Triple therapy in uncontrolled asthma: a network meta-analysis of phase III studies

Discussion

The overall results of this network meta-analysis provide the high-quality evidence that triple combination therapies including an ICS administered at HD have greater beneficial impact than MD ICS/LABA/LAMA FDC and both MD and HD ICS/LABA FDCs in reducing the risk of moderate to severe exacerbation and improving trough FEV₁ in patients suffering from uncontrolled asthma, regardless of whether the monocomponents were combined in the formulation as FDC or free combination.

Such a superiority on the risk of exacerbation of triple combinations including an ICS at HD was further confirmed by the NNT analysis. In fact, while ∼40 patients had to be treated for 1 year with HD ICS/LABA/LAMA FDC to prevent one moderate or severe asthma exacerbation compared with MD ICS/LABA/LAMA FDC, the NNT was ∼33 and only ∼12 when compared with HD and MD ICS/LABA FDCs, respectively. Unexpectedly, the treatment with MD ICS/LABA/LAMA FDC resulted in a clinically appreciable NNT value of ∼17 when compared with MD ICS/LABA FDC, but not versus HD ICS/LABA FDC as the NNT value was ∼179. Since Kaplan–Meyer curves were not available for all the studies that included free combinations of HD ICS/LABA FDC+TIO [19], the available data were spurious and not suitable to adequately assess the NNT of this triple combination. In any case, looking at the specific relative risks of triple combinations including HD ICS, we can postulate that the NNT of HD ICS/LABA FDC+TIO could be similar to that of HD ICS/LABA/LAMA FDC.

Overall, the superiority of triple combination therapies with an ICS at HD over MD ICS/LABA/LAMA FDC, and of MD ICS/LABA/LAMA FDC over MD and HD ICS/LABA FDC, was confirmed in the subset analysis on severe asthma exacerbations. Conversely, all the combinations, excluding MD ICS/LABA FDC, were equally effective in reducing the risk of moderate asthma exacerbation. These findings suggest that triple combinations including an ICS administered at HD may represent the first treatment choice in patients with a history of severe asthma exacerbation, whereas in those patients with a history of moderate asthma exacerbation either MD ICS/LABA/LAMA FDC or HD ICS/LABA FDC, but not MD ICS/LABA FDC, may be used as first-line treatment.

Despite the unpredictable nature of asthma exacerbations regardless of disease severity, recent evidence indicates that the past history and severity of exacerbations in the year prior may predict the risk and severity of future asthma exacerbations [21]. In this respect, the results of the subset analysis on exacerbation severity may provide the rationale for a tailored therapy based on the severity of exacerbation that each patient experienced in the previous year, thus leading to the optimisation of the dose of ICS and the number of bronchodilators included in the FDC.

Concerning the secondary end-points investigated in this network meta-analysis, triple combination therapies were equally effective in improving ACQ score, and no safety concerns resulted with respect to the risk of SAEs, pneumonia and serious CV AEs.

Indeed, the studies included in this network meta-analysis enrolled symptomatic patients with uncontrolled asthma [17–20], for which the current Global Initiative for Asthma (GINA) document [1] suggests using MD ICS/LABA FDC at Step 4 and HD ICS/LABA plus either TIO or biological therapy depending on the phenotypic assessment at Step 5 as preferred controller therapies to control symptoms and prevent exacerbations. Furthermore, GINA [1] recommends HD ICS or add-on TIO at Step 4 as an alternative controller option.

Clearly, the evidence raised by this quantitative synthesis provides new horizons for the treatment of severe asthma, in which poorly controlled symptomatic patients could be effectively treated with triple combination therapies including different doses of an ICS according to the severity of previous exacerbations. The recent approval of the first once-daily single inhaler triple therapy for the treatment of asthma by the US Food and Drug Administration (FDA) [22] and the current data resulting from the studies included in this network meta-analysis [17–20] support the clinical benefit of adding a LAMA to ICS/LABA, an effect of class not related to a specific antimuscarinic agent. In the light of this evidence, it is expected that the next iteration of asthma recommendations and guidelines will also include ICS/LABA/LAMA FDCs as effective pharmacological strategies for symptom control and risk reduction. Moreover, and no less important, this meta-analysis allows us to clear the assumption that in the treatment of uncontrolled asthma a LAMA should be combined only with an ICS administered at HD, as well as a LABA, because MD ICS/LABA/LAMA FDC was effective on both symptoms and moderate asthma exacerbations.

