TORCH and UPLIFT: What Has Been Learned from the COPD “Mega-Trials”?

Abstract

TORCH and UPLIFT are amongst the largest and most ambitious COPD trials ever undertaken. In terms of the primary outcomes, both trials were negative. Compared with placebo, combined salmeterol and fluticasone therapy did not significantly reduce all cause mortality over 3 years in TORCH, and tiotropium did not slow the decline in lung function over 4 years in UPLIFT. Secondary outcomes from these studies strongly confirmed findings from previous trials. Monotherapy with all three drugs provided small improvements in respiratory health status and reductions in exacerbation rates with some additive effect from the salmeterol/fluticasone combination. Both salmeterol/fluticasone and tiotropium also reduced COPD hospitalization rates. The trials provide very strong evidence that the long-acting bronchodilators, salmeterol and tiotropium, are not associated with increased risk of death or major cardiovascular adverse events.

Key words: :

• COPD
• TORCH
• UPLIFT
• salmeterol/fluticasone
• propionate
• tiotropium

Inhaled pharmacotherapy has long been the mainstay of COPD therapy and in recent years long-acting bronchodilators and inhaled corticosteroids have assumed a more prominent role. Numerous randomized, controlled clinical trials have assessed the efficacy and safety of inhaled agents. The authors of one recent review synthesized findings from 42 randomized trials and 8 meta-analyses that had been published between 1997 and 2007, and other publications have appeared in the intervening period ([1]). Results from two of the more recent trials merit special consideration because of their large size and ambitious goals.

The “Towards a Revolution in COPD Health” (TORCH) trial evaluated the inhaled long-acting beta-agonist, salmeterol, and the inhaled corticosteroid, fluticasone, in 6,112 randomized COPD subjects with a planned follow-up of three years ([2]). The “Understanding Potential Long-Term Impacts on Function with Tiotropium” (UPLIFT) trial evaluated tiotropium in 5,993 randomized COPD subjects with a follow-up of four years ([3]). The patient-years of experience in each of these two trials exceeds that of most previous trials by an order of magnitude or more and the combined total of more than 40,000 patient years represents a substantial proportion of total trial exposure with these three classes of drugs. Significant discrepancies between pooled analyses of smaller trials and very large confirmatory trials are not uncommon and in some instances may be sufficiently large to alter recommendations and practice patterns ([4]). Hence, it is of interest to examine the results of TORCH and UPLIFT in light of previous trials and determine how they might impact the treatment of COPD.

TORCH was a four-arm randomized, placebo-controlled trial comparing the combination of salmeterol and fluticasone against the two individual components and against placebo over a three-year period ([2]). A systematic review of prior trials along with observational studies suggested that fluticasone as monotherapy, and perhaps its combination with salmeterol, might reduce mortality in COPD patients ([5], [6]). Those findings provided the rationale for a confirmatory trial, comparing the difference in all cause mortality between the combination of salmeterol and fluticasone treatment with placebo being the primary outcome. The majority of subjects were male (76%), the mean age was 65 years, and the mean post-bronchodilator FEV₁was 44% of predicted. The primary analysis was performed on a strict intent to treat basis, and of the 6112 patients enrolled in the trial, vital status at three years was ascertained in all but a single subject. Compared with placebo, the combined salmeterol and fluticasone treatment regimen reduced all cause mortality in relative terms by 17.5%, but the statistical significance of this difference (p = 0.052) fell short of the prespecified level required to reject the null hypothesis (p = 0.050).

UPLIFT was a two-arm randomized placebo-controlled trial comparing tiotropium with placebo over a four-year period in patients with characteristics very similar to those in TORCH ([3]). A retrospective analysis of two previous one-year, placebo-controlled trials suggested that tiotropium might slow the accelerated decline in lung function that characterizes COPD ([7]). UPLIFT represents an effort to confirm that observation. Co-primary outcomes were the mean rates of decline in the FEV₁ measured at regular intervals before and after bronchodilation over the four-year period. Compared with placebo, tiotropium had no or minimal effect on the mean annual decline for either the pre-bronchodilator FEV₁ (30 ± 1 ml vs 30 ± 1 ml, respectively) or the post-bronchodilator FEV₁ (42 ± 1 vs 40 ± 1, respectively).

As neither TORCH nor UPLIFT achieved its principal objective, it would appear unlikely that those particular results will have much effect on clinical practice. However, both trials collected much additional information and these secondary outcomes are of considerable interest, because they tend to support most, but not all, previous findings.

TORCH and UPLIFT confirmed that all drugs evaluated in the two trials improve respiratory health status to a statistically significant extent; mean unit change in the St. George's Respiratory Questionnaire (SGRQ), compared with placebo, being −2.7, −2.2, −1.2, and −3.1 for tiotropium, salmeterol, fluticasone, and the combination of salmeterol and fluticasone, respectively ([2], [3]). These results are similar to those found in previous trials; active drug treatment statistically significantly improved the SGRQ, but average improvement generally failed to achieve the 4-unit change that is deemed clinically meaningful ([1]). Compared with placebo, more patients receiving active study medications did experience a 4-unit improvement, but that proportion was small and the overall effect of these inhaled drugs on respiratory health status must be viewed as being rather disappointing.

