Combination Treatments for Psoriasis: Title and subTitle BreakA Systematic Review and Meta-analysisCombination Treatments for Psoriasis

With an increasing number of topical and systemic treatments available for psoriasis, choosing a therapy often depends on multiple factors, including disease severity, patient preference, practitioner experience, and medical insurance.^{1 - 3} Combination topical therapies are frequently used in clinical practice for patients with mild to moderate psoriasis, whereas systemic combination treatments are reserved for patients with recalcitrant disease. Data from a number of clinical trials suggest that combination therapies may have greater efficacy, tolerability, and, perhaps, fewer combined adverse effects compared with monotherapies.^{4 - 5} However, to our knowledge, no comprehensive meta-analysis has been performed to synthesize data on combination therapy in psoriasis; such a meta-analysis will be useful in guiding evidence-based practice for clinicians. We applied meta-analysis methods to various combinations of topical and systemic therapies and analyzed and summarized efficacy data for combination therapy compared with monotherapy regimens for psoriasis.

We conducted a systematic search of PubMed (using MeSH and free-text terms), Cochrane Review, CINAHL, and EMBASE for randomized clinical trials of psoriasis combination therapy, most recently updated in July 2010. Searches performed are listed in eAppendix 1. A total of 2213 potentially relevant reports were initially identified; among them, 129 randomized controlled trials met study inclusion criteria (Figure 1).

Study inclusion criteria were (1) use of between-patient (parallel group), crossover, or left-right design; (2) designs with patient self-randomization by lesion if they were a left-right design without the presence of multiple plaques treated with more than 2 different therapeutic combinations; and (3) reported measures of effectiveness and inclusion of at least 10 study participants. These criteria were modeled after those used in the Cochrane Review by Mason et al.⁶ Studies were not limited by disease severity, area of involvement, skin area treated, or type of psoriasis.

All data were extracted by at least 2 of 3 independent investigators (E.E.B., E.H.F., and M.T.H.) using a standardized extraction form with disagreements resolved by consensus between investigators. Study design and population characteristics were extracted. All corticosteroid treatments were categorized by potency using the criteria outlined by Jacob and Steele.⁷ The proportion of patients who achieved clearance, the definition of clearance used, means and standard deviations of the baseline disease symptom score, and final symptom score were also extracted. For any study in which any of the aforementioned outcomes data were not available, additional data were requested from the original study investigators.

Baseline psoriasis severity was defined as mild, mild to moderate, moderate, moderate to severe, or severe (see eAppendix 2). A trial was classified as following an intention-to-treat analysis if all subjects who were randomized to a treatment arm and received study drug or placebo were included in the efficacy analysis. Patients who were lost to follow-up or who stopped using the study drug could not be excluded from the efficacy analysis.

The primary outcome of this study was the number of subjects who experienced disease clearance as defined by the individual study investigators, analyzed by therapeutic combination type. This outcome was adapted from a similar outcome measure used in the systematic review by Griffiths et al.⁸ The proportion of patients with disease clearance was measured on a risk difference scale, which was calculated as the risk of clearance in the combination group minus the risk of clearance in the monotherapy group. A random-effects model with inverse variance and DerSimonian and Laird weighting was used to account for within-study and between-study variability, which does not assume that the studies are homogeneous in patient population or study characteristics.

When sufficient data were available, we also reported the standardized mean change in the clinical severity score (such as the Psoriasis Area and Severity Index, the Psoriasis Severity Index, or other symptom severity scores used in the included trial) as a secondary outcome. If multiple measures were used, the measure used as the primary outcome measure for the original study was used. This analysis was performed for any subgroup in which more than 1 study reported sufficient data to calculate a mean change in disease score and standard deviation for each arm. These results were pooled from the trials using a standardized mean difference statistic in a random-effects model with Cohen d weighting.⁹

Heterogeneity testing for the primary outcome was performed using the I² statistic, and meta-regression was used to determine whether study duration, dosage/treatment potency, type of randomization, blinding, use of intention to treat, baseline disease severity, presence of past treatment failure inclusion criteria, number of dropouts, and number of adverse events predicted between-study variation. All predictors that were significant at P = .05 were used in a further multivariate meta-regression, and, if a predictor was significant after controlling for all other predictors, subgroup analyses were performed. Subgroup analyses were also performed for any known significant differences between studies within a treatment subgroup based on type of disease treated or the treatment used. For groups with more than 2 studies, we performed Egger and Begg tests for publication bias and sensitivity analyses to investigate the implications of individual study exclusion and study year. A P value of less than .05 was considered statistically significant. All analyses were performed with commercially available software (STATA, version 10; StataCorp LP, College Station, Texas).

