Clinical efficacy and safety profile of handheld narrow band ultraviolet B device therapy in vitiligo – Systematic review and meta-analysis

Search strategy

Two reviewers (SS and DP) independently conducted systematic searches to identify articles published from 1 January 2002 to 28 November 2023 in the Embase, PubMed and Scopus databases using specific keywords (narrow band ultraviolet B, handheld, targeted, home, vitiligo) and MeSH terms (ultraviolet therapy, phototherapy, vitiligo, home environment) with appropriate Boolean operators [see Appendix S1 for more details]. Additionally, each reviewer performed a manual search, examined references of various articles, and searched for ongoing trials in ClinicalTrials.gov.

Study selection criteria

Inclusion criteria:

• 1.

  Monotherapy with a handheld narrow band UVB device with a lamp system for treatment of vitiligo.

• 2.

  Studies which explicitly reported an absolute number of patients in terms of corresponding percentages of re-pigmentation or the incidence of side effects.

The eligible studies encompassed a range of research designs, including clinical trials, case series, systematic reviews, randomised controlled trials, and meta-analyses. Studies mentioning the specific NB-UVB devices by name were included, after meticulous verification of the technical specifications to confirm their nature as hand-held, portable home-based devices through product documentation; among such studies, those with sample size equal or greater than ten, were included. In situations where multiple studies emanated from a single centre, precedence was given to that which exhibited the largest sample size. No restrictions were imposed on the prior application of alternative therapeutic modalities in the vitiligo patient cohort studied. Geographical and linguistic factors were not deemed as limiting criteria. In the event of any discrepancies or inconsistencies, it was addressed through a process of consensus-building among the investigators or if required, by arbitration by a third reviewer (SK).

Data extraction

A comprehensive dataset encompassing study characteristics was generated, comprising parameters such as the primary author’s nomenclature, year of publication, study typology, detailed demographics of the study cohort (including vitiligo subtype, activity, body surface area [BSA] involvement, total sample size, average or median age of the study population), treatment particulars (initial therapeutic dosage, treatment frequency, dosage escalations and minimal dose erythema calculation), treatment duration and assessment of outcome quality.

The clinical efficacy data included a proportion of patients with re-pigmentation (> 25%, > 50% and > 75%) of skin over discrete time intervals spanning from 3 months to 12 months. Additionally, the percentages of re-pigmentation achieved in distinct anatomical regions, such as the head and neck, extremities, trunk, and acral areas, were meticulously catalogued. The data also included incidence of adverse events following therapy such as erythema, pruritus, burning sensation, dryness, skin atrophy, hyperpigmentation (abnormal skin darkening), herpetic lesions, blistering, the Koebner’s phenomenon and oedema (localised swelling).

The QoL outcomes were derived from studies employing distinct scoring systems, notably VitiQoL, Skindex 29 score, DLQI (Dermatology Life Quality Index) and Tjioe M questionnaire.⁸.

Statistical analysis

Statistical analyses were performed using MedCalc Statistical Software version 22.005 (MedCalc Software Ltd, Ostend, Belgiumhttp://www.medcalc.org; 2023). We determined the pooled proportions of re-pigmentation exceeding 25%, 50%, and 75%, achieved over a maximum duration of therapy with the device. To evaluate heterogeneity among the studies, we utilised the I² statistics, where a value greater than 50% indicated significant heterogeneity. Random effect model was used for I² statistics greater than 50% and the fixed model for I² statistics less than 50%. Additionally, we conducted a sensitivity analysis to further explore heterogeneity. Subgroup analysis was carried out based on factors such as the number of patients, MED calculation, sessions per week, disease stability/activity, and response adjusted to time. Meta-analysis was used to generate a forest plot summarising the re-pigmentation response rate. Publication bias was evaluated using a Funnel plot and Egger’s regression test, taking into consideration the percentage of re-pigmentation. A p-value less than 0.05 was considered statistically significant in this context.

