AML/MDS Decitabine Trial — Washington University (2016)
Overview
A single-arm prospective clinical trial of 10-day decitabine cycles in adults with AML or MDS at Washington University in St. Louis (NCT01687400, March 2013 – November 2015), with an extension cohort of 32 patients from the University of Chicago and WashU on 5-day schedules (combined N=116). The study paired serial enhanced exome and amplicon-panel sequencing with clinical response assessment to identify predictors of hypomethylating-agent response and to map clonal dynamics under therapy. Exome data are deposited in dbGaP (phs000159); methylation arrays in GEO (GSE80762). The TCGA AML cohort (laml_tcga_pub) was used as a comparator to benchmark TP53 mutation spectrum and methylation patterns. PMID:27959731
Composition
- Discovery cohort: 84 adults on the WashU 10-day decitabine protocol (AML ≥60 yr, relapsed AML, or transfusion-dependent MDS; ECOG PS ≤2). PMID:27959731
- Extension cohort: 32 additional patients — 24 with relapsed AML (10-day decitabine, University of Chicago, 2005–2010) and 8 on a 5-day schedule at WashU (5 single-agent, 3 combined with panobinostat). PMID:27959731
- Combined cohort: 116 patients; 99 had any sequencing performed. Cancer types: AML and MDS. PMID:27959731
- Drug: decitabine 20 mg/m²/day on days 1–10 of 28-day cycles (discovery cohort). PMID:27959731
Assays / panels (linked)
- Enhanced whole-exome sequencing (Illumina HiSeq 2000/2500, NimbleGen v3 capture) supplemented with 264-gene biotinylated-probe panel (aml-264-gene-panel) — 39 patients, 157 exomes across serial timepoints. PMID:27959731
- Ion AmpliSeq 8-gene amplicon sequencing panel (aml-8-gene-ampliseq-panel) covering TP53, DNMT3A, IDH1, IDH2, ASXL1, SRSF2, U2AF1, SF3B1 — 45 additional patients. PMID:27959731
- Illumina HumanMethylation450 BeadChip on day 0 and day 10 of cycle 1. PMID:27959731
Papers using this cohort
- PMID:27959731 — Welch et al., N Engl J Med 2016: 10-day decitabine in AML/MDS; TP53 mutations predict blast clearance (100% vs 41%, P<0.001) but not survival benefit relative to TP53 wild-type.
Notable findings derived from this cohort
- TP53 mutations predict decitabine response: 21/21 (100%) TP53-mutant patients achieved bone-marrow blast clearance (<5% blasts) vs 32/78 (41%) wild-type-TP53 patients (P<0.001); effect replicated in discovery and extension cohorts independently. PMID:27959731
- Unfavorable cytogenetics also predicted response: blast clearance in 29/43 (67%) with unfavorable-risk karyotypes vs 24/71 (34%) with intermediate/favorable risk (P<0.001); 20/21 TP53-mutant patients had unfavorable-risk karyotypes. PMID:27959731
- Response does not translate to survival advantage: median overall survival 12.7 months (TP53-mutant) vs 15.4 months (TP53 wild-type), P=0.79; notably better than the 4–6 month survival with cytotoxic induction in TP53-mutant AML. PMID:27959731
- Mutation clearance is universal but incomplete: only TP53 and SF3B1 mutations consistently dropped to VAF <5%; founding-clone VAF at maximum clearance ranged from 0.06% to 18.43% in the 20 best responders; bone-marrow blast clearance preceded mutation reduction in 15/54 patients. PMID:27959731
- Clonal hematopoiesis in remission: 7/22 responders developed nonleukemic rising clones; 2 carried mutations in DNMT3A or PPM1D. PMID:27959731
- All evaluable relapses (9/9) arose from preexisting subclones detectable before therapy, confirming selection rather than de novo mutation as the resistance mechanism. PMID:27959731
- TP53 mutation spectrum (missense-dominant, hotspot distribution) was indistinguishable from the TCGA AML cohort (laml_tcga_pub); no TP53-driven methylation signature identifiable in either dataset. PMID:27959731
- Allogeneic SCT had the largest survival benefit by Cox stepwise regression (P<0.001), regardless of TP53 status. PMID:27959731
Sources
- ClinicalTrials.gov: NCT01687400
- dbGaP: phs000159 (exome data)
- GEO: GSE80762 (methylation arrays)
- cBioPortal study ID: mnm_washu_2016
- Welch et al., N Engl J Med 2016 — DOI: 10.1056/NEJMoa1609227
This page was processed by crosslinker on 2026-05-14.