Randomized Double-blind Trial of the Treatment of Chronic Plaque Psoriasis: Efficacy of Psoralen–UV-A Therapy vs Narrowband UV-B Therapy

Objective  To compare the efficacy of oral psoralen–UV-A (PUVA) therapy with that of narrowband UV-B (NB-UVB) therapy in patients with chronic plaque psoriasis.

Design  Double-blind randomized study.

Setting  Phototherapy unit in a university hospital.

Patients  Ninety-three patients with chronic plaque psoriasis.

Interventions  Twice-weekly NB-UVB or PUVA therapy, starting at 70% of the minimum phototoxic or erythema dose, with 20% incremental increases. Patients were treated until clearance, up to a maximum of 30 sessions; those with clearance were followed up until relapse or for 12 months.

Main Outcome Measures  Proportion of patients achieving clearance, number of treatments to clearance, and, among those with clearance, the proportion remaining in remission at 6 months.

Results  Patients with skin types V and VI had a lower rate of clearance than those with skin types I through IV (24% vs 75%; P = .001). In patients with skin types I through IV, PUVA was significantly more effective than NB-UVB at achieving clearance (84% vs 65%; P = .02). The median number of treatments to clearance was significantly lower in the PUVA group (17.0 vs 28.5; P<.001). More patients treated with PUVA vs NB-UVB were reported to have erythema at some stage during treatment (49% vs 22%; P = .004), although this difference may have been due to ascertainment bias. Six months after the cessation of therapy, 68% of PUVA-treated patients were still in remission vs 35% of NB-UVB–treated patients.

Conclusion  Compared with NB-UVB, PUVA achieves clearance in more patients with fewer treatment sessions and results in longer remissions.

The combination of oral 8-methoxypsoralen and psoralen–UV-A (PUVA) is a well-established and effective treatment for psoriasis. It has more recently been claimed that narrowband UV-B (NB-UVB) therapy (311-312 nm, TL-01) has similar efficacy,^1-3 although this has been disputed.^4-6 In view of the greater safety of NB-UVB and its convenience (there is no need for previous oral medication use or eye protection after therapy sessions), this modality would be preferable to PUVA if it is of similar efficacy. We, therefore, performed the first (to our knowledge) randomized double-blind study comparing PUVA with NB-UVB for the treatment of chronic plaque psoriasis.

This trial was approved by the Research Ethics Committee at St Thomas' Hospital (London, England), and all the patients gave informed and signed consent. Sample size calculations suggested that for 90% power to detect a 25% decrease in the median number of exposures at the 5% significance level, 80 patients with psoriasis were needed. Therefore, to allow for dropouts, 93 patients with psoriasis were recruited from among those attending the Phototherapy Clinic of St John's Institute of Dermatology between April 1, 2002, and March 31, 2004; given the 12-month follow-up period, the study terminated March 31, 2005. The inclusion criterion was chronic plaque psoriasis with at least 8% body surface area involvement as assessed by the rule of nines and with a Psoriasis Area and Severity Index (PASI)⁷ score of at least 8; this corresponds to moderate-to-severe disease. Exclusion criteria included being younger than 18 years or older than 70 years, previous skin malignancy, photo(chemo)-therapy in the preceding 3 months or more than 150 sessions in the patient's lifetime, administration of a drug known to frequently cause photosensitization, topical antipsoriatic treatment in the previous 4 weeks or systemic antipsoriatic treatment in the previous 3 months, pregnancy, lactation, renal or hepatic disease, and a history of photosensitivity.

After study enrollment, patients were randomly allocated to either NB-UVB or PUVA therapy by means of a sequentially numbered list. The dermatologist conducting the assessments was blinded to the treatment allocations; nursing and pharmacy staff conducting the treatment were necessarily aware of it. To ensure full patient blinding, the phototherapy cabinet labels were concealed, and all the patients ingested identical-appearing tablets 2 hours before treatment: the NB-UVB group received placebo and the PUVA group received 10-mg 8-methoxypsoralen tablets (Deltapsoralen; Crawford Pharmaceuticals, Milton Keynes, England). The dose of the latter was determined by patients' body surface area, namely, 25 mg/m², and ranged from 30 to 60 mg. All the patients were advised that nausea could occur with use of the medication. During the study, patients intolerant of 8-methoxypsoralen because of nausea were instead given 20-mg 5-methoxypsoralen tablets (Pentaderm; Crawford Pharmaceuticals) at a dose of 50 mg/m² 3 hours before phototherapy, with doses ranging from 60 to 120 mg.

