Efficacy and safety profile of doxofylline compared to theophylline in asthma: a meta-analysis

This meta-analysis has been submitted to the international database of prospectively registered systematic reviews (PROSPERO, registration number: CRD42019119849), and performed in agreement with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocols (PRISMA-P) [15]. The PRISMA flow diagram is reported in Fig. 1a. This quantitative synthesis satisfied all the recommended items reported by the PRISMA-P checklist [15].

Two reviewers performed a comprehensive literature search for randomized controlled trials (RCTs) evaluating the influence of doxofylline and theophylline in adolescent and adult asthmatic patients. The PICO (Patient problem, Intervention, Comparison, and Outcome) framework was used to develop the literature search strategy, as previously described [16]. Namely, the “Patient problem” included subject affected by asthma; the “Intervention” regarded the administration of doxofylline and theophylline; the “Comparison” was performed with regard to placebo and across each active treatment; the “Outcomes” were the asthma events, adverse events (AEs), lung function expressed as forced expiratory volume in 1 s (FEV₁), and use of salbutamol as rescue medication.

The terms “doxofylline” AND “theophylline” AND “asthma” were searched in Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, Scopus, Web of Science, ClinicalTrials.​gov and EU Clinical Trials Register databases in order to provide for relevant studies available up to December 7, 2018. No language restriction was applied. The following Query Translation was used: (“asthma” [MeSH Terms] OR “asthma”[All Fields]) AND (“theophylline” [MeSH Terms] OR “theophylline” [All Fields]) AND (“doxofylline” [Supplementary Concept] OR “doxofylline”[All Fields]).

Citations of previously published analyses and relevant reviews were examined to identify further pertinent studies, if any [6, 8, 17‐19].

Published RCTs involving asthmatic patients treated with oral formulations of doxofylline and theophylline were included in this meta-analysis.

Two reviewers independently checked the relevant studies identified from literature searches obtained from the already mentioned databases. The studies were selected in agreement with the above-mentioned criteria, and any difference in opinion about eligibility was resolved by general consensus.

The primary endpoints of this meta-analysis were the impact of doxofylline on the change in asthma events and the risk of AEs, compared to theophylline and placebo. Asthma events were defined as episodes of increased symptoms or increased airflow limitation responsive to as needed short-acting β₂ agonists or resulting in either unscheduled medical attention, unscheduled use of oral corticosteroids, hospital admission [20]. The secondary endpoints were the impact of doxofylline on the change in FEV₁ and salbutamol use, compared to theophylline.

The Jadad score, with a scale of 1 to 5 (score of 5 being the best quality), was used to assess the quality of the RCTs concerning the likelihood of biases related to randomization, double blinding, withdrawals and dropouts [21]. A Jadad score ≥ 3 was defined to identify high quality studies. Two reviewers independently assessed the quality of individual studies, and any difference in opinion about the quality score was resolved by consensus.

In the pairwise meta-analysis moderate to high levels of heterogeneity between-studies were considered for I² > 50%; the risk of publication bias was assessed by applying the funnel plot and Egger’s test, as previously described [21]. Evidence of asymmetry from Egger’s test was considered to be significant at P < 0.1, and the graphical representation of 90% confidence bands have been presented [21]. The risk of bias in the network meta-analysis was checked via the consistency/inconsistency analysis to assess whether the outcomes resulting from the consistency and inconsistency models fit adequately with the line of equality, as previously described [22].

The quality of the evidence was assessed for the primary endpoint in agreement with the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system, indicating ++++ for high quality of evidence, +++ for moderate quality of evidence, ++ for low quality of evidence, and + for very low quality of evidence [23].

Data from included RCTs were extracted and checked for study characteristics and duration, enrolled patients, treatments and doses, disease characteristics, age, gender, asthma events, previous hospitalizations, lung function, and Jadad score. Data have been extracted in agreement with DECiMAL recommendations [24].

A pairwise meta-analysis was performed to quantify the impact of doxofylline compared to theophylline on primary and secondary endpoints.

Results of the pairwise meta-analysis are expressed as mean difference (MD) or relative risk (RR), and 95% confidence interval (95%CI). Since data were selected from a series of studies performed by researchers operating independently, and a common effect size cannot be assumed, binary random-effects model was used in order to balance the study weights and adequately estimate the 95%CI of the mean distribution of drugs effect on the investigated variables [21].

A network meta-analysis was also carried out to perform a comparison across doxofylline, theophylline, and placebo with respect to the primary endpoints, and to rank their efficacy in reducing asthma events and the risk of AEs.

The network meta-analysis was carried out by including RCTs that introduced no significant heterogeneity and bias in the effect estimates of primary endpoint. Since heterogeneity and bias may propagate through a network of RCTs, and thus affect the estimates differentially across regions of the network, this approach permitted to identify those studies that might alter the correct results of the network meta-analysis [25].

