High-risk and silent clonal hematopoietic genotypes in patients with nonhematologic cancer

Authors

Aaron J. Stonestrom

Kamal N. Menghrajani

Sean M. Devlin

Sebastia Franch-Expósito

Ryan N. Ptashkin

Swara Y. Patel

Barbara Spitzer

Xiaodi Wu

Justin Jee

Pablo Sánchez Vela

Jennifer H. Milbank

Ronak H. Shah

Abhinita S. Mohanty

A. Rose Brannon

Wenbin Xiao

Michael F. Berger

Simon Mantha

Ross L. Levine

Doi

10.1182/bloodadvances.2023011262

PMID: 38147626 · DOI: 10.1182/bloodadvances.2023011262 · Journal: Blood Advances (2024)

TL;DR

Stonestrom et al. analyzed matched-normal blood sequencing from 42,714 MSK patients with nonhematologic cancer profiled by MSK-IMPACT to characterize clonal hematopoiesis (CH) genotypes associated with hematologic malignancy risk. They identified JAK2, RUNX1, and XPO1 mutations as high-risk CH genotypes, and showed that “silent” CH (synonymous / noncoding variants), especially when multiple variants were present or VAF was high, also predicted hematologic malignancy. Tracking clones into 26 subsequent hematologic malignancies revealed that JAK2 and TP53 VAFs consistently expanded at disease onset, whereas DNMT3A and silent CH VAFs mostly decreased PMID:38147626.

Cohort & data

  • 42,714 patients with nonhematologic cancer who underwent matched normal peripheral-blood sequencing as the germline comparator for MSK-IMPACT tumor profiling; 39,510 were alive without a hematologic diagnosis within 2 weeks of sequencing and had follow-up beyond that PMID:38147626.
  • Dataset: msk_ch_2023 (“Cancer Therapy and Clonal Hematopoiesis, MSK, Blood Adv 2023”), available on cBioPortal.
  • Assay: MSK-IMPACT hybridization-capture targeted sequencing, 341–505 genes depending on panel version PMID:38147626.
  • Hematologic malignancy events were identified via a hybrid NLP (CEDARS) + manual chart-review pipeline with final diagnoses confirmed by board-certified hematologists/oncologists PMID:38147626.

Key findings

  • Any CH was associated with increased risk of subsequent hematologic malignancy (HR = 1.93, 95% CI 1.45–2.57, P < .001) after adjustment for age, sex, and solid cancer type PMID:38147626.
  • Gene-level hazards identified JAK2, RUNX1, and XPO1 as CH genotypes conferring particularly high hematologic malignancy risk PMID:38147626.
  • Chronic hematologic diseases were predicted better than acute diseases by baseline CH, consistent with length bias in detection PMID:38147626.
  • “Silent” CH — synonymous coding and nonsplice noncoding variants — was associated with increased hematologic malignancy risk, especially when multiple silent variants co-occurred or VAF was high, indicating that hematopoietic clonality predicts risk independent of mutational function PMID:38147626.
  • In 26 patients who developed a hematologic malignancy sequenced on the same platform, JAK2 and TP53 CH VAFs consistently expanded at disease onset, whereas DNMT3A and silent CH VAFs largely decreased, implying divergent clonal trajectories by driver class PMID:38147626.

Genes & alterations

  • JAK2 — high-risk CH genotype; VAF expands at onset of hematologic malignancy PMID:38147626.
  • RUNX1 — high-risk CH genotype for subsequent hematologic malignancy PMID:38147626.
  • XPO1 — high-risk CH genotype for subsequent hematologic malignancy PMID:38147626.
  • TP53 — CH VAF consistently expands at hematologic malignancy onset, consistent with preleukemic selection under therapy PMID:38147626.
  • DNMT3A — most prevalent CH gene; VAFs mostly decreased at hematologic malignancy onset in this cohort PMID:38147626.
  • TET2, ASXL1 — prevalent CH drivers used as reference genotypes PMID:38147626.

Clinical implications

  • CH interpretation in patients with solid tumors should weight genotype: JAK2, RUNX1, and XPO1 mutations anticipate hematologic malignancy and may warrant closer hematologic surveillance PMID:38147626.
  • Silent CH (synonymous / noncoding), particularly with high VAF or multiple variants, is itself a signal of elevated risk and should not be dismissed as noise when found on tumor-normal panels PMID:38147626.
  • Divergent VAF trajectories (JAK2/TP53 expand vs. DNMT3A contract) argue for genotype-aware longitudinal monitoring in cancer patients receiving genotoxic therapy PMID:38147626.

Limitations & open questions

  • Length bias: chronic hematologic diseases are more readily predicted than acute ones, limiting sensitivity for aggressive therapy-related myeloid neoplasms PMID:38147626.
  • Cohort is single-center (MSK) and defined by nonhematologic cancer patients, so generalizability to healthy populations or other assay platforms is uncertain.
  • Only 26 patients had paired post-diagnosis sequencing to track VAF expansion, limiting statistical power of clonal trajectory claims PMID:38147626.
  • The degree to which specific prior therapies drive expansion of high-risk versus silent clones was not resolved.

Citations from this paper used in the wiki

  • “we analyzed data from 42714 patients who underwent blood sequencing as a normal comparator for nonhematologic tumor testing using a large cancer-related gene panel.”
  • “some CH genotypes including JAK2, RUNX1, and XPO1 variants were associated with high hematologic malignancy risk.”
  • “silent CH, particularly when multiple variants were present or VAF was high, was associated with increased risk of hematologic malignancy.”
  • “JAK2 and TP53 VAF consistently expanded at disease onset, whereas DNMT3A and silent CH VAFs mostly decreased.”
  • “CH HR = 1.93 (95% CI 1.45-2.57), P < .001.”

This page was processed by crosslinker on 2026-04-10.