After curative-intent treatment for solid tumors, molecular residual disease (MRD) testing is increasingly used to look for residual tumor signals that may not yet be visible on imaging. [1]
Current ctDNA MRD strategies generally take one of two approaches. Some use tumor tissue sequencing to identify patient-specific mutations and build a personalized assay for follow-up testing. Others use a fixed panel that can be applied across patients without designing a new panel for each case. [3]
These two approaches address the same problem in different ways. In this article, we look at the trade-off between personalized and fixed-panel MRD testing, and at how combining mutation, copy number variation (CNV), and fragmentomic signals may help make better use of the tumor-related information captured through fixed-panel sequencing.
1. Beyond Imaging: Why Postoperative MRD Matters
Postoperative imaging is essential, but it only detects recurrence once disease becomes visible on a scan. ctDNA provides a different type of information by looking for tumor-derived DNA in the bloodstream. [1]
In the TRACERx study, serial postoperative ctDNA testing identified impending relapse, including additional cases that were not detected at the first postoperative blood draw. [1] A separate multicenter study found that postoperative ctDNA positivity was independently associated with recurrence, with molecular detection preceding radiologic recurrence by a median of 6.6 months. [2]
This makes sensitivity especially important. At the point when imaging is still negative, the amount of tumor-derived DNA in plasma may be extremely small. An MRD assay has very little signal to work with. Therefore, how to detect it reliably when very little tumor DNA is present remains one of the major challenges in today’s MRD testing. [1,3]
2. Personalized and Fixed-Panel MRD: Different Routes to Sensitivity
There are two main approaches for MRD (minimal residual disease) tracking.
2.1. Personalized tumor-informed assay
It starts by profiling the patient’s tumor. Tumor tissue is sequenced using WES, WGS, or another broad sequencing method to identify mutations that are specific to that tumor. A subset of those variants is then selected for high-depth tracking in plasma. [3]
This approach has two clear advantages. First, the assay already knows what to look for. Instead of searching broadly for an unknown tumor signal, it can focus on dozens or hundreds of patient-specific variants. Second, that focus can translate into strong sensitivity. In a recent head-to-head comparison, a tumor-informed personalized assay achieved higher landmark MRD sensitivity than both tumor-informed and tumor-naïve fixed-panel approaches.[3]
In contrast, there are three practical limitations: a longer workflow requiring tumor sequencing to identify mutations and custom assay design; difficulty in standardization for kit-based commercialization; and limited ability to capture newly acquired mutations outside the baseline-selected tracking targets. [3,5]
2.2 Fixed-panel MRD
Fixed-panel MRD takes the opposite approach. The same predefined genomic regions are sequenced for every patient. There is no need to design a new panel each time, which makes the workflow easier to standardize, scale, and potentially implement in a kit-based format for clinical use. [5]
Fixed-panel sequencing can also provide information beyond mutation calls. The same sequencing dataset may contain other tumor-associated signals, including copy number and fragmentomic features. [6] Incorporating these additional dimensions offers an opportunity to make broader use of the data generated from each sample and potentially improve MRD detection while retaining the standardization and scalability of a fixed-panel workflow.
3. How Can Fixed-Panel MRD Improve Sensitivity?
To improve sensitivity, copy number variations (CNVs) and fragmentomic features (e.g., fragment size, fragmentation patterns, and features related to nucleosome or chromatin organization) can be considered. [4-6]
Copy number variations (CNVs) capture gains or losses across larger genomic regions. Rather than looking for a single-base change, CNV analysis looks for abnormal patterns in the amount of DNA represented across different parts of the genome.
