Summary of Available Evidence
The evidence presented in this review was acquired from four manufacturer-sponsored phase III RCTs. In each trial, patients were randomized to receive treatment with weekly subcutaneous injections of 300 mg dupilumab following a loading dose of 600 mg on day one, treatment every other week with subcutaneous injections of 300 mg dupilumab following a loading dose of 600 mg on day one, or weekly subcutaneous injections of placebo. Patients in the SOLO trials were included if topical AD treatment was inadvisable or provided inadequate treatment. In LIBERTY AD CHRONOS, patients were included if topical treatment provided inadequate treatment, and excluded patients who experienced important side effects to topical medications (e.g., intolerance, hypersensitivity). The inclusion and exclusion criteria in LIBERTY AD CHRONOS were also reflected in the criteria for LIBERTY AD CAFÉ, with the additional inclusion criteria of either a history of prior CSA exposure and either inadequate response to CSA or intolerance and/or unacceptable toxicity, or patients had to have a history of being CSA-naive and not eligible for CSA due to medical contraindications or other reasons. All patients in LIBERTY AD CHRONOS and LIBERTY AD CAFÉ were required to use medium-potency TCS on active lesions. In the SOLO trials, use of any TCS was classified as use of rescue medication.
Across all studies, the proportion of patients with EASI-75 at week 16 was the primary efficacy end point. The proportion of patients with IGA 0 or 1 (on a five-point scale) and a reduction from baseline of two or more points at week 16 was an additional primary end point for the SOLO trials and LIBERTY AD CHRONOS, and a secondary end point for LIBERTY AD CAFÉ. The primary efficacy analysis used intention-to-treat methodology. Patients were classified as nonresponders for the time points following study withdrawal or use of rescue treatment. If a patient had a missing value at week 16, they were counted as a nonresponder at week 16. Hierarchical testing was applied to the primary and secondary end points at a two-sided significance level of 0.025 for the comparison between each dupilumab dose regimen and placebo. The trials also included the same primary efficacy analysis using per-protocol methodology. Sensitivity analyses were included that utilized alternative methods to account for missing data (last observation carried forward), and to assess all patient data, regardless of use of rescue medication with and without imputation (via multiple-imputation methodology). An AD-specific MCID was found for the EASI but not for the IGA although, according to the clinical expert consulted for this review, AD severity and response to therapy is typically assessed on a continuous basis at the discretion of the patient’s physician and not by using a tool such as the EASI or IGA.
Secondary end points assessing AD severity (i.e., SCORAD), AD symptoms (i.e., Pruritus NRS, POEM), and health-related quality of life (i.e., DLQI, EQ-5D) were consistent across all trials. Continuous end points used multiple imputation using the Markov-chain Monte Carlo algorithm and ANCOVA to account for missing data. The covariates included in the ANCOVA model included treatment group, baseline value, and randomization strata. Hierarchical testing was applied to secondary end points at a two-sided significance level of 0.025 for the comparison between each dupilumab dose regimen and placebo. Sensitivity analyses for secondary end points included analysis based on all observed data, regardless whether rescue treatment was used or if data were collected after withdrawal using the multiple-imputation method and mixed-effect model repeated measures, including factors (fixed effects) for treatment, baseline strata, visit, baseline value, treatment-by-visit interaction, and baseline-by-visit interaction as covariates. Sensitivity analyses using alternate methods to handle missing data were also conducted; these included the worst observation carried forward method, the last observation carried forward method, and no imputation.
Across the trials, the baseline characteristics and baseline disease severity were similar across treatment groups. One inclusion criteria in the LIBERTY AD CAFÉ trial required patients to have a baseline EASI of 20 or greater, while the other studies required a baseline EASI of 16 or greater. This specific inclusion criterion did not appear to select a more severe set of patients, as the mean EASI score for LIBERTY AD CAFÉ (mean = 33.5) was similar to that of the other studies (mean range = 32.6 to 34.4). Over the course of the studies, the greatest proportions of patients who discontinued the trial were most commonly from the placebo groups. AEs, including those related to the disease itself (i.e., AD flares), were cited as the main cause of discontinuation. While no major safety issues were noted in the trials, more patients in the dupilumab groups experienced AEs relating to eye disorders than those in the placebo group.
Interpretation of Results
Efficacy
Across the studies, the same tools were used to assess AD severity. These tools included the EASI, the IGA, and the SCORAD.
Consistency in results between the primary end points (using the EASI and IGA) and the secondary end point (using the SCORAD) was seen. Regardless which measure was used, the severity of AD showed a statistically significant (P < 0.0001) decrease in the dupilumab group compared with the placebo group at week 16. LIBERTY AD CHRONOS showed a consistent statistically significant (P < 0.0001) decrease in disease severity for all three end points at week 52. Sensitivity analyses showed minor numerical differences, but statistical significance remained consistent.
