A random-effects Mantel-Haenszel model was used. For each subgroup, the sum of the statistics, along with the summary OR, is represented by the middle of the solid diamonds. The width of the diamond represents the pooled means, with 95% CIs at the points. Squares represent mean values, with error bars representing 95% CIs. OR indicates odds ratio.
A random-effects Mantel-Haenszel model was used. Squares represent mean values with error bars representing 95% CIs; diamonds, pooled means with the 95% CIs at the points. OR indicates odds ratio.
A random-effects inverse variance model was used. Squares represent mean values with error bars representing 95% CIs; diamonds, pooled means with the 95% CIs at the points. HR indicates hazard ratio.
eTable. Search Strategy for CDSR, ACP Journal Club, DARE, Embase and Ovid Medline Electronic Databases
eFigure 1. PRISMA Flowchart of Search Strategy for Present Systematic Review
eFigure 2. Risk of Bias of Included Studies Assessed Using Cochrane Tools
eFigure 3. Forest Plot of Partial Response in Early-Stage MF Treated by PUVA vs NBUVB
eFigure 4. Forest Plot of Failed Response in Early-Stage MF Treated by PUVA vs NBUVB
eFigure 5. Funnel Plot for Publication Bias of Meta-analysis of Any Response in Early-Stage MF Treated by PUVA vs NBUVB
Customize your JAMA Network experience by selecting one or more topics from the list below.
Phan K, Ramachandran V, Fassihi H, Sebaratnam DF. Comparison of Narrowband UV-B With Psoralen–UV-A Phototherapy for Patients With Early-Stage Mycosis Fungoides: A Systematic Review and Meta-analysis. JAMA Dermatol. 2019;155(3):335–341. doi:10.1001/jamadermatol.2018.5204
How does the efficacy of narrowband UV-B compare with that of psoralen–UV-A for early-stage mycosis fungoides?
In this systematic review and meta-analysis of 7 studies with 778 patients with early-stage mycosis fungoides, response rates to narrowband UV-B were similar to those for psoralen–UV-A. No significant difference was found between narrowband UV-B and psoralen–UV-A in terms of adverse effects.
The findings suggest that narrowband UV-B is a viable and safe alternative to psoralen–UV-A for treatment of early-stage mycosis fungoides.
Phototherapy is one of the mainstays of treatment for early mycosis fungoides (MF). The most common modalities are psoralen–UV-A (PUVA) and narrowband UV-B (NBUVB).
To compare the efficacy and adverse effects of PUVA vs NBUVB in early-stage MF.
A systematic review was performed by searching Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Ovid Medline, PubMed, Cochrane Library, American College of Physicians ACP Journal Club, and Database of Abstracts of Review of Effectiveness from inception to March 30, 2018. UV A, PUVA, mycosis fungoides, Sézary syndrome, cutaneous T-cell lymphoma, UV B, and UVB were used as either key words or MeSH terms.
Studies of cohorts with histologically confirmed early-stage MF, defined as stages IA, IB, and IIA, that compared PUVA vs NBUVB, had at least 10 patients in each comparator group, and reported outcomes of response to therapy. Exclusion criteria were studies with patients with stage IIB or higher MF, pediatric patients, fewer than 10 in each comparator group, noncomparative studies, case reports, and abstract studies.
Data Extraction and Synthesis
The Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline was followed. Data were pooled using a random-effects model with odds ratio (OR) as effect size.
Main Outcomes and Measures
Main outcomes were complete response rate, partial response rate, disease recurrence, and adverse effects, including erythema, nausea, pruritus, phototoxic effects, dyspepsia, and pain.
Seven studies were included with a total of 778 patients (405 of 724 [55.9%] men; mean age, 52 years); 527 were treated with PUVA and 251 with NBUVB. Most of the included studies were of poor to moderate quality. Any response was found in 479 of the 527 (90.9%) patients treated with PUVA vs 220 of 251 (87.6%) treated with NBUVB (OR, 1.40; 95% CI, 0.84-2.34; P = .20). Complete response was found in 389 of 527 (73.8%) patients who received PUVA vs 156 of 251 (62.2%) who received NBUVB, which was statistically significant (OR, 1.68; 95% CI, 1.02-2.76; P = .04). Partial response was similar (90 of 501 [18.0%] vs 64 of 233 [27.5%]; OR, 0.58; 95% CI, 0.33-1.04; P = .07). No significant difference was found between PUVA and NBUVB in terms of adverse effects of erythema (38 of 527 [7.2%] vs 17 of 251 [6.7%]; P = .54), nausea (10 of 527 [1.9%] vs 3 of 251 [1.2%]; P = .72), pruritus (2 of 527 [0.4%] vs 4 of 251 [1.7%]; P = .26), phototoxic effects (7 of 527 [1.4%] vs 2 of 251 [0.9%]; P = .72), dyspepsia (6 of 527 [1.2%] vs 0 of 251 [0%]; P = .59), or pain (0 of 527 [0%] vs 2 of 251 [0.9%]; P = .50).