The PUVA therapy was given in a Waldmann 6002 cabin containing 40 Waldmann 100-W UV-A fluorescent tubes (Waldmann Medizintechn, Villingen-Schwenningen, Germany). The UV-A irradiance at the surface of the patient's skin was approximately 17.4 mW/cm². The NB-UVB treatments were given in a Waldmann UV5000 cabin containing 24 Philips 100-W NB-UVB fluorescent tubes (Waldmann) emitting predominantly in the wavelength range of 311 to 313 nm. The UV-B irradiance in the cabin at the surface of the patient's skin was typically 7 to 8 mW/cm². The exact irradiance of both sources was checked monthly.

The minimum erythema dose (MED) and minimum phototoxic dose (MPD) were assessed on unaffected upper buttock skin. In the NB-UVB group, the doses tested were 100, 140, 200, 280, 390, 550, 770, and 1100 mJ/cm²; in the UV-A group, the doses were 0.5, 1.0, 1.5, 2.0, 3.0, 5.0, 7.0, and 9.0 J/cm². The first 6 doses were used for skin types I and II and the last 6 for types III through VI. The MED and MPD were defined as the lowest dose that caused just-perceptible erythema 24 hours (NB-UVB) and 96 hours (PUVA) after exposure; all the patients were checked at both times to maintain blinding.

Patients were treated twice weekly, initially with a radiation dose corresponding to 70% of the MED or MPD, followed by 20% incremental increases at each visit, if tolerated. No further incremental increases were applied once a dose of 5 J/cm² (NB-UVB) or 15 J/cm² (PUVA) was reached. Doses were, however, adjusted according to the occurrence of any erythema after treatments (Table 1).

All the patients were instructed to apply aqueous cream (30% emulsifying ointment, 1% phenoxyethanol) twice daily and to use a bath emollient (Hydromol; Ferndale Pharmaceuticals Ltd, Thorp Arch Estate, England) (13% isopropyl myristate, 37.8% light liquid paraffin) daily throughout therapy and follow-up. For 12 hours after treatment, all the patients wore eye protection as for oral PUVA therapy. During treatment sessions, unaffected skin was covered as far as practically possible with clothing, and male patients wore genital protection.

All the patients were assessed immediately before the commencement of treatment and after 8, 16, 24, and 30 sessions. Between assessments, if nursing staff noted at routine examination before exposure that a patient was clear of psoriasis, the subsequent assessment by the study investigator was brought forward. All assessments were made by a blinded investigator (S.S.Y.) and consisted of estimation of the PASI⁷ score and of the Physician's Global Evaluation score as indexes of psoriasis severity. The latter is a 7-point scale (6 = severe, 5 = moderate to severe, 4 = moderate, 3 = mild to moderate, 2 = mild, 1 = almost clear [minimum residual activity, with all lesions flat and free of scale], and 0 = clear). In addition, patients completed the Dermatology Life Quality Index (DLQI)⁸ and a visual analog scale. The latter was a visual scale from 0 to 10 prefaced with the following question: “At the moment, how would you grade or rate your psoriasis on a scale from 0 to 10? Please circle as appropriate between 0 (no psoriasis) and 10 (the worst psoriasis you have ever had).”

Treatment was terminated in the event of any of the following: clearance of psoriasis (a Physician's Global Evaluation score of 0 or 1, ie, either complete clearance or minimum residual activity), absent or minimal improvement after 16 treatments or very slow progress thereafter, intolerance to therapy, or the completion of 30 treatments.

Patients whose psoriasis had cleared continued to be assessed each month for a year, or until relapse, whichever was sooner. Relapse was defined as recurrence of psoriasis with a PASI score of 50% or more of the baseline score.