As monotherapy, each of the three drugs evaluated in TORCH and UPLIFT reduced exacerbation rates to a similar extent; the relative reduction, compared with placebo, was 14%, 15%, and 18% for tiotropium, salmeterol, and fluticasone, respectively ([2], [3]). Of note, the combination of salmeterol and fluticasone in TORCH reduced exacerbations by 25% in relative terms compared with placebo, a reduction that is statistically significantly larger than that seen with either monotherapy. Tiotropium modestly reduced COPD hospitalizations in UPLIFT, while the combination of salmeterol and fluticasone, but not the individual components, had a similar effect in TORCH. Overall, these results offer no surprises and they are fully in line with previous trial experience ([1]).

While the primary analysis in UPLIFT failed to show any significant beneficial effect of tiotropium on decline in lung function, a secondary analysis of the TORCH data yielded positive effects ([3], [8]). Compared with placebo, salmeterol, fluticasone, and their combination slowed the annual rate of decline in FEV₁ by statistically significant annual rates of 13 ml, 13 ml, and 16 ml, respectively. The apparently discrepant results between TORCH and UPLIFT might be explained partly by differences in study design between the two trials. First, salmeterol and fluticasone given as monotherapy in TORCH slowed FEV₁ to the same extent with little additivity from the combination. Second, TORCH disallowed long-acting bronchodilators and inhaled corticosteroids apart from study medications, whereas UPLIFT permitted long-acting beta agonists or inhaled corticosteroids or both as part of usual therapy, and 74% of subjects reported use of at least one of those classes of drugs at baseline. In the remaining 26% who were using neither class of drug, annual rate of FEV₁ decline was smaller by 7 ml in the tiotropium arm, as compared with placebo (p = 0.046). These results suggest that monotherapy with any of the three study drugs might slow lung function decline in COPD. These results are provocative and of potential clinical value, but they cannot be viewed as conclusive.

A major weakness of both TORCH and UPLIFT is the large amount of missing data. Approximately 40% of randomized patients discontinued study drug prior to the planned completion date in each of the trials and information about health status, exacerbations, and lung function was not collected subsequent to stopping the study drug. As has been observed previously with these drug classes, differential drop out rates occurred according to treatment assignment with the highest losses occurring in the placebo arms. These losses were probably not random, as it has been shown that poor lung function is a strong predictor of study drug termination ([9]). Differential and nonrandom subject losses could minimize treatment differences under some conditions, but might exaggerate them under other scenarios. Missing information always introduces some residual uncertainty about the appropriateness of analysis plans and the interpretation of results.

All cause mortality data from TORCH and UPLIFT may be the most interesting and clinically important information obtained in these trials, particularly as they relate to safety ([2], [3]). Safety has been a recurrent issue with inhaled bronchodilators since soon after their introduction 50 years ago and safety issues have been raised about inhaled corticosteroids as well. This subject has been reignited with recent claims, based on systematic reviews, that COPD patients treated with either an inhaled beta-agonist or an inhaled anticholinergic experience an excess of major cardiovascular events and possibly an increase in mortality ([10], [11], [12]).

The TORCH and UPLIFT investigators made concerted efforts to ascertain vital status at the protocol-defined termination date for all randomized patients, whether or not they prematurely stopped study drug. These efforts were successful for all but a single patient in TORCH and for all but 2–3% of subjects in UPLIFT, amongst whom 875 patients in TORCH and 921 patients in UPLIFT were known to have died. Compared with previous trials of long-acting bronchodilators, the TORCH and UPLIFT data sets relating to mortality are unmatched in terms of size and quality, and they provide strong evidence that neither salmeterol nor tiotropium causes excess deaths. A post hoc factorial analysis of the TORCH data indicate a statistically significant 19% all cause morality reduction in the two treatment arms that received salmeterol, as compared with the two arms that did not receive salmeterol ([13]). Similarly, as compared with placebo, subjects who received tiotropium in UPLIFT experienced a statistically significant 13% reduction in all cause mortality at the conclusion of the protocol-defined study period ([3]). Also, adverse event reports while patients were on study drug indicated no excess of major cardiovascular events with either salmeterol or tiotropium.

One unanticipated safety concern emerged from TORCH and that was an increased number of adverse events reported as pneumonia in patients who received fluticasone. These events were seen in both fluticasone treatment arms and they were not uncommon, occurring at a rate of about 3 events per 100 patient-years of therapy. The clinical importance of these events remains uncertain and the matter remains under investigation.

In summary, the TORCH and UPLIFT investigators and sponsors are to be commended for successful completion of these ambitious projects, but the results are unlikely to cause major changes in current prescribing practices. The small improvements in health status and exacerbation frequency seen with active treatments are closely in line with results from previous trials. The apparent slowing of FEV₁ decline associated with the three monotherapies is intriguing but the findings can be viewed as hypothesis generating only. For a number of reasons it would seem unlikely that confirmation trials will ever be attempted. Secondary analyses suggest that both salmeterol and tiotropium may reduce all cause mortality, but these findings also require confirmation. However, we can fairly conclude that it is highly unlikely that either of these drugs causes excess mortality in COPD patients, and that is valuable clinical information.