Secondary efficacy end points that assessed the symptoms of AD (Pruritus NRS and POEM) provided efficacy results similar to those of the AD severity end points. Regardless which measure was used, the intensity of AD symptoms showed a statistically significant (P < 0.0001) decrease in the dupilumab group compared with the placebo group at week 16 across studies. Patients were most concerned about symptoms such as pruritus. Additionally, patients were concerned about health-related quality of life. Regardless which quality-of-life tool was used (DLQI or EQ-5D-3L), the improvement in quality of life was statistically significant (P < 0.0001). ▬▬▬▬▬. In both the EASI and IGA efficacy end points, greater efficacy was seen in the LIBERTY trials’ placebo groups compared with the placebo groups in the SOLO trials. This was expected due to concomitant treatment with medium-potency TCS in the LIBERTY trials. This difference in treatment highlights the issue with comparing the trials but is important, as dupilumab is indicated for use with or without TCS. Subgroup data for some of the efficacy end points by geographic region were included and showed numerically variable results by region. With the data distributed between four regions (North and South America, Asia Pacific, Western Europe, Eastern Europe) sample sizes for each treatment group were small (N < 30) in some groups and the CIs were large; this precludes the ability to make specific quantitative conclusions about the treatment effect according to geographic region.
▬▬▬▬▬. Each of the four trials evaluated in this review were placebo-controlled; therefore, there is no evidence to assess the efficacy of dupilumab compared with other drugs. However, the availability of Health Canada–approved products to treat AD in patients who are suboptimally controlled on disease-specific skin care measures (irritant avoidance, emollients, bleach baths, etc.), TCS, and/or TCIs is lacking.
A matching-adjusted indirect comparison (MAIC) was identified in the company evidence submission template for the single technology appraisal of dupilumab by NICE.36 The company indicated that a MAIC was performed for the comparison between dupilumab and ciclosporin (the only EMA licensed immunosuppressant therapy available for patients with AD) due to the lack of active comparator trials and the lack of common comparators to enable an indirect treatment comparison of dupilumab versus ciclosporin. In the MAIC, the company included patient-level data from the CHRONOS trial to inform the efficacy and safety of dupilumab, and used aggregate-level data from two trials (Haeck, 2011 and Jin, 2015) to inform the efficacy and safety of ciclosporin. Improvements in absolute SCORAD values (after weighting) were significantly higher for dupilumab in comparison to ciclosporin suggesting that dupilumab + TCS has superior efficacy compared to low-dose cyclosporine and high-dose ciclosporin initiation followed by low-dose maintenance ciclosporin. However, the Evidence Review Group (ERG) indicated that the results of the MAIC were limited by small sample sizes, heterogeneity, and limited prognostic factors and agreed with the submitting company’s decision “not to place much emphasis on the result of the MAIC and would recommend interpreting the results with caution.”36
Harms
SAEs were reported in 1.7% to 4.7% of patients in the dupilumab group and 3.5% to 9.3% in the placebo group across trials. Regardless of treatment group, patients in LIBERTY AD CAFÉ had the highest frequency of SAEs. ▬▬▬▬▬. The greater number of SAEs present in the placebo group were not unexpected, as they relate to the condition itself. Two deaths occurred in SOLO 2 and one death occurred in LIBERTY AD CHRONOS. It was reported that the deaths were unrelated to the study drug. Consistently across trials, conjunctivitis (and general eye disorders) affected more patients in the dupilumab group compared with the placebo group. This may relate to the mechanism of action, as dupilumab binds specifically to the IL-4Rα subunit of the IL-4 and IL-13 receptor complex and inhibits the signalling related to the immune response. Although there are some harms associated with dupilumab, patients indicated the desire to try alternative medications, as many of them stated that none of the currently available treatments work. Current second-line treatment for moderate-to-severe AD includes off-label therapies such as methotrexate, cyclosporine, azathioprine, and mycophenolate mofetil. These therapies are generally used for a short duration, as they are associated with negative side effects and highlight the need for novel second-line therapies.
Potential Place in Therapy2
Dupilumab, an IL-4 and IL-13 antagonist that limits type 2 T helper–driven inflammatory activity, is the first biologic drug approved for treatment of moderate and severe AD in Canada.
Dupilumab has, in phase III trials, demonstrated efficacy in AD through 52 weeks of treatment.6 There is evidence that dupilumab is effective in patients who have failed.7 Safety analyses through 52 weeks have not shown serious concerns.
Presently, patients achieving suboptimal disease control with appropriate disease-specific skin care measures (irritant avoidance, emollients, bleach baths, etc.), TCS and/or TCIs, and narrow-band ultraviolet B (NB-UVB) phototherapy are offered treatment with off-label immunosuppressive drugs. In Canada, the most commonly chosen immunosuppressive drug is methotrexate followed by cyclosporine, azathioprine, and mycophenolate mofetil. Because of their potential toxicities, these drugs are generally prescribed as intermittent courses in AD. There are patients for whom some or all of these drugs are contraindicated or for whom toxicities limit their use. There are also patients who do not respond to these drugs.
In practice, dupilumab will likely offer a useful alternative for those patients who have contraindications to, experience adverse effects from, or are nonresponsive to immunosuppressive drugs. It will also be useful to that subset of patients who respond to immunosuppressive drugs, but who require continuous long-term systemic therapy.
