9.1.2.1. Evidence statements

In people with psoriasis there was no statistically significant difference between 3-times weekly selective BBUVB and 3-times weekly NBUVB for:

• Clear at the end of treatment [1 between-patient study; 85 participants; moderate quality evidence]¹⁸⁸.
• Remaining clear at 3 months post treatment [1 between-patient study; 43 participants; moderate quality evidence]¹⁸⁸.
• Remaining clear at 6 months post treatment [1 between-patient study; 32 participants; very low quality evidence]¹⁸⁸.
• Withdrawal due to toxicity [1 between-patient study; 89 participants; low quality evidence]¹⁸⁸.

Evidence statements for individual studies where no statistical analysis could be performed comparing 3–5-times weekly BBUVB and 3–5-times weekly NBUVB:

• One within-patient study found that both sides improved at 8 weeks although the improvement was slightly greater on the NBUVB-treated side [1 study; 29 participants (58 randomised units); very low quality evidence]²¹². This study was randomised by order of exposure and not for which side of the body received which treatment.
• Two within-patient studies found that NBUVB was more effective than BBUVB
  • 1 study found that 3–5-times weekly NBUVB resulted in greater improvement in PASI than 3–5- times weekly BBUVB after 5–15 treatments [1 study; 10 participants (20 randomised units); very low quality evidence]³⁹².
  • 1 study found that the average reductions in PASI at 10 weeks were 78.5% and 73.9% for NBUVB and BBUVB (both 3-times weekly), respectively, which was a statistically significant difference [1 study; 15 participants (30 randomised units); very low quality evidence]³¹². Note that this study did not use equi-erythemogenic dosing.

9.1.3.1. Oral PUVA (between patient randomisation)

9.1.3.2. Evidence statements

In people with psoriasis two- or three-times weekly oral PUVA was statistically significantly better than two- or three-times weekly NBUVB for:

• Clear or nearly clear on PGA at the end of treatment (maximum 30–40 treatments) [2 between-patient studies; 167 participants; low quality evidence]^124,429
• Relapse rate for clearers after 6–12 months [4 between-patient studies; 196 participants; moderate quality evidence] ^{56,124,237,429}
• Mean time to clearance after a maximum follow-up of 3 months [1 between-patient study; 60 participants; moderate quality evidence] ⁷⁰

In people with psoriasis three-times weekly NBUVB was statistically significantly better than three-times weekly oral PUVA for:

• Final PASI score (three-times weekly UV) after a maximum of 20 treatments [1 between-patient study; 50 participants; very low quality evidence]³⁶²

In people with psoriasis there was no statistically significant difference between two- or three-times weekly NBUVB and two- or three-times weekly PUVA for:

• PASI75 (skin type II – III or IV – V) at 3–4 months or after a maximum of 20 treatments [3 between-patient studies; 153 participants; moderate quality evidence] ^56,70,362
• Final PASI score (twice-weekly UV) at 3 months [1 between-patient study; 60 participants; very low quality evidence] ⁷⁰
• Mean time to PASI75 after a follow-up of 4 months [1 between-patient study; 43 participants; very low quality evidence]⁵⁶
• Withdrawal due to toxicity after a maximum 16–30 treatments [2 between-patient studies; 164 participants; very low quality evidence] ^124,429

Evidence statements for individual studies where no original analysis could be performed comparing narrowband UVB and PUVA:

• One study found that there was a longer time to relapse with twice weekly PUVA compared with twice weekly NBUVB after a maximum follow-up of 12 months [1 between-patient study; 57 participants; low quality evidence]⁴²⁹
• One study found that there was no significant difference in time to relapse with twice weekly PUVA compared with three-times weekly NBUVB after a maximum follow-up of 12 months [1 between-patient study; 45 participants; low quality evidence]^236,237
• Two studies found that there was a greater mean or median change in PASI with two- or three-times weekly PUVA than two- or three-times weekly NBUVB at 8–10 weeks [2 between-patient studies; 109 participants; low to very low quality evidence]^11,429
• One study found that there was a no significant difference in median time to clearance between twice weekly PUVA and three-times weekly NBUVB [1 between-patient study; 45 participants; low quality evidence]²³⁷

9.1.3.3. Subgroup analysis and heterogeneity

Data were available for different skin types based on the Fitzpatrick classification between studies and as a post-hoc subgroup analysis in one study.

• There was significant heterogeneity for the outcome of final PASI between two studies^70,362. This could be explained by pre-defined subgroups based on skin type (II–III³⁶² and IV–V⁷⁰). However, it was felt to be more likely that the heterogeneity was due to differences in treatment frequency between the studies as skin type variation would have been accounted for in the calculation of the minimal erythrogenic dose. One study³⁶² using 3-times weekly administration (optimal for UVB but higher than usual for PUVA) and the other⁷⁰ twice-weekly administration (sub optimal for NBUVB but usual for PUVA) of both interventions. There was no significant heterogeneity between these two studies for the outcome of PASI75.
• One study⁴²⁹ presented a post-hoc subgroup analysis for different skin types for the outcome of clear or nearly clear on PGA. The samples sizes in the type V-VI subgroup were very small making it difficult to draw any conclusions about the relative difference in effectiveness of NBUVB and PUVA. There was a high, but not statistically significant, degree of difference between the subgroups (I² = 47.6%) and the proportion responding to either kind of light treatment was markedly lower in the skin type V-VI subgroup (23.5%) than the I-IV subgroup (74.6%).

9.1.3.4. Bath PUVA

9.1.3.5. Evidence statements

In people with psoriasis there was three-times weekly NBUVB was statistically significantly better than twice weekly bath PUVA for:

• Time-to-remission (clearance or minimal residual activity) after a maximum of 30 treatments [1 within-patient study; 28 participants (56 randomised units); moderate quality evidence]⁶⁸
• Mean change in PASI at 10 weeks [1 within-patient study; 14 participants (28 randomised units); moderate quality evidence]³⁷⁶
• Mean days to relapse after a maximum follow-up of 6.5 months [1 within-patient study; 21 participants (36 randomised units); low quality evidence]⁶⁸

In people with psoriasis there was no statistically significant difference between three-times weekly NBUVB and two-or three-times weekly bath PUVA for:

• Withdrawal due to toxicity at 10 weeks [1 within-patient study; 15 (30 randomised units) participants; low quality evidence]³⁷⁶
• Burn after a maximum of 30 treatments [1 within-patient study; 28 participants (56 randomised units); very low quality evidence] ⁶⁸

9.1.4.1. NBUVB five-times vs three-times weekly

9.1.4.2. Evidence statements

In people with psoriasis there was no statistically significant difference between 3-and 5-times weekly NBUVB for:

• Clearance at 23 weeks or after a maximum of 30 treatments [2 studies (one between-patient and one within-patient); 64 participants (83 randomised units); moderate quality evidence]^69,137

In people with psoriasis there were no events with either 3- or 5-times weekly NBUVB for:

• Burn after a maximum of 30 treatments [1 between-patient study; 65 participants; moderate quality evidence] ¹³⁷
• Withdrawal due to toxicity at 23 weeks [1 within-patient study; 19 participants (38 randomised units); moderate quality evidence] ⁶⁹