Patients who enrolled in the study but withdrew before starting treatment were excluded from statistical analysis, whereas all those who commenced therapy were included regardless of withdrawal for any reason thereafter (“on-treatment analysis”). Comparisons between the 2 treatment groups were made by means of the χ² test for nominal data and the Mann-Whitney test for ordinal data. Binary logistic regression was used to assess the effect of possible predictors (treatment modality, skin type, age, sex, previous phototherapy, and MPD or MED) on binary outcome variables (clearance, erythema, and remission). To examine the role of the MPD or MED, 5 categories of each were formed and then combined so that all the patients were in 1 of 5 groups. For those who achieved clearance, Kaplan-Meier survival analysis using the log-rank test was used to compare remission in the 2 treatment groups. The Kendall τ-b correlation coefficient was used to assess, separately in the PUVA and NB-UVB groups, the association between skin type (in 6 groups) and the development of erythema during the study (yes/no). To investigate for possible correlations between PASI and DLQI scores, the Spearman ρ rank correlation coefficient was used. Analysis was performed using statistical software (SPSS version 11.5; SPSS Inc, Chicago, Ill). All tests of significance were 2-sided, and statistical significance was assumed at P<.05.

Ninety-three patients were enrolled (Figure 1). Five patients withdrew before the commencement of treatment and were not considered further. Eighty-eight patients commenced therapy and are included in the results (Table 2). There were 14 patients with skin type V and 3 with skin type VI. The 2 treatment groups were well matched for age, sex, and baseline PASI score. There were slightly more patients in the PUVA group who had been treated previously with photo(chemo)therapy (PUVA, broadband UV-B, or NB-UVB), but this difference was not statistically significant (χ²₁ = 2.1; P = .15). Clearance rates among those who had been previously treated and those who had not were similar (69% and 63%, respectively; χ²₁ = 0.35; P = .56). Before the commencement of treatment, 3 patients did not have their MPD checked and 3 did not have their MED checked for the following reasons: assessment of erythema is difficult in patients with skin type VI (n = 3), logistic factors (n = 2), and the absence of unaffected skin in the area of testing (n = 1). These patients began therapy using a skin type–based starting dose, which ranged from 100 to 500 mJ/cm² (NB-UVB) and from 0.5 to 3.0 J/cm² (PUVA).

Seventy-one patients had skin types I through IV and 17 had types V and VI. The clearance rate was significantly lower in patients with skin types V and VI compared with I and II and III and IV (24% vs 74% and 75%, respectively). We constructed a binary logistic model with skin type (in 2 groups, ie, I-IV and V-VI), treatment modality (PUVA and NB-UVB), age, MPD or MED group, previous phototherapy, and sex as predictor variables for clearance of psoriasis; the latter 4 factors were not, however, significant predictors and were sequentially removed from the model. Skin type (P = .001) and modality (P = .02) had significant effects on clearance rates. Of the 71 patients with skin types I through IV, PUVA achieved clearance in 31 (84%) of 37 compared with 22 (65%) of 34 using NB-UVB; the median number of treatments to clearance was 17.0 in the PUVA group and 28.5 in the NB-UVB group (Z = −3.7; P<.001).

In the 71 patients with skin types I through IV, the median PASI score at baseline in the PUVA group was 11; this score declined to 4.2 after 8 sessions. Corresponding scores for the NB-UVB group were 9.6 and 5.7. These reductions were significantly greater in the PUVA group (Z = −3.2; P = .001) (Figure 2A). Similarly, after 8 sessions, there were significantly greater reductions in Physician's Global Evaluation scores (Z = −3.2; P = .001), visual analog scale scores (Z = −2.5; P = .01), and DLQI scores (Z = −2.4; P = .02) (Figure 2B) for patients undergoing PUVA therapy compared with NB-UVB therapy.

There was no evidence that the superiority of PUVA over NB-UVB varied according to the severity of the initial psoriasis; in the 44 patients with an initial PASI score below 10.8, the clearance rate was 79% with PUVA and 52% with NB-UVB, whereas in those with an initial PASI score of 10.8 or greater, clearance rates were 79% and 50%, respectively.

In patients treated with PUVA, low skin types had a higher incidence of the development of erythema during the study (P = .005) (Table 2). In patients treated with NB-UVB, we did not detect a corresponding significant effect (P = .45) (Table 2). In a binary logistic model (incorporating skin type as a potentially confounding variable), modality (P = .004) and age (P = .04) were significant predictors of the development of erythema during therapy, with patients treated with PUVA and older individuals being more likely to develop erythema. Twenty-one (49%) of 43 patients treated with PUVA compared with 10 (22%) of 45 treated with NB-UVB experienced erythema (of either grade) at some stage during treatment. Considering only grade 2 erythemas, the corresponding figures were 14% and 7%. Two patients changed from taking 8-methoxypsoralen to taking 5-methoxypsoralen during the study because of nausea.