Full Bayesian evidence network was used in the network meta-analysis (chains: 4; initial values scaling: 2.5; tuning iterations: 20.000; simulation iterations: 50.000; tuning interval: 10). The convergence diagnostics for consistency and inconsistency was assessed via the Brooks-Gelman-Rubin method, as previously described [26]. Results of the network meta-analysis are expressed as relative effect (RE) and 95% credible interval (95%CrI). The probability that each intervention arm was the most effective was calculated by counting the proportion of iterations of the chain in which each intervention arm had the highest mean difference, and the surface under the cumulative ranking curve (SUCRA), representing the summary of these probabilities, was also calculated. The SUCRA is 1 when a treatment is certain to be the best, and 0 when a treatment is certain to be the worst [27].

Sensitivity analysis was performed to identify the studies that introduced heterogeneity and bias in the effect estimate of primary endpoints, if any. Meta-regression analysis was performed to identify the factors that might modulate the efficacy and safety profile of the investigated agents with regard to primary endpoints.

The specisafety profile of the investigated treatments was investigated through a pooled analysis of the frequency of specific AEs.

OpenMetaAnalyst [28] and GeMTC [29] software were used for performing the meta-analysis, GraphPad Prism (CA, US) software to graph the data, and GRADEpro GDT to assess the quality of evidence [23]. The statistical significance for the effect estimates resulting from the pairwise and network meta-analyses was assessed for P < 0.05.

Data obtained from 696 asthmatic patients (44.68% treated with doxofylline, 31.62% treated with theophylline, and 23.71% treated with placebo) were selected from 4 RCTs published between 2015 and 2018 [8‐10]. The relevant characteristics of studies, disease, and patients are described in Table 1; Fig. 1b shows the network across the treatments.

All the RCTs were published as full-text papers [8‐10]. The DOROTHEO 1, DOROTHEO 2, and the study of Margay et al. were published as high quality studies (Jadad score ≥ 3) [8, 9], whereas the study of Lal et al. as low-quality study (Jadad score = 1) [10]. The duration of treatment ranged from 6 weeks to 12 weeks.

The pairwise meta-analysis indicated that doxofylline was significantly (P < 0.05) more effective than theophylline in reducing the daily asthma events (MD -0.14, 95%CI -0.27 – 0.00) (Fig. 2a). The effect estimate was not affected by heterogeneity (I² 0%, P = 0.53). Although a certain level of asymmetry resulted by the visual inspection of funnel plot, the Egger’s test found not significant publication bias (Fig. 2b and c). The risk of AEs was significantly (P < 0.05) lower in patients treated with doxofylline than in those that received theophylline (RR 0.76, 95%CI 0.59–0.99) (Fig. 2d). The substantial but not significant heterogeneity (I² 59%, P = 0.05) was confirmed by the visual inspection of funnel plot (Fig. 2e), and mainly related with the study of Lal et al. [10]. However, the analysis performed via Egger’s test indicated that, although the study of Lal et al. [10] was small and characterized by low quality, it introduced no significant publication bias in the effect estimate concerning the risk of AEs (Fig. 2f).

The results of the network meta-analysis indicated that doxofylline, but not theophylline, significantly (P < 0.05) reduced the daily asthma events (RE: − 0.33, 95%CrI -0.62 - -0.04, − 0.33; − 0.23 95%CrI -0.54 - 0.04; respectively, compared to placebo). Doxofylline was also safer than theophylline, reporting a significant (P < 0.05) reduction in the risk of AEs (RE 0.53, 95%CrI 0.27–0.87; doxofylline vs. theophylline). The network meta-analysis also showed that doxofyline was the most effective treatment (upper quartile in the SUCRA ranking) with respect to both theophylline (second quartile in the SUCRA ranking) and placebo (lower quartile in the SUCRA ranking) (Fig. 2 g). As expected, placebo was the safest arm (upper quartile in the SUCRA ranking), followed by doxofylline (second quartile in the SUCRA ranking); conversely, theophylline was the less safe treatment (lower quartile in the SUCRA ranking) (Fig. 2 h).

The consistency/inconsistency analysis showed that all points fit adequately with the line of equality (efficacy: R² 0.99; slope 1.05, 95%CI 0.96–1.15), indicating that the network meta-analysis was not affected by significant bias.

Meta-regression analysis confirmed that the patient demographics, baseline and study characteristics did not modulate the efficacy and safety profile of the investigated agents.

The GRADE analysis indicated high quality of evidence (++++) for doxofylline vs. theophylline with respect to efficacy and safety profile in both pairwise and network meta-analysis.

There was no significant difference between doxofylline and theophylline on the change in change in FEV₁ (P > 0.05) (Fig. 3a). A trend of superiority (P = 0.058) was detected for doxofylline over theophylline with respect to the reduction in the use of salbutamol as rescue medication (Fig. 3b).

The pooled analysis of safety profile showed that the AEs with a frequency greater than 5% were headache (doxofylline 20.61%, theophylline 23.64%), nausea (doxofylline 10.96%, theophylline 21.82%), nervousness (doxofylline 4.39%, theophylline 11.36%), and dyspepsia (doxofylline 6.58%, theophylline 8.18%). AEs were generally mild in severity, and detailed frequencies of further specific AEs are reported in Table 2.