The PRP skin transcriptome differs from other immune-mediated inflammatory diseases (IMIDs)
To differentiate PRP from other related IMIDs, we performed a targeted transcriptomic assessment of lesional skin from patients with adult-type PRP (active and post-PRP), psoriasis, AD, and healthy controls (HC) (Fig. 1A, Table S1). Analysis of differentially expressed genes (DEG) revealed a distinct inflammatory pattern associated to PRP (Fig. 1B, Figure S1-2) with significantly increased expression of genes encoding the cytokines, IL1A, IL1B, IL23A, IL17A/F, TNF, chemokines CCL18, CCL20, and defensin/alarmins S100A8/9, and DEFB4A (Fig. 1C). The transcriptional signature for PRP included CCL20, consistent with previous reports6, and IL1B as the predominant DEGs. Their gene expression correlated significantly with clinical severity of PRP measured by PASI (p = 0.0018 and > 0.0001 for IL1B and CCL20, respectively), indicating a pathogenic relevance (Fig. 1D). In summary, PRP embodies a unique transcriptional profile compared with related IMIDs (Fig. 1E), thereby highlighting its distinct pathophysiology.
IL-1β is a key regulator of PRP and linked to keratinization
To confirm our findings, we then constructed a regulatory network of PRP, further dissecting its molecular signature. We analyzed publicly available data of PRP patients with paired lesional vs. non-lesional skin biopsies5. Ingenuity Pathway Analysis (IPA) identified IL1B as the second strongest predictor for upstream regulation of the PRP transcriptomic signature (Fig. 1F). A subsequent gene-gene co-expression analysis identified ten co-expressed gene modules (Fig. 1G, S3A) where IL1B-containing module 1 strikingly overlapped with the PRP signature and demonstrated the highest percentage of significant PRP-DEGs amongst all modules (Figure S3B). Furthermore, module 1 genes exhibited a notable log-fold increase in all genes that were shared with PRP DEGs (Figure S3C), and their enrichment/predominance in PRP was confirmed by gene set enrichment analysis (GSEA) (Fig. 1G, S3D).
Gene-Set over representation analysis of the module 1 exposed connections with immune and inflammatory signals, including Interleukin-1 Receptor Binding, and with keratinocyte differentiation (Fig. 1H, S3E). Module 1 was subsequently dissected into 8 functional clusters (Figure S3F, S4A) using STRING functional network analysis (protein-protein interaction, PPI). The two primary subclusters relate to keratinization (orange) and IL1/inflammation (blue, Fig. 1I). The IL1/inflammation cluster includes 8 genes from the IL-1 family: IL1A, IL1B, IL36A, IL36G, IL36RN, IL1RN, IL1F10, and IL36B, as well as CCL20, IL23A, CARD14, and S100A8. Both subclusters exhibit strong interaction with each other. Enrichment of each module confirmed these annotations (over representation analysis, Figure S4B-C), which contained a substantial number of genes overexpressed in PRP (Figure S4D-E). This analysis provides strong evidence that the immune pathways involving IL-1 are closely intertwined with keratinization mechanisms in PRP. These findings advocate the application of IL-1-targeting biologics in the management of PRP.
Treatment of PRP with IL-1 inhibitors rapidly ameliorates disease
To validate the potential of IL-1β blockade for PRP treatment, we treated three therapy-refractory PRP patients (two males, ages 53 and 59 and one female, age 53) with the IL-1 receptor antagonist anakinra (Fig. 2A-B, Figure S5A-D, Table S2).
Patient 1 responded well initially to a standard anakinra dose of 100 mg/day. Skin severity improved by 50% PASI (PASI50) at week 2 in patients 1 and 2 and at week 3 in patient 3 (Fig. 2C). When the disease worsened at week 6 in patient 1, the dose was doubled to 200 mg/day. The same dose regimen was administered to patients 2 and 3. PASI75 was reached by week 8 in patients 1 (PASI 11.4 to 2.6, ΔPASI 77%) and 2 (PASI 21.4 to 5.7, ΔPASI 73%) and by week 12 in patient 3 (PASI 34.2 to 9.4, ΔPASI 73%). All patients tolerated the treatment well (Supplementary Appendix) and reduced the concomitant use of topical steroids.
