A systematic review of randomized controlled trials for prevention or treatment of atopic dermatitis in dogs: 2008-2011 update
Background – The management of atopic dermatitis (AD) in dogs relies mainly on the use of interventions to reduce pruritus and skin lesions.
Objectives – To provide a critical analysis of recent clinical trials reporting the efficacy and safety of interventions for canine AD.
Methods – Systematic review of randomized controlled trials (RCTs) published, presented or completed between 2008 and 2011, which enrolled dogs with AD. The search was done using electronic databases, reviewing published meeting abstracts and sending queries to professional email lists. Trials reporting the efficacy of interventions aimed at treating, preventing or reducing glucocorticoid usage in atopic dogs were selected.
Results – Twenty-one RCTs were included. We found further moderate-quality evidence of efficacy and safety of oral glucocorticoids and ciclosporin for treatment of canine AD. There was additional moderate-quality evidence of the efficacy of a topical glucocorticoid spray containing hydrocortisone aceponate. Low-quality evidence was found for the efficacy and safety of injectable recombinant interferons, a budesonide leave-on conditioner, a ciclosporin topical nano-emulsion and oral fexofenadine. There is low-quality evidence of efficacy of oral masitinib, with a need for monitoring for protein-losing nephropathy. Finally, we uncovered low-quality evidence of efficacy of a commercial diet as a glucocorticoid-sparing intervention and of a glucocorticoid spray as a flare-delaying measure. Very low-quality evidence was found for the efficacy of other interventions.
Conclusions and clinical importance – Topical or oral glucocorticoids and oral ciclosporin remain the interventions with highest evidence for efficacy and relative safety for treatment of canine AD.
Atopic dermatitis (AD) is a common allergic skin disease of dogs that is currently defined as a genetically predisposed inflammatory and pruritic skin disease, with characteristic clinical features and an association with IgE antibodies most commonly directed against environmental allergens.1 This disease has recently been shown to have a strong impact on the quality of life of both affected dogs and their owners2,3 and, as such, it is probably one of the most important chronic skin diseases of dogs.
Recent international practice guidelines have highlighted the need for a multifaceted line of attack for the management of canine AD.4 Approaches to consider currently include the avoidance of flare factors, an increase in skin and coat hygiene and care, the control of skin infections and the use of pharmacotherapy to alleviate skin lesions and manifestations of pruritus.4 To reduce signs immediately during acute flares of AD, topical and/or oral glucocorticoids are suggested.4 For long-term pharmacological treatment of chronic or recurrent signs of canine AD, oral and/or topical glucocorticoids, topical tacrolimus, oral ciclosporin and injectable interferons are currently recommended.4 Finally, additional strategies are also used to prevent the recurrence of clinical signs.4
The recommendations for specific drugs included in the 2010 guidelines derived principally from two systematic reviews of interventions to treat dogs with AD. The first review, published in 2003,5 analysed results of clinical trials testing pharmacological interventions, whether the trials were randomized or not; it did not discuss studies evaluating the efficacy of essential fatty acid (EFA) formulations or allergen-specific immunotherapy, and it was limited to clinical trials published in peer-reviewed journals.5 In 2010, an international collaboration published a second systematic review using the more stringent methodology and the support of the Cochrane Skin Group.6 That study was limited to randomized controlled trials (RCTs), and there were no restrictions on publication type and status, languages or types of interventions.6 In that review, database searches were done in 2005 for trials from 1980 to 2004, and RCTs published in 2005, 2006 and 2007 were included in a prospective fashion. In all, 49 RCTs were scrutinized.6
To determine the efficacy of interventions to treat or prevent skin lesions and/or pruritic manifestations of canine AD, we carried out a systematic review of recent RCTs that had enrolled atopic dogs of any age and disease severity. These trials had to be completed, presented or published between 2008 and 2011. This paper serves, therefore, as an update of the previous systematic review.6 This article is written according to the reporting standards for systematic reviews and meta-analyses set up by the latest 2009 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.7
Protocol and registration
With minor changes in outcome measures and study selection criteria, the protocol for this systematic review was nearly identical to that of a recent publication.6 Owing to the lack of availability of online repositories accepting the advanced publication of veterinary systematic review protocols, it could not be published before this review was conducted.
Types of studies.
