The Cancer Testis Antigen (CTA) superfamily of genes have long emerged as potential targets for immunotherapy and response monitoring due to their relative restrictive expression in testis and variable expression in some cancers. Much of the data on CTA expression in cancer have been in solid tumors with little data on CTA expression in leukemias. Identification of leukemia associated antigens (LAA) may provide attractive targets for immunotherapy or monitoring of residual disease.

In order to discover AML associated CTAs, we performed transcriptome sequencing on 446 de novo pediatric AML samples and 20 normal bone marrows (NBM) from healthy donors and interrogated the expression of 245 candidate CTAs obtained from the CT database (http://www.cta.lncc.br). Differential expression using Limma (https://bioconductor.org/packages/release/bioc/html/limma.html) identified 32 CTA genes to be significantly differentially expressed with 7 being up-regulated compared to NBM and 15 genes with expression in NBM which are silenced in AML. The upregulated genes include: CCNA1, PRAME , CT45A5, IGF2BP3, SPAG6, TEX14, and ZNF165 and are collectively expressed in 330 of 446 patients (75%) with the highest contributors being PRAME and CCNA1 (Fig. 1A, Fig. 1B) . These 2 CTAs are expressed in every fusion group and constitute 65% of the total cohort, with a notable enrichment in core binding factor (CBF) patients (80%). High CCNA1 expression are present in 2/3 of patients with Inv(16), 71% of patients with a NUP98 fusion, 88% of DEK - NUP214 carriers, and 64% of cytogenetic normal cohort. PRAME is expressed in 79% of patients with a RUNX1 - RUNX1T1 fusion, 57% of CBFA2T3 - GLIS2 positives, and 20% of Inv(16) patients. While the oncoprint (Fig. 1A) shows that most patients express 1 or 2 CTAs, with some genes showing exclusivity, patients with a KMT2A fusion tend to express multiple CTAs. As a group, 77% of patients with KMT2A fusions express at least 1 CTA.

To determine the underlying mechanism of dysregulated CTA expression in AML, we inquired whether promoter methylation status might be associated with CTA expression. We correlated the expression level of the 7 upregulated CTA genes with the methylation of their corresponding promoters. Overall there was a strong signal demonstrating anti-correlation of expression of the above mentioned genes with their promoter methylation. In specific, for PRAME, differences in methylation at the region chr22:22899991-22901267 explain over 87% of the variation in expression, and this locus is normally methylated in NBM (Fig. 1C). Likewise, for CT45A5, it's normally methylated in NBM, and demethylation of this locus explains about 61% of the variance in expression (Fig. 1C).

Additionally, we evaluated the clinical implications of CTA expression, and demonstrated that CCNA1 expression was significantly associated with clinical outcome. The range of expression for CCNA1 in the AML samples was 0-561.4 TPM (mean = 31.8 TPM), whereas expression for NBMs ranged from 0.3-2.5 TPM (mean = 0.9) (p=8.45E-08). After binning expression values to determine the best cutpoint for high and low expressors of CCNA1, patients classified with high expression had statistically significant worse event-free (EFS, Fig.1D) and overall survival (OS) (p=<0.0001, p=0.0393 respectively, Log-Rank test). As a sub-group, patients with KMT2A rearrangements and high CCNA1 expression also show prognostic significance with correlation of poor outcome (EFS, p=0.0009, Fig. 1E). In addition, cytogenetically normal (CN-AML) patients with FLT3 ITD had a much higher overall expression of CCNA1 than did CN-AML without FLT3 ITD (p=0.009, Mann-Whitney test).

In this analysis we identify candidate CTA genes that are cumulatively expressed in the majority of AML cases and can provide novel therapeutic options by targeting with modified T-cell receptor therapies. They can also be used as biomarkers for response monitoring. Further, as aberrant expression of these genes are regulated by altered methylation, use of hypomethylating agents might enhance therapeutic efficacy by increasing target expression.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.