Questionable research and marketing of a combination drug for smoker’s lungs

In 2007, a pivotal trial sponsored by GlaxoSmithKline (GSK) was published in the New England Journal of Medicine. ¹ The trial’s name, ‘Towards a Revolution in COPD Health’ (TORCH), indicated that it had an ambitious aim, which was to find out whether it was beneficial to give corticosteroids to patients with chronic obstructive pulmonary disease (COPD). The primary outcome was total mortality.

GSK randomised 6184 patients to four groups: placebo; salmeterol; fluticasone; and both drugs together. By definition, this design is factorial. It is powerful, as it allows the investigators to study three research questions with a sample size that would usually only allow one question to be answered. Such a trial can tell us whether the two drugs are effective, and whether the combination is better than any of its components.

However, nowhere in the 15-page trial report was the correct factorial analysis to be found, and the abstract gave the readers the clear impression that the combination was better than any of its components ¹ :

The hazard ratio for death in the combination-therapy group, as compared with the placebo group, was 0.825 (95% confidence interval [CI], 0.681 to 1.002; p = 0.052, adjusted for the interim analyses), corresponding to … a reduction in the risk of death of 17.5%. The mortality rate for salmeterol alone or fluticasone propionate alone did not differ significantly from that for placebo.

The criticisms

The authors of a letter to the editor showed the correct factorial analysis. ² The effect of the combination was entirely due to salmeterol and this effect, a hazard ratio of 0.81 (0.70 to 0.94), or a reduction in mortality of 19%, was significant (p = 0.004). In contrast, the hazard ratio for fluticasone was 1.00 (0.87 to 1.15), p = 0.99.

Another letter pointed out the worrisome finding that pneumonia occurred more frequently in the two groups receiving fluticasone (in an average of 19% of the patients) than in the placebo or the salmeterol-only groups (13%) and that the number needed to harm was 17 (p < 0.001). ³

In their reply, the TORCH trial authors, one of whom was an employee of GSK, dismissed the factorial analysis by saying that it assumes that each treatment has the same additive effect in the absence and presence of the other treatment and that this was not the case. ⁴ However, they did not document their reservation. Other researchers have documented that a factorial analysis was indeed appropriate, as the interaction term was not statistically significant (p = 0.32). ⁵

The trial authors also noted that their ‘data show the clear clinical superiority of combination treatment with salmeterol and fluticasone, including fewer exacerbations and better health status’. ⁴ However, these secondary outcomes were not particularly reliable because many patients discontinued treatment, and the analyses only included data up till that point, i.e. the intention-to-treat principle was not adhered to. ¹

Misleading marketing

In 2013, GSK published a full-page advertisement in a Danish industry-funded throw-away journal that claimed that Seretide (called Advair in the UK), the combination of salmeterol and fluticasone, reduces the decline in lung function in patients with chronic bronchitis. ⁶ The advertisement encouraged doctors to give their patients with moderate (or worse) chronic bronchitis more air and a better life with Seretide with the argument that Seretide reduces the annual decline in lung function from 55 mL to 39 mL per year.

The source of these data was a post-hoc analysis of 86% of the patients who participated in the TORCH trial. The paper showed that the adjusted annual decline in FEV₁ was 55 mL in the placebo group, 42 mL in both the salmeterol and the fluticasone groups and 39 mL in the combination group. ⁷ The paper noted that the tiny difference of 3 mL (healthy people lose 30 mL per year) between the combination and each treatment alone was not statistically significant (p = 0.44 and 0.45, respectively).

However, the advertisement did not mention these results. It only said that the difference between the Seretide group and the placebo group was statistically significant (p < 0.05). Thus, the advertisement encouraged doctors to use the combination product while it concealed the fact that the combination is no better than any of its components.

The post-hoc paper listed 10 people as authors, of which three were GSK employees. The other seven were on GSK payroll in various functions, e.g. advisory board members, speakers and consultants. Two additional people, a person from GSK and a professional medical writer, were not listed as authors but were acknowledged for ‘technical support’.

I recently drew attention to the misleading marketing and research of Advair. ⁸ GSK did not address any of my criticisms but stated that the company has confidence that the Danish authorities and specialists are fully capable of making decisions and guidelines that benefit the patients. ⁹ One of the trial authors argued that one should not do analyses that were not prespecified in the trial protocol. ¹⁰ But the published protocol stated: ‘The other objectives of the study include comparisons of mortality in the SFC group with that seen in the salmeterol and FP groups, and in the salmeterol and FP groups compared with the placebo group’. ¹¹ Both GSK and AstraZeneca have not performed the appropriate analysis in other similar trials as well. ⁵

The role of the medical journal

The New England Journal of Medicine published a misleading analysis in the abstract that only included half of the patients, thereby spoiling the advantage of the factorial design. ¹ Favourable articles in prestigious journals generate sales of reprints, and are a powerful marketing tool for the company. ¹²

Although a recent increase in sales of asthma drugs was driven by patients with chronic bronchitis, it was not because they increased in number, but because they were shifted from cheap to expensive drugs. ¹³ In 2012, the sales of the combination was four times larger than the combined sales of the two components. ¹⁴

Two drugs are more harmful than one, and the increased incidence of pneumonia with fluticasone is worrying. Further, the combination costs double as much as salmeterol.