TAG : Treatment Action Group

• Cynical Swiss Saquinavir Scam: Roche Admits Licensed Dose Suboptimal
• How Many Became Cross-Resistant in Roche's Rush to Market?
• New Formulation Gets Blood Levels Eightfold Higher, Safely
• Why Would the Weakest Protease Inhibitor Cost the Most?
• Company Pulls "Strategy" Ads After Community Balks
• Can SQV Soft Gel Caps Save the Day?

by Mark Harrington
7 June 1997

"Cynical", "greedy", "manipulative", "opportunistic", "penny-pinching", "short-sighted", "slipshod" -- do these words come to mind when you think of Hoffmann-LaRoche's AIDS drug development efforts? They should. Examples of such behavior are legion, from the fiasco that was ddC to the joke that is the current formulation of saquinavir to the curious decision by the Basel-based pharmaceutical giant to drastically curtail development of valganciclovir, the oral ganciclovir prodrug which offered the hope for finally being an effective prophylaxis for cytomegalovirus (CMV) disease. As one high-placed Federal official noted (off the record, of course) of the latter decision, "Roche finally has a decent drug, and they're thinking of dropping it." Roche's HIV program gives the "ethical" pharmaceutical industry a bad name.

Saquinavir, like ddC, is the most potent drug of its class -- in vitro. However, only 4% of the drug gets into the bloodstream. Hoffmann-LaRoche was in such a hurry to get its drug licensed as the first protease inhibitor that it never bothered doing the dose-ranging studies which could have defined a maximum tolerated dose (MTD) for saquinavir.

I was a member of the ACTG's Primary Infection Committee when Roche approached them to conduct the phase II study, dubbed ACTG 229. The dose chosen was 600 milligrams (mg) thrice daily, based, they claimed, on three European phase I studies, or, as others thought, on a limited drug supply which made higher doses impractical -- or not worth Roche's investment. While the Primary Infection Committee was never known as a bastion of open scientific debate, ACTG 229 was swaddled in a secrecy unusual even for them. Roche declined to present the results of its phase I studies to the committee as a whole. Rather, they allowed Thomas Merigan of Stanford University and Ann Collier of the University of Washington at Seattle to take a peek at the alleged phase I virological response to saquinavir.

The study would take place in AZT-experienced patients, then the favorite population for trials of new antiretrovirals (remember ACTG 155?). They would be randomized to receive either AZT and ddC, AZT and saquinavir, or AZT, ddC and saquinavir. This was one of the first of the so-called "incestuous combination" studies recently pilloried by Joep Lange, in which a company's own drugs are studied together as much as possible, regardless of the scientific rationale for doing so.

Preliminary review of the study design by the Division of AIDS (DAIDS) and Harvard's Statistics & Data Analysis Center (SDAC) raised several concerns, which I mentioned to Dr. Collier (the principal investigator of ACTG 229) in a letter on 30 September 1992:

I remain perplexed about the current design of ACTG 229. In particular, I share the CTRC's concern "about the selection of 600 mg tid as the dose of Ro 31-8959 [saquinavir] since there is no established maximum tolerated dose" [NIAID Clinical Trials Review Committee letter, 27 August 1992]. Doses as high as 1200-1800 mg tid have been tested in HIV-negative patients and found to be safe... but people with HIV have only been given doses up to 600 mg tid. I would concur with the CTRC that "the need for the pharmaceutical sponsor to be forthcoming with data from their European trials" is pressing as we proceed towards opening ACTG 229...

I became even more concerned when I read David Schoenfeld's SDAC review. His bottom line was that "the proposed study will not be able to detect whether Ro 31-8951 has moderate activity."

Needless to say, the ACTG brushed aside the concerns of statisticians and activists and conducted the study as Roche wished it to. 300 AZT-experienced individuals were enrolled and followed for 18 months. By June 1994, Roche had detected the surrogate marker response it hoped for (triple drug combination proved superior to either two drug combination as measured by CD4 cell response and, less impressively, by viral load). As Schoenfeld predicted, the study failed to show whether saquinavir was any more potent than ddC, the weakest of the nucleoside analogues in vivo. Undeterred by this minor annoyance, Roche promptly petitioned the FDA to consider an accelerated new drug application (NDA) for saquinavir.