The limitations of this study are related to the intrinsic characteristics of the Bayesian network approach [23–25], mainly due to the indirect comparison across treatments and to the fact that only five studies [17–20] were included in the quantitative synthesis. However, as shown in figure 2, the resulting network included several direct comparisons across the investigated treatments, leading to solid network loops also in the subset analyses carried out in more homogenous asthmatic populations. Therefore, also considering that data from a large number of patients (more than 9500 subjects) were analysed, the resulting effect estimates were free from any within- and across-studies risk of bias, at least for the co-primary end-points.

Furthermore, as correctly stated by Mauger and Apter [26], while some degree of variation across study populations could be acceptable in a pairwise meta-analysis, this can lead to biased results in a network meta-analysis. Thus, considering that four [18–20] of the five studies included in the Bayesian network enrolled frequent exacerbators, whereas only the CAPTAIN study [17] enrolled patients reporting less than one exacerbation in the previous year, we performed a sensitivity analysis by considering exclusively those studies in which populations of asthmatic patients with frequent exacerbations were investigated. Considering also the severity of exacerbation, the sensitivity analysis basically confirmed the main findings of the overall analysis, indicating that the results are robust. The small changes in the level of significance detected in the sensitivity analysis could be due by the fact that in the overall analysis some of the 95% credible intervals were lying close to the line of equality, thus the reduction in the study population due to the exclusion of the CAPTAIN study [17] changed the 95% credible intervals with no relevant modification of the effect estimates. Interestingly, in the sensitivity analysis we found greater consistency in the level of statistical significance between the risk of moderate to severe exacerbation and trough FEV₁ than in the overall analysis. This evidence corroborates the hypothesis that exacerbation-prone asthma may be a specific phenotype with implications for the targeting of exacerbation prevention strategies along with lung function improvement [27].

Certainly, the CAPTAIN study [17] also provided further important information concerning the potential confounder of disparate study populations. Specifically, Lee et al. [17] demonstrated that increasing the ICS dose resulted in improved outcomes among patients with high type 2 inflammatory biomarkers. On the other hand, in patients with low type 2 inflammatory biomarkers the increase in ICS dose provided no further disease improvement, while adding a LAMA was efficacious. This is an important finding supporting an alternative approach by adding a LAMA instead of escalating the dose of ICS in type 2-low asthma.

International recommendations and guidelines [1, 28] provide general definitions of asthma exacerbation; however, despite the attempt to make the definition of asthma exacerbation uniform [29, 30], to date there is no consensus on a standardised definition of moderate and severe asthma exacerbation [31–33]. Such a clinical unmet need led to the last intrinsic limitation of this network meta-analysis: apart from the TRIMARAN and TRIGGER RCTs [20], there was not consistency among the investigated studies in the definition of either moderate or severe asthma exacerbation. Unfortunately, this is a limitation that cannot be solved by a sensitivity analysis and that may potentially introduce some unquantifiable bias in the exacerbation outcome.

In conclusion, both ICS/LABA/LAMA FDC and free combination of TIO added to ICS/LABA FDC are effective and safe therapeutic strategies in patients suffering from uncontrolled asthma, with the level of the ICS dose representing the discriminating factor to treat patients with a history of moderate or severe exacerbation. Furthermore, here we provide the clinical evidence that adding either a LAMA or increasing ICS dose on a background of ICS/LABA/LAMA FDC triple FDCs may reduce the risk of severe exacerbation and improve lung function, and that adding a LAMA along with escalating ICS provides incremental effects. Indeed, the evidence raised by this quantitative synthesis may help to solve the inconsistencies across the primary publications [17–20] with respect to the beneficial impact of triple combination therapy against asthma exacerbation. However, there remains the question concerning the correct positioning of triple combination therapy in the GINA stepwise approach for adjusting treatment for individual patient needs [1]. In this respect, MD and HD ICS/LABA/LAMA FDCs should be tested in well-designed phase III RCTs enrolling separately asthmatic patients at Step 4 and 5 in order to guide clinicians to correctly practice personalised medicine. In any case, the decision of whether or not to first add a LAMA or escalate the dose of ICS, or both, in a poorly controlled patient on MD ICS/LABA FDC remains a clinical matter that may be driven by the overall level of disease control, available biomarkers or concerns over potential AEs.