Mutation burden of narrowband ultraviolet B phototherapy (NB-UVB) in human skin: relevance to NB-UVB lifetime exposures and skin cancer surveillance

Authors

Joanna C Fowler

Roshan K Sood

George Coltart

Chester Lai

Noeline Nadarajah

John W Holloway

Matthew J J Rose-Zerilli

Brian Diffey

Philip H Jones

Eugene Healy

Doi

10.1093/bjd/ljaf173

PMID: 26950094 · DOI: 10.1093/bjd/ljaf173 · Journal: British Journal of Dermatology (2025)

TL;DR

Fowler et al. used nanorate sequencing (NanoSeq) on epidermal DNA from paired pre- and post-treatment skin biopsies of 16 patients with psoriasis to quantify the somatic mutation burden induced by a single course of narrowband ultraviolet B (NB-UVB) phototherapy. Median mutation-burden increase was 0.55 substitutions/Mb in infrequently sun-exposed buttock skin (n=14) and 0.89 substitutions/Mb in frequently sun-exposed forearm skin (n=10), driven predominantly by UVR-associated COSMIC signatures SBS7a and SBS7b plus a DBS1 (CC>TT) increase. Modelling Δ–mutation burden/dose against minimal erythema dose (MED), the authors estimate that for an MED of 2 SED, skin-cancer surveillance should begin after 422, 165, or 58 NB-UVB exposures for cautious, typical, and enthusiastic sun-exposure behaviour respectively — far earlier than the 500-exposure threshold currently recommended by British Association of Dermatologists guidelines.

Cohort & data

  • 21 patients with psoriasis recruited at a single UK dermatology centre; 16 provided paired pre/post NB-UVB biopsies (6 women, 10 men; median age 33, range 20–73). UK census ethnicity: 13 White British, 1 White Irish, 1 Asian Indian, 1 Asian other.
  • 6-mm punch biopsies of nonlesional skin obtained from buttock (infrequently sun-exposed, n=14 paired) and dorsal forearm (frequently sun-exposed, n=10 paired). Venous blood used as germline reference.
  • NB-UVB delivered in upright cabinets at 311 nm, dosed against MED. Patients received 14–35 exposures (median 28); total course dose 84–756 SED (median 219). MED 1.7–5.1 SED (median 2.9).
  • Epidermal DNA isolated by microdissection and sequenced with NanoSeq per Abascal et al. 2021 (error rate <5/billion bp); roughly one-third of the genome covered by NanoSeq libraries — sufficient for mutation-burden / signature analysis but not for driver-gene calling.
  • Ancillary cell-culture experiments treated HaCaT keratinocytes with 8-methoxypsoralen (methoxsalen) plus UVA, then re-sequenced with NanoSeq to validate a PUVA-associated signature.
  • Sequencing data deposited in the European Genome-Phenome Archive (whole-genome EGAD0000101525; NanoSeq EGAD00001015249) and HaCaT data in ENA (ERP156525).

Key findings

  • Pre-treatment baseline. Buttock mutation burden (median 0.52 substitutions/Mb, n=14) was significantly lower than forearm (3.13 substitutions/Mb, n=10; P=0.001, Kolmogorov–Smirnov). Within-patient buttock and forearm burdens were positively correlated (Pearson r=0.867, P=0.003).
  • Prior phototherapy elevates baseline. Buttock burden was higher in patients with prior phototherapy (range 0.195–2.740, n=10) than in phototherapy-naive patients (range 0.072–0.207, n=4; P=0.01, Kolmogorov–Smirnov), indicating mutations from earlier courses persist long-term.
  • NB-UVB increase. A single course raised mutation burden in buttock (median post-treatment 1.09 substitutions/Mb; P<0.001, Wilcoxon signed-rank) and forearm (median 3.77 substitutions/Mb; P=0.01) skin. Fold change ranged 1.16–10.50 in buttock and 0.93–2.33 in forearm.
  • Dose-normalised burden (Δ–mutation burden/dose). 0.0002–0.0066 substitutions/Mb/SED in buttock (median 0.0021) and 0.0000–0.0069 in forearm (median 0.0034); no significant difference between sites (P=0.55, Mann–Whitney U). The two lowest buttock values were both Asian patients (median 0.0003 substitutions/Mb/SED).
  • Signature attribution. UVR-associated SBS7a/SBS7b signatures appeared in all post-NB-UVB samples and most pre-treatment samples; the post-NB-UVB increase in buttock skin was predominantly attributable to SBS7a–d (vs. ageing signatures SBS1/SBS5; P<0.001, Mann–Whitney U), again with Asian patients an exception.
  • DBS1 corroboration. The double base substitution signature DBS1 (primarily CC>TT) increased significantly in post-NB-UVB buttock skin (P=0.001, Wilcoxon signed-rank). Δ-DBS1 correlated tightly with Δ-SBS7 (Pearson r=0.897, P<0.001, n=14).
  • PUVA signature recovery. Patients 8 and 15 carried a mutational signature matching the PUVA signature reported by Olafsson et al. 2023; the lab reproduced it in HaCaT keratinocytes treated with 8-MOP+UVA, deriving SBS288A and SBS288B (the latter combining APOBEC SBS2/SBS13 features with T>A/T>C/T>G changes consistent with PUVA).
  • Genetic background. Polygenic risk scores for “very tanned” propensity (PGS001246, 4130 variants) and “mildly/occasionally tanned” (PGS001244, 958 variants) tracked with low Δ–mutation burden/dose in the two Asian participants; PGS001247 (“never tan, only burn”, 3159 variants) was also evaluated. Δ–mutation burden/dose was negatively associated with MED (P=0.007, Spearman; combined buttock+forearm).
  • No effect of timing. Δ–mutation burden/dose did not associate with the days between final NB-UVB exposure and second biopsy (range 7–105 days; P=0.23 buttock, P=0.13 forearm), implying induced mutations persist beyond the epidermal turnover time.
  • Surveillance modelling. Fitting Δ–mutation burden/dose as S_MED = A·exp(−b·MED²) (A=0.0042, b=0.12±0.03), and assuming linear accumulation against the cSCC mean burden of 50 substitutions/Mb (Inman et al. 2018) and BCC mean of 65 substitutions/Mb (Bonilla et al. 2016), the authors compute lifetime exposure ceilings (Table 3). For MED=0.35 J cm⁻² (2 SED): 422 (95% CI 365–487) cautious, 165 (137–194) typical, 58 (49–67) enthusiastic NB-UVB exposures.