Worried by the precedent this would set for the protease inhibitors as a class, TAG then wrote to FDA Commissioner David Kessler requesting that accelerated approval for saquinavir be placed on hold until a full and open public debate could take place to assess how much data would be required for accelerated approval of protease inhibitors, and how post-marketing confirmatory studies should be designed.

In the controversy that ensued, Roche quietly agreed to double the size of its pivotal efficacy trials, thereby increasing their ability to determine whether saquinavir provided any clinical benefit. Unfortunately, the study which was eventually to provide such evidence -- Roche NV14256B -- compared saquinavir to ddC to the combination in AZT-experienced patients. Since the role of ddC in this population is far from clear, and its benefit dubious in any population, such a control arm must be regarded as questionable. Nonetheless, to no one's surprise, the combination of these two drugs, each the weakest in its class, proved to be more potent than either one alone.

This led to accelerated approval for saquinavir, now dubbed INVIRASE, by the FDA in November 1995. The drug was licensed at the dose studied in ACTG 229, 600 mg thrice daily, despite the fact that there was already evidence at the time that a dose twice as high was more potent and equally tolerable. Moreover, it was already known at the time that suboptimal doses of protease inhibitors might predispose HIV towards the development of resistance and possibly even cross-resistance to other protease inhibitors.

Thus, ever since saquinavir's licensure at the end of 1995, Roche has known that the licensed dose was suboptimal and that its use could well result in widespread cross-resistance to multiple protease inhibitors.

Had saquinavir rapidly become the drug of choice for people who were failing on nucleoside analogue monotherapy or double therapy, a public health disaster might well have resulted. If cross-resistance became widespread through broad and prolonged use of saquinavir, many people would not have been able to benefit from the later introduction of more potent protease inhibitors.

Luckily, help was not long in coming. Within three months, both Abbott's NORVIR brand ritonavir and Merck's CRIXIVAN brand indinavir were licensed, at doses which were able, when given in combination with new reverse transcriptase inhibitors, to drive viral load beneath the limit of detection in over 75% of patients who could tolerate them for up to one year (Merck 035, etc.), and could prolong health and life when compared with standard of care (Abbott study, ACTG 320). Of note, Roche's survival study used ddC monotherapy, which no one, even then, regarded as standard of care. None of this deterred Roche from charging $7,000 for a year's supply of INVIRASE, a higher price for a drug weaker than either of its competitors.

Yet Roche faced a quandary. Despite its slipshod, post-haste development plan, two more potent protease inhibitors reached the market within three months of its own accelerated NDA, and even those unversed in the intricacies of retrovirology could tell that they were far more potent. How could Roche redeem its drug?

Two opportunities presented themselves. The first was to use the ability of other protease inhibitors -- and particularly ritonavir -- to inhibit cytochrome p450 metabolism, thereby increasing the bioavailability, exposure, half-life, and maximum concentration of saquinavir to therapeutic levels. The other, more prosaic, approach was to finally begin addressing the need for a more bioavilable formulation and higher dose of saquinavir itself, unassisted by complex hepatometabolic pathways. Roche proceeded to follow both leads.

As for those participants lucky enough to survive ACTG 229, they were given the chance to enroll in ACTG 333, the first-ever randomized study in protease failures. ACTG 333 randomized 72 SQV-experienced individuals to continue on hard gel cap (HCG) saquinavir at 1.8 grams/day, switch to the more bioavailable soft gel capsule (SGC) formulation at 3.6 grams/day, or switch to indinavir at 2.4 grams/day. They were asked not to switch underlying nucleoside analogues for the first eight weeks of the study. The primary endpoint was virologic response. The study would stop early if no arm achieved greater than a 0.7 log10 reduction in HIV RNA. After an interim analysis conducted when 72 patients reached 8 weeks of follow-up showed that no arm did in fact achieve such a reduction, ACTG 333 was terminated. Participants had received an average of 112 weeks of prior saquinavir therapy. 86% were male, 75% white, non-Hispanic, and the median age was 43. Median baseline HIV RNA was 20,911 copies/mL; 6% had fewer than 200 RNA copies/mL at entry. Median baseline CD4 was 220 cells/mm3. Follow-up for the first 72 subjects was a median 18 weeks (range 12-22 weeks).