Evidence statements for individual studies where no original analysis could be performed comparing narrowband UVB 3- vs 5-times weekly:

• 2 studies showed that 5-times weekly NBUVB resulted in a shorter time to clearance that 3-times weekly NBUVB after a maximum of 23 weeks [2 studies (one between-patient and one within patient); 52 participants (71 randomised units); low to moderate quality evidence]^69,137
• 1 study showed that there was no significant difference in time to relapse with 3- and 5-times weekly NBUVB after a maximum follow-up of 12 months [1 within-patient study; 19 participants (38 randomised units); very low quality evidence]⁶⁹

9.1.4.3. Narrowband UVB two times vs three times weekly

9.1.4.4. Evidence statements

In people with psoriasis there was no statistically significant difference between 2- and 3-times weekly NBUVB for:

• Clearance [1 between-patient study; 92 participants; moderate quality evidence]⁵⁰
• Withdrawal due to toxicity [1 between-patient study; 87 participants; very low quality evidence]⁵⁰
• Severe UV erythema (burn) [1 between-patient study; 113 participants; very low quality evidence] ⁵⁰

Evidence statements for individual studies where no original analysis could be performed comparing narrowband UVB 2- vs 3-times weekly:

• 1 study showed that 3-times weekly NBUVB resulted in a shorter time to clearance that 2-times weekly [1 study; 113 participants; low quality evidence]⁵⁰
• 1 study showed that 2-times weekly NBUVB resulted in a longer time to relapse that 3-times weekly after a maximum follow-up of 12 months post-treatment [1 study; 113 participants; low quality evidence]⁵⁰

9.1.4.5. Different oral PUVA treatment frequencies (3 vs 2 times weekly)

9.1.4.6. Evidence statements

In people with psoriasis there was no statistically significant difference between 2-and 3-times weekly oral PUVA for:

• Clear/nearly clear on IAGI at 12 weeks [1 between-patient study; 19 participants; low quality evidence]⁸³
• Percentage change in PASI at 12 weeks [1 between-patient study; 19 participants; low quality evidence]⁸³
• Burn at a maximum of 25 treatments [1 within-patient study; 23 participants (46 randomised units); very low quality evidence]⁴⁰⁶

9.1.4.7. Subgroup analysis and heterogeneity

Data were available for percentage change in PASI up to 25 treatments for different skin types based on the Fitzpatrick classification and for different psoriasis phenotypes (see Glossary).

Table 84

Summary of non-analysed data for PUVA 2 vs 3 times weekly.

• 1 study showed that there was no significant difference for median change in PASI between oral PUVA 2-and 3-times weekly after a maximum of 25 treatments [1 within-patient study; 28 participants (56 randomised units); low quality evidence]⁴⁰⁶
• Oral PUVA 2-times weekly resulted in a greater median decrease in PASI after a maximum of 25 treatments for skin types III-IV and for the ostraceous subtype of psoriasis (this is an infrequently used term to describe plaque-type psoriasis that is particularly hyperkeratotic, typically with relatively concave centres, similar in shape to oyster shells) [1 within-patient study; 28 participants (56 randomised units); very low quality evidence]⁴⁰⁶

9.1.4.8. Oral PUVA vs no treatment for palmoplantar pustulosis

9.1.4.9. Evidence statements

In people with palmoplantar pustulosis oral PUVA 3 or 4 times weekly for hand and foot palmoplantar pustulosis was statistically significantly better than no treatment for:

• Clearance at 7.5–12 weeks [2 within-patient studies; 34 participants (68 randomised units); moderate quality evidence]^259,341
• Improvement at 7.5–12 weeks [2 within-patient studies; 34 participants (68 randomised units); moderate quality evidence] ^259,341

In people with palmoplantar pustulosis there was no statistically significant difference between 3- or 4-times weekly oral hand and foot PUVA and no treatment for:

• Withdrawal due to toxicity at 7.5–12 weeks [2 within-patient studies; 35 participants (69 randomised units); very low quality evidence] ^259,341
• Burn at 7.5–12 weeks [2 within-patient studies; 34 participants (68 randomised units); low quality evidence] ^259,341

9.1.4.10. Cream PUVA vs narrowband UVB for hand and foot palmoplantar pustulosis

9.1.4.11. Evidence statements

In people with palmoplantar pustulosis three-times cream hand and foot PUVA was statistically significantly better than NBUVB three-times weekly for:

• Clear or nearly clear at 9 weeks [1 within-patient study; 21 participants (42 randomised units); moderate quality evidence]³⁶⁴

In people with palmoplantar pustulosis there was no statistically significant difference between cream hand and foot PUVA three-times weekly and NBUVB three-times weekly for:

• Withdrawal due to toxicity at 9 weeks [1 within-patient study; 22 participants (43 randomised units); very low quality evidence]³⁶⁴
• Relapse 10 weeks after treatment [1 within-patient study; 21 participants (42 randomised units); very low quality evidence]³⁶⁴

9.1.4.12. Home vs hospital NBUVB for psoriasis

9.1.4.13. Evidence statements

In people with psoriasis there was no statistically significant difference between 3- or 4-times weekly NBUVB home and 2- or 3-times weekly hospital NBUVB for:

• Clear/nearly clear (PASI90) after a maximum of 46 treatments [1 between-patient study; 185 participants; low quality evidence]^190,191
• PASI75 after a maximum of 46 treatments [1 between-patient study; 185 participants; low quality evidence] ^190,191
• PASI50 after a maximum of 46 treatments [1 between-patient study; 185 participants; high quality evidence] ^190,191

Evidence statements for outcomes where no original analysis could be performed comparing 3- or 4- times weekly NBUVB home and 2- or 3-times weekly hospital NBUVB for:

• There was no meaningful difference between the number of participants experiencing severe UV erythema, blistering or a burning sensation after a maximum of 46 treatments [1 between-patient study; 185 participants; low quality evidence] ^190,191.

9.1.5.1. Unit costs

In the absence of recent UK cost-effectiveness analysis, relevant unit costs are provided below to aid consideration of cost effectiveness.

9.1.5.2. Economic evidence statements

• No cost-effectiveness analyses were identified comparing all three interventions of interest – broadband UVB, narrowband UVB and PUVA – in the treatment of patients with psoriasis.
• One partially applicable study with potentially serious limitations found that in a population with psoriasis eligible for treatment with phototherapy, narrowband UVB delivered in the home was more costly and more effective than narrowband UVB delivered in an outpatient setting, with an ICER of £ 34,967 during treatment and £ 7,432 in the year following treatment. There is considerable uncertainty as to whether narrowband UVB delivered in the home would be cost effective.
• One partially applicable study with very serious limitations found that in a population with mild to moderate psoriasis, oral PUVA is more costly and more effective than broadband UVB with an ICER of £ 20 per additional day in remission. This was roughly translated to an incremental cost per QALY ratio of £ 39,167.
• One partially applicable study with very serious limitations found that in a population with moderate to severe psoriasis, oral PUVA is more costly and more effective than narrowband UVB with an ICER of £ 67 per additional 1% of patients achieving a PASI 75 or total body clearance. Based on this evidence alone, it is impossible to conclude whether PUVA would represent a more or less cost-effective use of NHS resources compared to narrowband UVB.