After week 12, patients 1 and 2 stopped treatment due to lack of cost coverage through health insurance. Patient 2 continued to improve further and reached complete resolution without any other treatment. Patient 1 was subsequently switched to biologics targeting TNF, IL-23, or IL-17, none of which adequately controlled the disease (Table S2). However, when switched to canakinumab (anti-IL-1β), the patient improved significantly within 8 weeks (ΔPASI 85%) (Fig. 2D). Patient 3 continued treatment with anakinra. From baseline to week 8, itch severity significantly reduced in patients 1–2, whereas patient 3 showed no symptoms (Supplementary Appendix).
The clinical improvement in all patients was confirmed histologically (Fig. 2E-F, S5E-F) including normalization of IL-1β expression at week 8, reaching similar levels to those in non-lesional tissue (Fig. 2G-H, S5G-H).
We report here a rapid and successful response to the IL-1 antagonists anakinra and canakinumab in 3 PRP patients, underscoring the importance of IL-1β in PRP disease pathogenesis.
Anakinra reverses the PRP transcriptional signature in patients with PRP
We sought to dissect the mechanisms involved in the development of PRP by exploring the transcriptional signature following treatment and confirm IL-1β as a potential target. All top 5 positive upstream regulators in PRP (Fig. 1F) were significantly downregulated upon IL-1β-targeting treatment (Fig. 3A). Overall, predicted activation scores of IPA disease, pathways, and upstream regulators annotations exhibited significant negative correlation between PRP and the treatment signal (Figs. 3B, S6A-B), which is defined as the DEGs from analysis of lesional samples after vs. before treatment. Positive DEGs in PRP showed significant negative enrichment in the treatment signal (Fig. 3C). IPA mechanistic network analysis further showed that gene regulation arising from IL1B could itself activate the regulation of TNF, IFNG, IL6 and TGFB1, with downstream activation of NFkB and STAT (Fig. 3D), which was reversed with treatment (Fig. 3E). The interaction of these cytokines is also supported by the IPA summary network of PRP and treatment DE analysis, showing a central role of IL-1β (Figure S6C-D). Furthermore, GSEA analysis of selected pathways of interest, such as TNF-alpha Signaling via NF-kB, showed positive enrichment with PRP and negative enrichment following IL-1 antagonist treatment (Fig. 3F). In summary, these results highlight a reversion of PRP transcriptional signals and NF-kB inhibition upon anakinra treatment (Fig. 3G).
IL-1β drives PRP-specific signature in keratinocytes in vitro
Next, we assessed the impact of IL-1β signaling and activation in keratinocytes. DE analysis of IL-1β-stimulated keratinocytes showed a significant upregulation of genes encoding CCL20, IL-1β, TNF, IL-23A, IL-36γ, NFKB1, as well as enrichment in pathways of inflammation, IL-1, and NF-kB (Figure S7A-C). DEGs in IL-1β stimulated keratinocytes were enriched in PRP which showed similar activated pathways. Many significantly overexpressed genes overlapped between IL-1β-stimulated keratinocytes and lesional PRP, with a strong enrichment of NFkB-related signals (Figure S7D-G). The in vitro data overlapping with the PRP molecular signature demonstrates the central role of IL-1β and hints at the involvement of previously described downstream players such as such as CCL20, TNF, IL-23A, IL-36γ, DEFB4A, IL-17C, CXCL8 and NOD2.
The strongest upregulated chemokine in the transcriptome analysis was CCL20 (Fig. 1B). TNF and IL-1β were also the strongest inducers of CCL20 in keratinocytes in vitro (Figure S8A-B) and inhibition of IL-1 signaling with anakinra decreased CCL20 expression (Figure S5I), suggesting a major role of an IL-1β-CCL20 axis in disease pathogenesis.
Additionally, our analysis identifies NOD2 and CARD14 as the most prominent CARD proteins in active and lesional PRP. These proteins signal via NF-κB and display a strong correlation with IL1B (Figure S8; detailed in Supplementary Material), which underscores the role of CARD14 and NOD2 in keratinocyte interactions within PRP-affected skin. Moreover, Caspase-1 expression is diminished in PRP lesional skin during anakinra treatment (Figure S5J; detailed in Supplementary Material).
Ultimately, the overlap of the gene signature in active PRP with IL-1β-stimulated keratinocytes suggests a crucial role of keratinocytes and IL-1β in triggering major inflammatory reactions, with NF-kB being a key factor involved, all of which is central to PRP pathogenesis.