ELV.
|Women and Latino/a associated with greater BMI increases.|
Bedimo R et al.
|Retrospective analysis of treatment naive pts 2009 – 2017.||n=4,048.
53% black, 28% latino, and 16% non-Hispanic Whites.
Mean baseline BMI 27.0 kg/m2 (SD 6.4). Median follow-up on ART 6.7 years (IQR 2.8 – 11.2).
|BMI, ≥ 25 increased from 51% to 65% at year 3 (p<0.001).
BMI slope per year was 0.22, 0.24 and 0.32 on NNRTI, PI and INSTI, respectively.
|All PIs >BMI women > men, but no difference by race/ethnicity.
EVG >BMI vs DTG/RSL. No difference by sex or race/ethnicity.
DTG and RAL associated with greater BMI gains in women, and DTG with greater gains in black & Latino/a.
|Oral Abs 30.
Landovitz R et al.
USA, Brazil and sub-Saharan Africa.
|Phase 2a randomised placebo-controlled study of two dose regimens of cabotegravir vs placebo in HIV negative.||n=199 (n=134 active CAB).
26% white, 40% black, 25% Latino/a.
|Median weight change over 41 weeks was +1.1 kg (IQR -0.9, 3.0) in the CAB arm and +1.0 kg (IQR -1.2, 3.2) in the PBO arm (p=0.66).|
A systematic review, conducted earlier this year, concluded that it is currently unclear whether INSTI cause clinically significant changes in body weight or whether these changes are statistically significant but small. 
Data shown at CROI 2019 showed no distinct patterns.
RCT data, expected at IAS 2019 from the ADVANCE and NAMSAL studies (comparing dolutegravir-based regimens to efavirenz-based ones in two African countries) where weight was looked at systematically will be important.
Unless stated otherwise, all references are to the programme and abstracts of the Conference on Retroviruses and Opportunistic infections (CROI) 2019, Seattle, 4–7 March 2019.
- Lake JE et al. Risk factors for excess weight gain following switch to integrase inhibitor–basedART. Poster abstract 669.
- Bourgi K et al. Greater weight gain among treatment-naive persons starting integrase inhibitors. Poster abstract 670.
- McComseyGA et al. Weight gain during treatment among 3,468 treatment-experienced adults with HIV. Poster abstract 671.
- Kerchberger AM et al. Integrase strand transfer inhibitors are associated with weight gain in women. Poster abstract 672.
- Palella FJ et al. Weight gain among virally suppressed persons who switch to InSTI-based ART. Poster abstract 674.
- Bedimo R et al.Differential BMI changes following PI- and InSTI-based ART initiation by sex and race. Poster abstract 675.
- LandovitzRJ et al. Cabotegravir is not associated with weight gain in HIV-negative individuals: HPTN 077. Oral abstract 34.
- Hill A et al. Are new antiretroviral treatments increasing the risks of clinical obesity? Journal of Virus Eradication 2019;5: e45–e47.