9.2.2.1. Evidence profile

9.2.2.2. Evidence statements

In patients with psoriasis, acitretin plus BBUVB was statistically significantly better than acitretin for:

• Clear/nearly clear on IAGI after a maximum of 30 exposures [1 study; 9 participants (18 randomised units); low quality evidence]¹⁵⁹

In patients with psoriasis, there was no statistically significant difference between acitretin and acitretin plus BBUVB for:

• Withdrawal due to drug toxicity after a maximum of 30 exposures [1 study; 9 participants (18 randomised units); very low quality evidence]¹⁵⁹

9.2.3.1. Evidence profile

9.2.3.2. Evidence statements

In patients with psoriasis, there was a statistically significant difference favouring the use of acitretin plus BBUVB compared to a placebo plus BBUVB for:

• Clear/nearly clear on IAGI at 8 weeks [1 between-patient study; 78 participants; low quality evidence]³⁴⁵

In patients with psoriasis, there was no statistically significant difference between acitretin plus BBUVB and placebo plus BBUVB for:

• Withdrawal due to drug toxicity at 8 weeks [1 between-patient study; 66 participants; very low quality evidence]³⁴⁵

9.2.4.1. Evidence profile

9.2.4.2. Evidence statements

In patients with psoriasis, there was no statistically significant difference between acitretin plus NBUVB and acitretin plus PUVA for:

• PASI 75 at 8 weeks [1 between-patient study; 60 participants; low quality evidence] ³⁰⁰
• PASI50 at 8 weeks [1 between-patient study; 60 participants; moderate quality evidence]³⁰⁰
• Number of UV treatments after a maximum of 8 weeks [1 between-patient study; 60 participants; moderate quality evidence] ³⁰⁰
• Maintenance of remission at 3 months [1 between-patient study; 36 participants; high quality evidence] ³⁰⁰
• Burns at 8 weeks [1 between-patient study; 60 participants; low quality evidence] ³⁰⁰
• Withdrawal due to drug toxicity at 8 weeks [1 between-patient study; 60 participants; low quality evidence] ³⁰⁰

The data for the number of UV treatments was not reported clearly. The figures given were assumed to be a standard deviation rather than a standard error of the mean. If using the SEM the SD would have been greater than the mean number of UV treatments.

9.2.5.1. Evidence profile

9.2.5.2. Evidence statements

In patients with psoriasis, there was a statistically significant difference favouring the use of acitretin plus PUVA compared to placebo plus PUVA for the:

• Clear/nearly clear on IAGI at 8–12 weeks [3 between-patient studies; 169 participants; very low quality evidence] ^352,379,397
• Mean number of UV treatments (studies using a Completers Analysis) after a maximum of 8 weeks [2 between-patient studies; 86 participants; moderate quality evidence] ^352,397
• Time to remission after a maximum of 12 weeks [1 between-patient study; 33 participants; moderate quality evidence]³⁵²

A statistically significant difference favouring the use of a placebo plus PUVA compared to acitretin plus PUVA was found for:

• Severe adverse events at 12 weeks [2 between-patient studies; 125 participants; moderate quality evidence] ^352,379.

No statistically significant associations were found for:

• Withdrawal due to drug toxicity at 8–12 weeks [3 between-patient studies; 159 participants; very low quality evidence] ^352,379,397)
• Mean number of UV treatments after a maximum of 8 weeks [1 between-patient study; 83

9.2.5.3. Heterogeneity

There was heterogeneity between the three studies for the outcome of number of UV treatments. The studies did not report the mean number of UVB treatments require for clearance in those who achieved remission but rather the total mean number in the analysis set; however, those who achieved remission before the end of the study did stop treatment early. It is likely that this was because the Sommerburg study included all patients randomised while the other two studies only reported an available case analysis, but it could also have been due to the higher proportion of people in the Sommerburg study with non-plaque type psoriasis: both the Tanew and Saurat studies had primarily patients with chronic plaque psoriasis (100% and 93% respectively) whereas Sommerburg had a mixed population (acitretin arm: guttate 12.5%, nummular 27.5%, plaque 57.5%, guttate and nummular 2.5%; placebo arm: guttate 9.3%, nummular 23.3%, plaque 65.1%, guttate and nummular 2.3%)- figures are acitretin plus PUVA and placebo plus PUVA respectively. The lower proportion with plaque psoriasis in the acitretin arm could have meant that the psoriasis was more resistant and took relatively longer to clear than that in the placebo arm.

9.2.6.1. Evidence statements

• One partially applicable study with very serious limitations found that in a population with moderate to severe psoriasis, narrowband UVB is more costly and more effective than acitretin (25 mg/day), with an ICER of £40 per additional 1% achieving PASI75 or total body clearance. However, based on this evidence alone, it is unclear whether this represents good value for the UK NHS.
• One partially applicable study with very serious limitations found that in a population with moderate to severe psoriasis, oral PUVA is more costly and more effective than narrowband UVB with an ICER of £67 per additional 1% of patients achieving a PASI 75 or total body clearance. Based on this evidence alone, it is impossible to conclude whether PUVA would represent a more or less cost-effective use of NHS resources compared to narrowband UVB.

9.3.2.1. Evidence profile

9.3.2.2. Evidence statements

In people with psoriasis, calcipotriol combined with NBUVB was statistically significantly better than calcipotriol alone for:

• PASI 50 at 3 months [1 between-patient study, 30 participants, very low quality evidence]³⁴³
• Mean reduction in PASI at 3 months [1 between -patient study, 30 participants, very low quality evidence]³⁴³

In people with psoriasis, there was no statistically significant difference between vitamin D analogues combined with NBUVB versus vitamin D analogue alone for:

• Clearance (PASI100) at 3 months for calcipotriol [1 between -patient study, 30 participants, very low quality evidence]³⁴³
• Withdrawal due to adverse events for tacalcitol [1 within-patient study, 23 participants (45 randomised units), very low quality evidence]³³⁷

Evidence statements for individual studies where no original analysis could be performed comparing vitamin D analogue plus NBUVB versus vitamin D analogue alone:

• Mean PASI improved significantly more at 3 months with calcipotriol combined with NBUVB versus calcipotriol alone [1 between -patient study, 30 participants, very low quality evidence]³⁹
• Mean final PASI at 3 weeks was a statistically significantly lower with tacalcitol combined with NBUVB versus tacalcitol alone [1 within-patient study, 22 participants (44 randomised units), very low quality evidence]³³⁷

9.3.3.1. Evidence profile

9.3.3.2. Evidence statements

In people with psoriasis, there was no statistically significant difference between calcipotriol combined with BBUVB and calcipotriol alone for:

• Clearance at 8 weeks [1 within-patient study, 18 participants (36 randomised units), very low quality evidence]¹⁹⁴

9.3.4.1. Evidence profile

9.3.4.2. Evidence statements

In people with psoriasis, there was no statistically significant difference between calcipotriol combined with NBUVB versus NBUVB plus placebo for:

