Researchers at UC Davis, Genentech and Foundation Medicine are the first to show that a blood-based test to assess tumor mutational burden (TMB) accurately identifies non-small cell lung cancer (NSCLC) patients who could benefit from immunotherapies called checkpoint inhibitors. The blood test offers a much less invasive and more repeatable alternative to tissue testing. The study was published online today in Nature Medicine.
“We wanted to know if we could transfer this TMB assay from tissue to blood,” said David Gandara, who directs the Thoracic Oncology Program at the UC Davis Comprehensive Cancer Center and is first author on the paper. “We succeeded, establishing a TMB level in blood that correlates well with similar levels in tissue and was associated with favorable patient outcomes.”
Checkpoint inhibitors take the molecular brakes off T cells, allowing them to attack tumors. However, they work best in patients who exhibit certain tumor biomarkers. One of these is the PD-L1 protein. More recent is tumor mutational burden – the number of mutations found in specific genomic sequences in tumor cells of an individual patient. Patients with higher TMB are often better candidates for immunotherapy.
Translating these findings into clinical practice is now feasible. The initial laboratory research methods used to identify these biomarkers, such as exome sequencing, take a long time and are not always scalable for clinical care. In addition, as many as 30 percent of NSCLC patients have too little tumor tissue to facilitate these tests. A fast, minimally invasive blood test would be the ideal solution.
“There are patients for whom the biopsy is inadequate from the start, or the tissue is used for routine pathology and we don’t have enough tissue left to do either genomic testing or tissue TMB,” Gandara said. “If we can do it in blood in one test, that offers many advantages for patients who have had an inadequate biopsy.”
In addition, because it’s much less invasive, a blood test could be repeated to determine if a treatment is effective or provide an additional option for patients who might not tolerate a traditional biopsy in the first place.
To determine whether blood could produce TMB results as well as tumor tissue, the researchers examined more than a thousand blood samples from patients with advanced NSCLC (two or more lines of treatment) in two studies, OAK and POPLAR.
This retrospective study compared these blood samples with tumor tissue and found a strong, though not perfect, TMB correlation between the two. This was not unforeseen, as tumor tissue is heterogeneous, and the blood test is actually more sensitive.
Despite these differences, the blood test performed well, consistently predicting which patients would benefit – with improved response and progression-free survival – from the PD-L1 inhibitor atezolizumab (Tecentriq). The assay proved both accurate and reproducible.
Development of a bTMB assay and correlation with tTMB
a, A schematic depicting the development of the bTMB assay (see Supplementary Fig. 1 for details regarding sample flow). Briefly, ctDNA sequencing data were obtained and a bTMB score was determined by counting all SNVs with allele frequencies of ≥0.5%, excluding driver mutations and identifiable SNPs. POPLAR samples were used as a training set and OAK samples as a validation set. b, Pairwise comparison of tTMB and bTMB in patients from the POPLAR and OAK studies with adequate data (SNP matched and passed quality control) from both platforms (N = 259 patient samples). c, A comparison of TMB (the number of mutations) calculated using the F1 assay versus the bTMB assay from split ctDNA samples (N = 69 cell lines). d, The PPA represents the fraction of the somatic variants from the FACT assay that was also detected in the bTMB assay. e, The VAF for matching variants detected in the overlapping regions for FACT and bTMB. For undetected variants, the VAF is 0.0. Known artefacts are excluded. The left panel (N = 202) represents the full distribution of matched variants in bTMB and FACT; the right panel shows a close up of low VAFs (N = 62). f, A comparison of the TMB count between tissue and blood samples is shown for patients with a high (>30) total mutation count derived from tissue. Samples are arranged in order of increasing tTMB. Subs, substitutions. g, A Venn diagram showing the overlapping (59%) and mutually exclusive (25% tissue only; 26% blood only) variants shown individually in f.
This blood-based approach seems poised to move rapidly into the clinic. Foundation Medicine is now seeking FDA approval to incorporate it into their FoundationACT (FACT) liquid biopsy. In addition, interim data from the prospective BFIRST study presented at the recent ASCO conference confirmed that blood samples are a viable way to test TMB.
“This blood TMB assay, and the associated FACT test, have already received Breakthrough Designation, the first step towards full FDA approval,” Gandara said. “This is very important for practicing physicians and for patients. What they hear about as research one day could very shortly be standard of care.”
Source – UC Davis