Tumor-Informed ctDNA Sequencing Could Identify Osteosarcoma Relapse Months Before Radiographic Detection

Utilizing tumor-informed deep sequencing of circulating tumor DNA (ctDNA) for minimal residual disease (MRD) surveillance represents one of the most compelling demonstrations to date that molecular monitoring can identify osteosarcoma patients at high risk for relapse and detect recurrence approximately 3 months before conventional imaging, though prospective validation is needed before ctDNA can replace or supersede standard radiographic surveillance, according to R. Lor Randall, MD, FACS.

In an interview with OncLive®, Randall discussed a longitudinal, prospective study evaluating a tumor-informed MRD approach in patients with osteosarcoma at the First Affiliated Hospital of Sun Yat-sen University in Guangzhou, China.¹ Among the 59 patients with successful panel customization and available blood samples, postoperative ctDNA positivity was identified in 13 patients and emerged as a significant independent predictor of event-free survival (EFS) on both univariate and multivariate Cox regression analysis (P < .05), with ctDNA detection preceding positive imaging in 5 patients at a mean lead time of 92.6 days. Specifically, positive ctDNA results were associated with HRs of 7.43 (95% CI, 2.67-20.62) and 7.74 (95% CI, 1.55-38.75), according to findings from the respective analyses.

"The ability to identify patients at high risk for relapse and potentially detect recurrence months before imaging picks it up is really an exciting development for our field. [Although] additional validation is needed before routine clinical implementation, these data suggest that ctDNA may eventually become an important component of precision surveillance in osteosarcoma," said Randall, the David Linn Endowed Chair for Orthopedic Surgery, the chair of the Department of Orthopedic Surgery, and a professor at the University of California Davis Comprehensive Cancer Center in Sacramento.

In addition to reviewing the major findings from this study, Randall explained the biological rationale for a tumor-informed, personalized ctDNA approach in osteosarcoma and addressed the clinical implications and limitations of these data.

OncLive: What was the biologic rationale taking a tumor-informed approach to ctDNA detection, and what distinguished the study’s methodology from others using generic off-the-shelf ctDNA panels?

Randall: Despite decades of progress in surgery, chemotherapy, imaging, and supportive care, osteosarcoma surveillance remains quite a challenge. We still rely heavily on radiographic imaging to identify recurrence, often after disease has already become clinically evident. This study addresses an important question: Can ctDNA serve as a sensitive biomarker of minimal residual disease and identify recurrence earlier than conventional imaging?

What makes this study particularly relevant is the biology of osteosarcoma itself. Unlike some malignancies that have a relatively small number of recurrent driver mutations, osteosarcoma has a shattered genome and is extraordinarily heterogeneous genetically. This means generic off-the-shelf ctDNA panels may miss clinically important disease signals. Instead, the investigators used what is called a tumor-informed MRD panel. In the [simplest] terms, they first sequenced each patient's tumor and created a personalized molecular fingerprint based on that individual's unique mutations. They then tracked those same mutations in blood samples over time. The goal was to determine whether MRD could be detected in the bloodstream before recurrence became visible on conventional imaging. This personalized approach is particularly attractive in osteosarcoma because of the remarkable genetic diversity we see from one patient to the next.

What were the major findings?

Several findings stand out. The investigators successfully generated individualized MRD panels for [85.5%] of patients [n = 71 of 83]. Among the 59 patients ultimately included in the analysis, postoperative ctDNA positivity emerged as a powerful predictor of outcome. Patients who remained ctDNA positive after surgery experienced significantly worse EFS than patients whose ctDNA was negative; [upon] multivariate analysis, postoperative ctDNA positivity carried HRs of nearly 8 for [disease] relapse or progression.

Perhaps even more intriguing, ctDNA identified relapse before conventional imaging in several patients. Among those cases, the average lead time was approximately 3 months, earlier than radiographic detection. For a disease where early intervention may matter, this is potentially an important finding.

How could these data be utilized in clinical practice?

I think the most immediate application is risk stratification. Today, 2 patients may look identical on imaging after surgery and chemotherapy, yet their underlying biologic risk may be very different. ctDNA may help identify patients who harbor microscopic residual disease despite appearing disease-free radiographically. Those patients could potentially benefit from more intensive surveillance, earlier imaging, enrollment in clinical trials, or future ctDNA-guided therapeutic strategies.

However, I don't necessarily believe that this study supports changing treatment solely on the basis of ctDNA results in its present form. Rather, I view ctDNA as an emerging adjunct to conventional surveillance, but certainly not a replacement for conventional imaging. The technology does appear promising, but prospective validation remains necessary before it becomes standard of practice.

What are the strengths of this study?

There are several [strengths] worthy of recognition. First, this represents one of the largest tumor-informed ctDNA cohorts reported in osteosarcoma to date. Second, the investigators used a prospective longitudinal sampling strategy, allowing them to follow molecular changes throughout treatment and surveillance. Third, the study addresses a clinically meaningful end point: relapse detection and EFS. Finally, the biologic rationale is strong: the personalized assay design aligns well with the marked heterogeneity that characterizes osteosarcoma.

What are the limitations of this work?

First, this remains a relatively small, single-center study. The follow-up was relatively short, with a median follow-up of just over 8 months. The sensitivity was certainly not perfect; some patients ultimately relapsed despite negative ctDNA testing, reminding us that a negative assay does not completely eliminate risk.

Most importantly, this study demonstrates prognostic value, but not yet clinical utility. The critical unanswered question is whether acting on ctDNA information actually improves patient outcomes. Detecting recurrence 3 months earlier is only meaningful if that earlier detection ultimately changes survival, quality of life, or treatment effectiveness, and that remains unproven.

What is your overall take-home message regarding this research?

This study represents one of the most compelling demonstrations to date that tumor-informed ctDNA can function as a MRD biomarker in osteosarcoma. For decades, orthopedic oncologists and pediatric and medical oncologists have relied largely on imaging to monitor osteosarcoma patients after treatment. This study suggests that molecular surveillance may offer a glimpse into disease biology long before recurrence becomes radiographically apparent. If validated in larger prospective studies, tumor-informed ctDNA could help us move beyond a one-size-fits-all surveillance strategy and toward a more personalized approach—one in which we identify high-risk patients earlier and monitor them more intelligently and ultimately intervene at a time when disease burden is lowest and opportunities for successful treatment may be greatest.

Reference

Fu Y, Xu Y, Liu W, et al. Tumor-informed deep sequencing of ctDNA detects minimal residual disease and predicts relapse in osteosarcoma. EClinicalMedicine. 2024;73:102697. doi:10.1016/j.eclinm.2024.102697