In this systematic review, we included solely RCTs of interventions aimed at treating or preventing AD in dogs. As our latest systematic review focused on trials published or presented from the early 1980s until the end of 2007,6 we limited our present analysis to trials published, presented or completed between 2008 and 2011, both years included. There were no language or publication status restrictions. Finally, we excluded studies that had been presented at meetings before 2007, and which had already been discussed in our 2010 review.6
Types of participants.
As done previously,6 dogs had to be diagnosed with AD based on (as minimal criteria) the presence of characteristic clinical signs and the exclusion of pruritic dermatoses of similar appearance.8,9 The fulfilment of published diagnostic criteria such as those of Willemse10, Prélaud et al.11 or Favrot et al.12 was considered acceptable if non-AD pruritic diseases had also been excluded according to current standards. If the RCT had enrolled dogs with different diagnoses (e.g. AD and other diseases or pruritic dogs without further diagnostic characterization), the study was excluded. We also eliminated four trials that enrolled laboratory dogs with experimentally induced atopic skin lesions.
Types of interventions.
Randomized controlled trials had to report either the treatment or prevention of manifestations of pruritus and/or skin lesions of canine AD. There was no restriction on the route or type of intervention. Comparators could either consist of a relevant placebo or an active medication, be it a different dose of the same drug or an intervention already recommended for treatment of canine AD.4
As done in our preceding systematic review,6 RCTs were categorized at ‘short-term’ if they lasted 8 weeks or less and ‘long-term’ if their duration extended beyond 8 weeks. We also separated studies aimed at preventing flares of canine AD from those designed to relieve existing signs (i.e. treatment sensu stricto).
Types of outcome measures.
As in our recent review,6 included studies had to report an assessment of the extent and/or intensity of pruritus and/or skin lesions after a preventive or therapeutic intervention. Trials solely reporting the safety of an intervention were not reviewed further.
Primary outcome measures were similar to those used recently, while secondary outcome measures were expanded from previous ones.6 Primary outcome measures consisted of the proportion of dogs with a good-to-excellent improvement at study end using a categorical global assessment scale assessed by either investigators (primary outcome 1a) or dog owners (primary outcome 1b).
As secondary outcome measures, we determined the percentage of dogs with complete – or near complete – remission of signs, as estimated by a reduction of 90% or more from baseline investigator-graded lesional (secondary outcome 1a) or owner-rated pruritus scores (secondary outcome 1b). We also extracted from the RCTs the percentage of dogs with a 50% or greater reduction from baseline of investigator-graded lesional (secondary outcome 2a) or owner-rated pruritus scores (secondary outcome 2b). In the absence of universally accepted validated severity scales for evaluating skin and pruritus in dogs with AD, the outcome measures listed above were determined from values assessed with any scoring scheme used by the study authors, but only if there were more than 10 possible grades of severity in the scales utilized.
Furthermore, whenever RCTs employed validated scales, such as the Canine Atopic Dermatitis Extent and Severity Index version 3 (CADESI-03)13,14 and Hill’s Pruritus Visual Analog Scale (PVAS),15,16 for which thresholds for normal dogs have been established, we added four other secondary outcome measures. For these RCTs, we calculated the percentage of dogs that, at trial’s end, had CADESI-03 values in the range of those of normal dogs (0-15; secondary outcome 3a) or of dogs with mild AD (16-59; secondary outcome 4a). Likewise, we determined the percentage of dogs with a PVAS in the range of that of normal dogs (0-1.9; secondary outcome 3b) or dogs with mild pruritus (i.e. the value anchored by the third lowermost descriptor, 2-3.5; secondary outcome 4b). We also calculated these outcome measures for other scales if there were clearly indicated benchmarks for the absence of signs (or normal dog status) and/or for mild AD.
Finally, for glucocorticoid-sparing effect trials and prevention studies, we used different outcome measures that appeared to be more clinically relevant for these unique study designs. These ad hoc outcome measures were the glucocorticoid dosage at study end and the time to relapse, respectively.
Whenever possible and to provide a better comparison of the efficacy between different interventions tested in placebo-controlled RCTs, we calculated numbers needed to treat (NNT) based on each of the available outcome measures. However, to limit the lack of relevance of NNTs because of a high random chance placebo effect in RCTs with small group sizes,17 NNTs were calculated only for large trials with more than 50 dogs per group.