The 57 of 88 patients whose psoriasis cleared were followed up until relapse or for a maximum of 12 months, except for 3 who were lost to follow-up and were not included in the analysis of relapse. Binary logistic regression showed that at 6 months, the only significant predictor of remission was treatment modality (P = .02), with more patients treated with PUVA still having clearance (23/34; 68%) than patients undergoing NB-UVB therapy (8/23; 35%). Kaplan-Meier survival analysis also showed that patients treated with PUVA had significantly longer remissions (P = .03), with the median time to relapse being 8 months vs 4 months among patients treated with NB-UVB (Figure 3).

At baseline, PASI scores were not correlated with DLQI scores (ρ_s = 0.10; P = .35), but after phototherapy, quality of life was correlated with the extent of improvement in PASI scores resulting from the phototherapy (DLQI score compared with reduction in PASI score: ρ_s = −0.43; P<.001). Furthermore, at the time of relapse, quality of life was lowest in patients with the greatest deterioration in PASI scores (DLQI score compared with increase in PASI score: ρ_s = 0.52; P<.001).

To our knowledge, this is the first double-blind randomized trial of oral PUVA vs NB-UVB for the treatment of chronic plaque psoriasis. Oral PUVA was superior to NB-UVB in all the main outcome measures: clearance rate (among patients with skin types I-IV: 84% for PUVA and 65% for NB-UVB), median number of treatments to clearance (17.0 vs 28.5), and duration of remission (median time to relapse: 8 vs 4 months). Psoralen–UV-A was also more efficacious in analyses relating to PASI, Physician's Global Evaluation, visual analog scale, and DLQI scores. In terms of clearance rates, age, sex, and MPD or MED did not have a significant effect. In accordance with the results of the study by Markham et al,⁵ but unlike those of Tanew et al,² we found that the superiority of PUVA over NB-UVB was of similar magnitude regardless of initial psoriasis severity.

Two patients changed from taking 8-methoxypsoralen to taking 5-methoxypsoralen because of severe nausea. It is possible that these patients, and possibly others, particularly alert patients experienced in PUVA and NB-UVB therapy, may have become aware of their treatment allocation. However, overall, it is believed that most patients remained blinded to this throughout therapy, particularly because any differences in degree or type of pigmentation, or erythema, were not discernable to the investigator, who himself, therefore, remained blinded to group allocation. This is further expected in that observer blinding was successful in 3 previous comparisons of PUVA and NB-UVB.^1,3,4 In 2 of those studies,^1,3 comparisons were made between 2 sides of the same patient, maximizing the chance of detecting differences, which were nevertheless not noticed.

We used MED and MPD testing to ensure that both treatments were given as effectively as possible.^9-11 We administered PUVA twice a week rather than more frequently because PUVA erythema peaks at approximately 96 hours¹²; it has been shown that thrice weekly administration may produce unacceptable burning episodes and does not seem to be more effective.^13-15 Therefore, to enable effective masking of group allocation, we also used NB-UVB twice per week, an acceptable frequency because in a randomized trial¹⁶ of 113 patients concerning NB-UVB frequency, the mean number of treatments to clearance in those treated twice weekly was 24.4 and in those treated thrice weekly was 23.0, a difference that was not statistically significant. Treatment with NB-UVB 5 times weekly is not more effective than thrice weekly.¹⁷ It might, despite this, be argued that we administered NB-UVB at a suboptimal frequency, but this seems unlikely to have affected our conclusions because of the large difference between PUVA and NB-UVB in the median number of treatments to clearance.

Among patients who undergo phototherapy, those that have been treated with phototherapy in the past could perhaps be considered more likely to respond well than those who have never had phototherapy, and our PUVA group had a slightly higher proportion of patients who had previously been treated (44% vs 29%). In fact, however, the clearance rates between previously treated and untreated patients were only minimally dissimilar (69% and 63%, respectively), and in the binary logistic regression model, previous phototherapy was not a predictor of clearance, so this issue is not relevant in the interpretation of the results.