ACTG 333: 8 Week RNA + CD4 Results

	HIV RNA (log10) reduction	% undetectable (<200/mL) ever*	CD4 at week 8 (/mm3)	change
SQV-HGC	+0.04 log	2/24 (8%)	2/22 (9%)	- 0.4 cells
SQV-SGC	-0.23 log	4/22 (18%)*	2/20 (10%)	+ 37 cells
IDV	-0.58 log	9/21 (43%)	7/19 (37%)	+ 22 cells

* Undetectable at one or more of the week 2, 4, 6, or 8 timepoints.

The study team commented that, "while there was variability in the RNA responses in individual subjects in both the IDV and SQVsgc arms, the mean decreases in RNA and mean CD4 cell increases in both arms was [sic] less than seen in other trials of protease inhibitor[s] used in combination with nucleosides."

Based on these disappointing results, accrual to ACTG 333 was terminated. Already enrolled patients were allowed to remain on assigned therapy or switched based on virological response. Genotypic resistance analyses are underway. The study will end formally on 14 July 1997 (Bastille Day).

Several things are notable about ACTG 333:

1. These were sequential monotherapy patients, many given first AZT, then AZT/ddC or AZT/saquinavir (in ACTG 229), then given SQV-HCG, SQV-HCG or indinavir, without regard to treatment history or virological status at baseline. Certainly the trial would be designed differently if it were begun today.
2. ACTG 333 participants had almost two years (112 weeks) of previous saquinavir experience upon enrolling into 333.
3. Most participants switched to SQV-SGC did not experience much of an antiretroviral benefit. The minority who did probably had not been receiving therapeutic doses of SQV-HGC, and hence had not developed SQV resistance.
4. Most participants switched to indinavir experienced far less of a viral load reduction than typical with this drug when given as a first protease inhibitor. However, results are given for indinavir patients as a group. Most likely they fall into three subgroups:
   • fully susceptible to indinavir;
   • partially susceptible to indinavir (as suggested by the group average); and
   • wholly resistant to indinavir.

What proportion of patients fall into each category is an intriguing question which may be answered, at least in part, by the ongoing resistance analyses.

After the ACTG 333 fiasco, Roche called various community groups in a series of anxious conference calls to try and squelch doubts raised by the study. Roche's whole marketing campaign for INVIRASETM was based on the drug's alleged tolerability and the presumption that you could use it as a first-line protease inhibitor and then go on to use others without fear of cross-resistance. ACTG 333 called this notion into doubt. Moreover, on one of these calls, Roche representatives revealed that saquinavir HGC, when used with AZT and 3TC in antiretroviral-naive individuals, lowered viral load beneath the limit of detection in fewer than 40% of patients -- less than AZT/ddI/nevirapine in INCAS/BI 1046.

Roche's anxieties were deepened when it apparently received a preliminary draft of the HHS Clinical Practice Guidelines for Treatment of HIV Infection and discovered that -- quelle surprise! -- saquinavir did not make the cut as a strongly recommended first-line protease inhibitor.

Spurred by the prospect of being left off formularies across the country, Roche decided to accelerate its filing for FDA approval of the new saquinavir formulation.

Thus it was that on 14 May 1997 Roche convened a conclave of treatment advocates from the East Coast and the Midwest to hear the exciting new data on its new formulation, soft gel capsule (SGC) saquinavir. The meeting took place at the chic, sleek, postmodern Soho Grand Hotel in lower Manhattan.

They had a new team of eager young investigators and public relations experts who, they earnestly explained, wanted to "open doors", "start an ongoing dialogue" -- even "set up a community advisory board". Gasps emanated from the activists who remembered the fiasco of Roche's previous CAB, which resigned en masse amidst screams and spilled shrimp cocktail at a melee at the Times Square Marriot Marquis in summer 1992 over ddC. Roche's new team, unaware of its predecessors' plight, quickly redubbed the proposed CAB a "task force".

Clinical team manager Laurent Fischer, M.D., presented preliminary data on new (SGC) saquinavir and asserted that SGC provided eight to nine times the exposure of the licensed hard gel capsule (HGC) formulation.

Activists at the meeting were skeptical, assailing Roche's failure to define an MTD before bringing saquinavir to market, and said since the drug company had made its bed, now it must lie in it. Some asked the company to reduce the price of the current formulation by 7/8 (to approximately $875 per year) to reflect Roche's new assessment of its potency.

