Results from the largest prospective analysis that evaluated differences between tissue and circulating tumour DNA (ctDNA) next-generation sequencing (NGS) with a large cancer gene panel and comparison of their respective impact in terms of molecular tumour board recommendation and treatment guidance based on actionable alterations classified according ESCAT for patients with advanced solid tumours are published by Prof. Antoine Italiano and colleagues from the French National Center for Precision Medicine on 16 September 2022 in the Annals of Oncology. They analyzed data from December 2020 to November 2021 among 1021 patients who were enrolled in two ongoing precision medicine studies, BIP and STING. Findings provide sufficient evidence of clinical utility of ctDNA NGS approach for capturing actionable alterations in patients with advanced cancer as a complementary or even an alternative to a tissue-based strategy when tissue is not available.
In a letter to the editor, the authors wrote that tissue-based genomic profiling is still considered as a gold standard to assist decision-making for genomics-driven treatment in patients with advanced cancer. However, it has several limitations including screening failures due to limited tissue availability, and inability to capture intratumour spatial and temporal heterogeneity, which may impair accurate treatment selection. Furthermore, tumour biopsy is challenging.
The authors explained that NGS of ctDNA is an increasingly used method for the genomic profiling of cancer and has several advantages in comparison with NGS of tissue biopsies, as a non-invasive method, easily achievable and repeatable, and representative of the whole molecular landscape of the patient’s tumour. Several studies have demonstrated the potential of ctDNA to detect genomic alterations at high accuracy compared with tissue analysis. However, no study has comprehensively evaluated the differences between tissue and ctDNA NGS results with a large cancer gene panel and compared their respective impact in terms of molecular tumour board recommendation.
Among enrolled patients in two ongoing precision medicine studies, BIP and STING, the most frequent tumour types were colorectal (13%), non-small cell lung cancer (13%), breast (12%), prostate (8%) and pancreatic adenocarcinoma (6%). Genomic analysis was performed by using the FoundationOne®Liquid CDx assay and the FoundationOne®CDx covering analysis of 324 genes, including calculation of tumour mutational burden (TMB) and microsatellite instability (MSI). The results were discussed during a weekly molecular tumour board and actionable alterations were classified according ESCAT classification.
Median time difference between archival tumour acquisition and ctDNA collection for NGS testing was 15.0 months. Median time elapsed between request and assay results was 12 days for ctDNA and 28 days for tissue. Testing failure with no result was 15% for tissue and 3.9% for ctDNA (p < 0.001). Overall, 824 patients (81%) had evaluable results for both tissue and ctDNA sequencing.
After exclusion of genes univocally associated with Clonal Hematopoiesis of Indeterminate Potential (CHIP), the five most frequent cancer-related alterations identified through ctDNA sequencing were: TP53 (16.8%), ATM (7.8%), APC (6.1%), KRAS (4.3%) and PIK3CA (3.9%) mutations. Most frequent cancer-related alterations through tissue sequencing were: TP53 mutation (11.5%), APC mutation (5%), KRAS mutation (4.3%), CDKN2A copy number variation (3.9%) and PIK3CA mutation (3.81%).
Proportion of patients with a higher number of cancer alterations identified in ctDNA compared with tissue increased in parallel with the time elapsed between the tissue and ctDNA sampling. Among patients whose results could be evaluated for both tissue and ctDNA sequencing, the proportion of genes with copy number variations (CNVs) was higher in tissue than in liquid (47% versus 22%). In contrast, the proportion of mutations and rearrangements was higher in liquid (17% versus 13% and 36% versus 21%, respectively). In total, 58 patients (7%) had non-detectable ctDNA alterations despite cancer-related alterations identified in tissue.
Regarding CHIP, the study team found 409 patients (42%) with at least one alteration in the genes known to be associated univocally with CHIP and 52 patients (5%) had only genes associated with CHIP in their liquid biopsy. DNA repair genes such as ATM and CHEK2 have been also described as genes associated with CHIP. Their potential actionability was assessed on a case-by-case basis based on allele frequency, the presence or absence of other concomitant CHIP variants, and the presence or absence of other variants associated with tumourigenesis.
MSI status and TMB were evaluable for 97% and 95% of patients through ctDNA versus 90% and 92% through tissue sequencing. A total, 9% of patients had blood-TMB high with a cut-off of 16 mutations/Mb, versus 14% in tissue with a cut-off of 10 mutations/Mb. MSI and TMB status were concordant for 71% and 64% of patients. A patient with a blood-TMB ≥10 mutations/Mb through ctDNA sequencing had a probability to have a TMB ≥10 mutations/Mb through tissue sequencing of 51.9%. This probability was of 43.0% when considering a cut-off of 16 mutations/Mb.
Number of actionable alterations was identical in 42% of cases, whereas it was higher in tissue than in ctDNA. A ctDNA profiling allowed the identification of an ESCAT I/II or III or IV alteration not observed in tissue for 9%, 14% and 6% of patients, respectively. Overall, molecular tumour board recommended a matched therapy for 52% of patients. Such a recommendation would not have been made for 17% of patients without the results of tissue and for 15% of the patients without the results of ctDNA NGS.
In this study, only 10% of all alterations found in ctDNA were CNVs whereas they represented 38% of all alterations in tissue, which is in line with previous studies showing a lower sensitivity of ctDNA for CNV detection. The authors commented that further developments will be key to better differentiate true from false negative results and improve sensitivity for copy number aberrations.