Therapy-Related Myeloid Neoplasms with Balanced Chromosome Rearrangements Frequently Arise from Pre-Existing Clonal Haematopoiesis

Introduction

Therapy-related myeloid neoplasms (tMN) are an increasing healthcare problem resulting from rising long-term survival from primary cancers. Approximately 80% of cases are associated with an adverse karyotype and have a dismal prognosis; pre-existing clonal haematopoiesis (CH) appears to be a predisposing factor and mutations in genes including TP53, PPM1D and DNMT3A have been detected prior to chemo- or radiotherapy exposure. In contrast ~20% of tMN are characterised by balanced chromosome rearrangements and have a relatively favourable outcome; chemotherapy agents targeting topoisomerase II have been implicated in generating these rearrangements however the involvement of CH has not previously been described.

Methods and results

We performed whole exome sequencing (WES) of samples taken at diagnosis (D) and molecular complete remission (mCR) followed by targeted capture and error-corrected deep sequencing (ECS) on ≥2 mCR samples for 11 patients with therapy-related acute promyelocytic leukemia (tAPL) and 20 patients with de novo APL (dnAPL). All patients had t(15;17) / PML-RARA at diagnosis with no additional cytogenetic abnormality.

The median age of patients with tAPL was 46.7y (range 30-78); primary cancer types were breast (4), colorectal (2), lymphoma (3), CNS (1) and testicular (1). Ten patients had received chemotherapy and 4 radiotherapy. The median latency between primary cancer treatment and tAPL diagnosis was 3.9y (range 2.2-6.1). Patients received a mixture of chemotherapy- and arsenic-based treatments.

The median age of patients with dnAPL was 40.3y (range 18-69); all patients were treated with the AIDA regimen and none developed tMN subsequently.

There were no significant differences between the number or type of mutations between dnAPL and tAPL however disease associated somatic mutations were detectable in mCR samples by ECS in 4/11 tAPL samples compared to 0/20 dnAPL samples (p=0.04).

Mutations detected in mCR were UPN1: PPM1D exon (e) 6 1bp deletion (del, variant allele fraction, VAF, D 32% mCR 11.3%); UPN2: DNMT3A e8 1bp del (VAF D 40.2% mCR 0.48%); UPN3: DNMT3A e10 1bp del (VAF D 35.4% mCR 2.9%); UPN4: MYCN e3 6bp insertion (ins, VAF D 38% mCR 0.37%).

We screened samples from these patients and a further 39 dnAPL patients for a panel of genes with known CH associated mutations (CH-M) using ECS and detected additional mutations in 2 tAPL patients (UPN4, DNMT3A G104R, VAF D 0% mCR 3% and UPN5 DNMT3A R693H VAF D 2% mCR 25%, who subsequently developed tMN with complex karyotype). We did not detect CH-M in diagnostic samples from any patient with dnAPL and publicly available NGS datasets encompassing 220 patients only showed 1 APL case with a CH-M (DNMT3A e8 ins, prior cytotoxic exposure unknown). We detected treatment-emergent CH clones in mCR by ECS in 6/59 dnAPL patients treated with AIDA (DNMT3A n=3, PPM1D n=2, TP53 n=1, SF3B1 n=1).

Applying the same techniques to six patients with therapy-related core-binding factor AML, we identified a persistent CH-M in mCR in one (DNMT3A e15 del, VAF D 38%, mCR 11.8%).

DNA samples taken at the time of primary cancer diagnosis were available from UPN 1-3. In UPN1 we detected the PPM1D mutation in a lymph node (LN) involved with T-cell lymphoma (LN) (VAF 2.1%) and an uninvolved staging bone marrow (VAF 3.3%). In UPN3 the DNMT3A mutation was detected in a breast biopsy (VAF 0.7%) and LN involved with carcinoma (VAF 0.9%). In UPN2 the DNMT3A mutation was not detected in the LN biopsy diagnostic for Hodgkin lymphoma.

We used FACS to isolate T, B, monocyte, granulocyte and CD34+ cells from complete remission samples from UPN1-4 with >99% purity and detected the persistent mutations in each cell compartment e.g. UPN3 VAF: T cell 1.3%, B cell 8.6%, monocyte 4%, granulocyte 3.7%, CD34+ 3.9%.

Diagnostic material from UPN1 and 4 was injected into irradiated NSG mice. After 12 weeks we detected multilineage human engraftment in both samples by FACS in 1/3 mice from UPN1 and 2/3 mice from UPN4. We detected the PPM1D and MYCN mutations respectively in bone marrow samples from each engrafting mouse.

Conclusions

Together these findings indicate that tMN with balanced chromosome rearrangements can occur on a background of non-malignant CH. Using ECS we observed this phenomenon in 5/17 (29%) patients with therapy-related APL or CBF AML. This has important implications for planning curative therapy, notably for tAPL where effective cytotoxic-free regimens are available.