• Clearance at 6 weeks [1 between-patient study, 28 participants, very low quality evidence] ³³³
• Mean number of UVB treatments [2 between-patient studies, 103 participants, high quality evidence]^40,425
• Mean number of UVB treatments (extremities or trunk) [1 between-patient study, 28 participants, very low quality evidence]³³³
• Percentage change in PASI [1 between-patient study, 53 participants, very low quality evidence]⁴⁰
• Change in PASI after a maximum of 20 sessions [1 between-patient study, 50 participants, high quality evidence]⁴²⁵
• Mild to moderate burn at 6 weeks [1 between-patient study, 28 participants, very low quality evidence] ³³³
• Withdrawal due to adverse events at 6 weeks or a maximum of 20 sessions [3 between-patient studies, 131 participants, very low quality evidence] ^40,333,425

Evidence statements for individual studies where no original analysis could be performed comparing vitamin D analogue plus NBUVB versus NBUVB alone:

• Mean PASI improved significantly more at 3 months with calcipotriol combined with NBUVB versus NBUVB alone [1 between-patient study, 30 participants, very low quality evidence] ³⁹

9.3.5.1. Evidence profile

9.3.5.2. Evidence statements

In people with psoriasis, there was a statistically significant difference favouring a vitamin D or vitamin D analogue combined with BBUVB versus BBUVB plus placebo for:

• Clear or nearly clear on IAGI at 8 weeks for calcitriol [1 between-patient study, 102 participants, moderate quality evidence]³³⁴
• Number of UV treatments to clearance after a maximum follow-up of 3 months for calcipotriol [1 between-patient study, 159 participants, low quality evidence]³²⁴
• Number of UV treatments to modified PASI80 after a maximum follow-up of 3 months for calcipotriol [1 between-patient study, 159 participants, low quality evidence]³²⁴

In people with psoriasis, there was no statistically significant difference between vitamin D or vitamin D analogue combined with BBUVB versus BBUVB plus placebo for:

• Clearance at 3 months for calcipotriol [1 between-patient study, 159 participants, very low quality evidence] ³²⁴
• Modified PASI 80 at 3 months for calcipotriol [1 between-patient study, 159 participants, very low quality evidence] ³²⁴
• Percentage change in modified PASI at 3 months for calcipotriol [1 between-patient study, 159 participants, low quality evidence] ³²⁴
• Relapse post-treatment among clearers after a maximum follow-up of 12 weeks post-treatment for calcipotriol [1 between-patient study, 95 participants, very low quality evidence] ³²⁴
• Burn/erythema/pruritus at 3 months for calcipotriol [1 between-patient study, 159 participants, very low quality evidence] ³²⁴
• Withdrawal due to adverse events at 8 weeks for calcitriol [1 between-patient study, 102 participants, very low quality evidence] ³³⁴

Evidence statements for individual studies where no statistical analysis could be performed comparing vitamin D plus BBUVB versus placebo plus BBUVB:

• Percentage change in PASI at 8 weeks was greater with calcitriol compared with placebo [1 between-patient study, 102 participants, low quality evidence]³³⁴

9.3.5.3. Heterogeneity

There was statistically significant heterogeneity between the two studies for the outcome of clear/nearly clear^324,334. It was not possible to conclusively determine the cause of this inconsistency, which could have been due to different vitamin D agents being used, different definitions of response or different follow-up times.

9.3.6.1. Evidence profile

9.3.6.2. Evidence statements

In people with psoriasis, there was no statistically significant difference between LCD combined with NBUVB versus NBUVB for:

• Clearance at 12 weeks [1 within-patient study, 12 participants (24 randomised units), very low quality evidence]²²
• Moderate burn at 12 weeks [1 within-patient study, 12 participants (24 randomised units), very low quality evidence]²²

In people with psoriasis, there were no events with either LCD combined with NBUVB or NBUVB for:

• Withdrawal due to adverse events at 12 weeks [1 within-patient study, 12 participants (24 randomised units), moderate quality evidence]²²
• Serious adverse events at 12 weeks [1 within-patient study, 12 participants (24 randomised units), moderate quality evidence]²²

Evidence statements for individual studies where no original analysis could be performed comparing LCD plus NBUVB versus NBUVB:

• There was no statistically significant difference reported between the median number of weeks to clearance/minimal disease after a maximum followup of 12 weeks [1 within-patient study, 12 participants (24 randomised units), low quality evidence]²²

9.3.7.1. Evidence profile

9.3.7.2. Evidence statements

In people with psoriasis, there was no statistically significant difference between tar oil with suberythemogenic BBUVB versus maximally erythemogenic BBUVB with placebo for:

• Clearance at 12 weeks [1 between-patient study, 49 participants, very low quality evidence] ²⁵⁰

Evidence statements for individual studies where no original analysis could be performed comparing tar oil plus suberythemogenic BBUVB versus placebo plus maximally erythemogenic BBUVB:

• There was a statistically significant reduction in mean number of UVB treatments for clearance with tar oil + suberythemogenic BBUVB versus placebo + maximally erythemogenic BBUVB after a maximum follow-up of 12 weeks [1 between-patient study, 33 participants, very low quality evidence] ²⁵⁰

9.3.8.1. Evidence profile

9.3.8.2. Evidence statements

In people with psoriasis, dithranol (micanol) plus BBUVB was statistically significantly better than dithranol alone for:

• Clear or nearly clear (≤1% BSA, ≤1 on all severity scores) at 8 weeks [1 within-patient study, 24 participants (48 randomised units), low quality evidence]¹¹⁸

In people with psoriasis, there was no statistically significant difference between dithranol (Micanol) plus BBUVB versus dithranol alone for:

• Irritation (requiring adjustment of dithranol) at 8 weeks [1 within-patient study, 24 participants (48 randomised units), very low quality evidence]¹¹⁸

Evidence statements for individual studies where no statistical analysis could be performed comparing dithranol (micanol) plus BBUVB versus dithranol alone:

• The median number of weeks to achieve clear or nearly clear status was shorter with the combination regimen after a maximum follow-up of 8 weeks [1 within-patient study, 15 participants (22 randomised units), low quality evidence]¹¹⁸

9.3.9.1. Evidence profile

9.3.9.2. Evidence statements

In people with psoriasis, there was no statistically significant difference between dithranol (micanol) plus BBUVB versus placebo plus BBUVB for:

• Clear or nearly clear (≤1% BSA, ≤1 on all severity scores) at 8 weeks [1 study, 24 participants (48 randomised units), low quality evidence] ¹¹⁸

Evidence statements for individual studies where no statistical analysis could be performed comparing dithranol (micanol) plus BBUVB versus placebo plus BBUVB:

• The median number of weeks to achieve clear or nearly clear status was the same with both treatments after a maximum follow-up of 8 weeks [1 within-patient study, 15 participants (26 randomised units), low quality evidence] ¹¹⁸

9.3.9.3. Dithranol (short-contact) plus coal tar plus BBUVB vs dithranol

The short-contact dithranol intervention included salicylic acid in the formulation and is likely to have been administered in a day-care setting, unlike micanol, which is suitable for home use.

