How do we predict which patients with lung cancer will attain favorable clinical outcomes with immunotherapy? Our research focuses on answering this challenging question through assessing the dynamics of circulating tumor DNA (ctDNA) during immunotherapy in patients with metastatic non-small cell lung cancer (NSCLC).
Oncologists use predictive and prognostic biomarkers – clues that guide choice of therapy and predict clinical outcomes, such as PDL1 expression and tumor mutation burden (TMB). These tumor tissue-based biomarkers, however, do not always predict clinical outcomes. In our recent study published in Clinical Cancer Research (https://doi.org/10.1158/1078-0432.CCR-23-1469) our group investigated blood-based characteristics of clinical response and immune-related adverse effects (irAEs) for individuals with lung cancer.
But let’s backtrack for a second. It’s important to note that a sizable fraction of patients with metastatic NSCLC receive an immunotherapy-containing regimen, introducing an unmet need to rapidly and accurately identify which patients are most likely to do well on immunotherapy – and potentially spare them of unnecessary interventions – as well as which patients are at higher risk for disease progression, necessitating further interventions
Physicians conventionally and predominantly use imaging techniques to assess a tumor’s response to treatment. These methods, however, have technical limitations regarding the level of detection, as imaging cannot capture microscopic disease. In addition, imaging may be less effective in capturing heterogenous responses with immunotherapy which can impact patients whose cancer appears to be stable on imaging. The challenge here is identifying the true disease status of these patients.
Liquid biopsies (LB) and analyses of circulating cell-free tumor DNA (ctDNA) have emerged as groundbreaking technologies aiming at addressing and tackling these challenges. These methods offer a minimally invasive alternative to traditional tissue biopsies. But what exactly is ctDNA? As most tumors rapidly divide, necrosis and apoptosis occur. This essentially means that cancer cells die. When that happens most cancer cells “shed” small fragments of their DNA in the bloodstream. Scientists are able to isolate and detect these “cell-free” DNA fragments through LBs. Here’s how it works in simple terms: they take a small sample of blood, separate the liquid part (plasma), and then pick out the tumor’s circulating DNA from it. After that, they sequence it, meaning they zoom in on the ctDNA and read through it, looking for any genetic changes in the building blocks of the DNA, known as mutations.
Through liquid biopsies, we can estimate the cell-free tumor load (cfTL), meaning how much ctDNA was detected in the bloodstream, as well identify specific mutations in ctDNA. Through this process, physicians are able to see whether molecular progression is occurring behind the scenes in patients with radiographic stable disease. This would appear as persistence or rise of ctDNA levels on LB during immunotherapy. Another way this could manifest is as emergence of specific mutations detected on LB that can be characterized as mechanisms of primary or acquired resistance to treatment.
Based on clearance or persistence of cfTL, patients in our study were mainly placed into one of two groups – the molecular response group (mR), or the molecular progressive disease group (mPD). Importantly, our study found that molecular responses were associated with longer survival, highlighting the potential of ctDNA dynamics as a predictive tool for treatment outcomes. Patients in the mR group attained longer overall survival (OS) and progression free survival (PFS) than patients in the mPD group. Interestingly, a correlation was observed between lower ctDNA levels at baseline and more favorable clinical outcomes for patients receiving single-agent immunotherapy.
But what about a negative LB result? What does it mean when ctDNA is not detected? Can we confidently trust this result and de-escalate treatment? Despite the advances made in the technology behind LBs, there are still significant limitations that make physicians air on the side of caution when making therapeutic decisions based on negative LB results.
On the one hand, are the technical limitations of LBs, which mainly revolve around the level of detection, and the specifics behind how ctDNA is isolated and sequenced, which we call “technical” noise. On the other hand, LB results can often be confounded by “biological” noise. This refers to some mutations identified in liquid biopsies that are not truly tumor-derived and could lead to a false positive result. Some of the mutations picked up on LBs do not come from the tumor, but from blood cell precursors. This phenomenon is called clonal hematopoiesis (CH). Understanding which mutations are from the tumor itself is a big challenge that we addressed in our study through additionally sequencing matched normal DNA from white blood cells. This allowed our research team to exclude CH mutations and identify tumor-derived mutations.
Aside from assessing patient outcomes to immunotherapy, we additionally investigated the role of LBs in detecting clues that may predict toxicity. To do so, we evaluated immune cell repertoire dynamics in relation to clinical response and the emergence of immune-related toxicities. Through T-cell receptor (TCR) sequencing, we were able to see which TCR clones were increasing, and which were decreasing between baseline and on-treatment timepoints. These TCR clones were then clustered based on similarity. Significant expansions and regressions of specific TCR clusters were observed in patients who developed immunotherapy toxicity. These patients were also found to have elevated plasma protein expression of pro-inflammatory mediators, both at baseline and during treatment. Monitoring T cell dynamics and plasma proteomic profiles could help identify patients at higher risk of severe toxicities early on, allowing for timely intervention.
Taken together, our research group concludes that despite its limitations, ctDNA-based molecular response is a robust predictor of clinical outcomes in patients with lung cancer receiving immunotherapy and can be particularly informative for patients with stable disease on imaging. We anticipate that using liquid biopsies to longitudinally track ctDNA as well as T cell repertoire monitoring, could enhance clinical decision-making and improve patient outcomes during immunotherapy. Our work has already been implemented in ctDNA-driven clinical trial design for patients with lung cancer receiving immunotherapy (https://doi.org/10.1038/s41591-023-02598-9, NCT04093167).