Fifty trials were used in the clearance efficacy analysis, which yielded 66 combination vs monotherapy arms because several studies included multiple monotherapy or combination arms. The 66 arms included efficacy analysis on 8325 trial subjects. Ten trials were used in standardized mean difference analysis for change in clinical severity score, including 11 combination vs monotherapy arms. These analyses included results from 1728 trial subjects. Vitamin D derivative and immunomodulator combination therapy was not included for summary analysis because of a lack of similarity between immunomodulator treatments. Combination therapies consisting of coal tar derivative and UV-B and of tar treatment and coal tar derivative were not included in the analysis owing to a lack of sufficient data for analysis. These studies are included in eTables 1, 4, and 5. No publication bias was apparent based on the Begg and Egger tests for any treatment combination or overall, although many subgroups had a small number of studies, which limited bias ascertainment. Study year did not have a significant effect on study results for any therapeutic subgroup.

Twelve randomized controlled trials examined the efficacy of vitamin D derivative–corticosteroid combinations (eTable 1).^{10 - 21} Two studies examined the treatment of only nail or scalp psoriasis and were not used in the statistical analysis.^{16 ,21} Of the remaining 10 studies, 7 had sufficient data reported to be used in an analysis of clearance efficacy,^{11 ,13 - 15 ,17 - 19} and 2 of these 7 studies had sufficient data to analyze efficacy by disease severity score reduction.^{11 ,15}

A total of 4593 subjects underwent analysis in these 7 trials, with 151 dropouts in the combination arm and 276 dropouts in the monotherapy arms. Compared with vitamin D derivative monotherapy, combination vitamin D derivative–corticosteroid therapy led to a 22% increased likelihood of clearance (95% CI, 12%-33%) (Figure 2). This effect was robust to the removal of individual studies. There was significant heterogeneity between studies, indicating a statistically significant difference in outcomes between individual studies (I² = 92.3% [P < .001]). There was some variation in the potency of the topical corticosteroid used, with 5 studies using class 1 corticosteroids,^{11 ,13 ,15 ,17 - 18} 2 studies using class 2 corticosteroids,^{14 ,19} and 1 study arm using class 3 corticosteroids.¹⁴ The corticosteroid class used introduced significant heterogeneity (P = .04) (Figure 3). When the results were stratified by corticosteroid class, patients had a 28% increased likelihood of disease clearance using a class 1 corticosteroid combination (95% CI, 16%-41%) and a 14% increased likelihood of disease clearance using a class 2 corticosteroid combination (95% CI, 5%-22%) compared with vitamin D derivative monotherapy. Use of a class 3 corticosteroid in combination with a vitamin D derivative did not lead to increased disease clearance compared with vitamin D derivative monotherapy. Combination therapy with any corticosteroid class decreased disease severity by 1.52 units of standard deviation (95% CI, −2.56 to −0.48) more than the vitamin D derivative monotherapy arm (data not shown).

Compared with corticosteroid monotherapy, vitamin D derivative–corticosteroid treatment led to a 20% increased likelihood of clearance (95% CI, 15%-24%) (Figure 2). There was no significant heterogeneity between studies within the group (I² = 0.0% [P = .84]).