Study characteristics

Through a systematic exploration of databases using predetermined search keywords, 250 articles were obtained. After a screening process, 13 studies were found suitable for inclusion, encompassing a total of 557 patients [Figure 1].^8,10–21 Among these 13 studies, eight had a randomised controlled design,^3,14,16–21 three were open-label single-arm studies,^10–12 and two were open-label non-randomised comparative trials.^8,15 The median age of patients was 30.4 years (range: 22.5–48.4 years). In the 13 studies, eight included patients of non-segmental vitiligo,^{8,10,13,14,16,19–21} two each included all types^11,12 and focal/localised vitiligo,^17,18 and one study included generalised vitiligo patients.¹⁵ Notably, seven studies recruited patients with only stable vitiligo.^{8,10,15,18–21} The extent of vitiligo involvement varied among the studies, with a reported BSA involvement from less than 2% to 25%,^{8,10,13,14–16} with other studies not specifying BSA involvement.^{11,12,17–21} Four studies used a starting dose of therapy based on MED calculation,^13,15,17,20 two studies^8,10 employed a fixed dose of 500 mJ/cm², while the remaining studies used energy based on skin type or arbitrary doses that ranged from 50 mJ/cm² to 400 mJ/cm² and studies with predominantly patients of skin types IV, V and VI used higher initial energy doses. The increment of energy per sessions was 10–20% in five studies, > 20% in two studies, and < 10% in two studies [Table 1].

Close

Figure 1:

Study selection process. The preferred reporting items for systematic reviews and meta-analysis (PRISMA) flow chart.

Export to PPT

In the majority of studies, the treatment regimen involved three sessions per week,^{10,12–16,18,19} whereas three studies employed twice-weekly sessions^11,20,21 and one study each employed once-weekly¹⁷ and daily sessions.⁸ The median duration of treatment across the evaluated studies was six months (range: 3–12 months). The studies utilised various tools (VitiQoL, Skindex 29 score, DLQI, Tjioe M questionnaire,⁸ VIS-22) to assess QoL, which consistently demonstrated improvement compared to baseline.^8,10,13–16 The characteristics of each study are mentioned in Table 1.

Pooled analysis of efficacy

The response to therapy was graded as a percentage of re-pigmentation in the 13 studies, of which 9, 12, and 13 studies had provided proportions of patients attaining > 25%, > 50%, and > 75% re-pigmentation respectively, compared to baseline [see Appendix S2 for more details]. The pooled proportions of patients achieving > 25%, > 50%, and > 75% re-pigmentation compared to baseline were 63.6% (95% CI: 51.0–75.3%) of 297 patients, 40.8% (95% CI: 30.4–51.6%) of 370 patients and 15.4% (95% CI: 7.6–25.3%) of 506 patients over a median duration of six months of therapy. The pooled proportions of patients achieving > 25%, > 50%, and > 75% re-pigmentation compared to baseline in the non-randomised studies were 73.82% (95% CI: 67.59–79.4%) of 221 patients, 51.1% (95% CI: 44.4–57.8%) of 221 patients and 16.50% (95% CI: 83.2–95.2%) of 221 patients, respectively. The pooled proportions of patients achieving > 25%, > 50%, and > 75% re-pigmentation compared to baseline in the randomised studies were 52.38% (95% CI: 39.2–65.3%) of 59 patients, 27.54% (95% CI: 9.9–49.9%) of 95 patients and 13.03% (95% CI: 5.9–22.5%) of 231 patients, respectively. The details of the response to treatment in terms of the percentage of re-pigmentation in randomised and non-randomised studies are summarised in Table 2.

Re-pigmentation based on anatomical sites

Three studies found > 50% re-pigmentation in vitiligo patches over the face and neck in 50% (45/90) of their cases, over the limbs in 50% (53/106) cases, and on the trunk in 39.1% (38/97) cases, but acral regions had only 2.5% (2/80) improvement.^12,14,15 In the study by Thomas et al., > 75% re-pigmentation was seen in 32% (20/63) of head and neck lesions, 11% (7/79) of hands and feet macules, and 17% (16/92) of the patches present on the rest of the body.¹³ In the study by Zhang et al.,¹⁵ no significant differences in re-pigmentation based on site were observed, although face, trunk, and extremities showed better response than acral sites.¹⁵ In 11 studies, a total of 38 (11.3%) out of 334 patients had no response to therapy.^{8,10–12,14–18,20,21}

