Poster Session I: Myeloma and other monoclonal gammopathies - Biology & translational research
In multiple myeloma (MM), several factors impair anti-tumour immunity, providing a rationale for immunotherapeutic agents in treatment protocols. Selection of optimal agents requires a systematic examination of suppressive factors operative in specific therapeutic situations.
To map T cell subsets and immune checkpoints (IC) in the bone marrow (BM) of NDMM, and explore associations with clinical outcomes of first line treatment with novel agents.
Pre-treatment BM was obtained from 78 NDMM patients, and 21 healthy donors (HD). BM mononuclear cells were stained for surface (CD3,CD4,CD8,PD-1,LAG-3,CD25) and intracellular markers (GzmB,CTLA-4,TNF-a,FoxP3). Data were acquired on a BD LSR Fortessa and analysed using FlowJo. Progression free survival (PFS) was defined from start of therapy to first progression or death, and variables assessed using Cox regression model. CDNA libraries were prepared from RNA using SMART-Seq v4 and sequenced on HiSeq 3000 (Illumina).
Median age was 59yrs (35-86), 64% were male. All patients commenced active treatment; 92% with proteasome inhibitors, 8% thalidomide or lenalidomide, 31.3% had autologous stem cell transplant (ASCT).
MM BM contained higher frequencies of Treg (CD4+FoxP3+) compared to HD (0.51% live MNCs, vs 0.11%; p < 0.0001, 3.33% CD4+ vs 1.47%; p = 0.001), while CD4 and CD8 frequencies were similar. CD4effector (CD4FoxP3-, CD4eff):Treg ratio was lower in MM patients (22.2 vs 79.7; p = 0.0001); as was CD8:Treg ratio (p = 0.0003). Treg in MM patients expressed more CD25(% positive and MFI) compared to HD (p = 0.01), indicative of activation. Frequency of PD-1 and LAG-3 was higher on CD4eff and CD8 T cells in MM(p = 0.03, p = 0.004, and p = 0.002 for both, respectively cf. HD). Frequency of Treg correlated with frequency of PD-1+ on CD4eff and CD8 T cells (p < 0.01 for both).
Responses to treatment were ORR 87%, CR/VGPR, 53.8%. With median follow up of 23 months (1-43), median PFS was not reached(NR). Higher frequency of Treg (>median %CD4) was associated with shorter PFS (median 23mo vs NR, p = 0.02), as was low CD4eff:Treg ratio (≤median; 25mo vs NR, p = 0.004), but not CD8:Treg ratio. Patients with high PD-1+ CD4eff(>median; %positive, CD4-PD-1Hi) had shorter PFS compared to CD4-PD-1Lo (23mo vs NR, p = 0.008), but PD-1+ CD8 did not influence PFS. In multivariate analysis, CD4eff:Treg ratio was independently associated with PFS (HR: 4.1; 95%CI 1.5-16.02; p = 0.043), as were PD1+ CD4eff, and depth of response.
We examined the function of CD4eff in the 2 patient groups, CD4-PD-1Hi and CD4-PD-1Lo. The PD-1+ fraction of CD4eff from CD4-PD-1Hi patients co-expressed CTLA-4 and LAG-3 more frequently compared to CD4-PD-1Lo patients (p = 0.03, p = 0.02). Upon stimulation with anti-CD3 and anti-CD28, significantly more CD4eff from CD4-PD-1Lo patients produced TNF-a compared to CD4-PD-1Hi patients (p = 0.002). Finally, there was enrichment of previously described transcriptional profiles of dysfunction in CD4+ and CD8+ T cells comparing CD4-PD-1Hi(n = 5) and CD4-PD-1Lo (n = 5).
The BM of ND MM patients contains higher frequency of activated CD25-hi Treg that associate with accumulation of dysfunctional CD4 effectors and shorter PFS in ND MM. CD4, and not CD8, effectors appear to correlate with clinical outcomes, hence may be more important for anti-tumour response. Our study provides a rationale for exploring anti-Treg and PD-1 strategies to restore T cell function in NDMM patients.