Eight randomized controlled trials examined the efficacy of combined vitamin D derivative–UV-B (UV-B) therapies (eTable 1).^{22 - 29} Of these 8 studies, 4 had sufficient data to be used in comparative analysis of clearance efficacy of combination therapy to UV-B monotherapy,^{24 ,26 - 28} and 2 of these 8 studies had sufficient data to be used in analysis of clearance efficacy compared with vitamin D monotherapy^{23 - 24} ; thus, 5 studies in total were used in clearance efficacy analysis. Two studies had sufficient data to be used in analysis of efficacy by disease severity score reduction compared with UV-B monotherapy.^{26 ,28}

A total of 493 subjects underwent analysis in these 5 studies, with 19 dropouts in the combination therapy arms and 20 dropouts in the monotherapy arms. Compared with UV-B monotherapy, vitamin D derivative–UV-B combination therapy did not lead to a statistically significant increase in the likelihood of clearance (11%; 95% CI, −2% to 24%) (Figure 2). There was a statistically significant effect when the study by Molin et al²⁴ or the one by Ramsay et al²⁶ was removed from the analysis, which both had null results in the original studies. There was no significant heterogeneity among studies (I² = 56.9% [P = .07]). There were no significant differences in reduction of disease severity scores after vitamin D derivative–UV-B combination therapy compared with UV-B monotherapy (−0.92; 95% CI, −3.86 to 2.02).

Two studies had sufficient data to compare clearance efficacy of combination therapy with vitamin D derivative monotherapy.^{23 - 24} Compared with vitamin D derivative monotherapy, vitamin D derivative–UV-B combination therapy led to a 34% increased likelihood of clearance (95% CI, 22%-47%) (Figure 2). There was no significant heterogeneity between studies (I² = 11.5% [P = .29]).

Seven trials examined the efficacy of vitamin A derivative–psoralen-UV-A (PUVA) combinations (eTable 2).^{30 - 36} All 7 studies had sufficient data to be used in an analysis of the clearance efficacy of combination therapy compared with PUVA monotherapy, and 2 studies had sufficient data to analyze the clearance efficacy of combination therapy compared with vitamin A derivative monotherapy.^{30 ,33} No studies provided sufficient data to analyze efficacy by the reduction of disease severity score. All studies used oral vitamin A derivative treatments in combination with PUVA. Four studies used etretinate,^{30 - 31 ,33 - 34} 2 studies used acitretin,^{35 - 36} and 1 study used beta carotene.³²

A total of 265 patients were enrolled in these 7 trials. There were 15 dropouts from the combination arms and 13 dropouts from the monotherapy arms. Compared with PUVA monotherapy, vitamin A derivative–PUVA combination treatment led to a 22% increased likelihood of clearance (95% CI, 7%-38%) (Figure 4). This effect was robust to the removal of individual studies. There was significant heterogeneity, indicating statistically significant differences between individual study results (I² = 59.3% [P = .02]). The blinding status of the study introduced significant heterogeneity (P = .048), adjusting for intention-to-treat status (Figure 5). When results were stratified by blinding status, patients using combination therapy in the unblinded studies had a 35% increased likelihood of clearance (95% CI, 16%-55%), whereas patients using combination therapy in the double-blinded studies were not more likely to experience disease clearance compared with PUVA monotherapy (6%; 95% CI, −8% to 20%).

Additional analyses were performed, stratified by type of psoriasis (Figure 6) and with the removal of the study by Macdonald et al³² in which beta carotene was used. Among patients with palmoplantar pustulosis and hyperkeratotic psoriasis of the palms and soles, those using combination therapy had a 50% increased likelihood of clearance compared with those receiving monotherapy (95% CI, 26%-74%). Patients with all other types of psoriasis were not more likely to experience disease clearance using combination treatment than when using monotherapy (15%; 95% CI, −1% to 31%). Additional results excluding studies of patients with palmoplantar pustulosis found an effect of blinding status similar to that in the original analysis. When the study by Macdonald et al³² was excluded and the remaining studies were stratified by blinding status, all study heterogeneity was removed (I² = 0.0% [P = .85] for unblinded studies; I² = 0.0% [P = .70] for double-blinded studies).

Compared with vitamin A derivative monotherapy, vitamin A derivative–PUVA combination therapy led to a 47% increased likelihood of clearance (95% CI, 25%-69%). Both studies were unblinded. There was no significant heterogeneity between studies (I² = 0.0% [P = .68]).