The results presented herein and in previous randomized between-patient comparison studies^4-6 on the treatment of chronic plaque psoriasis are highly consistent with each other (Table 3). In all 4 studies, the median number of treatments to clearance with PUVA was 16 to 19 and with NB-UVB was 25 to 28, and in 3 the clearance rates were 84% to 85% with PUVA and 60% to 65% with NB-UVB. In contrast, within-patient side-to-side comparison studies^1-3 have had more equivocal results (Table 3). In such studies it is possible that the effect of treatment on one side is transmitted to the other, thereby minimizing the difference between the 2 sides, but any such effect is likely to be small.¹⁸ One of these studies² found slight differences in favor of PUVA, but they were not statistically significant. Another study,³ in which bath PUVA was used, showed a higher clearance rate with NB-UVB, although the median number of treatments to clearance was lower in the PUVA group. The third study¹ included only 10 patients. Overall, therefore, PUVA seems to be more efficacious than NB-UVB in the treatment of chronic plaque psoriasis.

Forty-nine percent of the patients treated with PUVA vs 22% of those treated with NB-UVB experienced erythema at some stage during treatment. This occurred despite that both groups were following the same dose protocol (70% of the MPD or MED followed by 20% incremental increases). Although the assessment of erythema involves not only examination of the patient but also inquiry of the patient regarding any redness since the preceding visit, in practice, this inquiry is unlikely to be 100% sensitive. Therefore, more erythemal episodes may have been recorded with PUVA use because of bias in that PUVA erythema persists for much longer (approximately 96 vs 12-15 hours)¹² and, therefore, is more likely to be evident at the time of examination at the subsequent treatment session. Also, the absorption of psoralen and the timing of its administration may vary, leading to subsequent erythema. However, only 1 patient treated with PUVA withdrew because of burning, and it is probably acceptable for 49% of patients to develop mild erythema at some stage during treatment; indeed, some regimens deliberately try to provoke mild erythema.^1,2

We found a much lower clearance rate for patients with skin types V and VI compared with types I through IV (24% compared with 75%). This may well be because the development of facultative pigmentation, most marked in skin types V and VI, may interfere with efficacy by causing, in effect, underdosage.¹⁹ However, we did not detect differences between skin types in rates of erythemal episodes that would have supported this hypothesis.

The existence of a perhaps surprisingly weak association between the clinician's assessment of severity and the DLQI score has previously been reported,²⁰ and before phototherapy, we could not detect any association at all. However, after phototherapy, quality of life was correlated with the extent of psoriasis improvement, suggesting that, at least in the context of phototherapy, quality of life may be associated more with recent changes in psoriasis severity than with the severity of psoriasis itself.

Both NB-UVB and broadband UV-B offer definite advantages over oral PUVA that make these modalities preferable for many patients. In particular, PUVA may cause nausea, requires the use of eye protection after treatment sessions, cannot be used during pregnancy, is contraindicated in patients with significant hepatic impairment or taking warfarin or phenytoin, and requires the somewhat inconvenient previous administration of psoralen. Of greatest concern, however, is the potential of PUVA to cause nonmelanoma skin cancer after 160 to 200 lifetime treatments, which almost certainly exceeds by some margin that of a comparable number of NB-UVB sessions^9,21 and certainly of broadband UV-B sessions.²² However, our results suggest that PUVA compared with NB-UVB tends to clear psoriasis more reliably, with fewer treatments, and for longer and should, therefore, still be used in appropriate patients.

Correspondence: Sami S. Yones, MSc, FCD, Photobiology Unit, St John's Institute of Dermatology, St Thomas' Hospital, Second Floor, London SE1 7EH, England (yones5@yahoo.com).

Financial Disclosure: None reported.

Accepted for Publication: January 21, 2006.

Author Contributions: Dr Yones had full access to all data in the study and takes responsibility for the integrity of the data and accuracy of the data analysis. Study concept and design: Yones, Garibaldinos, and Hawk. Acquisition of data: Yones and Hawk. Analysis and interpretation of data: Yones and Palmer. Drafting of the manuscript: Yones and Palmer. Critical revision of the manuscript for important intellectual content: Yones, Palmer, Garibaldinos, and Hawk. Statistical analysis: Yones and Palmer. Obtained funding: Yones. Administrative, technical, and material support: Yones and Hawk. Study supervision: Hawk.

Acknowledgment: We thank the staff of The Dowling Day Treatment Unit (including Maria Csazar), the staff of the Skin Therapy Research Unit, and Jonathan Barker, MD, Mohammed Ben-Gashir, PhD, Paul Seed, MSc, and Manal Bugossa, Dip.