Genes & alterations

  • The authors explicitly state that NanoSeq, with ~⅓-genome coverage on each library, “is not appropriate to look for mutations in specific genes or driver selection” — there are no gene-level alteration claims in this paper. Findings are framed entirely at the genome-wide mutation-burden and mutational-signature level.

Clinical implications

  • Surveillance threshold should drop. Current BAD/BPG guidelines trigger skin-cancer surveillance after 500 whole-body NB-UVB treatments (Goulden et al. 2022). This paper argues surveillance should start much earlier — at 422 (cautious), 165 (typical), or 58 (enthusiastic) exposures for MED=2 SED, and progressively lower for more sun-sensitive (lower-MED) patients. Tables S3/S4 provide analogous numbers for 60- and 40-year sun-exposure horizons.
  • Personalised by MED + sun behaviour. The recommended action is to use each patient’s MED (already measured at NB-UVB initiation) together with self-reported sun-exposure habit (cautious / typical / enthusiastic, with assumed annual solar exposures of 10 / 30 / 90 SED to trunk and limbs) to set the per-patient surveillance threshold from Table 3.
  • Prior phototherapy adds persistent risk. Because pre-treatment buttock burden was significantly higher in patients who had previously received phototherapy and persists across decades (PUVA signature recovered in patient 8 from 2016 PUVA, and in patient 15 from likely psoralen-containing suntan products dating to the 1970s), cumulative-exposure counting must include all prior NB-UVB / PUVA courses, not just the current one.
  • Genetic background modifies risk. Polygenic tan-propensity scores correlated with lower Δ–mutation burden/dose, suggesting future surveillance models could incorporate PRS in addition to MED.

Limitations & open questions

  • Small, single-centre cohort. Only 16 patients with paired biopsies, all from one UK dermatology centre; only 2 Asian and no Black participants, limiting generalisability across skin phototypes.
  • Variability not fully explained. Δ–mutation burden/dose varied substantially for a given MED, so per-patient surveillance thresholds derived from Table 3 should be treated as guides rather than precise cut-offs.
  • No driver-gene resolution. NanoSeq’s single-molecule approach is unbiased by clonal expansion but cannot identify which mutations strike known driver loci or which cells are stem cells; the link from genome-wide burden to clonally expanded driver mutations is assumed, not measured here.
  • Linear extrapolation. Surveillance estimates assume a linear relationship between cumulative UVR mutation burden and skin-cancer risk, motivated by epidemiological observations (Armstrong & Kricker 2001; Rivas et al. 2012) but not directly demonstrated for NB-UVB.
  • Face is excluded. Models target trunk and limbs and assume the face is shielded by a UV-opaque visor; risk estimates do not cover the face, scalp, or other commonly affected sites.
  • Open question for cross-paper synthesis. How does the NB-UVB-induced SBS7 burden in normal skin compare quantitatively with SBS7 burden in matched cSCC/BCC tumours from the same anatomic sites — does the per-exposure increment scale linearly into the tumour regime, or saturate?

Citations from this paper used in the wiki

  • “Median increase in mutation burden was 0.55 substitutions per Mb in infrequently sun-exposed (buttock; n = 14) skin and 0.89 substitutions per Mb in frequently sun-exposed (forearm; n = 10) skin (P < 0.001).” — Abstract.
  • “Change in mutation burden due to NB-UVB ranged from 1.16- to 10.50-fold in buttock skin and from 0.93- to 2.33-fold in forearm skin.” — Abstract / Table 2.
  • “for patients with a MED equal to 2 standard erythemal doses (SEDs), skin cancer surveillance should be offered at 422, 165 and 58 NB-UVB exposures for those receiving low, typical and high levels of sun exposure, respectively.” — Abstract / Table 3.
  • “approximately one-third of the genome is covered by NanoSeq libraries, so although this technique provides accurate information on mutation burden, it is not appropriate to look for mutations in specific genes or driver selection.” — Results, p.721.
  • “the average mutation burden of cSCC and basal cell carcinoma (BCC) from exome sequencing data are 50 and 65 substitutions per Mb, respectively.” — Methods / Discussion.
  • “Δ–mutation burden/dose for buttock skin ranged from 0.0002 to 0.0066 (median 0.0021) substitutions per Mb per SED … and from 0.0000 to 0.0069 (median 0.0034) substitutions per Mb per SED in forearm skin.” — Results.
  • “In patients 8 and 15, we noted a mutational signature similar to that reportedly associated with historical psoralen + UVA (PUVA) treatment … De novo signature analysis revealed two signatures, SBS288A and SBS288B, the latter of which was associated with combined 8-MOP and UVA.” — Results, p.723–724.
  • “British Association of Dermatologists and British Photodermatology Group guidelines for narrowband ultraviolet B phototherapy recommend ‘offer[ing] skin cancer surveillance at appropriate intervals to people identified as having received more than 500 whole-body NB-UVB treatments’.” — Discussion, p.726.

This page was processed by crosslinker on 2026-05-14.