Six trials examined the effects of vitamin A derivative–corticosteroid combination therapy compared with a monotherapy (eTable 2).^{37 - 42} Of these 6 studies, 4 had sufficient data to be used in an analysis of clearance efficacy compared with vitamin A derivative monotherapy.^{37 ,39 - 40 ,42} No studies provided sufficient data to analyze the clearance efficacy compared with corticosteroid monotherapy. No studies had sufficient data to be used in an analysis based on the disease severity score reduction. Two of the 4 studies included several combination arms with topical corticosteroid treatment of differing potency. In the 4 studies, there were a total of 4 arms using class 2 topical corticosteroids,^{37 ,39 - 40} 4 arms using class 3 topical corticosteroids,³⁹ 2 arms using class 4 topical corticosteroids,^{40 ,42} and 1 arm using a class 5 topical corticosteroid.⁴⁰ Three of the 4 studies used topical tazarotene gel,^{39 - 40 ,42} whereas 1 study used oral etretinate treatment.³⁷

A total of 653 patients were enrolled in the 4 trials. There were 65 dropouts in the combination arm and 49 dropouts in the monotherapy arm. Compared with patients receiving vitamin A derivative monotherapy, patients receiving vitamin A derivative–corticosteroid combination therapy had a 19% increased likelihood of clearance (95% CI, 11%-27%) (Figure 4). The estimate was robust to the removal of individual studies and was similar when the trials were not analyzed with separate treatment arms based on the type of corticosteroid used (23%; 95% CI, 14%-31%). There was significant heterogeneity (I² = 53.9% [P = .02]); the number of patient dropouts, baseline disease severity, blinding status, and intention-to-treat status were significant sources of heterogeneity when evaluated individually. However, there was not sufficient trial size to analyze the independent effect of each factor.

Four trials examined the effects of vitamin A derivative–UV-B combination therapy compared with UV-B monotherapy (eTable 2).^{43 - 46} Three trials had sufficient data to be included in the analysis of clearance efficacy,^{43 - 44 ,46} and only 1 trial had sufficient data to be included in efficacy analysis based on reduction in disease severity score.⁴⁵ Two of the 3 studies used oral vitamin A derivative treatments, etretinate⁴³ or acitretin.⁴⁶ The third study used a topical tazarotene therapy.⁴⁴

One-hundred sixty-two patients enrolled in these 3 trials. There was 1 dropout from the combination therapy group and 3 dropouts from the UV-B monotherapy group among the parallel group studies and 14 dropouts from the left-right study by Koo et al.⁴⁴ Compared with UV-B monotherapy, patients using vitamin A derivative–UV-B combination therapy had a 21% increased likelihood of clearance (95% CI, 5%-36%) (Figure 4). This estimate was somewhat unstable because the effect lost statistical significance when the study by Ruzicka et al⁴⁶ was removed. There was no significant heterogeneity (I² = 42.9% [P = .17]).

Five randomized controlled trials examined the efficacy of UV-B–balneotherapy combinations (eTable 3).^{47 - 51} All except 1 trial compared UV-B–balneotherapy combination therapy with UV-B monotherapy. Léauté-Labrèze et al⁵¹ used UV-B and balneotherapy monotherapy arms. Four studies had sufficient data reported to be used in an analysis of clearance efficacy,^{47 - 49 ,51} and 2 studies had sufficient data to analyze efficacy by the disease severity score reduction.^{49 - 50}

A total of 465 patients enrolled in these 5 trials. There were 27 dropouts from the combination arms and 55 dropouts from the monotherapy arms. Compared with patients using UV-B monotherapy, patients using UV-B–balneotherapy combination therapy were not more likely to achieve disease clearance (14%; 95% CI, −3% to 31%) (Figure 7). There was significant heterogeneity in this estimate, with I² = 84.9% (P < .001). Several differences between studies (ie, study duration, baseline disease severity, type of randomization, blinding status, and use of intention to treat) were significant sources of heterogeneity when analyzed independently. These differences all correlated with differences between the 2 studies conducted by Brockow et al^{47 - 48} and the 2 studies by other investigators.^{49 ,51} There was instability in the estimate when individual studies were removed. Examining the effect on reduction of the disease severity score, there was not a significant effect of UV-B–balneotherapy combined therapy compared with UV-B monotherapy (−0.68; 95% CI, −1.87 to 0.51).