9.3.9.4. Evidence profile

9.3.9.5. Evidence statements

In people with psoriasis, there was no statistically significant difference between dithranol plus coal tar plus BBUVB versus dithranol for:

• Clearance at 3 weeks [1 between-patient study, 53 participants, very low quality evidence]³⁰⁵
• Mean number of days to clearance after a maximum of 3 weeks [1 between-patient study, 53 participants, low quality evidence]³⁰⁵
• Relapse rate post treatment [1 between-patient study, 36 participants, very low quality evidence]³⁰⁵

Evidence statements for individual studies where no statistical analysis could be performed comparing SCDT plus coal tar plus BBUVB versus dithranol:

Mean time to relapse among those who cleared was longer with SCDT + BBUVB + coal tar versus dithranol alone [1 between-patient study, 53 participants, low quality evidence]³⁰⁵

9.3.10.1. Unit costs

In the absence of recent UK cost-effectiveness analysis, relevant unit costs were sourced to aid consideration of cost effectiveness. In the case of dithranol and crude coal tar, costs are quite variable. Products listed in the BNF, are typically of lower concentrations and are intended for home use and application. Dithranol and crude coal tar products that are used in specialist day centres are of higher concentrations and are available as ‘specials’ from licensed ‘special-order’ manufacturers. Table 109 presents unit costs for the home use products included in the BNF and Table 110 presents unit costs of ‘specials’ from a selection of licensed NHS hospital manufacturing units.

Table 109

Costs of medications for home use.

Table 110

Costs of medications for specialist day centre use.

The unit costs for ‘specials’ are dependent on the ingredients, quantities, pack size and batch size, with the most significant drivers being concentration (due to ingredients) and batch size. Based on personal communications with pharmacy technicians and directors at a variety of NHS hospital manufacturing units (Calderdale & Huddersfield NHS Foundation Trust, Colchester Hospital University NHS Foundation Trust, Eastbourne Pharmaceuticals at Eastbourne District General Hospital, Guy’s & St Thomas’ NHS Foundation Trust, Royal Free Hospital), dithranol and crude coal tar produced in batches are quite modest in cost (between £5 and £22 per 100 g depending on concentration); however, when prepared extemporaneously (individually compounded products) the cost is significantly greater (£70 to £150 per 100 g depending on concentration). Several NHS hospital manufacturing units also indicated that they had either reduced preparation of these ‘specials’ or had stopped making them altogether due to low demand or increasing difficulty in sourcing suitable raw materials. Based on this information, it seems reasonable to conclude that outside of very busy specialty dermatology units, it is very likely that dithranol and crude coal tar ‘specials’ will be prepared extemporaneously and therefore have high unit costs.

Table 111

Unit cost of phototherapy and psoriasis-related day case hospital visit.

9.3.10.2. Evidence statements

• No cost-effectiveness analyses were identified comparing narrowband UVB combined with dithranol, coal tar, or vitamin D or its analogues compared with narrowband UVB, dithranol, coal tar or vitamin D or vitamin D analogue alone.

9.4.2.1. Review protocol

A literature search was conducted for RCTs, prospective cohort studies or systematic reviews that addressed whether the risk of skin cancer is increased in people with psoriasis and whether there are subgroups of the psoriatic population who are at particularly high risk.

No time limit was placed on the literature search. The sample size was required to be sufficient to result in at least 10 cancer cases per covariate and studies were restricted to those with an average of at least 12 months follow-up since first treatment. Indirect populations were excluded but retrospective studies were included if no prospective data were available for a particular intervention that may be a risk factor for cancer.

The outcomes considered were:

• Melanoma
• Non melanoma skin cancer
  • Stratified in to squamous cell carcinoma and basal cell carcinoma if data were available

Subgroup analysis was considered for the following prognostic factors (in addition to the stated interventions that were considered to be potential risk factors):

• Skin type
• Concomitant or previous immunosuppressive treatments
• Duration of previous systemic treatment
• Cumulative exposure to previous systemic treatment or coal tar
• Previous exposure to ionising radiation
• Disease severity
• Previous skin cancer
• Age at first exposure
• Smoking
• Alcohol consumption
• Family history of skin cancer

Any interactions between the prognostic factors indicating whether there was additive risk were also extracted.

9.4.2.2. Included studies

Nineteen studies^{144,221,231,235,283,284,304,308,382–391,414} were found that addressed the question and were included in the review.

• No suitable RCT data were available owing to the limited duration of follow-up and insufficient sample sizes
• The majority of the studies reported on the same cohort followed-up at different time points^{221,235,283,284,382–391}
• Two studies^304,414 addressed the risk of skin cancers in people with psoriasis treated with biological therapies.
• One study³⁰⁸ compared the incidence of skin cancer in people with psoriasis treated with systemic treatments or coal tar and people with psoriasis not treated with these interventions. This allowed attribution of the increased risk to the interventions rather than any intrinsic risk associated with the psoriasis itself. The comparison of the incidence in a treated psoriasis cohort compared with a matched general population was also considered to be applicable. This provided indirect evidence from which inference can be made about the risk in people with psoriasis treated with systemic/phototherapy. However, the full treatment history remains unclear (and uncontrolled for). Because of this any difference in risk compared with the general population to the particular intervention being studied is difficult to determine. Note also that this comparison leads to risk of bias as the exposed and unexposed cohorts are selected from different sampling frames.
• No data (prospective or retrospective) were available for the biologics with follow-up of > 12 months.
• No data were available for the risk in children.

A summary of the characteristics of included studies is given in Table 112. Note that the number of patients given is the number of people in the psoriasis cohort, which was compared in the studies with the incidence rate of skin cancer among a matched general population sample (sample size not specified).

Table 112

Summary of characteristics of included studies.

Due to the design of the studies considered, GRADE could not be used to assess quality. Quality was assessed using a modified version of the Checklist for Prognostic Studies²⁷² (see Table 113). The quality rating was derived by assessing the risk of bias across 5 domains (selection bias; attrition bias; prognostic factor bias; outcome bias; and confounders and analysis bias) and although listed per study the adequacy of outcome measurement and controlling for confounders were considered per outcome; however, the rating was the same across outcomes unless otherwise stated.

Table 113

Study quality checklist.

For all studies the unexposed cohort was a general population sample and so would have included a proportion with psoriasis and potentially with exposure to the interventions beings assessed as risk factors (e.g., PUVA or ciclosporin). Also, in the Stern cohort 39 patients had a history of skin cancer before PUVA and this was not controlled for in all analyses. Across all studies there was high risk for outcome surveillance bias as there is likely to be more complete ascertainment of skin cancer cases among the exposed cohort who were actively followed-up and examined compared with the general population where diagnoses may be missed. None of the studies reported how missing data were handled or if imputation was used.

9.4.2.3. Confounding variables

In observational studies it is necessary to control or adjust for confounding variables, other than the stated intervention, that may also vary between the comparison groups and cause any observed differences. Therefore, in assessing study quality the adequacy of controlling for confounders was assessed.

Table 114 summarises which of the key confounders have been controlled for and by what method in each of the included studies. This information does not relate to the comparison of the risk of skin cancer in people with psoriasis versus the general population, which in all cases was based on an age-matched and sex- matched analysis, without controlling for other key confounders. The Stern cohort also matched for geographic location. The Hearne, Papp and van Lumig papers are excluded from Table 114 as they only provided data comparing observed rates with those expected in a matched general population sample.

Table 114

Adequacy of controlling for key confounders.

Pooling the results of observational studies is inappropriate owing to inconsistencies in design, comparison and potential confounders. All observational study data have been considered individually.

