Formulary committee evidence sheet. Historical exercise, evidence dated February 8, 2001

Rofecoxib and naproxen in VIGOR: the label should disclose the cardiac comparison

Policy call

Disclose a cardiovascular warning: Yes: disclose

Yes. As of February 8, 2001, if VIGOR results appear in rofecoxib labeling, the label should disclose the myocardial infarction comparison in the study description and carry a cardiovascular statement in the Warnings section. The gastrointestinal benefit does not warrant ignoring the cardiac comparison, and the naproxen protection explanation cannot be used to explain it away because this trial cannot distinguish it.

Basis: the FDA medical reviewer wrote, It would be difficult to imagine inclusion of VIGOR results in the rofecoxib labeling without mentioning cardiovascular safety results in the study description as well as the Warnings sections. (VX2, PDF p. 39).

Unknown counterfactual: Myocardial infarction risk for the same participants with no anti-inflammatory therapy is unknown. The absolute effect of either drug is not identified.

Cardiac versus stomach, each in its own unit

Stomach benefit, as published

confirmed events per 100 patient years

Rofecoxib: 2.1 per 100 patient years
Naproxen: 4.5 per 100 patient years
Reported relative risk, rofecoxib to naproxen: 0.5 (95% CI 0.3 to 0.6)
Follow up: median 9.0 months; 8,076 randomized

Source: VX1, abstract. Event counts are not given in this excerpt.

Myocardial infarction, FDA counts

crude risk per 1000 randomized participants

Rofecoxib: 20 of 4,047 = 4.94 per 1000
Naproxen: 4 of 4,029 = 0.99 per 1000
Crude risk ratio, rofecoxib to naproxen: 4.98
Excess with rofecoxib: 3.95 more per 1000 participants over trial follow up

Source: VX2, PDF p. 20.

The stomach figures are rates per 100 patient years; the cardiac figures are risks per participant. They are different units and are not subtracted or netted here. A lower bleeding rate does not cancel a higher infarction risk; a committee needs both disclosed.

Myocardial infarction by arm

Bars start at zero. Crude events per 1000 randomized participants: rofecoxib 4.94 (20 of 4,047), naproxen 0.99 (4 of 4,029). Source VX2, PDF p. 20, adjudicated sponsor counts discussed by the FDA reviewer. Gastrointestinal rates stay separate above with VX1.

Showing: reviewer reasoning, plus a chart row for the comparison VIGOR did not measure.

Could naproxen protection explain the gap?

Verdict: not distinguishable in this trial. Not established, not ruled out.

The VIGOR study was a large study with a longer drug exposure and follow-up than the two smaller studies (085 and 090). The cardiovascular thrombotic event rates, while not high, were significantly different between the two groups; most striking were the myocardial infarction event rates.

As there is no placebo group, it will be difficult to assess the CV thrombotic risk with rofecoxib use compared with no therapy at all.

FDA reviewer, VX2, PDF p. 36.

VIGOR compared two active drugs with no placebo group, so the same crude ratio fits higher risk with rofecoxib, lower risk with naproxen, or both. These sources give no direct evidence that measures a naproxen effect.

The dashed chart row marks that missing comparison: with no placebo group, the ratio of 4.98 cannot be split into a rofecoxib effect and a naproxen effect.

Article versus FDA table

Myocardial infarction, as printed and as counted
Measure	Article	FDA counts	Reading
Rofecoxib	0.4%	20 of 4,047 = 0.49%	FDA updated counts round to 0.5%, not 0.4%; these sources do not explain the gap.
Naproxen	0.1%	4 of 4,029 = 0.10%	Consistent.
Ratio	0.2, naproxen to rofecoxib (95% CI 0.1 to 0.7)	4.98, rofecoxib to naproxen	Same finding, opposite direction: 1 / 4.98 = 0.20.
Denominators	8,076 randomized	4,047 + 4,029	Sum matches.

Sources: VX1, abstract and VX2, PDF p. 20.

Other rows in the FDA table, audited one at a time

Crude events per 1000 randomized participants
FDA row	Rofecoxib	Naproxen	Ratio, rofecoxib to naproxen
Any Event	47 of 4,047 = 11.61	20 of 4,029 = 4.96	2.34
Cardiac Events (Fatal/Nonfatal)	28 of 4,047 = 6.92	10 of 4,029 = 2.48	2.79
Cardiovascular Death	6 of 4,047 = 1.48	6 of 4,029 = 1.49	1.00

The FDA note reads Patients may be counted in more than 1 row, but are only counted once within a row. (VX2, PDF p. 20). These rows overlap with each other and with the infarction row, so they are never added together. Cardiovascular deaths were equal (6 and 6), matching the article statement that cardiovascular death rates were similar (VX1).

What this evidence does not show

This study does not prove that rofecoxib causes myocardial infarction.
Naproxen is not shown to be a safe alternative.
No population-wide number of harmed people is estimated.
Gastrointestinal rates per patient-year are not netted against myocardial infarction risks per participant.
Excess is per 1000 trial participants randomized to rofecoxib rather than naproxen, over a median 9.0 months of follow-up. It is not a population count of harmed people.

Source ledger

VX1. Comparison of Upper Gastrointestinal Toxicity of Rofecoxib and Naproxen in Patients with Rheumatoid Arthritis. New England Journal of Medicine, dated 2000-11-23. Location: abstract; DOI linked in this ledger. Links: PubMed abstract record; Publisher article via DOI.
VX2. Review of cardiovascular safety database; NDA 21-042 S-007. FDA Division of Cardio-Renal Drug Products, dated 2001-02-01. Location: original FDA memorandum pages 1 and 18 and 34-37; preserved in an archival FDA correspondence scan; FDA Feb 8 2001 meeting briefing original docket. Links: FDA correspondence scan, PDF; FDA advisory briefing docket, PDF.

Counts in VX2 are sponsor submitted adjudicated VIGOR counts as discussed by the FDA reviewer, transcribed from PDF p. 20 of the archival scan.