Abstract
Immunotherapeutic approaches to treat patients with myeloid cancers show considerable promise. Delineation of the immunologic milieu in these patients is necessary to develop rationally-designed strategies. We have previously reported a phase I trial in which intermediate/high-risk(I/HR) MDS patients were vaccinated against NY-ESO-1 tumor antigen (anti-DEC-205-NY-ESO-1 fusion protein with poly-ICLC adjuvant; CDX-1401) in combination with decitabine (20 mg/m2/day x 5 days x 4 cycles). Vaccination of MDS patients induced NY-ESO-1-specific immune responses but there was significant intra-patient variability in the quantity and quality of these responses. CDX-1401 targets multiple populations of antigen presenting cells (APCs) including CD141Hi conventional dendritic cells (cDCs). Compared to other APCs, CD141Hi cDCs express higher levels of DEC-205, an antigen uptake receptor, and TLR3, the poly-ICLC receptor. These cells regulate anti-tumor immune responses in solid tumor models and are necessary for the efficacy of checkpoint inhibitors. Thus, we hypothesized that the observed variations in immune response to vaccine resulted from quantitative and/or qualitative differences in CD141Hi cDCs in MDS. Overall, in MDS patients compared with healthy donors, there was a 9-fold decrease in the frequency of CD141Hi cDCs (0.001 ± 0.002% versus 0.009 ± 0.006% respectively; p < 0.001). In contrast with this finding, frequencies of the related CD1c+ cDC population were not different (0.18 ± 0.11% versus 0.27 ± 0.61% respectively; p = 0.12). Strikingly, MDS patients with the highest frequency of CD141Hi cDCs demonstrated a greater functional response to vaccination (both humoral and T-cell mediated). Expression of DEC-205 in cDCs from MDS patients was not lower than in those from healthy donors, suggesting that response to vaccination was not mediated by differential expression of the vaccine target. To further analyze gene expression in cDCs, we performed RNA-seq analysis of DCs isolated from healthy donors (n = 3) and a vaccine responsive MDS patient from the study. Principal component analysis showed that gene expression profiles were similar in CD141Hi and CD1c+ cDCs from MDS and normal patients, and further these profiles were distinct from plasmacytoid DCs and CD34+ progenitors. This suggests that for MDS patients with detectable CD141Hi populations, cDC function is intact. We expanded our observations to a larger cohort of untreated HR MDS patients and confirmed a selective decrease in the baseline number of bone marrow CD141Hi cDCs (n = 17; 0.002% ± 0.006%) compared to healthy donors (n = 5; 0.14 ± 0.01%; p < 0.001). We further characterized cDC progenitors in a subset of these patients using flow cytometry (n = 8). MDS patients demonstrated a population of bipotent cDC progenitors (early pre-DC) with an expected ratio of differentiation to committed CD1c+ cDC progenitors and terminally differentiated CD1c+ cDCs (p < 0.01). In contrast, the expected relationships between early pre-DCs and CD141Hi cDC committed progenitors or terminally differentiated CD141Hi cDCs was lost (p= 0.33 and 0.79 respectively). These data suggest a defect in the differentiation program for CD141Hi cDCs in MDS patients. We then hypothesized that decitabine treatment might promote a more normal cDC differentiation pattern. Using an in vitro model of cDC differentiation, we found that decitabine treatment increased differentiation of MDS CD34+ progenitors to CD1c+ cDCs by an average of 1.72 ± 0.96 fold (n = 3). However, we were unable to detect a similar effect on CD141Hi cDC differentiation. Since in vitro models do not reliably recapitulate the human environment, we analyzed serial bone marrow samples isolated from patients treated on study with decitabine and vaccination (n = 4). In these assays, 2/4 and 3/4 patients showed increased numbers of both CD141Hi and CD1c+ cDC at the end of study compared to pre-treatment. Together, these data suggest that CD141Hi cDCs are selectively reduced in patients with HR MDS and that decitabine treatment promotes cDC differentiation in vitro and in MDS patients. Optimal immunotherapeutic approaches for MDS patients may therefore benefit from strategies that address defects in APC differentiation and function.
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.