Calreticulin exposure by malignant blasts correlates with robust anticancer immunity and improved clinical outcome in AML patients

Abstract

Cancer cell death can be perceived as immunogenic by the host only when malignant cells emit immunostimulatory signals (so-called "damage-associated molecular patterns," DAMPs), as they die in the context of failing adaptive responses to stress. Accumulating preclinical and clinical evidence indicates that the capacity of immunogenic cell death to (re-)activate an anticancer immune response is key to the success of various chemo- and radiotherapeutic regimens. Malignant blasts from patients with acute myeloid leukemia (AML) exposed multiple DAMPs, including calreticulin (CRT), heat-shock protein 70 (HSP70), and HSP90 on their plasma membrane irrespective of treatment. In these patients, high levels of surface-exposed CRT correlated with an increased proportion of natural killer cells and effector memory CD4⁺ and CD8⁺ T cells in the periphery. Moreover, CRT exposure on the plasma membrane of malignant blasts positively correlated with the frequency of circulating T cells specific for leukemia-associated antigens, indicating that ecto-CRT favors the initiation of anticancer immunity in patients with AML. Finally, although the levels of ecto-HSP70, ecto-HSP90, and ecto-CRT were all associated with improved relapse-free survival, only CRT exposure significantly correlated with superior overall survival. Thus, CRT exposure represents a novel powerful prognostic biomarker for patients with AML, reflecting the activation of a clinically relevant AML-specific immune response.

Figure 1.

Chemotherapy-independent emission of danger signals by malignant AML blasts. (A-C) Percentage of ecto-CRT⁺ (A), ecto-HSP70⁺ (B), or ecto-HSP90⁺ (C) CD45⁺CD33⁺ cells from 10 healthy donors (HD) or 50 patients with AML before (Prior) or after (Post) the initiation of induction chemotherapy. Median values are reported. NS, nonsignificant. (D) Correlation between the percentage of ecto-CRT⁺, ecto-HSP70⁺, and ecto-HSP90⁺ CD45⁺CD33⁺ cells measured in 50 patients with AML before induction chemotherapy. R, Pearson correlation coefficient. (E) Correlation between CALR mRNA levels and ATF4, HSP5A, or DDIT3 mRNA levels in 34 patients with AML before induction chemotherapy. (F) Correlation between CALR mRNA levels and ATF4, HSP5A, or DDIT3 mRNA levels in 173 patients with AML from the TGCA public database. (G) Circulating HMGB1 levels from 10 HD or 50 patients with AML before the initiation of induction chemotherapy (Prior), 12 hours after induction chemotherapy (Post), or at the reestablishment of normal hematopoiesis (Recovery). Median values are indicated.

Figure 2.

Transcriptional signatures of PBMCs in patients with AML exhibiting robust vs weak CRT exposure on blasts. (A-B) Expression levels of genes from the Human Immune Panel TaqMan low-density array in PBMCs from 13 CRT^Hi vs 13 CRT^Lo patients with AML prior to the initiation of induction chemotherapy (A), or at recovery of normal hematopoiesis (B). The Mann-Whitney test was employed to assess intergroup variations. (C) qRT-PCR–assisted quantification of CD8A, CD28, CXCR3, IFNG, IL2, IL2RA, TBX21, FASLG, GNLY, GZMB, PRF1, CCL5, CCR4, CCR4, CCR5, CD40LG expression levels in PBMCs from 13 CRT^Hi vs 13 CRT^Lo patients with AML prior to the initiation of induction chemotherapy, or at recovery of normal hematopoiesis. Box plots: lower quartile, median, upper quartile; whiskers, minimum value, maximum value.

Figure 3.

Transcriptional and phenotypic signatures of PBMCs in patients with AML exhibiting robust vs weak CRT exposure on blasts. (A) Nanostring-assisted quantification of NK cell–related mRNAs in PBMCs from 6 CRT^Hi vs 6 CRT^Lo patients with AML at recovery of normal hematopoiesis. The Mann-Whitney test was employed to assess intergroup variations. (B) qRT-PCR–assisted quantification of KIR2DL1, KIR2DL2, and KIR2DL3 (cumulatively as CD158 family members), KLRF1, KLRC2, and NCR1 in PBMCs from 20 CRT^Hi vs 20 CRT^Lo patients with AML at recovery of normal hematopoiesis. Box plots: lower quartile, median, upper quartile; whiskers, minimum value, maximum value. (C-D) Percentage of circulating CD33⁻CD3⁻CD56⁺ NK cells (C), CD45⁺CD3⁺CD4⁺ T cells, and CD45⁺CD3⁺CD8⁺ T cells in 13 CRT^Hi vs 13 CRT^Lo patients with AML before induction chemotherapy (Prior, D) and at reestablishment of normal hematopoiesis (Recovery, C-D). Box plots: lower quartile, median, upper quartile; whiskers, minimum value, maximum value. (E) Percentage of IFN-γ–producing cells among CD45⁺CD3⁺CD4⁺ T cells and CD45⁺CD3⁺CD8⁺ T cells from 13 CRT^Hi vs 13 CRT^Lo patients with AML before induction chemotherapy (Prior) and at reestablishment of normal hematopoiesis (Recovery). Box plots: lower quartile, median, upper quartile; whiskers, minimum value, maximum value. (F) Distribution of circulating CD45⁺CD3⁺CD8⁺ T cells and CD45⁺CD3⁺CD4⁺ T cells in 20 CRT^Hi vs 20 CRT^Lo patients with AML before (Prior) and after (Post) induction chemotherapy. Mean percentage values are depicted as pie charts. Central memory: CD45RA⁻CD45RO⁺CCR7⁺CD62L⁺; effector memory: CD45RA⁻CD45RO⁺CCR7⁻CD62L⁻; naive: CD45RA⁺CD45RO⁻CCR7⁺CD62L⁺; terminally differentiated: CD45RA⁺CD45RO⁻CCR7⁻CD62L⁻.

Figure 4.

Antigen-specific immune responses in patients with AML exhibiting ecto-CRT⁺vs ecto-CRT⁻blasts. (A-B) Representative dot plots (A) and quantitative data (B) of IFN-γ–secreting CD45⁺CD3⁺CD4⁺ T cells and CD45⁺CD3⁺CD8⁺ T cells from 10 HD or 15 CRT^Hi vs 15 CRT^Lo patients with AML before induction chemotherapy (Prior, B) and at reestablishment of normal hematopoiesis (Recovery, A-B), upon exposure of the corresponding PBMCs to peptide mixture spanning BIRC5, CCNB1, PRAME, or WT1. Box plots: lower quartile, median, upper quartile; whiskers, minimum value, maximum value.

Figure 5.

Prognostic value of CRT, HSP70, and HP90 exposure in patients with AML. (A-C) RFS and OS among 50 patients with AML stratified in 2 groups based on median percentage of circulating ecto-CRT⁺ (A), ecto-HSP70⁺ (B), or ecto-HSP90⁺ (C) blasts measured prior to induction chemotherapy. Patients at risk are reported. The Wilcoxon test was employed to assess statistical significance. (D-E) Overall survival of 30 patients with AML from our cohort (D) and 173 patients with AML from the TCGA public database (E) stratified in 2 groups based on median CALR mRNA levels. Patients at risk are reported.