Personalized ctDNA detected residual disease signals across the perioperative course in localized sarcoma
Personalized circulating tumor DNA analysis identified measurable disease biology in a setting that has long resisted blood-based monitoring: localized soft tissue sarcoma. In the SU2C-SARC032 phase II randomized trial, 313 plasma samples from 106 patients were examined before treatment, after preoperative radiotherapy with or without pembrolizumab, and after surgery. The assay design was deliberately individualized, tracking a median of 46 somatic mutations per patient, a strategy suited to a tumor type that is both rare and genetically heterogeneous.
Baseline ctDNA was detected in 85% of patients, a strikingly high yield for a disease often considered technically difficult for plasma-based profiling. The pretreatment ctDNA burden was not random background noise. It correlated with larger tumors, grade 3 disease, and expression of genes tied to hypoxia and cell-cycle activity. Those links fit the biology of aggressive sarcoma: larger, high-grade tumors tend to shed more DNA, and hypoxic, proliferative tumors are more likely to be genomically unstable and clinically aggressive.
One sentence captures the central practical point. Blood-based detection was sensitive enough to follow sarcoma biology before a single incision.
Across the treatment course, ctDNA levels before therapy, after radiotherapy with or without immunotherapy but before surgery, and after surgery were strongly associated with disease-free survival and overall survival on both univariable and multivariable analyses. The signal persisted after adjustment for clinical risk factors, giving the blood-based measure independent prognostic weight rather than merely reflecting stage or grade. In a disease where radiographic response can be difficult to interpret and microscopic residual disease may escape detection, that kind of serial molecular readout has obvious clinical appeal.
The perioperative pattern also tracked treatment effect. Patients who received pembrolizumab and had better disease-free survival showed a greater drop in ctDNA after neoadjuvant therapy. In mechanistic terms, a falling ctDNA fraction likely reflects reduced viable tumor burden after combined local and systemic treatment. For an immunotherapy-containing regimen, a sharper decline may signal deeper tumor clearance before surgery, even when standard imaging cannot quantify that depth with equal precision.
Pembrolizumab was linked to a deeper ctDNA decline and better disease-free survival
The most clinically relevant pairing in this analysis was not simply ctDNA positivity at baseline, but the trajectory after treatment. Personalized ctDNA analysis detected quantitative shifts after preoperative therapy, and those shifts aligned with outcome. Patients who benefitted from pembrolizumab had a more pronounced post-neoadjuvant ctDNA reduction, paralleling their significantly better disease-free survival. In a localized sarcoma population receiving multimodality therapy, that relationship supports the idea that blood-based molecular response may serve as an early surrogate for treatment efficacy.
The postoperative samples added another layer of risk stratification. Patients with multiple tumor clones detected by ctDNA after surgery had inferior outcomes, implying that residual molecular heterogeneity carries clinical consequence. Clonal complexity after resection may represent persistent microscopic disease, incomplete eradication of subpopulations, or both. Either way, the finding suggests that surgery can remove the dominant lesion while leaving behind biologically distinct disease that continues to threaten relapse.
This is where personalized ctDNA differs from a generic tumor marker. It can do more than answer whether tumor DNA is present. It can follow subclones over time, show whether the molecular burden contracts after therapy, and suggest whether surgery has truly cleared the active disease reservoir. For rare sarcomas, where tissue is often limited and repeat biopsy is rarely attractive, serial plasma profiling offers a noninvasive window into response and recurrence risk.
The study design also strengthens the interpretation. Because the analysis came from a randomized trial platform, the ctDNA observations were embedded in a clinically relevant treatment sequence that included radiotherapy, surgery, and pembrolizumab in a high-risk localized population. The investigators adjusted for clinical risk factors in multivariable models, so the prognostic value was not confined to crude baseline differences alone. The result is a coherent story: ctDNA was detectable in most patients, reflected aggressive disease biology at baseline, declined with effective neoadjuvant therapy, and retained prognostic value after surgery.
One sentence stands out for bedside practice. Persistent or multifocal ctDNA after surgery marked a population with worse outcomes.
What this means for sarcoma care and prospective trial design
Localized soft tissue sarcoma has always challenged response assessment because the disease is uncommon, heterogeneous, and often managed with curative intent. Pathology and imaging remain essential, but neither fully captures minimal residual disease after multimodality treatment. Personalized ctDNA testing directly addresses that gap by offering a patient-specific molecular assay with enough sensitivity to monitor disease across the perioperative timeline.
The biology behind the assay’s performance is intuitive. Tumors with higher burden, higher grade, and hypoxia-related transcriptional programs should shed more DNA into plasma, and successful treatment should reduce that signal. When the signal fails to clear, especially after surgery, the likelihood of residual disease rises. In this trial, those principles translated into measurable prognostic separation.
The findings support prospective ctDNA-guided interventional trials in soft tissue sarcoma. Such studies would need to answer a concrete clinical question: can treatment escalation, treatment de-escalation, or intensified surveillance based on plasma molecular persistence improve outcomes without adding unnecessary toxicity? The current data do not answer that question, but they establish a strong rationale for testing it.
For now, the most defensible conclusion is narrower and clinically grounded. Personalized, tumor-informed ctDNA profiling is feasible in localized sarcoma, detects disease in most patients, tracks response to radiotherapy and pembrolizumab, and provides independent prognostic information before and after surgery. The next step is to determine whether acting on those molecular signals changes relapse risk rather than merely describing it.