Abstract

The molecular changes responsible for the recurrence of APL after achieving complete remission are largely unknown. The Phase III trial E2491 provided a venue for investigating whether treatment type had an effect on the changes in gene expression that occur from PTx to Rel, since ¼ of patients were randomized to receive only chemotherapy (C-set patients), while ¾ of patients additionally received all-trans retinoic acid (ATRA) either during the induction and/or maintenance phases of treatment (A-set patients). High-quality, ethanol-stored RNA was available from matched PTx and Rel samples with ≥85% blasts from 3 C-set and 3 A-set patients, who had the following characteristics:

Patient-SetPTx WBCFAB ClassP-Rα TypePTx MutationsRel MutationsDays on ATRADays off ATRA
1-C 68.6 M3 None Flt3ITD & D835 N/A 
2-C 2.4 M3 Flt3ITD & D835 Flt3ITD N/A 
3-C 0.4 M3v Flt3ITD & D835 Flt3ITD N/A 
4-A 77.4 M3v Flt3ITD PML-RARα 142 Maintenance 
5-A 3.0 M3 365 Maintenance 344 
6-A 10.9 M3 PML PML 19 Induction 410 
Patient-SetPTx WBCFAB ClassP-Rα TypePTx MutationsRel MutationsDays on ATRADays off ATRA
1-C 68.6 M3 None Flt3ITD & D835 N/A 
2-C 2.4 M3 Flt3ITD & D835 Flt3ITD N/A 
3-C 0.4 M3v Flt3ITD & D835 Flt3ITD N/A 
4-A 77.4 M3v Flt3ITD PML-RARα 142 Maintenance 
5-A 3.0 M3 365 Maintenance 344 
6-A 10.9 M3 PML PML 19 Induction 410 

Gene expression analysis utilized the Affymetrix Human Genome U-133 Plus 2.0 chip. Of 54,613 gene probe sets (gps), 6220 gps for named genes with a mean value >100 for all samples and a ≥100 absolute difference between PTx and Rel values were selected for study. By unsupervised hierarchical cluster analysis, the overall samples segregated into 2 major groups, PTx or Rel with single exceptions. The PTx samples co-segregated into 2 groups according to the co-incident parameters of high WBC count and V-form PML-RARα (Patients 1,4,6 vs 2,3,5). Most notably, the Rel samples segregated by treatment type, C-set and A-set. In order to identify the gps most significantly contributory to differences between the C- and A-set gene expression profiles, we made two linked analyses. First, 575 gps were selected based on the criteria of an ≥1.5-fold difference, p <0.05, between the average expression level of each gps in the 3 relapse samples of the A-set vs the C-set. Second, we identified 92/575 genes with a statistically-significant fold-change (p ≤0.05, paired T-test) from PTx to Rel. After editing out redundant gps, 21 unique gene transcripts (ugt) were upregulated (Up) and 54 ugt were downregulated (Dn) from PTx to Rel. There was a high degree of concordance between the fold-difference values in the Rel samples and the fold-change values from PTx to Rel. Remarkably, for 19/21 Up ugt and 49/54 Dn ugt, the selection was based on greater change in A-set than C-set samples. Prominent among the selected A-set ugt were those affecting DNA repair/replication (Up: NPN; Dn: RAD17, RPA1, PARP1, MCM3) and signal transduction (Up: STAT3, IRAK3, LIMK2, GRB10; Dn, STS-1, CSNK2A1, PAK2, OPN3). These results indicate that the addition of ATRA to chemotherapy-based treatment of APL (all patients received 2 courses of consolidation chemotherapy) had a major impact on the gene expression profile at relapse, which further suggests that ATRA treatment made an important and unique contribution to the molecular progression process leading to relapse.

Author notes

Disclosure: No relevant conflicts of interest to declare.