Two trials examined the efficacy of UV-B–alefacept combination therapy compared with alefacept monotherapy, and 1 trial examined the efficacy of UV-B–alefacept combination therapy compared with UV-B monotherapy (eTable 3).^{52 - 55} The 2 trials examining UV-B–alefacept combination therapy compared with alefacept monotherapy each provided sufficient evidence to analyze the clearance efficacy but did not provide sufficient data to evaluate the efficacy of the treatment based on reduction in the disease severity score.

A total of 74 patients were included in these 2 trials. There were 4 dropouts from the alefacept–UV-B combination treatment arm and 5 dropouts from the alefacept monotherapy arm. Patients using UV-B–alefacept combination therapy had a 68% increased likelihood of clearance than did patients receiving alefacept monotherapy (95% CI, 51%-85%) (Figure 7). There was no significant heterogeneity (I² = 39.9% [P = .19]).

Two trials examined the efficacy of UV-B–methotrexate sodium combination therapy compared with UVB monotherapy (eTable 3).^{56 - 57} They each provided sufficient evidence to analyze the efficacy at disease clearance, and only 1 study provided sufficient data to evaluate the efficacy of the treatment based on a reduction in the disease severity score.⁵⁶

A total of 64 patients were included in these 2 trials. There were 2 dropouts from the methotrexate–UV-B treatment arm and 5 dropouts from the UV-B monotherapy treatment arm. Patients using UV-B–methotrexate combination therapy had a 36% increased likelihood of disease clearance than did patients receiving UV-B monotherapy (95% CI, 10%-63%) (Figure 7). There was no significant heterogeneity (I² = 48.8% [P = .16]).

Two trials examined the efficacy of UV-B–psoralen combination therapy compared with UV-B monotherapy (eTable 3).^{58 - 59} They each provided sufficient data to analyze the disease clearance efficacy and to evaluate efficacy by the degree of disease severity score reduction.

Thirty-one patients were included in these 2 trials. There were 5 dropouts from these left-right trials (eTable 3). Patients using UV-B–psoralen combination therapy were not more likely to experience disease clearance than patients using UV-B monotherapy (13%; 95% CI, –7% to 34%) (Figure 7). There was no significant heterogeneity (I² = 0.0% [P = .74]). When we examined the standardized mean difference in the clinical severity score, there was a significant effect of UV-B–psoralen combination therapy compared with UV-B monotherapy (−1.68; 95% CI, −2.37 to −0.99).

Two trials examined the efficacy of UV-B–tar combination therapy compared with UV-B monotherapy (eTable 3).^{60 - 61} They each provided sufficient evidence to analyze clearance efficacy. Only 1 study had sufficient data to evaluate the efficacy of treatment on the basis of a reduction in the disease severity score.⁶⁰

Sixty-one patients were included in these 2 trials. There were 5 dropouts from the combination and monotherapy arms, all from the study by Menkes et al.⁶¹ Patients using UV-B–tar combination therapy were not more likely to experience disease clearance than patients using UV-B monotherapy (−5%; 95% CI, −27% to 17%) (Figure 7). There was no significant heterogeneity (I² = 0.0% [P = .44]).

Three trials examined the efficacy of corticosteroid–hydrocolloid dressing combination therapy compared with monotherapy (eTable 4).^{62 - 64} All 3 trials provided sufficient data to analyze clearance efficacy compared with corticosteroid monotherapy. One trial also compared combination therapy with hydrocolloid occlusion dressing monotherapy.⁶² No studies provided sufficient data to analyze efficacy on the basis of a reduction in the disease severity score. Two studies used class 3 topical corticosteroid regimens,^{62 - 63} and 1 study used a class 1 topical corticosteroid regimen.⁶⁴

A total of 116 patients were included in these 3 trials, with 5 dropouts from the combination arms and 9 dropouts from the monotherapy arms. Compared with corticosteroid monotherapy, patients using corticosteroid–hydrocolloid occlusion dressing combination therapy had a 44% increased likelihood of disease clearance (95% CI, 23%-64%) (Figure 8). The estimate was stable when individual studies were removed from the analysis. There was no significant heterogeneity (I² = 56.7% [P = .10]).

