NEWS - Vioxx MI risk seen in 2000: drug should have been withdrawn then, says meta-analysis

[Guest guest]

Rofecoxib MI risk seen in 2000: drug should have been withdrawn then,

says meta-analysis

Rheumawire

Nov 4, 2004

Zosia Chustecka

London, UK - An independent meta-analysis of clinical trial data with

rofecoxib (Vioxx, Merck & Co) published by the Lancet online November 5,

2004 [1] shows that a doubling of the risk of myocardial infarction (MI)

with the drug could be seen by 2000, and the researchers suggest that

the drug could and should have been withdrawn then4 years before it was

taken off the market.

When Merck withdrew rofecoxib on September 30, 2004, the company said in

a press release: " Given the availability of alternative therapies, and

the questions raised by the data, we concluded that a voluntary

withdrawal is the responsible action to take. " The authors of the new

meta-analysis, Dr Juni (University of Berne, Switzerland) and

colleagues say that this " same statement could and should have been made

several years earlier, when the data summarized here first became

available. Instead, Merck continued to market the safety of rofecoxib. "

This discovery points to astonishing failures in Merck's

internal systems of postmarketing surveillance, as well as to lethal

weaknesses in the US FDA's regulatory oversight.

" This discovery points to astonishing failures in Merck's internal

systems of postmarketing surveillance, as well as to lethal weaknesses

in the US FDA's regulatory oversight, " blazes an accompanying editorial

by Lancet editor Dr Horton [2]. This report will " add

significant weight to ongoing litigation against Merck by patients who

believed they were harmed by this drug, " he adds.

The meta-analysis was funded by the Swiss National Science Foundation

and reviewed all randomized controlled clinical trails in adult patients

with chronic musculoskeletal disorders that compared rofecoxib with

placebo or with other nonsteroidal anti-inflammatory drugs (NSAIDs). In

total, 63 reports were reviewed, covering 18 clinical trials (involving

25 273 patients) with 5 extension studies and 11 observational studies.

All of the trials were sponsored by Merck & Co.

Fatal or nonfatal myocardial infarction was the primary end point, and

64 events were found: 52 in the rofecoxib groups vs 12 in the control

groups (combined relative risk 2.24, 95% CI 1.24-4.02), with little

evidence of between-trial heterogeneity.

Cumulative meta-analysis showed that an increased risk of MI became

evident in 2000; at that point, 14 247 patients had been randomized and

44 events had occurred. At the end of 2000, with 52 MIs in 20 742

patients, the relative risk was 2.30 (95% CI 1.22-4.33, p=0.01). " The

effect was both substantial and unlikely to be a chance finding, " the

authors write.

Juni, the lead author of the meta-analysis, tells rheumawire that the

study shows that " by the end of 2000, there was enough evidence

accumulated to show that the risk of myocardial infarction with

rofecoxib was more than doubled. "

After that, until the end of 2001 when more data on cardiovascular end

points were coming in, this point estimate remained stable, only the

statistical precision increased, he says. After 2001, no additional

cardiovascular safety data were reported from musculoskeletal trials

with rofecoxib.

By the end of 2000, there was enough evidence accumulated to

show that the risk of myocardial infarction with rofecoxib was more than

doubled.

In an interview, asked whether there was enough evidence at the end of

2000 to remove the drug from the market, Juni answered carefully, saying

the statement that Merck released recently when withdrawing the drug

could have " arguably " been made several years earlier.

Juni tells rheumawire that the meta-analysis also shows that this

increased risk of MI with rofecoxib did not vary with controls (similar

for placebo, naproxen, or a nonnaproxen NSAID), did not vary with the

daily dose of rofecoxib used (12.5 mg, 25 mg, and 50 mg), and did not

depend on length of treatment (it was similar for trials of duration

less than 6 months and those of more than 6 months). " So, the reassuring

statement from Merck that these risks occur only after 18 months of

treatment cannot be confirmed with our data, " he comments. " Our findings

indicate that patients are at risk even if rofecoxib is taken for a few

months only. "

This published meta-analysis of rofecoxib data is part of a larger

network meta-analysis of analgesics in osteoarthritis, Juni explains.

All of the data used for the rofecoxib meta-analysis were contained

within the FDA databases, and he estimates that it took about 1600

working hours and cost about CHF40 000 to perform.

" This is ridiculously little compared with the potential impact that

this study may have, " Juni comments. " So, one of the questions raised is

shouldn't these types of analyses be routinely performed by

drug-approval bodies, which have appropriate access to the original

data? " He adds that, as the data were all there, the FDA could have done

an analysis much earlier than they themselves did (they began in

April/May 2004).

As part of their review, Juni et al also combed through the trial data

to check for evidence of a cardioprotective effect of naproxen. This has

been proposed as the explanation for the difference in the MI findings

seen in the VIGOR trialie, that the fewer MIs seen in the naproxen group

were a result of its cardioprotective effect, rather than the increased

number of MIs in the rofecoxib group being an adverse effect of the

drug.

