The Course of Chronic Plaque-Type Psoriasis in Placebo Groups of Randomized Controlled Studies

Objective  To determine the outcome in placebo-treated patients with plaque-type psoriasis.

Data Sources  Online search of MEDLINE and EMBASE until January 2001 and the Cochrane Library (2001, issue 1), supplemented by references, reviews, guidelines, and textbooks.

Study Selection  Randomized controlled induction of remission trials of patients with chronic plaque-type psoriasis with systemic treatments with a placebo group not treated with antipsoriatic medication. Identified studies were examined by 2 independent reviewers. Through MEDLINE, 290 studies could be identified. Twenty-seven placebo-controlled studies were included (488 patients).

Data Extraction  Two independent reviewers extracted data on first author, year of publication, design, comparison, placebo treatment, number of patients, treatment duration, type of psoriasis and baseline severity in the placebo group, mean relative change in outcome measures, and/or percentage of patients with worsening of psoriasis; no change; minimal, moderate, or good improvement; or complete clearance.

Data Synthesis  Owing to substantial heterogeneity and differences in the way outcomes were reported, no summary estimates could be obtained. The outcome of placebo treatment was poor in most studies. Some reported a mean relative change of 11% to 47%. The highest percentages of patients ended up in the worsening, no change, or minimal improvement categories. Also, complete clearance was possible. No explanation for the differences in outcome between placebo groups could be found. Description of placebo groups was often insufficient.

Conclusions  The effect of treatment in placebo groups varied across studies in an unpredictable way. To evaluate the variability, improvement of the standardization of study designs, entry criteria, and outcome measures is necessary in psoriasis trials.

In measuring the specific effects of systemic therapeutic interventions, researchers keep in mind that the natural course of a disease as well as the placebo effects of an intervention can influence therapeutic outcome.¹ The natural course of psoriasis varies considerably from patient to patient. Variation may include chronic persistence of the lesions for many years, temporary remissions with or without exacerbation, and persistent or only temporary regression.² Endogenous and exogenous factors can initiate, aggravate, or provoke the clinical manifestations.

Placebo effects have been reported to influence treatment outcome in general in up to 35% of patients.³ These effects include patient expectations, the attitude and instructions given by the treating physician, the treatment mode, and even the color of drugs.⁴ The existence of variation in the natural course and the placebo effect are therefore good reasons for the inclusion of placebo control groups in clinical trials of new therapies and to blind patients and physicians from treatment allocation.

The influence of natural course and placebo on treatment outcome in psoriasis has been the subject of debate. In chronic plaque-type psoriasis, variations in clinical expression are considered to be limited. The need for placebo control groups is believed by many to be less urgent for this disease than it is for more variable diseases.⁵

Some data about the natural course of psoriasis and the percentage of patients with self-limiting psoriasis can be obtained from older epidemiologic studies.^6,7 Farber and colleagues^6,7 reported that nearly 40% of the patients experienced at least once in their life an episode of complete remission. These results were based on patient questionnaires, without specifying the duration and extent of the psoriasis or factors influencing these episodes. In that review, 29% of the patients claimed that their psoriasis went into remission without physician-directed therapy. Krueger⁸ wrote that after the onset, psoriasis tends to wax and wane, but spontaneous remission is rare. Greaves and Weinstein⁹ wrote that psoriasis plaques can regress spontaneously without scarring after weeks, months, or years.

The existence of a placebo effect itself has been challenged by the results of a study conducted by Hróbjartsson and Gøtzsche.¹⁰ Based on a review of trials comparing placebo with no treatment, these authors concluded that there was little evidence that placebos have powerful clinical effects.

Performing a placebo-controlled trial may be more difficult than performing comparison studies with active treatments for the following reasons: participants must be convinced of the necessity of using a placebo treatment, withholding an accepted treatment may be harmful, there should be a accurate resemblance of the drug, and the blinding procedure must be thorough.

