Identification of Actionable Genomic Alterations Using Circulating Cell-Free DNA

Nora S. Sánchez, PhD; Michael P. Kahle, PhD; Ann Marie Bailey, PhD; Chetna Wathoo, MD; Kavitha Balaji, PhD; Mehmet Esat Demirhan, MD, Dong Yang, PhD; Milind Javle, MD; Ahmed Kaseb, MD; Cathy Eng, MD; Vivek Subbiah, MD; Filip Janku, MD, PhD; Victoria M. Raymond, MS; Richard B. Lanman, MD; Kenna R. Mills Shaw, PhD; Funda Meric-Bernstam, MD


JCO Precis Oncol. 2019;2019(3) 

In This Article


Among the 118 patients with 6 months or longer follow-up in our study, only 13 (11%) with HPCA alterations underwent genotype-matched targeted therapy on the basis of cfDNA testing. Although we predicted higher enrollment because of cfDNA turnaround time, this is in line with our previous report from genomic testing with a targeted tissue panel, which was also 11%.[21] However, we have previously found that with routine implementation of decision support with variant annotation and trial matching, trial enrollment is much higher for patients with actionable alterations (20.6%).[18]

We retrospectively reviewed clinical management after cfDNA testing to understand the causes of low trial enrollment rates. On the basis of at least 6 months of follow-up, a major contributor to not acting on a genomic alteration was the PS of patients at the time of trial consideration. Reports from the University of California, Los Angeles, also found that biopsy samples required for trial eligibility led to delays and decline in PS, which resulted in reduced enrollment by approximately one half.[22] Similarly, a review of 55 non–small-cell lung cancer trials at Princess Margaret Hospital reported a significant reduction in enrollment because of declining PS or insufficient tissue biopsy sample material when required.[23] In addition, we evaluated a novel and important factor—initiation of alternative therapy when ordering cfDNA testing—that may have reduced the number of trial-eligible patients because of a decline of PS during next-line therapy. Of note, cfDNA testing has a short enough turnaround time to facilitate selection of next-line therapy (a median of 7 calendar days now is achieved at the high-volume laboratory used herein).[3,24] Thus, its use at the point of care may increase its clinical utility before a decline in PS. Furthermore, outside select diseases with drug approvals linked to a genomic marker, often genomic testing is initiated late in the treatment course. Testing earlier, when PS is better, also is likely to optimize patient outcomes from investigational therapies.

There have been several reports on cfDNA testing utility.[3,25–27] Laufer-Geva et al[28] reported that treatment decisions were changed in 23% to 32% of patients with non–small-cell lung cancer dependent on the clinical scenario. Of note, we accrued very few patients with lung cancer and accrued predominantly tumor types without standard-of-care, genomically informed treatment options. This may have affected overall actionability rates as well as the number of patients who received genomically matched therapy.

Previous studies have reported higher concordance rates for cfDNA results and tumor testing.[24,29] Leighl et al[30] reported that in untreated nonsquamous cell carcinoma, Food and Drug Administration–approved target (EGFR, ALK, ROS1, BRAF) concordance was greater than 98.2%, with a 100% positive predictive value for cfDNA versus tissue (34 of 34 patients with EGFR-, ALK-, or BRAF-positive mutations). Our study was not designed to compare cfDNA and tissue testing; thus, it had inherent limitations to investigate this issue, including intervening time between sample collections, intervening treatments and resulting tumor evolution, tumor heterogeneity of solid tumor testing versus global representation of primary and metastatic sites from cfDNA, and tumor type dependence on the extent of cfDNA shedding. However, cfDNA testing reported potentially clinically relevant alterations that otherwise may have been missed because solid tumor testing was not attempted or not technically feasible because of lack of tissue or did not have sufficient coverage.

Our study has other limitations. First, as previously discussed, we prospectively enrolled patients into a study that offered cfDNA testing and retrospectively assessed clinical utility. Although there may have been greater use of cfDNA results for treatment selection if patients were prospectively tested and treatment initiated after results, we believe that our study design gives insight into clinical practice patterns. Second, we enrolled patients with a variety of tumor types. This design allowed us to determine that the detection of cfDNA and the detection of actionable alterations vary by tumor type. However, the variable histologies made some clinical outcome assessments challenging. Third, action-ability was assessed at the time of testing by a designated decision support team, but trial matching was performed automatically on the basis of gene-drug associations. Where there were trial matches, additional clinical exclusion criteria possibly made patients ineligible for selected trials. Finally, for actionability, we focused on positive predictive biomarkers only. cfDNA also can identify actionable resistance mutations, and including those would have potentially enhanced clinical utility.[3]

In summary, we demonstrated that cfDNA testing detects actionable alterations in patients with advanced cancers across a variety of tumor types. Additional study is needed to determine how to enhance clinical utility and determine optimal testing timing, especially in patients interested in investigational therapeutics.