Pooled analysis of adverse events

Handheld NB-UVB therapya-ssociated adverse events were reported in 12 studies with a total 542 patients (see Appendix S3 for more details). The most common side effect was erythema, with the pooled proportion of patients having erythema being 33.4% (95% CI: 19.3–49.2%) of 411 patients. Grades 3 and 4 erythema were reported in 27 and 15 patients, respectively. Pruritus was reported in five studies, with a pooled proportion of 22.1% (95% CI: 8.6–39.8%) among 227 patients.^{8,11,12,14,15} Four studies mentioned burning sensation, with a pooled proportion of 16.4% (95% CI: 6.4–29.9%) among 210 patients.^8,12,15,16 Two studies reported dryness, with a pooled proportion of 9.8% (95% CI: 2.8–20.4%) among 112 patients.^12,14 Hyperpigmentation around the vitiligo patch was reported in two studies, with a pooled proportion of 19.1% (95% CI: 11–28.9%) out of 71 patients.^14,16 Blister formation was reported in two studies, with a pooled proportion of 9.7% (95% CI: 0.8–27.1%) out of 69 patients.^8,16 One study each had mentioned skin thinning (two patients),¹³ cold sore (one patient),¹⁴ Koebner’s phenomenon (one patient),¹⁶ and oedema (three patients)⁸ as side effects of handheld NB-UVB therapy.

Heterogeneity

There was clinical heterogeneity noted in the included studies in terms of the study design, disease activity, BSA involvement, MED calculation, frequency of administration, and energy dosing. There was no significant heterogeneity noted among the non-randomised studies for pooled estimates of re-pigmentation of > 25% (I² = 0%; 95% CI 0–%, 71.9, p = 0.49) and > 50% (I² = 19.73%; 95% CI 0–64.4%, p = 0.28), but for > 75% repigmentation, they showed significant heterogeneity (I² = 91.01%; 95% CI 83.2–95.2%, p < 0.01). There was significant heterogeneity noted among the randomised studies for pooled estimates of re-pigmentation of > 50% (I² = 77.13%; 95% CI 25.54–92.9%, p = 0.01) and > 75% (I² = 79.04%; 95% CI 59.01–89.28%, p < 0.01, but no such heterogeneity was noted for estimates of repigmentation of > 25% (I² = 42.82%, p < 0.18), expect for). Based on sensitivity analysis, no articles were found to be a specific source of heterogeneity. There was significant heterogeneity for pooled estimates of erythema (I² = 89.3%; 95% CI 81.9–93.7%, p < 0.001), burning sensation (I² = 80.60%; 95% CI 49–92.6%, p < 0.01) and pruritus (I² = 87.4%; 95% CI 72.9–94.1%, p < 0.01). There was insignificant statistical heterogeneity noted for dryness (I² = 44.2%%, p = 0.18), hyperpigmentation (I² = 0%, p = 0.82) and blistering (I² = 67.1%, p = 0.08).

Subgroup analysis

Subgroup analyses of the percentage of re-pigmentation achieved based on the number of patients, MED calculation, vitiligo activity, the number of sessions per week and response adjusted to time were conducted. In studies with > 20 patients, there were significantly higher rates of > 25% re-pigmentation response (70.1% [95% CI: 60.7–78.7%] vs 55.5% [95% CI: 27.7–81.5%]; p = 0.02) and > 50% re-pigmentation response (48.6% [95% CI: 20.5–64.3%] vs 29.6% [95% CI: 11.1–52.5]; p = 0.002) compared to studies with < 20 patients. However, the pooled proportions of > 75% re-pigmentation showed no difference (16.6% [95% CI: 6.01–31.1%] vs 13.9% [95% CI: 4.6–27.1%]; p = 0.54] between studies including less than 20 patients and those including more patients. The number 20 was chosen to reflect the minimum sample size needed for a phase 1 clinical trial.