To determine the outcome in the placebo groups in randomized controlled trials, we performed a systematic review of randomized placebo-controlled trials of systemic drugs for chronic plaque-type psoriasis. For that purpose, we tried to identify all placebo-controlled trials and extracted data on treatment outcome in the placebo-treated control groups.

An extensive systematic search was performed for randomized controlled trials of systemic psoriasis treatments. Articles were gathered with the assistance of a clinical librarian through an online search of the MEDLINE computer database from 1966 to January 2001 and EMBASE. As main search terms (including analogues and derivatives), psoriasis, placebos, and placebo effect were used to identify relevant clinical trials and comparative studies. The Cochrane Library (2001, issue 1) was also screened for controlled studies. Additionally, references of articles such as textbooks, reviews, editorials, letters to the editor, free/rapid communications, and guidelines concerning these systemic treatments were screened. Furthermore, international professionals with expertise in psoriasis were consulted and pharmaceutical industries were requested to provide us with additional references of published clinical studies. Finally, abstract books of symposia and congresses were screened to optimize the result of the search.

In our review, we included randomized placebo-controlled studies of systemic drugs concerning adult patients with chronic plaque-type psoriasis, in which induction of remission (maximally, 16 weeks of treatment) was studied in the active treatment groups vs placebo-controlled groups. Studies were excluded from our review when the study design did not allow an evaluation of the outcome of placebo treatment or when antipsoriatic comedication was used in the placebo group. We also excluded studies in which the data were insufficiently documented. Double publications were also excluded.

All articles on eligible placebo groups were independently evaluated by 2 reviewers. In case of disagreement on the criteria for inclusion and exclusion, a third investigator was consulted. The following data were extracted from each included report on a placebo group: year of publication, study design, placebo treatment (if available, substance, dose, frequency, color, and taste) and treatment in the comparison group, number of placebo-treated patients, type of psoriasis and initial severity, treatment duration, outcome measurements, and outcome (final severity).

The outcome in placebo groups was analyzed in 2 ways. In some studies, the outcome of treatment is described in terms of changes in a specific disease severity parameter at the end of the treatment compared with baseline. In those studies, the mean relative change was calculated for each placebo group.

Such changes are not always reported. In studies that did not report average changes, the effect of treatment was summarized in terms of the number of patients with either worsening of psoriasis, no change, or improvement. We used the following categories: patients with worsening of psoriasis, no change, minimal improvement (≤25%), moderate improvement (≤50%), good improvement (>50%), and clearance (complete remission).

Some studies in our review mentioned the outcome of placebo treatment in both ways. These studies were used in both analyses. To explore the reasons for heterogeneity, we examined the study results with respect to the initial severity of the psoriasis, differences in treatments, and study duration.

Twenty-seven studies could be included in this analysis. All were performed in a double-blind manner except for 1.¹¹ Excluded studies were double publications; studies in which it was not mentioned that the active treatment or placebo was allocated at random; studies concerning topical treatments in which a comedication was used that might have influenced the results in the placebo group; studies in which the design and outcome measures did not allow us to analyze the placebo group; and studies that were named a placebo-controlled study but in which the placebo group did not really exist. Of the 27 studies, 21 mentioned the mean relative change in outcome measure; 15 reported the percentages of patients with worsening of psoriasis, no change, minimal, moderate, or good improvement, or clearance; and 9 mentioned the outcome of treatment in both ways.

The 27 included studies (9% of the identified studies through MEDLINE) reported on 488 placebo-treated patients, with the number of patients per placebo group ranging from 6 to 50. The baseline severity of psoriasis varied from moderate to severe, described in terms such as recalcitrant, chronic, disabling, or resistant to topical drugs. Treatment duration in these studies ranged from 10.3 days to 16 weeks. Most studies mentioned the kind of placebo treatment. Two studies had used identically appearing tablets or tablets with the vehicle only. Nine studies did not mention anything about the kind of treatment used as placebo.