Two trials examined the efficacy of corticosteroid–salicylic acid combination therapy compared with monotherapy (eTable 4).^{65 - 66} Both trials evaluated combination therapy compared with corticosteroid monotherapy. Both studies used mometasone furoate ointment, 0.1% (a class 3 topical corticosteroid), in combination with salicylic acid ointment, 5%, twice daily. Both provided sufficient data to analyze disease clearance efficacy but did not provide sufficient data to evaluate the efficacy by disease severity score reduction.

Seven hundred seventy-six patients were included in these 2 trials. There were 33 dropouts from the corticosteroid–salicylic acid treatment arm and 37 dropouts from the corticosteroid monotherapy treatment arm. Patients using corticosteroid–salicylic acid combination therapy were not more likely to experience disease clearance than patients using corticosteroid monotherapy (3%; 95% CI, 0%-7%) (Figure 8). There was no significant heterogeneity (I² = 0.0% [P = .78]).

Two studies evaluated the efficacy of corticosteroid–UV-B combination therapy compared with UV-B monotherapy on the basis of clearance efficacy (eTable 4).^{67 - 68} Neither study provided sufficient data to analyze the efficacy by disease severity score reduction. One study used a class 1 topical corticosteroid regimen,⁶⁸ and 1 study used a class 2 topical corticosteroid regimen.⁶⁷

One hundred forty-three patients enrolled in these 2 trials. In the trial by Larkö et al,⁶⁸ there were 10 dropouts from the combination group and 7 dropouts from the monotherapy group. No dropouts were reported in the trial by Dover et al.⁶⁷ Patients using corticosteroid–UV-B combination treatment were not more likely to experience clearance than patients using UV-B treatment alone (–6%; 95% CI, –24% to 12%) (Figure 8). There was no significant heterogeneity in this estimate (I² = 0.0% [P = .91]).

Four randomized controlled trials examined the efficacy of vitamin A derivative–vitamin D derivative combination therapy (eTable 6).^{69 - 72} Each trial compared combination therapy with vitamin A derivative monotherapy. Three of the 4 studies had sufficient data reported to be used in an analysis of clearance efficacy,^{70 - 72} and 2 of these 4 studies had sufficient data to analyze the efficacy by disease severity score reduction.^{69 - 70}

A total of 301 subjects underwent analysis in these 4 trials, with 26 dropouts from the combination arm and 31 dropouts from the monotherapy arms. Compared with vitamin A derivative monotherapy, vitamin A derivative–vitamin D derivative combination therapy led to a 33% increased likelihood of clearance (95% CI, 22%-44%). This effect was robust to the removal of individual studies. There was no significant heterogeneity (I² = 28.3% [P = .25]). When we examined the effect on the reduction of the disease severity score, the combination therapy did not have a statistically significant effect compared with vitamin A derivative monotherapy (−12.0; 95% CI, −31.6 to 7.49).

Five randomized controlled trials evaluated the efficacy of an immunomodulator treatment in combination with another therapy.^{73 - 77} These studies examined the efficacy of tacrolimus in combination with salicylic acid,⁷³ cyclosporine in combination with a low-calorie diet,⁷⁴ sirolimus in combination with cyclosporine,⁷⁵ methotrexate in combination with folic acid,⁷⁶ and methotrexate in combination with etanercept.⁷⁷ The characteristics and key findings of these studies are described in eTable 7.

Many of the examined combination therapies were more effective than monotherapy regimens (eTable 8). Our analysis provides an important summary of the current evidence for many commonly used psoriasis combination treatments and is the first study, to our knowledge, to analyze statistically the evidence for topical and systemic combination therapies for all severities of psoriasis.