Key studies of the soft gel cap saquinavir include NV15107, a dose finding study which "identified 1200 mg three times daily as the preferred dose," NV15182, a safety study, and NV15355C, a virological equivalency study comparing hard to soft gel caps in 160 treatment naive patients in the USA and Canada. They will be randomized to receive (open-label):

• SQV-HGC 600 mg tid + 2 new nucleosides, or
• SQV-SGC 1200 mg tid + 2 new nucleosides

Note that, just to be sure, Roche is giving twice the dose of the new formulation compared with the old (1200 vs. 600 mg tid) -- which would likely make it superior even if the new formulation were no more bioavailable (remember Schapiro 1995?). The company claims SGC saquinavir is 12% bioavailable (compared with 4% for hard gel caps), and stated that a monotherapy study among 22 volunteers demonstrated a 1.43 log10 (96.3%) reduction in HIV RNA. The primary "efficacy' comparison in NV15355C will be HIV RNA and CD4 changes over the first 16 weeks, after which SQV-HGC patients will be rolled over to SQV-SGC and followed for a further 32 weeks. The 16 week analysis is due to be complete by summer 1997 and will presumably be the basis for the FDA filing.

Ongoing & Planned Studies of SQV-SGC

	PI + RTIs	2 PIs/2 RTIs*	PI/NNRTI/RTI*	N
Antiretroviral naive	2 studies	2 studies	1 study	437
RTI experienced, protease naive	2 studies	3 studies	2 studies	845
Protease experienced	--	2 studies	2 studies	370
N	330	692	630	1,652

Needless to say, several additional studies continue to follow patients on hard-gel cap saquinavir, including the European study in antiretroviral naive patients, SV14604C (AZT vs. AZT/ddC vs. AZT/saquinavir vs. AZT/ddC/saquinavir -- though I don't know if this outdated design is still being used).

At the May meeting, Roche had the effrontery to claim that in ACTG 333, "patients switching protease inhibitor showed benefit" and attributed the disappointing results to "evolving treatment strategies".

This evoked considerable outrage. In fact, at the New York meeting and at a subsequent one in California, activists demanded that Roche immediately stop running its "Strategy" advertisements for INVIRASE, and stop advertising the drug as first-line therapy until FDA approves the soft gel caps.

We look forward go seeing whether the FDA concurs with Roche's assessment of the potency of saquinavir SGC, and to its use in creative and novel antiretroviral combinations.

As for those who have believed Roche and taken saquinavir HGC at the approved dose, the company has announced no plans to compensate them for whatever options this therapeutic choice may have foreclosed.

In summary:

1. Roche went to market in November 1995 knowing that the dosage and formulation of saquinavir for which it sought approval were suboptimal and might lead to resistance or cross-resistance.
2. Roche promoted saquinavir as a first-line protease option for 18 months while studying higher doses and a new formulation.
3. ACTG 333 reveals that individuals who took saquinvir HGC are less likely to experience a maximal response from either saquinavir SGC or indinavir.
4. Individuals considering starting combination therapy with a potent protease inhibitor should avoid starting with saquinavir at least until the new formulation is licensed by FDA, and then only if data support Roche's assertion that it is much more potent than the HGC.
5. In the interim, the only way to achieve maximal doses of saquinavir (HGC) is to double the dose and take it with a potent cytochrome p450 inhibitor such as ritonavir or nelfinavir. Even among those whose insurers will cover this, it will cost $14,000 per year for the saquinavir alone (never mind the nukes), which is unconscionable.
6. Roche should consider some form of compensation for individuals who have taken saquinavir HGC and may have developed cross-resistance to other, more potent protease inhibitors from which they may not now benefit.

After the meeting, Roche invited the activists upstairs for cocktails and "refreshments". Let us hope that the Soho Grand's cocktails were more potent than those being hawked by the unscrupulous pharmaceutical giant. I wouldn't know; I didn't go.

treatment index

Last modified: 6/13/97
Treatment Action Group
350 Seventh Avenue, Suite 1603
New York, NY 10001
phone: (212) 971-9022
fax: (212) 971-9019
copyright © 1997 TAG