Various outcome measures were used in the included trials, such as medium differences in total body surface area (BSA); erythema, scaling, and induration; mean percentage of PASI (Psoriasis Area and Severity Index¹²) score reduction; decrease in average global score; investigator's overallassessment; and investigator's final judgment. Figure 1 shows the mean relative change in outcome of 21 included placebo groups, relative to sample size. The standard error could not be calculated for all studies. In 5 studies the average change from baseline in outcome parameters in the placebo groups was 0. In 13 studies there was worsening of psoriasis, no change, or minimal improvement (<10%). Three studies reported 11.0% to 18.1% improvement on average, 4 reported 22.0% to 28.7% improvement, 1 reported 36.4% improvement, and 1 reported 47% improvement (Figure 1). Jakubowicz and colleagues¹³ reported on 15 patients who had been treated for 4 weeks with placebo. Their median PASI score changed from 22.72 to 16.51.¹³ The European FK 506 Multicentre Psoriasis Study Group¹⁴ reported on 23 patients in whom a mean reduction of PASI score of 47% was observed at the end of week 9. Peeters et al¹⁵ analyzed 14 patients in whom the BSA changed from 2.4% at baseline to 4.8% after 16 weeks. There was no change in the infiltration score, but a worsening could be seen in the mean (SD) scaling score (scale from 0-8), which changed from 1.9 (1.1) to 2.3 (1.2) (a worsening of 21%).

The data in the second group of studies were difficult to summarize. In some cases, the percentages of patients with at least 25% improvement or less than 50% improvement were given. Others used wide ranges of improvement. Altmeyer and colleagues¹⁶ reported that 18% of the patients showed complete to slight improvement. Some studies only mentioned the percentages of patients with moderate to good response, without providing definitions of moderate and good. In all cases, the highest percentages of patients were seen in the minimal improvement, no improvement, or worsening categories. Nevertheless, some studies mentioned that a few patients achieved complete clearance while receiving placebo treatment (Table 1).^13,16-29

No explanation for differences in outcome in placebo groups could be detected in terms of the duration of the study (range, 15 days to 16 weeks), the initial severity of the psoriasis, or the treatments in the placebo groups (frequency, color, and taste). Table 2 givesmore details about the studies.^11,13-38

In this systematic review of placebo-treated groups in controlled trialsof systemic drugs for chronic plaque-type psoriasis, we found substantialand unpredictable variation in the outcome of treatment in placebo groups.The effects of placebo in 27 included studies varied from worsening or nochange to sizable reductions in the severity of psoriasis measured.

In the first group of studies of 21 placebo groups in this review, theaverage change in outcome parameters ranged from 0% to 47%. In 13 of thesestudies, the mean change could be categorized as worsening, no change, orminimal improvement. In the second group of studies that mentioned the percentagesof patients with or without improvement, the highest percentages of patientswere found in the minimal improvement, no change, or worsening categories,although patients with moderate to good improvement and even clearance werealso described. We were not able to identify factors that could consistentlybe associated with the size of the reduction.

As in any systematic review, consideration of the effect of publicationbias is appropriate. The researchers' willingness to submit a study reportto a medical journal as well as the editor's eagerness to publish it can bothbe influenced by the size of the treatment effect found. If the differencebetween active treatment and placebo treatment is small, this can be due toeither a small effect of the active drug or a large improvement in the placebo-treatedpatients. If this holds, studies with large improvements in placebo-treatedgroups would be underreported.³⁹

There was a considerable variability in the design of the studies inthis review and the way in which the outcome of treatment was documented.Some studies used more objective measurement for outcome such as BSA and PASI.Others relied on a global impression by the treating physician or the patient.This variability hampers the use of meta-analysis as a tool to obtain moreprecise estimates of the effects of treatment. The more subjective ways ofmeasuring outcome in psoriasis trials are not free from bias. If both thephysician and the patient expect improvement, the effects of treatment willtend to be overestimated. Although this will not so much affect the comparisonof active treatment vs placebo, provided treatment allocation was blinded,high expectations can lead to an overestimation of treatment effect in placebogroups. This cannot explain all of the positive results, since improvementsof up to 47% were also reported with more objective outcome measures, suchas the changes in PASI score.