Each NNT was calculated as follows.18
As an example, in the context of this review, an NNT of ‘n’ could be interpreted as follows: a veterinarian would have to treat ‘n’ dogs with AD with the active intervention to obtain one additional positive outcome over treatment with placebo. The lower the NNT, the stronger the treatment effect over placebo.
Finally, in this review, we extracted and reviewed adverse events following each nonplacebo intervention.
Studies were identified by searching three databases (Medline via PubMed, Thomson Reuters’ Web of Science and CAB Abstract via EBSCO Host) for the period between 1 January 2008 and 31 December 2011. Searches were done once on 2 January 2012.
Additionally, we searched online published abstracts from the three leading veterinary dermatology international congresses: the World Congress of Veterinary Dermatology (WCVD, 2008), the annual joint congresses of the European Society of Veterinary Dermatology (ESVD) and European College of Veterinary Dermatology (ECVD) of 2009, 2010 and 2011, as well as those of the North American Veterinary Dermatology Forum (NAVDF) held annually between 2008 and 2011.
To identify RCTs that had not yet been published or presented, we sent an email twice (18 December 2011 and 3 January 2012) to the three main veterinary dermatology lists (Vetderm, DipECVD and Dipderm) requesting colleagues to provide information on recently completed studies. Finally, we contacted, by email, representatives of three companies known to the authors to have completed relevant RCTs.
The same search was done with the three electronic databases, with the goal of having a simple yet very sensitive strategy that yielded a maximum of species- and disease-specific citations. The terms employed were as follows: (dog or dogs or canine) and (atopic and dermatitis). We added a date limit from 1 January 2008 to 31 December 2012, but there were no language or publication type restrictions. The search was done by one author (T.O.) and verified by the co-author (P.B.).
As done for our 2010 Cochrane-style systematic review,6 the titles of all electronic citations and meeting abstracts were first scanned for identification of clinical trials. Then, abstracts and/or article full texts were assessed to determine whether or not the study was an RCT, if it had enrolled solely dogs with AD, and if it had reported efficacy outcome measures. Furthermore, meeting abstracts and electronic article citations were matched for the identification of duplicate studies. The trial selection was not blinded, but it was done independently by the two authors; disagreements were resolved by consensus. Reasons for exclusion were recorded.
Data collection process
One of the authors first assessed study characteristics and extracted outcome measures, while the other assessed the risks of bias; then they reversed roles and verified each other’s extracted data. Discrepancies were identified and resolved by consensus.
When information was insufficient to assess outcome measures from the published or presented data, study authors were contacted by email. In the event of a lack of reply within 2 weeks, a second request was sent. When authors further declined to provide the requested information or original data, the mention ‘not provided upon request’ was added to the tables or the study was excluded from review, depending upon the amount of information available for analysis. The extracted data were entered in tables similar to those of the 2010 review.6
The following parameters were extracted from each article and/or from information obtained from the authors: objective of the study (treatment or prevention of AD); study design (parallel or crossover); duration of the trial; type of duration (short or long term); characteristics of study participants (number of dogs, their age range, type of AD etc.); type of intervention (treatment or prevention, including dose, duration and frequency); type of outcome measure; and funding source.
Risk of bias in individual studies
To determine the validity of eligible RCTs, the adequacy of several parameters known to affect bias was assessed as done previously.5,6 The following parameters were rated as ‘none’, ‘adequate’, ‘unclear’ or ‘inadequate’: (i) method of generation of randomization sequences; (ii) method of concealment of allocation to treatment groups; (iii) masking of intervention for observers (e.g. clinicians) and participants (e.g. dog owners); (iv) inclusion of cases lost to follow-up in intention-to-treat (ITT) analyses; and (v) degree of certainty that the participants were affected with AD, as judged by the author’s description.
The parameters ‘comparison of groups at baseline’ and ‘assessment of compliance’ were also added to the appraisal of study design.
Three of the parameters above (randomization method, masking and ITT) were used for an overall evaluation of study quality. When these parameters had been rated as ‘adequate or performed’, the RCT was graded as ‘high quality’; when they were all rated as ‘inadequate or unclear’, the study was graded as ‘poor’; and when only one or two of three parameters was assessed as ‘inadequate or unclear’, the RCT was graded as being of ‘intermediate’ quality.