The patients who had used handheld NB-UVB therapy based on MED calculation showed no significantly better response for > 25% (67.5% [95% CI: 42.3–88.2%] vs 61.5% [95% CI: 45.8–76.1%]; p = 0.39), > 50% (36.0% [95% CI: 9.3–68.8%] vs 41.5% [95% CI: 29.9–53.5%]; p = 0.43) or > 75% (11.5% [95% CI: 3.4–23.7%) vs 17.1% [95% CI: 6.6–31.1%]; p = 0.08) re-pigmentation compared to patients who were treated without MED calculation.

The pooled > 25% re-pigmentation response was observed to be better in the four studies,^8,10,15,20 which included only stable vitiligo than in the three studies^14,16,17 that included both active or slowly progressive vitiligo along with stable disease (66.5% [95% CI: 44.3–85.4%] vs 51.5% [95% CI: 21.8–80.6%]; p = 0.06). The pooled > 50% (37.3% [95% CI: 20.6–55.7%] vs 38.9% [95% CI: 16.1–64.7%]; p = 0.80) and > 75% (13.9% [95% CI: 4.5–27.4%] vs 18.5% [95% CI: 7.8–32.6%]; p = 0.25) re-pigmentation responses reported in seven studies^{8,10,15,18–21} which included only stable vitiligo were not significantly different from the figures reported in studies^13,14,16,17 which included both active or slowly progressive vitiligo along with stable disease.

The > 25% re-pigmentation response in patients who had received three to four sessions per week of therapy was not significantly different compared to that of patients receiving once/twice weekly and daily therapy (62.8% [95% CI: 40.0–85.8%] vs 65% [95% CI: 46.6–81.4%]; p = 0.72). However, > 50% (46.5% [95% CI: 33.9–59.2%] vs 30.7% [95% CI: 13.2–51.8%]; p = 0.006) and > 75% (19.5% [95% CI: 9.9–31.4%] vs 9.1% [95% CI: 0.34–27.76%]; p = 0.01) re-pigmentation were significantly better with three to four sessions per week when compared to once/twice weekly and daily therapy. However, there were no significant differences in > 25%, > 50% and > 75% re-pigmentation in those receiving three to four sessions per week when compared with those receiving daily therapy sessions (p-values of 0.08, 0.19 and 0.85, respectively).

The pooled proportions of patients achieving > 25%, > 50% and > 75% re-pigmentation compared to baseline at 12 weeks of therapy were 31.1% (95% CI: 9.6–58.3%; I² = 94.4%, p < 0.01), 12.9% (95% CI: 3.1–28.1%; I² = 88.8%, p < 0.01) and 6.50% (95% CI: 1.7–14.1%; I² = 73.7%, p < 0.01), respectively. At 24 weeks, the corresponding proportions were 53.2% (95% CI: 24.5–80.7%; I² = 96.5%, p < 0.01), 36.7% (95% CI: 15.8–60.5%; I² = 94.5%, p < 0.01) and 11.1% (95% CI: 2.9–23.7%; I² = 88.4%, p < 0.01).

Risk of bias

In the included studies, five studies were open-label, thereby inherently carrying the risk of selection bias^8,10–12,15 with none of them having a score of less than four on the Newcastle Ottawa scale adapted for single arm. Funnel plot of the articles based on re-pigmentation > 25%, > 50% and > 75% showed no evidence of publication bias on visual inspection [Figure 2]. Following Cochrane guidelines,²² all randomised studies were found to have moderate to good quality with low risk of bias for the relevant outcome of interest [Table 3 and Figure 3]. The p value for the Egger’s regression test was statistically insignificant for > 25% (p = 0.83), > 50% (p = 0.27) and > 75% (p = 0.80) re-pigmentation.

Close

Figure 2a:

Pooled percentage of re-pigmentation more than 50% and 75%. Forest Plots. Squares and horizontal lines represent individual study estimates and 95% CIs, respectively.

Export to PPT

Close

Figure 2b:

Funnel plots. Central line shows summary estimates and the lines on either side show 95% CI. Circles represent individual studies.

Export to PPT

Close

Figure 3:

Assessment of the risk of bias via the Cochrane RoB 2 tool displayed by means of a traffic light plot of the risk of bias of each included clinical study.

Export to PPT