In most of the studies in this review, the use of emollients was allowed.The fact that the severity of psoriasis remained unchanged in several placebogroups might indicate that a gradual worsening could have happened if emollientshad not been used. It is possible that in the studies with no improvementor even worsening of the psoriasis, more patients with exacerbating psoriasisthan patients with chronic stable psoriasis had been included.

Other explanations for variations in the outcome of placebo treatmentare possible. Whether the psoriasis at baseline was in a deteriorating, stable,or improving phase was an important factor in these analyses. Differencesbetween the study centers can also influence treatment outcome.

We can conclude that the outcome in placebo groups in studies on chronicplaque-type psoriasis is variable and unpredictable. This may be due to variationsin the natural course of psoriasis in the included placebo-treated patientsand/or in the effect of treatment with placebo treatment itself. Chronic stableplaque-type psoriasis is maybe less stable than many of us believe it is,at least in the 16-week treatment period that was analyzed in this review.Open studies may therefore be of limited value. For example, in 2 studieson ranitidine in psoriasis, one (an open prospective study) suggested that4 months of treatment with ranitidine can result in a mean improvement of67% in two thirds of the patients,⁴⁰ and theother (a subsequent double-blind placebo-controlled study, which was not includedin this review because of the 24-week treatment duration), no significantdifferences were observed between the 2 different dosages of ranitidine andplacebo at any stage of the study.⁴¹ To evaluatethe variability, improvement of the standardization of study design, entrycriteria, and outcome measures is necessary in psoriasis trials.

Although it may be ethically more justified to perform a study comparinga new therapy with one of the available therapies in psoriasis (known as anactive control clinical trial) than to perform a placebo-controlled trial,^42,43 placebo-controlled trials are essentialin chronic plaque-type psoriasis study designs. In phase 2 trials, such studiescan reduce the number of patients that are necessary to participate. Ineffectiveor minimally effective treatments can be detected more easily if they arecompared with a placebo treatment, and highly effective treatments can beidentified as such with limited numbers of patients. To combine the investigationsabout the efficacy and the determination of the accurate dosages or dosageschemes, dose-finding studies may incorporate a placebo arm.

Article

Accepted for publication October 28, 2003.

A cooperative effort of the Istituto Dermopatico dell'Immacolata–Istitutodi Ricovero e Cura a Carattere Scientifico (IDI-IRCCS) and the Archives of Dermatology

Corresponding author and reprints: Phyllis I. Spuls, MD, PhD, AcademicMedical Centre, University of Amsterdam, Department of Dermatology, PO Box22770, 1100 DE Amsterdam, the Netherlands (e-mail: pspuls@hotmail.com).

The authors of this study use data from the placebo groups in clinicaltrials to study the natural history of psoriasis by following the outcomeof placebo-treated patients. Most patients were in the minimal improvement,no improvement, or worsening categories, although there was considerable variabilityin the outcome of placebo groups. The study highlighted the difficulty incomparing studies or performing meta-analyses of psoriasis studies becauseof the variability in the design and reporting of trials (see Ashcroft DM,Wan Po AL, Williams HC, Griffiths CE. Clinical measures of disease severityand outcome in psoriasis: a critical appraisal of their quality. Br J Dermatol. 1999;141:185-191; Weisman S, Pollack CR, GottschalkRW. Psoriasis disease severity measures: comparing efficacy of treatmentsfor severe psoriasis. J Dermatolog Treat. 2003;14:158-165;Naldi L, Svensson A, Diepgen T, Elsner P, et al, for The European Dermato-EpidemiologyNetwork (EDEN). Randomized clinical trials for psoriasis 1977-2000: the EDENsurvey. J Invest Dermatol. 2003;120:738-741). Thisdifficulty may be addressed in an upcoming conference on designing studiesfor the treatment of psoriasis (Eden Symposium: How to Perform Effective ClinicalTrials for Psoriasis; April 1-2, 2004; University of Geneva, Geneva, Switzerland[for further information e-mail: luigi.naldi@gised.it]).

Michael Bigby, MD