Some treatment combinations in our study have been examined in previous meta-analyses.^{6 ,8 ,78} There were some differences in the studies included in our analysis and those in previous systematic reviews, likely owing to variability in search criteria, addition of more recent studies, and differences in criteria for study inclusion. The results found in our study were consistent with previously reported results.^{6 ,8 ,78}

Vitamin D derivative–corticosteroid combination therapy was more effective than vitamin D derivative monotherapy in our study, based on clearance efficacy and disease severity score reduction. These results are consistent with results from a previous meta-analysis that also reported the standardized mean difference in severity score.⁶ In our study, vitamin A derivative–PUVA and vitamin A derivative–corticosteroid combination therapies were each more effective than vitamin A derivative monotherapy based on likelihood of disease clearance. Vitamin A derivative–PUVA combination therapy was also more effective than PUVA monotherapy based on clearance efficacy, and vitamin A derivative–UV-B combination therapy was more effective than UV-B monotherapy based on clearance efficacy. A previous meta-analysis demonstrated similar results based on the likelihood of clearance but did not analyze an overall effect estimate for vitamin A derivative–corticosteroid or vitamin A derivative–UV-B combination therapy trials.⁸ A previous meta-analysis examining treatments for chronic palmoplantar pustulosis found similar results for the efficacy of vitamin A derivative–PUVA combination therapy compared with either monotherapy based on disease clearance.⁷⁸

We also showed that corticosteroid–hydrocolloid occlusion dressing therapy was more effective than corticosteroid monotherapy, based on likelihood of clearance. This result is similar to the result from a single trial of patients with palmoplantar pustulosis described in a meta-analysis by Marsland et al.⁷⁸

Based on the results from the vitamin A derivative–PUVA combination therapy subset, we also showed that blinding can significantly affect results. When stratified by blinding status, studies that used a double-blinded design did not show a significant increase in disease clearance with vitamin A derivative–PUVA combination therapy compared with PUVA monotherapy, whereas unblinded studies demonstrated a significant effect. This effect was not explored in the previous meta-analyses.⁸ The significant effect of study blinding may be due to differences in the evaluation or treatment of patients in unblinded studies compared with double-blinded studies.

In some combination subsets, such as the UV-B–psoralen combination therapy and vitamin A derivative–vitamin D derivative combination therapy, the analyses using the 2 different outcome measures (clearance efficacy and disease severity score reduction) led to different conclusions about efficacy. For the vitamin A derivative–vitamin D derivative combination subset, there is limited interpretability of the clinical value of one outcome measure compared with another because each set of analyses used a different set of trials. However, the analysis of UV-B–psoralen combination therapy used the same 2 studies for both trials. Combination therapy was not more effective than UV-B monotherapy based on the likelihood of disease clearance but was more effective based on change in disease severity score. These findings highlight that some treatments that effectively reduce disease severity may not lead to complete disease clearance. An analysis based on disease clearance could miss the efficacy of this type of treatment in reducing disease severity.

Because we analyzed the results of trials with many differences in design characteristics (ie, length of study, treatment dosing, and patient characteristics), these studies could not be pooled on a patient level and required pooling on a study level using random-effects analysis to take into account differences in design and quality between studies. This type of analysis is helpful for providing more robust treatment efficacy results using existing study data, particularly for looking at a number of different therapies when more homogeneous pooling data are not available. One limitation of this study is that there may be therapies for which we did not find a statistically significant difference in efficacy but for which clinically meaningful benefits might not have been captured owing to insufficient sample size or the limitations of the outcome measures available for this study.⁷⁹ However, we attempted to present the data available as completely and objectively as possible to guide clinical practice, as is appropriate given the limitations, and to direct further research.

In summary, the current literature on combination therapy in psoriasis has focused on the comparison between a combination regimen with a monotherapy. It appears that combination topical therapies are generally more effective than topical monotherapy. Overall, the addition of phototherapy to a topical regimen results in significant clearance of psoriasis compared with a topical regimen alone. However, compared with phototherapy alone, the addition of a topical treatment to phototherapy does not appear to significantly decrease disease severity. The literature is lacking in combination systemic therapies. Future research in combination therapy is necessary to devise new therapeutic strategies to maximize efficacy while minimizing adverse effects.