9.4.2.4. Summary statistics

A range of summary statistics are reported, some of which are specific to prognostic investigations. To aid interpretation, a summary of the definitions of these statistics is provided in Table 115. Estimates of the absolute risk are provided in Appendix Q.

Table 115

Defining summary statistics.

9.4.3.1. Risk vs. no PUVA exposure

One study³⁰⁸, primarily designed to assess the risk associated with ciclosporin use, also assessed the independent risk for any skin carcinoma associated with PUVA exposure compared with those who had no exposure to PUVA. Skin carcinoma included squamous cell carcinoma (SCC), basal cell carcinoma (BCC) or any skin malignancy (SCC, BCC or malignant melanoma (MM). In total, 47% of the cohort had received some treatment with PUVA (Table 116).

Table 116

Relative risk of skin cancer in PUVA patients compared with non-PUVA-treated patients.

9.4.3.2. Risk vs. general population

Studies from the PUVA follow-up cohort provided information on the relative risk of skin cancer among people with psoriasis who have been, or are currently being, treated with PUVA compared with an age-, sex- and geographic location-matched general population sample based on incidence data. The data were stratified into squamous cell carcinoma, basal cell carcinoma and malignant melanoma.

Evidence summary

All non-melanoma skin cancer

One study³⁹¹ reported the overall relative risk of non-melanoma skin cancer in the PUVA cohort compared with the matched population. Based on a method that only counted the first tumour of each type per person (effectively measuring time-to-first tumour), the observed incidence in the psoriasis cohort was 2.63-times that expected in the matched age-, sex- and geographic location-matched general population (Table 117).

Table 117

Relative risk of non-melanoma skin cancer in PUVA patients compared with the general population.

Squamous cell carcinoma

Six studies^382–387 reported the relative risk of squamous cell carcinoma in the PUVA cohort compared with the matched population (Table 118).

Table 118

Relative risk of SCC in PUVA patients compared with the general population.

When recording the annual incidence, by counting the first tumour of a given type observed that year, the observed incidence in the psoriasis cohort was 16.2-times that expected in 1984³⁸⁵ and 27.0-times times that expected in 1994³⁸⁴. The earlier study (1984) only recorded tumours occurring at least 22 months after first treatment, whereas the later study (1994) appeared to include tumours from all time-points after treatment. Both only included invasive tumours.

Based on a method that only counted the first tumour of each type per person (effectively measuring time-to-first tumour), the observed incidence in the psoriasis cohort was 9.3-times that expected in 1984³⁸⁵, 9.5-times in 1988³⁸⁶ and 11.9-times in 1994³⁸⁴ (Table 118). Additionally, calculation of the observed incidence starting from 10 years after first PUVA use demonstrated that the increased risk of developing SCC among PUVA-treated psoriasis patients persisted many years after PUVA treatment had been stopped in the majority of the cohort, with the relative risk being 17.6-times that expected in the period 1985–1998³⁸⁷.

Two of these six studies specifically reported the incidence of genital tumours in men treated with PUVA (Table 118). In the 1990 report³⁸², based on the total number of tumours observed, the incidence of invasive genital SCC in the psoriasis cohort was 95.7-times that expected.

In the second report in 2002³⁸³, when counting just the first tumour per person, the observed incidence of invasive genital SCC in the psoriasis cohort was 81.7-times that expected. The increased incidence again persisted after 1989 (the last date of surveillance for the 1990 report) at a level of 52.6-times that expected although use of PUVA had decreased and genital shielding in the cohort had increased. Similarly, the annual incidence of genital SCC observed in the psoriasis cohort in this study was 134.6-times that expected, and the increased incidence again persisted after 1989 at a level of 87.7-times that expected.

Basal cell carcinoma

Four studies^384–387 reported the relative risk of basal cell carcinoma in the PUVA cohort compared with the matched population (Table 119).

Table 119

Relative risk of BCC in PUVA patients compared with the general population.

When recording the annual incidence, by counting the first tumour of a given type observed that year, the observed incidence in the psoriasis cohort was 2.2-times that expected in 1984³⁸⁵ and 4.1- times times that expected in 1994³⁸⁴. The earlier study (1984) only recorded tumours occurring at least 22 months after first treatment, whereas the later study (1994) appeared to include tumours from all time-points after treatment. Both only included invasive tumours.

Based on a method that only counted the first tumour of each type per person (effectively measuring time-to-first tumour), the observed incidence in the psoriasis cohort was 1.7-times that expected in 1984³⁸⁵, 2.3-times in 1988³⁸⁶ and 2.5-times in 1994³⁸⁴ (Table 119). Additionally, calculation of the observed incidence from 10 years after first PUVA use demonstrated that the increased risk of developing BCC among PUVA-treated psoriasis patients persisted (and even increased) many years after PUVA treatment had been stopped in the majority of the cohort, with the relative risk of first BCC after 1985 being 4.1-times that expected in the period³⁸⁷.

Malignant melanoma

One study³⁸⁹ reported the overall risk of malignant melanoma in the PUVA cohort compared with the matched population. The observed annual incidence in the psoriasis cohort was 2.3-times that expected in the matched age-, sex- and geographic location-matched general population over the full follow-up period. A breakdown of the incidence into an early and a late follow-up period demonstrated that the incidence in the PUVA cohort increased after 1990 (Table 120).

Table 120

Relative risk of MM in PUVA patients compared with the general population.

9.4.3.3. Risk modification factors

Some studies from the PUVA follow-up cohort also gave information on additional prognostic factors that could modify the risk of skin cancer associated with PUVA treatment in people with psoriasis.

Evidence summary

A1. PUVA dose (stratified dose subgroups compared with lowest dose subgroup as reference strata)

Nine studies^{221,235,284,383–385,387,389,390} provided data (adjusted for at least age, sex and some relevant prior treatment exposure) regarding the relative risk of skin cancer in the PUVA treated cohort at various dose/exposure levels of PUVA compared with a reference strata, which was the lowest dose group, assumed to carry the lowest risk for skin cancer (see Appendix Q for definitions of high and low dose). A dose-risk relationship may suggest that PUVA can act as an independent carcinogen.

However, the statistics used to calculate the size of the effect varied (relative SMR^{384,385,389,390}, incidence rate ratio^{221,235,284,383}, odds ratio^387,391 or hazard ratio^283,284), making direct comparison between the studies difficult.

Squamous cell carcinoma

Seven studies^{221,235,284,383–385,387} provided data for the relative risk of SCC at different doses/levels of exposure to PUVA. Despite the different methods of analysis used, all of these studies showed a dose-response relationship, with increasing dose/levels of exposure showing incremental rises in the relative risk of skin cancer compared with the reference strata (Table 121).

Table 121

Adjusted relative risk estimates for SCC at different levels of exposure to PUVA.

Based on a method that only counted the first tumour of each type per person, compared with the low dose reference group the observed incidence was 5.7-times³⁸⁵ or 2.6-times³⁸⁴ higher in the medium dose group and 12.8-times³⁸⁵ or 5.9-times³⁸⁴ higher in the high dose group based on an adjusted standard morbidity ratio, which is linked to the ratio of observed-to-expected incidence. The reason for the reduction in risk between the time of the first and second studies is unclear, although only the later study³⁸⁴ adjusted for MTX exposure.

