A new phase 2 clinical trial has found that circulating tumor DNA (ctDNA) response after two cycles of pembrolizumab immunotherapy may serve as an early predictor of outcomes in patients with advanced non-small cell lung cancer (NSCLC). The study, published in October 2023 in Nature Medicine, suggests that ctDNA molecular response could potentially be used to guide treatment decisions and identify patients who may benefit from treatment intensification.
The Canadian Cancer Trials Group BR.36 trial was a multi-center, open-label, biomarker-directed phase 2 study conducted across six institutions in Canada and the United States. It was funded by the Cancer Research Institute, the Mark Foundation for Cancer Research, and Personal Genome Diagnostics. The trial aimed to evaluate ctDNA molecular response as an early endpoint for clinical outcomes in NSCLC patients receiving immune checkpoint blockade therapy.
The study enrolled 50 patients with previously untreated, PD-L1 positive (≥1% expression) advanced or metastatic NSCLC who were eligible to receive pembrolizumab monotherapy as standard of care. Key eligibility criteria included EGFR and ALK wild-type status, no prior systemic therapy for metastatic disease, and measurable disease by RECIST v1.1 criteria. Patients with symptomatic brain metastases or those requiring high-dose corticosteroids were excluded.
The median age of participants was 65.5 years, with 56% aged 65 or older. The cohort was 52% female and predominantly White (82%). Most patients had stage IV disease (98%), adenocarcinoma histology (76%), and high PD-L1 expression ≥50% (96%). The majority (76%) had an ECOG performance status of 1.
Patients received standard pembrolizumab dosing (200 mg or 2 mg/kg intravenously every 3 weeks). Serial blood samples for ctDNA analysis were collected at baseline and after cycles 1 and 2 of therapy. Radiographic response assessments were performed every 6 weeks initially.
The primary objectives were to establish the concordance of ctDNA molecular response with radiographic response by RECIST v1.1 criteria and define the optimal timepoint for assessing ctDNA response. Secondary endpoints included progression-free survival (PFS), overall survival (OS), and time to ctDNA response.
A tumor-agnostic, white blood cell-informed next-generation sequencing approach was used to analyze plasma ctDNA. Molecular response was defined as clearance of the maximum mutant allele fraction (maxMAF) of tumor-derived mutations in plasma after two cycles of therapy.
Of the 50 enrolled patients, 45 were evaluable for ctDNA response. Ten patients had undetectable ctDNA at baseline, leaving 35 patients (77.8%) with detectable ctDNA for response assessment. The ctDNA molecular response rate was 43% (15/35 patients).
The study met its primary endpoint, demonstrating high concordance between ctDNA and radiographic responses. The sensitivity of ctDNA molecular response for predicting RECIST response was 82% (90% CI: 52-97%) and the specificity was 75% (90% CI: 56.5-88.5%). Similar results were seen using iRECIST criteria.
Importantly, ctDNA response appeared to be a stronger predictor of long-term outcomes compared to radiographic response. Patients achieving ctDNA molecular response had significantly longer PFS (median 5.03 vs 2.6 months) and OS (median not reached vs 7.23 months) compared to those with molecular progression.
The median time to ctDNA response was 2.1 months. The depth of ctDNA reduction correlated with radiographic response, with an area under the receiver operating characteristic curve of 0.81 at cycle 3.
Interestingly, ctDNA dynamics provided additional prognostic information beyond radiographic assessment, particularly in patients with stable disease by imaging. Among patients with radiographically stable disease, those achieving ctDNA clearance had markedly better outcomes.
The study had several limitations. The sample size was relatively small at 50 patients. The overall response rate of 32% was lower than expected based on previous pembrolizumab trials, which may have reduced statistical power. As anticipated, about 20% of patients had undetectable ctDNA at baseline, limiting the evaluable population. The trial was not designed to assess the lead time between ctDNA and radiographic responses.
Despite these limitations, the authors concluded that ctDNA molecular response after two cycles of pembrolizumab appears to be a robust early predictor of outcomes in advanced NSCLC. They suggest it could potentially be used to identify patients unlikely to benefit from pembrolizumab monotherapy who may be candidates for treatment intensification.
The findings have been incorporated into the design of the planned second stage of the BR.36 trial. This adaptive phase will use ctDNA detection after two cycles of pembrolizumab to identify high-risk patients, who will then be randomized to either continued pembrolizumab or intensified therapy with the addition of chemotherapy.
If validated in larger studies, the use of early ctDNA response assessment could have significant clinical implications for the management of advanced NSCLC. It may allow for more personalized treatment approaches, potentially sparing some patients from ineffective therapy while escalating treatment earlier for others.
The ability to obtain this prognostic information from a simple blood test, weeks before traditional radiographic assessment, is particularly appealing. It aligns with the growing interest in using liquid biopsies to guide precision oncology approaches.
However, several questions remain before ctDNA response can be widely implemented in clinical practice. The optimal definition of molecular response, the best method for measuring circulating tumor burden, and the generalizability of these findings to other immunotherapy regimens and cancer types all require further study.
Additionally, while the tumor-agnostic sequencing approach used in this study has advantages in terms of broad applicability, tumor-informed assays may potentially offer greater sensitivity. Head-to-head comparisons of different ctDNA analysis methods will be important to standardize approaches.
The ongoing second stage of the BR.36 trial will provide valuable data on whether treatment intensification based on early ctDNA response actually improves outcomes. If positive, it could pave the way for ctDNA-guided adaptive immunotherapy protocols.
More broadly, this study adds to the growing body of evidence supporting the use of ctDNA analysis as an early endpoint in immuno-oncology trials. It may allow for more rapid assessment of new therapies and combinations, potentially accelerating drug development.
For clinicians, while it is too early to incorporate ctDNA response assessment into routine practice, these findings suggest it may soon play an important role in treatment decision-making for advanced NSCLC. Oncologists should be prepared for the integration of serial liquid biopsies into standard workflows in the coming years.
In conclusion, this well-designed prospective study provides compelling evidence that early ctDNA dynamics can capture the heterogeneity of responses to pembrolizumab in NSCLC more accurately than conventional imaging. If validated, it represents an important step towards more precise, biomarker-guided immunotherapy. Further research in this rapidly evolving field is eagerly anticipated.
References
Anagnostou V, Ho C, Nicholas G, et al. ctDNA response after pembrolizumab in non-small cell lung cancer: phase 2 adaptive trial results. Nat Med. 2023;29(10):2559-2569. doi:10.1038/s41591-023-02598-9
