Replacing a PI with atazanavir: separating swans from ducklings?

Mark Mascolini for

The SWAN study, a 48-week open-label trial that randomised people to continue an effective PI regimen or switch to atazanavir, found fewer viral rebounds and better lipid trends in the atazanavir group [2]. But this SWAN paddled primarily through the backwaters of antiretroviral history because nearly half the study participants entered the trial taking an outmoded unboosted PI, and almost everyone who switched to atazanavir took it without ritonavir.

Big trials take a long time to mount, run, and analyse, so no-one can blame the investigators for presenting anachronistic results. But with ritonavir-boosted atazanavir being the standard of care, SWAN ends up giving clinicians lots of data they can’t use.

Jos矇atell from Barcelona’s University Hospital Clinic and colleagues across Europe recruited 419 people taking their first or second PI regimen. Almost everyone had a viral load under 50 copies/mL, and no one had a PI failure on their charts. Baseline CD4 counts averaged 490 cells/mm3, and 30% had HCV coinfection.

PI experience averaged 3.4 years, and 193 people (46%) were taking an unboosted PI, usually nelfinavir (72% of the unboosted groups), but also indinavir (18%) and even saquinavir (6%). Among the 226 people who signed up for SWAN while taking a boosted PI, 153 (68%) were taking lopinavir/ritonavir.

SWAN investigators randomised these people in a 2-to-1 ratio to trade their PI for atazanavir or to keep their current PI. Of the 278 people randomised to atazanavir, only the 10% taking tenofovir used 300/100 mg of atazanavir/ritonavir once daily. The rest took 400 mg of atazanavir without ritonavir.

After 48 weeks, an intent-to-treat analysis of people who started their assigned regimen found significantly fewer confirmed rebounds above 50 copies/mL in the atazanavir group. The 95% confidence interval in this overall analysis allowed statisticians to say that atazanavir is virologically noninferior to the baseline PIs, and not that it is virologically superior.

Perhaps not surprisingly, the lower rebound rate after the tradeoff to atazanavir can be traced almost entirely to people who started SWAN taking a pass瞠 unboosted PI. Among people who entered the trial taking a ritonavir-propped PI, switching to atazanavir did not result in tighter viral control.

Table 1: 48 wk responses from switching to atazanavir from a boosted or unboosted PI

ATZ CPI p-value
Rebound >50 copies 19/278 (7%) 22/141 16% p<0.01
By RTV use in previous regimen (ITT):
Boosted PI 12/150 (8%) 8/76 (11%) NS
Unboosted PI 7/128 (5%) 16/65 (22%) p<0.001
Treatment failure, any reason (ITT):
59.278 (21%) 48/141 (34%) p<0.01
Treatment failure, any reason, by RTV use in previous regimen (ITT):
Boosted PI 31/150 (23%) 24/76 (32%) NS
Unboosted PI 25/128 (20%) 24/65 (37%) p<0.01

The same proportion in each treatment arm, 6%, quit because of side effects or other “adverse events.” Scleral icterus and jaundice affected 29 people (11%) taking atazanavir and none who stuck with their baseline PI. Diarrhoea troubled 8 people (6%) taking a control PI and 5 (2%) switching to atazanavir (P < 0.05).

In the atazanavir group 43% had grade 3 or 4 leaps in total bilirubin, compared with 3% in the control group. Among people with HCV coinfection, 53% who switched to atazanavir had grade 3 or 4 total bilirubin buildups, compared with 8% in the control arm. Liver enzyme jumps proved similar in the two study arms after 48 weeks, regardless of HCV coinfection.

Equivalent proportions in the two treatment groups had total cholesterol at or above 240 mg/dL and triglycerides at or above 250 mg/dL when the trial began. After 48 weeks significantly fewer people switching to atazanavir had cholesterol or triglyceride readings that high (8% versus 29% for total cholesterol, P < 0.0001; 9% versus 30% for triglycerides, P < 0.0001).

As with the virologic outcomes, these side effect results do not readily translate into current practice because atazanavir levels soar so much higher with a ritonavir kick. Much work remains in defining the toxicity profile of boosted atazanavir. A Bristol-Myers Squibb study reported at the Lipodystrophy Workshop just before EACS, for example, found equivocal evidence on boosted atazanavir and glucose regulation [2].

Bristol-Myers Squibb researchers had 24 healthy volunteers take standard doses of atazanavir/ritonavir or lopinavir/ritonavir for 10 days. After a week or more taking no PIs, the volunteers switched to the alternate regimen. Glucose area under the curve rose significantly with lopinavir/ritonavir but not atazanavir/ritonavir. But insulin climbed significantly with both regimens, and mean glucose response during an oral glucose tolerance test rose almost identically with atazanavir/ritonavir (111.1 mg/dL) and lopinavir/ritonavir (110.6 mg/dL) (P < 0.05 versus baseline for both).

Mulling these results at EACS, Andrew Carr from Sydney’s St. Vincent’s Hospital opined that atazanavir/ritonavir probably does cause insulin resistance. Whether one can blame these glucose upsets solely on the ritonavir boost hardly matters if that’s the way atazanavir gets prescribed.

Notably, this study in healthy volunteers without HIV infection also charted significant drops in “good” high-density lipoprotein cholesterol with 10 days of either lopinavir/ritonavir or atazanavir/ritonavir (P < 0.01), a blot on atazanavir’s heretofore unsmirched lipid record.


  1. Gatell J, Salmon-Ceron D, Lazzarin A, et al. Efficacy and safety of atazanavir based HAART in patients switched from a stable boosted/unboosted protease-inhibitor treatment: the SWAN Study. 10th European AIDS Conference. November 17-20, 2005. Dublin. Abstract PS1/1.
  2. Noor MA, Flint OP, Parker RA, et al. Evaluation of insulin sensitivity in healthy volunteers treated with low-dose ritonavir combined with atazanavir or lopinavir: a prospective, randomised study using hyperinsulinemic, euglycemic clamp and oral glucose tolerance testing. 7th International Workshop on Adverse Drug Reactions and Lipodystrophy. November 13-16, 2005. Dublin. Abstract 16.

Links to other websites are current at date of posting but not maintained.