Conclusions and Relevance
The findings suggest that PUVA is a potential alternative to NBUVB in the management of early-stage MF. These findings have implications for clinicians involved in the management of early-stage MF.
Mycosis fungoides (MF) is a form of cutaneous T-cell lymphoma (CTCL) that follows an indolent clinical course in most cases. It is a form of non-Hodgkin lymphoma characterized in its early stages by papulosquamous patches or plaques, which can rarely progress to become tumors with involvement of the lymphatic system and other viscera. Most patients have early-stage MF (stages IA-IIA), which is associated with a favorable prognosis. Skin-directed therapies are used as first-line treatment, including topical corticosteroids, topical chemotherapeutic agents, topical and systemic retinoids, methotrexate, and phototherapy. For patients with advanced-stage MF or Sézary syndrome (stages IIB-IVB), disease is often recalcitrant to treatment and more aggressive modalities, such as chemotherapy, radiotherapy, extracorporeal photopheresis, and allogenic hematopoietic stem cell transplant, may be required.1,2
Phototherapy is one of the mainstays of treatment for patients with early-stage MF. The most commonly used modalities are narrowband UV-B (NBUVB) radiation, which involves UV irradiation at a wavelength of 311 nm or psoralen–UV-A (PUVA) phototherapy, involving UV irradiation ranging from 320-340 nm in combination with systemic or topical psoralen. Psoralen-UV-A and NBUVB have different electromagnetic and biological properties and underlying mechanisms of action. Given its longer wavelength, UV-A radiation is capable of penetrating deeper within the dermis, and this is thought to contribute to better response initially and prolonged disease-free periods. As such, PUVA may be more suited for patients with thick plaques or those recalcitrant to NBUVB therapy. The psoralen in PUVA also directly interacts and damages DNA, generating oxygen-free radicals, which damage cell organelles.3-5 UV-A has also been hypothesized to trigger apoptosis cascades via p53-independent programmed cell death mechanisms.3,4 Phototoxic effects of PUVA have been shown to selectively target neoplastic T lymphocytes in the skin.3,6,7 The mechanisms of UV-B therapy in MF are less clear; however, they are thought to involve interference with immunity via Langerhans cells. The cells’ antigen presentation capacity is inhibited, and cytokines interleukin (IL)–2, IL-6, and tumor necrosis factor are upregulated.4,8 In addition, T-cell apoptosis may be involved, suppressing neoplastic activity in MF.9
Although both PUVA and NBUVA have been widely used during the past several decades, the volume of noncomparative data to evaluate their efficacy in inducing disease response in early-stage MF is small. In recent years, NBUVB has become popular given reports of similar efficacy to PUVA but with fewer adverse effects and lesser risk of carcinogenesis.10-12 Narrowband UV-B radiation has been associated with reduced erythema and irritation compared with broadband UV-B, thus requiring a reduced lesion clearance dose in the context of inflammatory skin disorders such as psoriasis.10,13,14 However, whether these benefits are relevant to MF is yet to be demonstrated because there has been limited evidence from direct comparison of these 2 forms of phototherapy. In addition, associations between MF stages or lesion types and phototherapy outcomes have not been well studied.
To our understanding, there has been no meta-analysis directly comparing PUVA and NBUVB in the treatment of early-stage MF. Therefore, to evaluate the relative benefits and risks of PUVA vs NBUVB therapy, we conducted a systematic review and meta-analysis of the available comparative literature.
We performed a systematic review and meta-analysis in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline to investigate the efficacy and safety of PUVA vs NBUVB in patients with early-stage MF. Electronic searches were performed using Ovid Medline, PubMed, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, American College of Physicians Journal Club, and Database of Abstracts of Review of Effectiveness from their dates of inception to March 30, 2018. To achieve the maximum sensitivity of the search strategy, we combined the terms UV A, PUVA, mycosis fungoides, Sézary syndrome, cutaneous T-cell lymphoma, UV B, and UVB as either key words or MeSH terms (eTable in the Supplement). The reference lists of all retrieved articles were reviewed to further identify potentially relevant studies, and our inclusion and exclusion criteria were applied.