When comparing multiple dose strata the relative risk or the time-to-first tumour (based on a hazard ratio) clearly increased with increasing numbers of exposures, whether using person counts, population rates or total tumour counts^{221,235,284,387}.

One study³⁸⁷ showed that the odds of first cancer at least 10 years after first PUVA use increased with increasing cumulative exposure to PUVA during those 10 years (before 1985), while the levels of more recent PUVA exposure had a modest impact on tumour risk.

The risk of genital tumours was also increased at high compared with low PUVA dose, but this effect size was less pronounced than total SCC³⁸³.

Basal cell carcinoma

Five studies^{221,284,384,385,387} provided data for the relative risk of BCC at different doses/levels of exposure to PUVA. Similarly to the data for SCC, despite the different methods of analysis used, all of these studies showed a dose-response relationship, with increasing dose/levels of exposure showing incremental rises in the relative risk of skin cancer compared with the reference strata, although the effect size was lower than that for SCC (Table 122).

Table 122

Adjusted relative risk estimates for BCC at different levels of exposure to PUVA.

Based on a method that only counted the first tumour of each type per person, compared with the low dose reference group the observed incidence was 2-times lower³⁸⁵ or similar³⁸⁴ in the medium dose group and 2-times higher³⁸⁵ or 1.7-times higher³⁸⁴ in the high dose group based on an adjusted standard morbidity ratio, which is linked to the ratio of observed-to-expected incidence.

When comparing multiple dose strata the relative risk or time-to-first tumour (based on a hazard ratio) increased with increasing numbers of exposures, whether using person counts, population rates or total tumour counts. However, this increase was more modest than that seen with SCC.

One study³⁸⁷ showed that the odds of first cancer at least 10 years after first PUVA exposure increased with increasing cumulative exposure to PUVA during those 10 years.

Malignant melanoma

Two studies^389,390 provided data for the relative risk of MM at different levels/durations of exposure to PUVA. Again, an increase in risk was observed with high vs low numbers of PUVA treatments, although this effect was not statistically significant for either all melanoma or invasive melanomas. However, there was a significant effect of increasing time since first treatment for both all and invasive melanomas (Table 123).

Table 123

Adjusted relative risk estimates for MM (invasive and in situ) at different levels of exposure to PUVA.

A2. PUVA dose (stratified dose subgroups compared with the matched general population)

Seven studies^{382–387,389} provided data regarding the relative risk of skin cancer in the PUVA treated cohort at various dose/exposure levels of PUVA compared with the risk in an age-, sex- and geographic location-matched general population. These data were not adjusted for other confounders, including exposure to other psoriasis treatments.

Squamous cell carcinoma

Six studies^382–387 provided data for the relative risk of SCC at different doses/levels of exposure to PUVA compared with the general population. All of these studies again showed a dose-response relationship, with increasing dose/levels of exposure showing incremental rises in the relative risk of skin cancer compared with the general population; however, in most cases, even the lowest dose group had a significantly increased risk of SCC compared with the general population (Table 124).

Table 124

Relative risk of SCC in PUVA patients stratified by exposure level compared with the general population.

The risk of genital tumours was also increased at all PUVA dose levels compared with the general population, with increasing risk at higher dose levels, although the number observed in each subgroup were low, making the precision if the estimate poor³⁸³.

Basal cell carcinoma

Four studies^384–387 provided data for the relative risk of BCC at different doses/levels of exposure to PUVA compared with the general population. Again, all of these studies showed a dose-response relationship, with increasing dose/levels of exposure showing incremental rises in the relative risk of skin cancer compared with the general population; however, as with SCC, even the lowest dose group had a significantly increased risk of SCC compared with the general population based on population rates (Table 125).

Table 125

Relative risk of BCC in PUVA patients stratified by exposure level compared with the general population.

Melanoma

One study³⁸⁹ provided data for the relative risk of melanoma at different doses/levels of exposure to PUVA compared with the general population. This study only found a significantly higher rate of melanoma in the PUVA cohort compared with the general population among those with the higher level of exposure. Additionally, during the first 15 years of follow-up the risk in the low exposure group was lower than that expected in the general population and was also non-significantly higher than the general population in the high dose group (Table 126).

Table 126

Relative risk of melanoma in PUVA patients stratified by exposure level compared with the general population.

B. Skin type

Two studies^221,284 provided data regarding the additional skin cancer risk of fair skin (Fitzpatrick phototype I–II) in people with psoriasis who have been treated with PUVA (Table 127).

Table 127

Adjusted relative risk estimates for SCC and BCC (invasive) for people with different skin types.

Both studies demonstrated an increased risk of both SCC and BCC in those with fairer skin. However, the later study²²¹ showed a less pronounced effect size, which was not statistically significant for BCC. This difference may have been due to the additional covariates adjusted for in this analysis (immunosuppressive therapies, UVB and ciclosporin). Another difference in the analysis was that the lower relative risks were based on population rates and the higher risks were based on total tumour counts. The increased risk was lower for BCC than SCC.

One study³⁹¹ provided data regarding the relative risk of any skin carcinoma in the PUVA treated cohort for different skin types compared with the risk in an age-, sex- and geographic location-matched general population (Table 128). Note that these data were not adjusted for other confounders, including exposure to other psoriasis treatments.

Table 128

Relative risk of any non-melanoma skin cancer in PUVA patients stratified by skin type compared with the general population.

This study showed that there was only a significantly increased risk of skin carcinoma among skin types I–II and not III–IV, although there was still a strong trend towards increased risk in this group.

C. History of skin cancer

One study provided data regarding the additive risk of prior skin carcinoma at least 3 years before first retinoid use in people with psoriasis who have been treated with both PUVA and retinoids (Table 129).

Table 129

Adjusted relative risk estimates for SCC and BCC determined by prior non-melanoma skin cancer.

One study³⁹¹ provided data regarding the relative risk of any skin carcinoma in the PUVA treated cohort for those with and without prior non-melanoma skin cancer compared with the risk in an age-, sex- and geographic location-matched general population (Table 130). Note that these data were not adjusted for other confounders, including exposure to other psoriasis treatments.

Table 130

Relative risk of any non-melanoma skin cancer in PUVA patients with and without prior carcinoma compared with the general population.

This study showed that there was a significantly increased risk of skin carcinoma among both those with and without prior skin carcinoma, but that the risk was much greater for those with a history of skin carcinoma.

D. Use of other psoriasis treatments

Seven studies^{221,235,283,284,384,385,387} provided information on the additional risk attributable to other psoriasis treatments among those treated with PUVA. This was presented as the output from a multivariable analysis adjusted for level of exposure to PUVA (and not for PUVA use per se), meaning that the risk estimates do not demonstrate the independent risk of these interventions in isolation from PUVA treatment. The results are summarised in Table 131.

Table 131

Adjusted relative risk estimates for SCC and BCC based on exposure to systemic agents or tar in addition to PUVA.

Squamous cell carcinoma

One study²³⁵ showed that using CSA (n=28) in addition to PUVA significantly increased the risk of SCC, but the risk with high level of exposure to CSA was not significantly higher than that for low levels of exposure in another study²²¹.

High levels of exposure to MTX^{221,235,284,384} and UVB²²¹ also increased the risk of SCC among PUVA-treated individuals; although the odds of first SCC 10 years after first PUVA exposure were non-significantly higher for high vs low MTX exposure³⁸⁷.

