Background: in a recent double-blind, placebo controlled RCT we demonstrated that selective inhibition of cyclo-oxygenase 2 (COX2) is an effective adjunctive strategy in treatment-resistant bipolar depression (TRBDD). To better clarify the mechanisms underlying TRBDD and treatment response, we conducted a retrospective exploratory analysis of the systemic inflammatory response index (SIRI = absolute neutrophils × absolute monocytes/absolute lymphocytes) in relation to other biomarkers and clinical outcomes after escitalopram (ESC), combined with the COX-2 inhibitor, celecoxib (CBX), versus placebo. Methods: Baseline measures of SIRI were compared between TRBDD and healthy controls (HC), and correlated with blood-based inflammatory cytokines, kynurenines, and growth factors. Post-treatment Hamilton Depression Rating Scale 17 (HAMD-17) total scores (clinical outcome) were modelled according to SIRI adjusting for demographics (including relevant interactions with SIRI), baseline depression, treatment arm, and treatment timepoint using multiple linear regression and robust linear mixed effects models. Results: Baseline SIRI did not distinguish TRBDD from HC groups. Baseline SIRI was significantly correlated with lower baseline MCP-1. The relationship between SIRI and HAMD-17 was significant at treatment week 8, in contrast to baseline. Finally, baseline SIRI predicted elevated post-treatment HAMD-17 scores, amongst patients with elevated depression scores at baseline. Significance: High pre-treatment SIRI may predict poorer depressive outcomes amongst TRBDD patients with baseline elevated depression.