At the 2026 international conference of the Society for Laboratory Automation and Screening (SLAS), Serena Silver, Chief Scientific Officer of Accent Therapeutics, laid out a precision oncology strategy built around a structural feature of cancer genomes rather than a single mutated gene.
The focus: whole-genome doubling (WGD) — a state in which tumor cells duplicate their entire chromosomal content — and a mitotic kinesin called KIF18A.
Accent’s lead program, ATX-295, is an orally bioavailable small-molecule inhibitor of KIF18A. According to a company press release, the first patient was dosed in a Phase 1/2 study on April 15, 2025. The FDA has granted Fast Track designation for ATX-295 in adult patients with advanced or metastatic platinum-resistant or refractory ovarian cancer.
The scientific premise, as Silver described it, is straightforward: genomic instability may not just be a hallmark of cancer — it may create druggable dependencies.
Whole-genome doubling as a vulnerability
WGD is common across tumor types. Large genomic analyses have estimated that roughly 30 percent of advanced cancers exhibit WGD, though prevalence varies by indication. WGD is associated with chromosomal instability, tumor evolution, and, in many contexts, poor prognosis.
Historically, WGD has functioned as a biomarker — a sign of aggressive disease biology. Accent is attempting to turn it into a therapeutic entry point.
Cells that have undergone whole-genome doubling carry extra chromosomes and often extra centrosomes. This amplifies the complexity of mitosis. Managing spindle architecture and ensuring accurate chromosome alignment becomes more demanding.
Published academic studies have shown that chromosomally unstable cancer cells can exhibit selective dependency on certain mitotic regulators. KIF18A is one of them.
Why KIF18A?
KIF18A is a member of the kinesin-8 family of motor proteins. It localizes to kinetochore microtubules and regulates microtubule plus-end dynamics, helping align chromosomes at the metaphase plate and ensuring orderly mitotic progression.
In foundational cell biology studies, KIF18A depletion disrupts chromosome congression and alters mitotic timing. More recent research has shown that chromosomally unstable (CIN) tumor cells are more sensitive to KIF18A loss than near-diploid cells, suggesting a context-specific dependency.
Accent’s development program builds directly on that biology.
In a company poster presentation, Accent reported that ATX-295 treatment in preclinical models resulted in mitotic spindle disruption, accumulation of G2/M markers such as phosphorylated histone H3, increased DNA damage markers including γH2AX, and induction of apoptosis in WGD-positive tumor models. The same materials state that near-diploid models exhibited less sensitivity, supporting the hypothesis of differential dependency.
Those findings, while preclinical, align with published literature indicating that CIN tumor cells may rely more heavily on KIF18A-mediated spindle regulation to complete mitosis successfully.
From dataset mining to drug candidate
Silver emphasized that the KIF18A program was supported by large-scale functional genomics analyses, not solely mechanistic reasoning.
Accent leveraged publicly available resources such as the Cancer Dependency Map (DepMap), which aggregates genome-wide CRISPR knockout screens across hundreds of cancer cell lines, and PRISM, a pooled cell-line drug screening platform that uses DNA barcoding to measure compound sensitivity.
According to Accent’s poster data, analyses integrating PRISM screening results with genomic annotations identified whole-genome doubling as a predictive enrichment biomarker for ATX-295 sensitivity. WGD-positive cell lines showed greater sensitivity in these datasets compared with WGD-negative counterparts.
Silver noted that these analyses were supported by AI-enabled approaches, including collaboration with Imagene AI. The company materials attribute elements of the WGD modeling and histopathology classifier development to that partnership.
Public DepMap resources independently confirm that KIF18A demonstrates variable dependency scores across cancer cell lines, with context-specific vulnerabilities observable in genomically unstable backgrounds. Accent’s claim is that WGD represents a consistent and reproducible genomic context for that vulnerability.
Moving into the clinic
ATX-295 entered clinical testing in 2025. According to Accent’s announcement, the ongoing Phase 1/2 trial is designed to evaluate safety, tolerability, pharmacokinetics, and preliminary anti-tumor activity.
The FDA’s Fast Track designation applies to platinum-resistant or refractory ovarian cancer, a setting where therapeutic options remain limited and prognosis is poor once platinum resistance develops.
High-grade serous ovarian cancers frequently exhibit extensive chromosomal instability and copy-number alterations. While not all such tumors are confirmed WGD-positive, the genomic architecture of the disease provides a rational context for evaluating KIF18A inhibition.
As of now, no clinical efficacy data have been publicly released. The therapeutic window in humans — particularly given historical toxicity associated with anti-mitotic agents — remains an open question that early-phase data will need to address.
AI histopathology as a digital biomarker
A central translational challenge is identifying patients whose tumors exhibit whole-genome doubling.
Genomic sequencing and copy-number analysis can determine WGD status, but such testing is not uniformly performed in all clinical settings. Accent is exploring a complementary strategy: AI-based digital pathology.
According to company poster data, Accent and collaborators trained machine learning models on digitized hematoxylin and eosin slides annotated with genomic WGD status. In a reported proof-of-concept analysis using TCGA breast cancer samples, the model achieved an area under the curve of 0.83 on a held-out test set.
The concept is that WGD-positive tumors may exhibit morphologic features — including nuclear enlargement and altered mitotic figures — that can be detected algorithmically, even if not obvious to the human eye.
Silver described this approach as a potential screening tool to enrich clinical trials for WGD-positive patients. The work remains exploratory and would require broader validation across tumor types before clinical implementation.
A structural view of precision oncology
The strategy presented at SLAS reflects a broader shift in how precision oncology is being conceptualized.
First-generation targeted therapies focused on dominant driver mutations. Subsequent waves exploited synthetic lethal interactions tied to specific genetic lesions. Accent’s KIF18A program targets a vulnerability created by genome architecture — the consequences of WGD and chromosomal instability.
WGD is neither rare nor tumor-type specific. If KIF18A inhibition proves clinically effective, it would represent a validation of targeting structural genomic states rather than individual mutations.
For now, the hypothesis is supported by preclinical data, functional genomics analyses, and early clinical initiation. The decisive test will come from human data: whether WGD-positive tumors in patients demonstrate selective sensitivity, and whether KIF18A inhibition can be administered safely at biologically active doses.
At SLAS, Silver positioned genomic instability not as background noise, but as an exploitable weakness. The coming clinical readouts will determine whether that premise holds in practice.
