medwireNews: Integration of clinical and protein biomarker data using a machine-learning approach may help to distinguish patients with psoriatic arthritis (PsA) from those with psoriasis, suggests research presented at the ACR Convergence 2020 virtual meeting.
“Although several candidate [biomarkers] have been demonstrated to be associated with PsA, individually their power to discriminate between PsA and [psoriasis without PsA] is poor,” said researcher Sara Rahmati (University Health Network, Toronto, Ontario, Canada).
Using serum samples and clinical data from 192 patients with PsA and 191 with psoriasis, Rahmati and team carried out a three-step process to identify, compare, and rank biomarker signatures derived from 16 protein markers and four clinical features. The resulting nine final signatures all involved nail psoriasis and the protein markers LEP, C-reactive protein, and SOST.
When these nine signatures were ranked based on their ability to discriminate patients with PsA from those with psoriasis, Rahmati noted that the two lowest-performing signatures were “those without SPP1 [osteopontin], highlighting it as a potential strong marker” for PsA.
“Each of the nine proposed alternative signatures can be used depending on availability of markers,” she said, adding that “the next step of this work involves testing the proposed signatures on independent data.”
medwireNews is an independent medical news service provided by Springer Healthcare Ltd. © 2020 Springer Healthcare Ltd, part of the Springer Nature Group
ACR Convergence virtual meeting; 5–9 November 2020