Juni et al report that the risk of MI was indeed greater in trials

comparing rofecoxib with naproxen than with other controls but say this

finding was " probably attributable to chance. " Taken together, data from

their analysis and from other observational pharmacoepidemiological

studies indicate that " if a protective effect of naproxen exists, it is

probably small, and, as pointed out earlier [by others], not large

enough to explain the findings of VIGOR. "

That rofecoxib could increase the risk was not discussed,

despite the fact that, since the mid-1990s, the drug has been known to

reduce prostacyclin, a vasodilator and inhibitor of platelet

aggregation.

One lesson for the future from this is that " we can never be sure we

know all there is to know about mechanisms, " say the researchers. " The

VIGOR study group presented the myocardial infarction data exclusively

as a 'reduction in the risk of myocardial infarction in the naproxen

group' on the basis of documented inhibition of platelet aggregation by

naproxen, but not rofecoxib. That rofecoxib could increase the risk was

not discussed, despite the fact that, since the mid-1990s, the drug has

been known to reduce prostacyclin, a vasodilator and inhibitor of

platelet aggregation. "

Another lesson to be learned is directed at the FDA and other

drug-licensing authorities, which have a duty to carefully and

continuously monitor the evidence on the adverse effects of drugs, say

the researchers. " Clearly this has not happened in the case of

rofecoxib: the most recent labeling information in the USA, for example,

mentioned only 3 clinical trials. Had accruing data been analyzed

cumulatively as soon as they became available, appropriate and timely

decisions could have been taken. "

In the accompanying editorial, the Lancet withdraws its earlier praise

of Merck for acting promptly in the face of new findings about the

safety of Vioxx and cites the revelations made earlier this week in the

Wall Street Journal that the company knew about the cardiovascular risk

of rofecoxib for years. This " disturbing contradiction, " where Merck

understood Vioxx's true risk profile but attempted to gloss over these

risks in its public statements at the time may spell the end for the

company, it suggests. " It has been financially disabled and its

reputation lies in ruins. It is not at all clear that Merck will survive

this growing scandal. "

This refusal to engage with an issue of grave clinical

concern illustrates the agency's built-in paralysis, a predicament that

has to be addressed through fundamental organizational reform.

The FDA also comes in for stinging criticism. " The agency was urged to

mandate further clinical safety testing after a 2001 analysis suggested

a 'clear-cut excess number of myocardial infarctions.' It did not do so.

This refusal to engage with an issue of grave clinical concern

illustrates the agency's built-in paralysis, a predicament that has to

be addressed through fundamental organizational reform, " Horton writes.

Researchers involved with rofecoxib are also blamed " allegedly credulous

specialists who should have asked tougher questions about the drug they

were prescribing. " The editorial comments that why clinical

investigators studying rofecoxib did not do more to raise concerns is a

" fair question that needs to be answered. But in doing so, we must not

diminish the importance of the covenant of trust that society has

established with powerful commercial and governmental institutions. "

" For with Vioxx, Merck and the FDA acted out of ruthless, shortsighted,

and irresponsible self-interest, " the editorial thunders. " The licensing

of Vioxx and its continued use in the face of unambiguous evidence of

harm have been public-health catastrophes. This controversy will not end

with the drug's withdrawal. "

Merck & Co has published a detailed scientific critique of

this Lancet meta-analysis on its web site [3], in which it explains why

it disagrees with the conclusions. In a press release, the company

reiterates that it was vigilant in monitoring and disclosing the

cardiovascular safety of rofecoxib and " absolutely disagrees with any

implication to the contrary. Until the APPROVE study, data from Merck's

clinical trials showed no significant difference in cardiovascular risk

between Vioxx and either placebo or nonnaproxen NSAIDs. "

Merck's main criticism of the meta-analysis is that it

violates the basic principle of combining " like with like, " as the

authors combined data from studies with 3 kinds of comparators (placebo,

naproxen, and nonnaproxen NSAIDs). " The authors justify combining the

data across the comparators because confidence intervals against

individual comparators were wide and the statistical test for

interaction was not significant. This use of an underpowered statistical

test as the sole justification for combining the data is scientifically

inappropriate. . . . There are known different biological effects of the

comparators on platelet function and the data demonstrate large

differences in relative risk between the comparator groups. " This

inappropriate combining of heterogeneous data " invalidates the results

and the conclusions of their meta-analysis, " the company says.

Sources

Juni P, Nartey L Reichenbach S, et al. Risk of

cardiovascular events and rofecoxib: cumulative meta-analysis Lancet;

published online November 5, 2004; available at http://www.lancet.com.

Horton R. Vioxx, the implosion of Merck, and

aftershocks at the FDA. Lancet; published online November 5, 2004;

available at http://www.lancet.com

Merck & Co. Response to article by Juni et al

published in the Lancet on Nov 5, 2004. November 5, 2004. Available at:

http://www.merck.com/statement_2004_1105/lancet.pdf

I'll tell you where to go!

Mayo Clinic in Rochester

http://www.mayoclinic.org/rochester

s Hopkins Medicine

http://www.hopkinsmedicine.org