Eligible studies for the systematic review and meta-analysis included those in which cohorts of patients who underwent PUVA phototherapy were compared directly with cohorts of patients who underwent NBUVB phototherapy. Inclusion criteria were cohorts with histologically confirmed early-stage MF (defined as stages IA to IIA), comparison of PUVA (bath or oral) with NBUVB, at least 10 patients in each comparator group, and outcomes of response to therapy. Exclusion criteria were studies of stage IIB and higher (tumors, erythroderma, or lymph node involvement requiring systemic agents were not investigated in this meta-analysis), cohorts including pediatric patients, fewer than 10 patients in each comparator group, noncomparative studies, case reports, and abstract studies. Language was not an exclusion criterion.
All data were extracted from article texts, tables, and figures. Two investigators (K.P., V.R.) independently reviewed each retrieved article. Discrepancies between the 2 reviewers were resolved by discussion and consensus. If the study provided medians and interquartile ranges (IQRs) instead of means and SDs, we imputed the means and SDs as described by Hozo et al.15 If Kaplan-Meier graphs were provided but not the hazard effect or 95% CI, these were estimated using methods by Tierney et al.16 Quality appraisal of studies was assessed using the Cochrane Collaboration tool for assessment.
Odds ratios (ORs) were used as summary statistics. A random-effects model was used to take into account the possible clinical diversity and methodological variation among studies. To study heterogeneity between trials, we used χ2 tests. We used the I2 statistic to estimate the percentage of total variation across studies owing to heterogeneity rather than chance, with values greater than 50% considered as substantial heterogeneity; I2 can be calculated as I2 = 100 × [(Q–df)/Q], with Q defined as the Cochrane heterogeneity statistic. If there was substantial heterogeneity, the possible clinical and methodological reasons for this were determined qualitatively by subgroup analysis. Specific analyses considering confounding factors were not possible because raw data were not available. All P values were 2-sided, and statistical significance was set at P < .05. All statistical analysis was conducted with Review Manager, version 5.3 (Cochrane Collaboration, Software Update).
A total of 438 references were identified through electronic database searches. After exclusion of duplicate or irrelevant references, 374 potentially relevant articles were retrieved. After applying the selection criteria, 7 articles were selected for analysis (eFigure 1 in the Supplement).17-23 The study and baseline characteristics of these studies are summarized in Table 1 and Table 2. There were a total of 527 patients treated with PUVA compared with 251 treated with NBUVB. All studies were retrospective observational studies except for El-Mofty et al,20 which was a prospective study. Study appraisal is summarized in eFigure 2 in the Supplement. The majority of included studies were of poor to moderate quality, with the major common critiques including lack of random sequence generation, lack of allocation concealment, lack of blinding of participants and personnel, and lack of blinding of outcome assessment.
The mean age of the 527 patients who underwent PUVA ranged from 33 to 70 years, and the mean age of the 251 patients who were treated with NBUVB ranged from 33 to 68 years. A total of 292 of 501 (58.3%) of the patients treated with PUVA were male compared with 113 of 223 (50.7%) in the NBUVB cohort, which was not significantly different (P = .14). Of the 213 patients with MF data available who received PUVA, 95 (44.6%) had patch-type MF compared with 119 of 197 (60.4%) who received NBUVB, based on data reported from 3 studies (P < .001). Of the 527 patients who received PUVA 246 (46.7%) had plaque type compared with 95 of 251 (37.8%) who received NBUVB. A total of 238 patients (45.3%) who received PUVA had stage IA MF compared with 137 who received NBUVB (54.5%) (P = .08), and 271 patients (51.4%) who received PUVA had stage IB disease compared with 108 (43%) who received NBUVB (P = .02). A total of 17 patients (3.3%) who received PUVA had stage IIA disease compared with 6 (2.2%) who received NBUVB (P = .25).
Any response was defined as the totality of complete and partial responses. For all pooled cases of early MF, any response was found in 479 of the 527 patients (90.9%) who received PUVA compared with 220 of the 251 (87.6%) who received NBUVB (OR, 1.40; 95% CI, 0.84-2.34; P = .20), with no significant heterogeneity (I2 = 0%) (Figure 1). Subgroup analysis was performed, with no difference in any response in patients with stage IA MF (134 of 139 [96.4%] vs 44 of 48 [91.7%]; OR, 2.54; 95% CI, 0.57-11.26; P = .22; I2 = 0%) or in those with stage IB MF (129 of 145 [89.0%] vs 26 of 32 [81.3%]; OR, 1.84; 95% CI, 0.51-6.65; P = .35; I2 = 21%). The number needed to treat was calculated as 30.8 for early-stage MF to have better response rates to PUVA compared with NBUVB.
