An Exploration of Blood Marker×Environment Interaction Effects on Pain Severity and Interference Scores in People With Acute Musculoskeletal Trauma.

Objectives: Explore the moderating effects of psychological or social variables on associations between biomarkers of inflammation/stress and clinical reports of pain.

Methods: This is a cross-sectional exploratory study. Data were drawn from the Systematic Merging of Biology, Mental Health and Environment (SYMBIOME) longitudinal study (clinicaltrials.gov ID no. NCT02711085). Eligible participants were adults who presented to an Urgent Care Centre in Ontario, Canada within 3 weeks of a noncatastrophic musculoskeletal trauma (no surgery or hospitalization). A questionnaire package was given that included the Brief Pain Inventory (capturing pain severity and pain interference) and relevant person-level variables. Blood samples were also drawn for serum analysis of 8 target biomarkers (brain-derived neurotrophic factor, transforming growth factor beta 1 [TGF-β1], c-reactive protein, tumor necrosis factor-α, interleukin [IL]-1β, IL-6, IL-10, and cortisol).

Results: Employment before trauma (employed for pay/not employed for pay) fully moderated the association between tumor necrosis factor-α and pain severity (∆R2=4.4%). Pre-existing psychopathology (yes/no) fully moderated the association between TGF-β1 and pain severity (∆R2=8.0%). Sex (male/female) fully moderated the association between c-reactive protein and pain severity (∆R2=6.3%). A pre-existing pain condition (yes/no) was significantly associated with worse pain interference (R2=7.2%), and partially moderated the effect of IL-1β on pain interference (∆R2=6.9%). Higher peritraumatic life stress significantly explained 8.9% of variance in pain interference alone, and partially moderated the effect of TGF-β1 on interference (∆R2=4.4%).

Discussion: Simple bivariate associations between blood-based markers and clinical symptoms are unlikely to reveal meaningful relationships. However, when stratified by existing person-level or "metadata" variables, an association may exist for at least 1 clinically relevant subgroup.