Fragmentomics examines the cfDNA fragments themselves. Tumor-derived DNA can differ from non-tumor cfDNA in features such as fragment size, fragmentation patterns, and signals related to nucleosome or chromatin organization. [4]
Mutations, CNVs, and fragmentomic features therefore capture different aspects of tumor biology. For example, a sample may contain only a weak mutation signal. That signal alone may not be enough for a confident MRD call. But if the same sample also shows abnormal copy-number patterns and fragmentomic features consistent with tumor-derived DNA, those additional features can provide independent supporting evidence. This is where multidimensional analysis becomes useful, as it provides more comprehensive insights into the tumor’s genetic and epigenetic landscape. [4-6]
Combining multiple data types of fixed-panel MRD also increases the importance of clinical validation. Mutation, CNV, and fragmentomic features can each be influenced by biological and technical variation, so robust MRD models need to be developed and tested across large and diverse clinical cohorts. This helps determine which features consistently carry tumor-related signal and how they should be weighted in the final MRD assessment. [4,6]
4. SHIELDING ULTRA™: A Multidimensional Approach to MRD
Geneseeq’s SHIELDING ULTRA™ was developed around this principle.
SHIELDING ULTRA™ uses a standardized fixed panel covering hotspot regions across 2,365 cancer-related genes together with ultra-deep unique molecular identifier sequencing. SHIELDING ULTRA™ combines a fixed panel with multiple molecular features (mutations, copy number variations, and fragmentomics features) and AI modeling to support sensitive MRD detection. [6]
The sequencing foundation is ATG-Seq™, Geneseeq’s ultra-deep sequencing and error-suppression technology. ATG-Seq is designed to reduce technical background and improve confidence in very low-frequency ctDNA signals.
In data presented at AACR 2026, SHIELDING ULTRA™ was evaluated across multiple clinical cohorts. The study demonstrated high analytical sensitivity and specificity, strong concordance between tumor-naïve multimodal and tumor-informed workflows, and prognostic performance in postoperative MRD assessment.[6] In another use case, SHIELDING ULTRA™ was also evaluated in advanced NSCLC. Among patients who achieved a partial response to immunotherapy, the assay identified 18 of 34 patients who subsequently experienced early disease progression. [5]
These findings illustrate the rationale behind a multidimensional fixed-panel approach. Instead of depending on a small number of mutation signals alone, SHIELDING ULTRA™ uses mutation, CNV, and fragmentomic information together to extract more tumor-related information from each cfDNA sequencing dataset, while retaining the standardization and scalability of a fixed-panel workflow.
Learn more about SHIELDING ULTRA™ and Geneseeq’s approach to ctDNA MRD monitoring.
References
- Abbosh C, Frankell AM, Harrison T, et al. Tracking early lung cancer metastatic dissemination in TRACERx using ctDNA. Nature. 2023;616:553–562. doi:10.1038/s41586-023-05776-4.
- Fu R, Huang J, Tian X, et al. Postoperative circulating tumor DNA can refine risk stratification in resectable lung cancer: results from a multicenter study. Molecular Oncology. 2023;17(5):825–838. doi:10.1002/1878-0261.13387.
- Cui Q, Zhang J-T, Yin H, et al. Comparison of circulating tumor DNA assays for molecular residual disease detection in non-small cell lung cancer. Journal of Thoracic Oncology. Published online July 14, 2026. Article 104090. doi:10.1016/j.jtho.2026.104090.
- Sirajee AS, Kabiraj D, De S. Cell-free nucleic acid fragmentomics: A non-invasive window into cellular epigenomes. Translational Oncology. 2024;49:102085. doi:10.1016/j.tranon.2024.102085.
- Zhang H, Li N, Tang W, et al. ShieldingUltra: A novel approach for enhanced minimal residual disease detection through the integration of mutation, copy number variation, and fragmentomics [abstract]. Cancer Research. 2025;85(8 Suppl 1) 4561. doi:10.1158/1538-7445.AM2025-4561.
- Shi Y, Zhang H, Lin Z, et al. Ultrasensitive AI-driven framework for MRD detection based on multidimensional cfDNA features [abstract]. Cancer Research. 2026;86(7 Suppl):3852. doi:10.1158/1538-7445.AM2026-3852.