The increased risk with tar and tar plus UVB use was not statistically significant^384,387 and prior exposure to ionising radiation only significantly increased the risk of SCC among those who had low exposure to tar³⁸⁵.

One study²⁸³ found that oral retinoid use significantly reduced the risk of SCC among PUVA-treated patients when comparing years of use (at least 26 weeks of retinoid treatment) with years of no use (<26 weeks of retinoid treatment) among a subgroup of the PUVA cohort who had been treated with retinoids (n=135). However, when examining the whole cohort, the risk reduction associated with years of high retinoid use was not statistically significant²²¹.

Basal cell carcinoma

The majority of the evidence suggested that there was no statistically significant increase in risk of BCC among the PUVA cohort linked to high levels of exposure to CSA, MTX, tar alone, tar plus UVB or ionising radiation. However, one study²⁸⁴ did find a significantly increased risk among those who had high levels of exposure to MTX compared with low exposure; although it should be noted that this study did not adjust for use of CSA. Additionally, the odds for first BCC at least 10 years after first PUVA exposure were significantly higher among those who had high exposure to tar and UVB or to ionising radiation³⁸⁷.

One study demonstrated a statistically significant increase in risk of BCC among those with high compared with low lifetime exposure to UVB²²¹.

E. Interactions among risk factors among the PUVA treated cohort

Five studies^{235,283,383,385,386} indicated whether multiple additional risk factors (as well as exposure to PUVA) interacted with each other to further increase risk of SCC or BCC (Table 132). This gives information about whether risk factors modify the effect of other risk factors.

Table 132

Interactions among risk factors for SCC and BCC among the PUVA-treated cohort.

• One study³⁸⁵ found an interaction between ionising radiation and tar for SCC.
• PUVA dose appeared to increase risk of SCC and BCC to a similar degree regardless of skin type, although skin types I–II are associated with a higher risk than types II–IV compared with the general population³⁸⁶.
• One study showed that use of CSA was only significantly associated with increased risk of SCC in patients who had high levels of exposure to PUVA²³⁵.
• When analysing only the subset of the PUVA cohort who had also received oral retinoids (n=135), one study found that high tar/UVB exposure, any ionising radiation exposure and high PUVA exposure all significantly increased the risk of both SCC and BCC²⁸³.
• Finally, one study³⁸³ showed that the risk of genital SCCs was increased by exposure to medium-or high-dose PUVA in combination with high dose topical tar/UVB compared with low dose exposure to PUVA and tar/UVB. However, there were very few events in each subgroup, making the precision of these effect estimates very low.

9.4.4.1. Risk vs. no/low exposure

One study³⁰⁸, which was primarily designed to assess the risk associated with ciclosporin use, also assessed the independent risk for skin malignancies associated with prior exposure to other psoriasis treatments compared with those who had no/low exposure to these treatments (Table 133). However, there were too few events to meaningfully analyse SCC and BCC separately and it is noteworthy that less than 50% of the cohort completed the full follow-up period. Also note that the duration of follow-up for exposures other that ciclosporin is unclear, but would have been longer than that for ciclosporin as they were administered prior to trial entry. However, 34% of the cohort received other systemic treatments for psoriasis during the follow-up period and these do not appear to be taken into account in the analysis.

Table 133

Adjusted relative risk estimates for skin cancer based on exposure to systemic agents.

9.4.4.2. Risk vs. general population

Two studies^144,308 provided information on the relative risk of skin cancer among people with psoriasis who have been, or are currently being, treated with CSA or NBUVB compared with an age-, sex- and geographic location-matched general population sample based on incidence data. Two studies^304,414 provided data on the relative risk of skin cancer among people with psoriasis who have been exposed to biologics compared with an age- and sex-matched general population sample based on incidence data. The data were stratified into squamous cell carcinoma, basal cell carcinoma and malignant melanoma.

CSA

Evidence summary

One study³⁰⁸ provided information about the risk of skin cancer among those treated with any level CSA compared with the risk in the general population, including the risk in high and low exposure groups (≤2 years vs >2 years treatment; Table 134). However, the observed numbers of BCC and MM were very low.

Table 134

Relative risk of skin cancer in CSA patients compared with the general population.

Evidence statements

In people with psoriasis treated with CSA:

• the risk of all skin cancer and the risk of SCC were both statistically significantly higher than that expected in the matched general population [1 study³⁰⁸; 1252 participants; very low quality evidence]
• the observed number of BCC and MM cases were low and no statistically significant difference in the risk of these types of skin cancer was found compared with that expected in the matched general population [1 study³⁰⁸; 1252 participants; very low quality evidence]
• the increased risk of SCC and all skin cancer was significant for those with both high and low levels of exposure to CSA [1 study³⁰⁸; 1252 participants; very low quality evidence]

NBUVB

Evidence summary

One retrospective study¹⁴⁴ provided information about the risk of skin cancer among those treated with NBUVB or NBUVB and PUVA compared with the risk in the general population (Table 135).

Table 135

Relative risk of skin cancer in NBUVB patients compared with the general population.

Evidence statements

In people with psoriasis treated with NBUVB only:

• There was no statistically significant difference in the risk of SCC, BCC or MM from that expected in the matched general population [1 study¹⁴⁴; 2130 participants; very low quality evidence]

In people with psoriasis treated with NBUVB and PUVA:

• The risk of BCC was statistically significantly higher than that expected in the matched general population [1 study¹⁴⁴; 2130 participants; very low quality evidence]
• There was no statistically significant difference in the risk of SCC or MM from that expected in the matched general population [1 study¹⁴⁴; 2130 participants; very low quality evidence]

Biological therapy

Evidence summary

One retrospective study of prospectively gathered data³⁰⁴ provided information about the risk of skin cancer among those treated with etanercept for up to 48 months compared with the risk in the general population (Table 136). However, general population reference data were only available from USA registries while the exposed group were from Canadian cohorts, so the exposed and unexposed cohorts were not match on geographic location, which will effect sun exposure and skin cancer rates. This confounding variable was not accounted for in the analysis. One prospective study⁴¹⁴ provided information about the risk of skin cancer among those treated with any biological therapy for psoriasis and followed-up for 5 years compared with the risk in the general population (Table 136). However, prior treatments were not controlled for and all of those who had an event had also been exposed to PUVA and most to ciclosporin. Additionally, the time to first tumour was shorter than a year in the majority of cases, indicating that the biological agent was not causative to the pathology.

Table 136

Relative risk of skin cancer in people treated with etanercept compared with the general population.

Evidence statements

In people with psoriasis treated with etanercept:

• There was no statistically significant difference in the risk of SCC or BCC from that expected in the general population matched for age and sex, but not geographic location [1 study³⁰⁴; 506 participants; very low quality evidence]
  • The effect estimate suggested an increase in risk for SCC compared with the rates in Minnesota, which may be a better match in terms of ambient UV exposure to the Canadian cohort than the Arizonan rates [1 study; 506 participants; very low quality evidence]³⁰⁴
• However, there was a statistically significantly higher risk for people with psoriasis exposed to biological therapies compared with the general population in another study [1 study; 173 participants; very low quality evidence]⁴¹⁴