Ultrahigh-Sensitivity Sequencing Exposes Hidden FLT3-ITD Subclones in Elderly AML
In elderly patients with acute myeloid leukemia, FLT3-ITD microclones were present often enough to matter, yet small enough to escape conventional testing. Ultrahigh-sensitivity next-generation sequencing identified these hidden clones in 21 of 201 patients aged 65 years or older, a rate of 10.4% in a cohort already classified as FLT3-ITD negative by standard fragment length analysis. The median variant allele frequency was only 0.035%, with some clones measuring as low as 0.007%.
Those numbers sit at the edge of routine clinical detection. They also explain why fragment analysis, even when performed carefully, missed most of the signal.
Conventional Fragment Analysis Missed Most Low-Level FLT3-ITD Events
The sequencing strategy used here was deliberately deep, with a median read depth of 161,392 reads per sample. Three detection methods were applied to improve analytic sensitivity and specificity, including filt3r, getITD and a custom script. Even with that level of scrutiny, only 6 of the 21 microclones detected by sequencing were also identified by fragment length analysis, and all of those were already at the higher end of the microclone range.
The biology behind this gap is straightforward. Small subclones can sit below the threshold at which a standard assay reads them as positive, yet still carry enough proliferative advantage to shape disease behavior. In AML, a low-level FLT3-ITD clone can represent an early branch of a more aggressive disease architecture rather than a benign incidental finding.
Most of the detected insertions were singular, but some patients carried two or even three to five distinct insertions. Insertion length ranged from 6 to 102 nucleotides, with a median of 33 nucleotides. Even within this small cohort, the molecular heterogeneity was clear.
The Microclone Group Looked Biologically Distinct at Diagnosis
Patients with FLT3-ITD microclones presented with a significantly higher white blood cell count at diagnosis than patients without microclones, 23.9 versus 6.95 × 10^9/l. Hemoglobin and platelet counts did not differ meaningfully between groups. Age, sex and AML subtype were also similar.
Several genetic patterns separated the groups. Under the ELN 2024 classification for less-intensive therapy, patients with microclones were more often placed in the intermediate-risk category and less often in the adverse-risk category. The microclone-positive group also showed a significantly higher prevalence of DNMT3A, SMC3 and NRAS mutations, with a numerical increase in BCOR mutations. A lower frequency of TP53 mutations and a lower incidence of complex karyotype appeared in the same direction, although those differences did not reach statistical significance.
This pattern matters because FLT3-ITD rarely acts in isolation. In AML, signaling lesions and epigenetic or cohesin abnormalities often cooperate, and a low-level FLT3-ITD clone may gain traction within an already primed leukemic background. Higher leukocyte counts may reflect that proliferative bias at presentation.
Survival Was Unchanged Overall, But the Best Supportive Care Group Diverged
Across the full cohort, overall survival did not differ between patients with and without FLT3-ITD microclones. The p value was 0.97, leaving no signal of a broad survival separation across all treatment approaches.
One subgroup told a different story.
Among patients treated with best supportive care only, FLT3-ITD microclones were associated with shorter overall survival, with a hazard ratio of 2.8 and a 95% confidence interval of 1.006 to 7.808. The p value was 0.04. In contrast, this survival difference was not seen in patients receiving HMA monotherapy or HMA-based combination therapy, including regimens with venetoclax or ivosidenib.
That treatment interaction points toward a plausible therapeutic explanation. Less-intensive leukemia-directed therapy may suppress or contain a small FLT3-driven subclone before it expands, whereas supportive care alone leaves the disease biology untouched. In a frail elderly population, that distinction can determine whether a microclone remains a molecular curiosity or becomes a driver of rapid progression.
Clonal Behavior Shifted With Treatment Pressure
Follow-up samples from seven microclone-positive patients showed that these clones were not static. In one patient who achieved complete remission after four cycles of azacitidine and venetoclax, the FLT3-ITD microclone disappeared at remission. Another patient had re-emergence of peripheral blasts after two years of decitabine, accompanied by expansion of the microclone from 1.15% to 9.52%.
Three patients with refractory disease showed expansion of their microclones after four or fewer cycles of HMA-based therapy. Two patients with extremely small baseline clones, 0.029% and 0.0074%, lost the microclone after refractory disease emerged on HMA monotherapy. Among 19 patients initially judged microclone negative who later had relapsed or refractory samples available, no new FLT3-ITD mutation was detected at relapse.
That pattern suggests early clonal presence rather than frequent new acquisition later in the disease course. In practical terms, the mutant clone was usually there from the start if it was there at all; the challenge was detection, not emergence.
What This Means for Elderly AML Testing
The study places a firm number on a previously uncertain question: ultrasensitive sequencing found FLT3-ITD microclones in 10.4% of elderly AML patients receiving less-intensive therapy, all of whom had been called FLT3-ITD negative by conventional fragment analysis. The assay also showed that fragment analysis captured only a minority of those cases, and only when the clone was larger.
For older patients with AML, this has immediate diagnostic relevance. A low-level FLT3-ITD clone may accompany a different mutational profile, a higher white blood cell count and, in those managed without anti-leukemic therapy, shorter survival. The data do not support a blanket survival penalty across all treatment groups, but they do show that the clone is biologically real and clinically visible under some circumstances.
The remaining uncertainty is narrower and more actionable: whether microclone detection should alter risk assignment, treatment selection, or longitudinal monitoring in elderly patients who are unfit for intensive therapy. This cohort suggests that the answer may depend less on the clone alone than on the treatment context in which it is found.