Complete response to phototherapy was pooled and found to be 73.8% (389 of 527) for patients who received PUVA compared with 62.2% (156 of 251) for those who received NBUVB, which was significantly different (OR, 1.68; 95% CI, 1.02-2.76; P = .04; I2 = 39%) (Figure 2). For the subgroup of patients with stage IA MF only, a significantly higher complete response was found in patients treated with PUVA (179 of 218 [82.1%] vs 77 of 124 [62.1%]; OR, 2.70; 95% CI, 1.25-5.87; P = .01) and in those with stage IB MF (165 of 244 [67.6%] vs 52 of 90 [57.8%]; OR, 1.93; 95% CI, 1.13-3.31; P = .02).
Partial response to phototherapy was found to be similar between PUVA and NBUVB for total number of patients (90 of 501 [18.0%] vs 64 of 223 [27.5%]; OR, 0.58; 95% CI, 0.33-1.04; P = .07) (eFigure 3 in the Supplement). No significant difference was found in patients with stage IA disease (18 of 139 [12.9%] vs 14 of 48 [29.2%]; OR, 0.45; 95% CI, 0.05-3.72; P = .46) or in those with stage IB disease (40 of 145 [27.6%] vs 9 of 62 [14.5%]; OR, 1.68; 95% CI, 0.55-5.20; P = .36).
Failed response was similar between PUVA and NBUVB when considering all included cases (44 of 527 [8.3%] vs 28 of 251 [11.1%]; OR, 0.78; 95% CI, 0.45-1.33; P = .36) (eFigure 4 in the Supplement). However, significantly fewer failed responses were found in the subgroups with stage IA MF (21 of 218 [9.6%] vs 31 of 124 [25%]; OR, 0.43; 95% CI, 0.22-0.83; P = .01) and stage IB MF (39 of 244 [16.0%] vs 29 of 90 [32.2%]; OR, 0.41; 95% CI, 0.23-0.75; P = .004).
The relapse-free interval for PUVA ranged from 10 to 45.2 months (median, 33.4) compared with 5.2 to 24.5 months (median, 14.9) for NBUVB. A significantly longer relapse-free interval for PUVA was reported by 3 of the studies,18,21,23 whereas the other studies reported similar relapse-free intervals. Kaplan-Meier analysis for relapse-free duration was reported by Almohideb et al18 and Nikolaou et al.21 When pooled, PUVA was associated with significantly higher hazard ratio of freedom from recurrence compared with NBUVB (hazard ratio, 1.93; 95% CI, 1.07-3.49; P = .03) (Figure 3).
No significant difference was found between PUVA and NBUVB for adverse effects of erythema (38 of 527 [7.2%] vs 17 of 251 [6.7%]; P = .54), nausea (10 of 527 [1.9%] vs 3 of 251 [1.2%]; P = .72), pruritus (2 of 527 [0.4%] vs 4 of 251 [1.7%]; P = .26), phototoxic effects (7 of 527 [1.4%] vs 2 of 251 [0.9%]; P = .72), dyspepsia (6 of 527 [1.2%] vs 0 of 251 [0%]; P = .59) or pain (0 of 527 [0%] vs 2 of 251 [0.9%]; P = .50).
Funnel plot analysis was produced for meta-analysis of any response to PUVA vs NBUVB therapy for early-stage MF (eFigure 5 in the Supplement). No significant funnel plot asymmetry was noted, suggesting that publication bias was not a significant influencing factor.
Management of MF is focused on optimizing quality of life. For patients with early-stage MF and widespread skin involvement, PUVA and NBUVB are widely used therapeutic options. However, despite the recent publication of the United States Cutaneous Lymphoma Consortium guidelines on phototherapy for MF,24 there is not a solid evidence-based understanding of the relevant efficacy of these 2 options. There is a lack of available evidence from randomized clinical trials on the effectiveness of PUVA or NBUVB. Whittaker et al25 prospectively reported results from a phase 3 randomized clinical trial conducted by the European Organisation for Research and Treatment of Cancer Cutaneous Lymphoma Task Force. Of 41 patients receiving PUVA monotherapy, 29 (71%) had complete or partial remission after a median of 27.5 phototherapy sessions across 9.7 months.
Our meta-analysis of 7 comparative studies demonstrated that both PUVA and NBUVB were effective at producing a partial or complete response when used for early-stage MF. However, PUVA was associated with a significantly higher rate of complete response compared with NBUVB. This was also seen in the subgroups with stages IA and IB MF. Proportions of patients with partial responses were similar, whereas failed responses were significantly lower for patients treated with PUVA regardless of any stage, stage IA only, or stage IB only. However, based on results from 2 studies,18,21 the hazard ratio of disease recurrence was significantly lower for PUVA than for NBUVB for early-stage MF. Our pooled analysis also showed no significant differences between PUVA and NBUVB in terms of adverse effects.