Planned methods of analysis
As no two studies appeared to test similar interventions and/or used sufficiently similar designs, pooling of data for meta-analysis was not attempted. As a result, between-RCT variability (i.e. heterogeneity or inconsistencies) was not calculated. Nevertheless, whenever available, we compared dichotomous primary outcome measures between interventions using the Mantel–Haenszel (M-H) test; results are presented as risk ratios (RRs) with 95% confidence intervals (CIs) in the figures. These analyses were done using RevMan 5.0 analysis software (The Nordic Cochrane Centre, Copenhagen, Denmark).
Risk of bias across studies
In the absence of availability of published clinical trial protocols before the final results were eventually reported, the authors could not compare whether the outcome measures published matched those planned at study onset. However, the outcome measures reported in the study methods were matched to those of the results section. We also assessed whether the reported outcome measures were consistent with those employed in recent RCTs enrolling dogs with AD and if they appeared clinically relevant.
Owing to the heterogeneity of interventions and study designs, sensitivity, subgroup and meta-regression analyses were not conducted.
The search of Medline, Web of Science and CAB Abstract databases yielded a total of 672 citations, while 406 congress abstracts were available after examination of their online publication (Figure 1). Five replies were received from queries sent to the three veterinary dermatology lists. Among all citations, there were 48 clinical trials with atopic dogs, of which 18 were excluded because of lack of fulfilment of inclusion criteria. We subsequently eliminated seven RCTs that had been presented at congresses and later published as full papers (Figure 1). Owing to insufficient data available for full review and analysis, we also removed one published and one presented RCT.19,20 The remaining 21 RCTs are reviewed herein.
Details of study methodology, participants, interventions and funding sources can be found in Table 1.
All but one study included in this review were RCTs reported in English; one was in Italian. At the time of writing, 16 trials had been published in full in peer-reviewed journals, three were published only as abstracts,21–23 one was accepted for publication at the time of selection24 and one had neither been published nor presented, but study details were provided for our review in response to an email request to the vetderm list (A. Puigdemont, personal communication 2012).
Of 21 RCTs, there were 17 that used a parallel design of two groups or more, while the remaining four were cross-over trials (Table 1). Fourteen studies had interventions that were shorter than 8 weeks, hence were categorized as ‘short term’; the other seven lasted from 2 to 9 months (i.e. ‘long-term’ studies). While the abstract from one relapse prevention study did not specify the duration of the trial,24 it clearly lasted more than 2 months.
The number of subjects varied among RCTs (average, 42; median, 30; range, 10-316; Table 1). The lowest number of dogs in a treatment group was five (T-cell receptor peptide group22) and the highest was 202 (masitinib group25). Only six of 21 RCTs (29%) performed or reported a power analysis to justify subject numbers in each treatment group.25–30 All but one trial enrolled dogs with nonseasonal AD, while the last looked at the effect of an intervention to prevent recurrence of signs in dogs with predefined seasonal AD.31
Several studies added further limitations to their enrolment criteria. One trial of allergen-specific immunotherapy (ASIT) selected only dogs with demonstrable hypersensitivity to Dermatophagoides farinae house dust mites.32 One RCT selected only dogs with pedal lesions,21 another selected dogs with at least mild signs,30 while four studies enrolled dogs with moderate to severe AD (A. Puigdemont, personal communication 2012).25–27 Finally, two trials needed dogs to have signs for a minimal duration before selection (6 months33 and 1 year22).
