The integration of tumor genomics and high-risk clinicopathologic factors performs better than traditional TNM classifications when predicting recurrence after surgical resection of early-stage lung adenocarcinoma, according to a recent study (JAMA Surg 2020; doi:10.1001/jamasurg.2020.5601).
With a focus on gaining a better understanding of patients who have a complete resection yet still recur, the research team sought to answer the question, "How do we identify these patients upfront?" noted study author David R. Jones, MD, Chief of Thoracic Service in the Department of Surgery at Memorial Sloan Kettering Cancer Center.
"Currently, the only way to do that is through TNM classification," he explained. "With the advent of advances in pathology, next-generation sequencing, and genomic profiling of tumors, it occurred to us that we may be able to leverage these tools to better identify patients with a higher likelihood of developing metastasis following surgery."
This is an important area of exploration, noted Jones, because it can help identify when additional treatments, such as immunotherapy or chemotherapy, are needed, or conversely, it could save patients from unnecessary therapy.
Study Details
This prospective cohort study included patients with completely resected stages I-III lung adenocarcinoma who were treated January 1, 2008 to December 31, 2017 and selected in consecutive samples. Other eligibility criteria included broad-panel next-generation sequencing data with matched clinicopathologic data, and no neoadjuvant therapy.
The research team sought to develop a computational machine-learning prediction model (PRecur) to establish whether the combination of genomic and clinicopathologic features could better predict risk of recurrence when compared with the TNM system. The Cancer Genome Atlas was used as external validation for the PRecur prediction model.
Relapse-free survival, estimated using the Kaplan-Meier approach, was the study endpoint. Cox proportional hazards regression was used to establish associations among clinicopathologic factors, genomic alterations, and relapse-free survival. The predictive ability of the PRecur model was assessed with a concordance probability estimate.
The analysis included 426 patients: 286 were women (87%), 140 were men (33%), and 318 (75%) had stage I disease. The median age at surgery was 69 years. Seventy-five patients (18%) developed a recurrence, 57 (76%) of which were distant.
"Data showed that alterations in SMARCA4 and TP53 and the fraction of genome altered (clinicopathologic-adjusted HR, 1.03; 95% CI, 1.10-1.04; P=.005) were independently associated with relapse-free survival," Jones and colleagues wrote. "The PRecur prediction model outperformed the TNM-based model for prediction of relapse-free survival."
"These findings demonstrate that, when you use TNM staging and add additional pathologic findings and next-generation sequencing of the tumor, you have a much better predictive capability compared to just using standard pathologic TNM staging," noted Jones, who emphasized the importance of the external validation. "This model worked not only at our facility, but also in The Cancer Genome Atlas database. This is a critically important step that confirms this is a viable approach moving forward."
Key Implications
These findings, which validate the predictive power of the PRecur model, have important implications for both physicians and their patients.
"For physicians, it is a step forward in our ability to predict recurrence after complete removal of the tumor in patients with lung cancer. We are moving beyond the size of the tumor and whether or not the lymph nodes were involved," Jones explained. "We are now using state-of-the-art next-generation sequencing to help us identify who has the highest likelihood of recurrence. As a result, in the future, the physician may be confident in recommending or not recommending therapy."
From a patient's perspective, this knowledge can offer comfort, Jones noted. "No one wants intensive treatment if it isn't beneficial," he said. "More precise risk stratification helps us determine if a high-risk patient would benefit from further therapy or increased observation."
Ongoing Research
As research in this area continues, Jones and colleagues are exploring how to predict where a tumor will metastasize.
"Lung cancers often metastasize to the brain, bone, liver, or other lung," he said. "So, we are looking at genomic predictors that could shed light on where the tumor will metastasize. We want to take a more granular view to better understand where specific types of lung cancers tend to spread, which may be very helpful for us in the future."
Reiterating the significance of this approach, Jones noted, "The integration of tumor genomics and clinical pathologic features is an important risk stratification tool that allows us to better predict recurrence after curative intent surgery for lung cancer.
"This will be just one of a number of efforts to better predict who is at greatest risk, which will ultimately allow us to offer additional therapy to patients who will receive the most benefit and withhold treatment when it is unnecessary," he concluded.
Catlin Nalley is a contributing writer.