Our results were consistent with earlier studies reported in the literature. Nikolaou et al21 performed a large study of 227 patients with early-stage MF. They found that 74.5% of patients who received PUVA reached complete remission compared with 55.9% of those who received NBUVB. Although the present meta-analysis was not able to analyze the effect of lesions, Nikolaou et al found that patients with patch disease had better complete response rates to PUVA compared with NBUVB (91.3% vs 56.7%) but no difference in interval to relapse rate. Almohideb et al18 performed a study of 267 patients, of whom 158 received PUVA and 109 received NBUVB. The authors found that any response rate was similar between both techniques; however, disease-free survival was significantly longer for PUVA (43.25 vs 14.9 months; P < .001). They concluded that both phototherapy techniques were viable options but PUVA was associated with greater disease-free survival.
Although both PUVA and NBUVB seem to be well tolerated, adverse effects associated with PUVA have been reported in the literature. The photosensitizing agent psoralen has been linked with nausea, vomiting, headache, phototoxic effects, and photoimmune suppression.26 The carcinogenic risk of PUVA has also been reported to be greater compared with NBUVB.12 In contrast, NBUVB has been associated with erythema, blistering, pruritus, xerosis, and herpes simplex virus reactivation.27 Other advantages of NBUVB include convenience for patients because they do not need to take or bathe in psoralen and do not require UV protective glasses after treatment.
The British Association of Dermatologists guidelines suggest using NBUVB for patch disease and PUVA for plaque disease.28 Because NBUVB has proved to be effective in patch-stage disease, it is the preferred modality in such cases owing to its less severe adverse effect profile.28 However, NBUVB still has a role in plaque cutaneous T-cell lymphoma. Treatment with NBUVB is widely used for plaque MF in settings where PUVA is not readily available, and it may still be an effective therapeutic option as our review demonstrates. Furthermore, NBUVB is more widely available; is less demanding of patients because they are not required to take oral psoralen or psoralen baths or maintain absolute photoprotection after treatments as required with PUVA; and is less likely to be associated with the development of cutaneous malignant neoplasm.29,30 The decision to select a type of irradiation treatment must not only balance risks and benefits but also involve consideration of factors such as stage or lesion, compliance, cost, availability, and ease of use.31
The present meta-analysis is constrained by several limitations. First, the analysis does not adjust for potential confounders. Furthermore, articles used in the review stratified patients by stage of MF and classification of patch vs plaque disease, limiting interpretation of results. Heterogeneity in phototherapy techniques, including type of PUVA (bath vs oral), and protocol in terms of frequency of therapy and dose and duration, which varies from center to center, also undermines the validity of the results. Follow-up periods similarly varied from study to study, which limited assessment of major adverse effects after phototherapy. Pooled analysis does not account for differences in patient compliance to therapy, as well as baseline differences in patient characteristics including age, sex, comorbidities, and duration of disease before diagnosis and treatment. Included studies were predominantly observational and retrospective in nature and thus susceptible to selection bias. Randomized clinical trials would be ideal to negate this bias; however, we acknowledge the challenge of performing such studies of rare diseases such as MF. Overall the quality of evidence available was judged to vary from poor to moderate.
Based on limited evidence, our study suggests that PUVA may be an effective alternative to NBUVB for phototherapy for early-stage MF. This finding is in the context of the difficulty of performing randomized clinical trials for MF, which is a rare disease. Current results should be tempered by consideration of the adverse effect profile of both modalities.
Accepted for Publication: November 15, 2018.
Corresponding Author: Deshan F. Sebaratnam, MBBS (Hons), MMed, FACD, Department of Dermatology, Liverpool Hospital, Sydney, Australia (email@example.com).
Published Online: January 30, 2019. doi:10.1001/jamadermatol.2018.5204
Author Contributions: Drs Phan and Sebaratnam had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Phan, Fassihi, Sebaratnam.
Acquisition, analysis, or interpretation of data: Phan, Ramachandran, Fassihi.
Drafting of the manuscript: Phan, Ramachandran, Sebaratnam.
Critical revision of the manuscript for important intellectual content: Ramachandran, Fassihi.
Statistical analysis: Phan, Ramachandran.
Administrative, technical, or material support: Ramachandran, Fassihi.
Supervision: Fassihi, Sebaratnam.
Conflict of Interest Disclosures: None reported.