All but four studies tested the efficacy – and safety – of various types of topical, oral or injectable interventions for treatment of canine AD (Table 1). Two trials investigated the effect of dietary interventions for their potential reducing effect on concurrently given oral glucocorticoids.23,27 Finally, two RCTs examined the outcome of interventions aimed at preventing the recurrence of flares of AD (i.e. prevention studies).24,31
As indicated in Table 1 and per selection criteria, all studies reported one or more outcome measures of efficacy. In general, these included the evolution, over the duration of the tested intervention(s), of pruritus and/or skin lesions scores (Table 1). To these were added, in five RCTs (Table 1), owner- and/or investigator-assessed global assessment ratings.21,25-27,30 In one study, the evaluation of quality of life (QoL) was also added.28 The two studies on glucocorticoid-sparing effect logically included the evaluation of the reduction in the dosage of oral glucocorticoids.23,27 Finally, one of the two prevention studies used ‘time to relapse necessitating treatment’ as a logical and relevant outcome measure for that particular design.24
Of the RCTs that tested the evolution of skin lesions over time, the validated third version of the CADESI score was used in 10 studies,21-24,26,27,29,30,34,35 while five trials employed the second version of this scale, which had limited validation.25,31,33,36,37 Four trials employed unvalidated scales modified from the first, second or third versions of the CADESI – the so called ‘modified CADESI’ or ‘mCADESI’ – having changed the type of lesions, the body sites and/or the severity rating used (A. Puigdemont, personal communication 2012).28,38,39 One trial employed a scale derived from the ‘six area six signs AD’ (SASSAD) used in human patients with AD even though this scale had not been validated beforehand for canine AD.32 Finally, one study used a simple but not validated six-point categorical scale.40
Of 20 RCTs that tested the efficacy of interventions on pruritus, 16 used the evolution of scores obtained on a visual analog scale (VAS), which varied between 5 and 20 cm; only three22,29,30 used the PVAS developed and validated by Hill and colleagues,15 even though it was first published in 2007 before some of these RCTs were designed. Finally, three trials used other ad hoc unvalidated pruritus scales.35,38,40
One trial was funded by a grant from the government of the Republic of Slovenia,37 while all others were funded by pharmaceutical companies.
Risk of bias in individual studies
Details on each study’s randomization method, intervention masking, loss to follow-up, diagnosis certainty, baseline group comparison, compliance assessment and overall quality rating can be found in Table 2.
In summary, out of 21 RCTs, only one had an overall quality rated as ‘poor’,33 seven were given an ‘intermediate’ rating,31,35-40 and the remaining 13 had the highest quality mark (Table 2). The most common reason for not attributing this ‘high-quality’ rating was a lack of performance of ITT analyses.33,35,36,38
Results of individual studies
As in our recent systematic review,6 individual trials will be discussed below in groups testing similar interventions within three larger subheadings: (i) treatment; (ii) glucocorticoid sparing; and (iii) prevention RCTs. Details on study design and quality assessment can be found in Tables 1 and 2, respectively.
Interventions for treatment of clinical signs
Two RCTs used an oral glucocorticoid as control for another intervention,37,39 while four tested the efficacy of a topical glucocorticoid (Table 3).21,26,28,30
One study reported the efficacy of methylprednisolone (Medrol; Pharmacia, Luxemburg, Luxemburg),37 while the second used prednisone (Encorton; Polfa Pharmaceuticals, Lublin, Poland).39 In both studies, drug dosages and protocols corresponded to those employed and recommended in recent practice guidelines.4 In these two small trials, the frequencies of dogs that achieved positive outcomes (Table 3) are in the range of those found after similar interventions (reviewed by Olivry et al.6). The results suggest treatment efficacy even though the glucocorticoids were not compared with placebo but were used as positive comparators for another intervention. Adverse effects reported in the publications were those expected for oral glucocorticoids.37,39
Three RCTs reported efficacy data for the same novel diester glucocorticoid spray that contains 0.0584% hydrocortisone aceponate (Cortavance; Virbac, Carros, France),21,26,30 while one tested the effectiveness of a novel glucocorticoid leave-on conditioner containing 0.025% budesonide (Barazone; Dermcare-Vet, Springwood, Queensland, Australia; Table 3).28 Three studies were placebo controlled,21,26,28 while one used ciclosporin as the active control.30 One trial focused on pedal pruritus and skin lesions;21 two had a crossover design.21,28
In two studies, the hydrocortisone aceponate spray appeared to be more effective than placebo for nearly all outcome measures (Table 3 and Figures 2 and 3).21,26 The magnitude of the effect was dampened by the small size of each treatment group, this being shown by the very large confidence interval bracketing the relative risk (Figures 2 and 3). In the largest trial, the outcome measures of efficacy were comparable between the glucocorticoid spray and ciclosporin.30 The safety of this diester glucocorticoid spray was found to be acceptable, with the lack of report of relevant adverse drug events.21,26,30 Remarkably, skin thinning was not reported with this formulation, even after 3 months of usage at frequencies varying from once daily to twice weekly.30