Bone Marrow T-Cell Transcript Content Predicts Survival in Relapsed and Refractory Multiple Myeloma and Accounts for the Apparent Prognostic Value of T-Cell Exhaustion Signatures
Ali Turunç, Birol GüvençIntroduction/Objective:
The Bone Marrow (BM) immune compartment is increasingly invoked to explain heterogeneous outcomes in Multiple Myeloma (MM), and literature- derived T-cell exhaustion signatures are widely applied to bulk transcriptomic cohorts as proxies for immune dysfunction. We asked whether BM T-cell transcript content is prognostic in its own right, in which disease setting, and whether the exhaustion signatures in current use measure T-cell state or simply T-cell quantity.
Methods:
Raw expression matrices and clinical annotation were analysed for three MM cohorts with time-to-event data: GSE9782 (APEX trial, relapsed/refractory, randomised to bortezomib or high-dose dexamethasone; n = 264, 156 deaths), GSE24080 (UAMS Total Therapy 2/3, newly diagnosed; n = 559, 172 deaths) and GSE4204 (n = 537, 93 deaths); 1,360 patients and 420 deaths in total. BM T-cell transcript content was quantified as the mean z-score of twelve pan- T-cell transcripts. A 20-gene exhaustion composite was computed both as an unweighted mean, as commonly reported, and with the memory/effector arm reverse-coded as its definition requires. Overall Survival (OS) was analysed via Cox regression with proportional-hazards testing, Harrell’s C-index, and likelihood ratio tests against International Staging System (ISS) stage.
Results:
In relapsed/refractory disease, BM T-cell transcript content stratified OS: median 23.8, 21.7 and 11.8 months across high, intermediate and low tertiles (high versus low HR 0.47, 95% CI 0.32–0.70, P = 2.0 × 10-4), with death rates of 47%, 56% and 75%. The effect was present in both randomised arms (bortezomib HR 0.69, 95% CI 0.57–0.83, P < 0.001; dexamethasone HR 0.64, 95% CI 0.42–0.98, P = 0.041; treatment interaction P = 0.67) and added to ISS stage (likelihood ratio P = 0.015; C-index 0.604 to 0.624). Higher transcript content was associated with a response to bortezomib (odds ratio of 1.53 per SD; 95% CI, 1.11–2.11; P = 0.009). In newly diagnosed disease, the association was weaker and did not add to ISS (P = 0.69). Pooled across cohorts, T-cell transcript content predicted OS (HR 0.67, 95% CI 0.56–0.81, P = 2.5 × 10-5). The exhaustion composite tracked T-cell transcript content at ρ = +0.52 to +0.72; reverse-coding its memory/effector arm removed the correlation and abolished the prognostic signal entirely (HR 0.90, P = 0.30; C-index 0.507). VCAM1 (HR 0.75, P = 3.8 × 10-8) and CXCR6 (HR 0.80, P = 1.0 × 10-6) were the most robust individual transcripts.
Discussion:
BM immune transcript content carries independent prognostic information in relapsed and refractory MM, where host immune reserve appears more consequential than in the newly diagnosed setting. Published exhaustion composites index this same quantity signal rather than T-cell dysfunction, which explains their prognostic behaviour and has consequences for any analysis built on them, including connectivity-based drug repositioning.
Conclusion:
BM T-cell transcript content is an independent prognostic factor in relapsed and refractory MM and adds to ISS staging. Exhaustion signatures applied to bulk marrow should be interpreted as infiltration indices unless directional scoring and abundance adjustment demonstrate otherwise.