https://orcid.org/0000-0003-0105-9760
Abstract
Real-world success rate of liquid and tissue-based comprehensive genomic profiling (CGP) is unknown. We analyzed real-world pan-tumor cohorts that underwent CGP during clinical care via FoundationOne CDx (F1CDx) and FoundationOne Liquid CDx (F1LCDx) to determine tissue and liquid sample adequacy based on tumor type. Pan-tumor presequencing adequacy was high (>90%) by both tissue-based F1CDx (median: 92.3%; range: 88.2%-96.9%) and liquid-based F1LCDx (median: 94.8%; range: 86.6%-96.7%). Similarly, postsequencing analysis revealed that most tissue and liquid samples yielded successful sequencing results with a median sequencing success rate of 97.9% and 98.1% for F1CDx and F1LCDx, respectively. One exception is central nervous system (CNS) tumors, for which F1CDx had dramatically higher sample sufficiency (96.9%) and postsequencing success rate (97.0%) compared with F1LCDx (86.6% and 92.9%, respectively). The pan-tumor median sample-to-success rate was 90.4% (range: 84.8%-94.4%) for F1CDx. The equivalent rate for F1LCDx was slightly higher at 93.2% (range: 80.4%-95.7%). Conversely, when examining the prevalence of F1LCDx results with high tumor fraction (TF≥1%), the sample-to-high TF results rate was dramatically lower (median: 37.7%, range: 2.1% [CNS tumors]-46.0%). In conclusion, except in CNS tumors or when accounting for liquid TF, success rates of F1CDx and F1LCDx are equivalently high. These results may guide informed decision on when to pursue tissue vs liquid testing of patients with cancer.
Introduction
Current guidelines recommend molecular testing of patients with advanced cancer for potential targeted therapy selection. While comprehensive genomic profiling (CGP) of tissue samples may be the gold standard to uncover potentially targetable biomarkers in cancer, liquid biopsy via peripheral blood is emerging as a viable alternative to tissue. The success rate of CGP may depend on several factors, including pre-analytical factors (ie, the type and quality of sample being analyzed such as small biopsies, resections, cytology, and liquid), the specific technology and methods used, and the clinical context in which it is performed.1 For instance, prior reports have suggested higher unsuccessful rates of solid tissue CGP in prostate cancer clinical trials than in standard clinical practice, real-world prostate cancer samples, in which most inadequate tissue prostate cancer samples did not meet pre-analytical criteria to advance into sequencing.2 Alternatively, liquid biopsies can be a valuable option in cases where solid tissue biopsies are difficult to obtain, and in prostate cancer, liquid CGP has been shown to be informative when tissue CGP was unsuccessful.2 Currently, pan-tumor success rates of tissue- vs liquid-based CGP are unknown.
Methods
Cohort selection
With IRB approval and waiver of HIPAA authorization (Western IRB Protocol No. 20152817), we analyzed real-world pan-tumor cohorts of tumors who underwent standard-of-care CGP during clinical care via FoundationOne CDx (F1CDx) and FoundationOne Liquid CDx (F1LCDx). Any tissue or liquid sample accessioned at the Cambridge, Massachusetts, Foundation Medicine laboratory location in 2022 was included. CGP was performed via hybrid capture-based next-generation sequencing (NGS) as a part of routine clinical care (Foundation Medicine, Inc., Cambridge, MA) with either the tissue-based F1CDx3 or the blood-based F1LCDx.4 Assay methods have been previously described.3,4
ctDNA tumor fraction
The level of ctDNA TF shed for each specimen was quantified as previously described.5 The primary analyses of this study treated TF as a binary variable, distinguishing only whether a specimen had TF ≥ 1% or TF < 1%, as this cutpoint had previously shown utility in informing negative liquid biopsy results.6,7
Results
Of all tissue samples received, 7.5% (3366 out of 44 608) were inadequate prior to DNA sequencing due to various reasons such as insufficient volume, insufficient % tumor nuclei, insufficient DNA yield following extraction, and inadequate quality control metrics for library construction or hybrid capture. As a comparison, of all liquid samples received, 3.0% (907 out of 30,561) were insufficient prior to DNA sequencing. Despite some differences, pan-tumor presequencing sample sufficiency was high (>90%) by both tissue-based F1CDx (median: 92.3%; range: 88.2%-96.9%; Supplementary Figure S1) and liquid-based F1LCDx (median: 94.8%; range: 86.6%-96.7%; Supplementary Figure S3).
We next examined postsequencing tissue and liquid success metrics. Similar to presequencing sufficiency, postsequencing analysis revealed that the vast majority (>97%) of both tissue and liquid samples entering sequencing yielded successful postsequencing results in the form of pass or qualified reports at 97.4% (40 183 out of 41 242) and 98.3% (28 604 out of 29 071), respectively. The pan-tumor median sequencing success rate of F1CDx and F1LCDx were 97.9% and 98.1%, respectively (Supplementary Figures S2 and S4). Comparison of successful postsequencing results of specific tumor types revealed that most tumor types had similar pass/qualified report rates of F1CDx (range: 96.2%-98.9%) and F1LCDx (range: 92.9%-99.0%; Figure 1; Supplementary Figures S2 and S4). One notable exception is central nervous system (CNS) tumors, for which F1CDx tissue profiling had dramatically higher sample sufficiency (96.9%) and postsequencing success rate (97.0%) compared with F1LCDx (86.6% and 92.9%, respectively; Supplementary Figures S1-S4).
Rate of successful reporting by liquid biopsy and tissue biopsy by disease group. Top: All disease groups show high success rates via both tissue and liquid. Bottom: Inset showing differences between disease groups. Success is defined as generating a report from a submitted sample with a Pass or Qualified quality control status.
The pan-tumor median sample-to-success rate was 90.4% (range: 84.8%-94.4%) for F1CDx. The equivalent rate for blood-based F1LCDx was slightly higher at 93.2% (range: 80.4%-95.7%). Conversely, when examining the prevalence of F1LCDx results with high ctDNA tumor fraction (TF ≥ 1%), the sample-to-high TF results rate was dramatically lower (median: 37.7%, range: 2.1% [CNS]-46.0%; Figure 2).
Rate of generating results via liquid and tissue biopsy accounting for ctDNA tumor fraction in blood samples. Top: all disease groups have higher rate of successful CGP via tissue than successful CGP via liquid biopsy and ctDNA tumor fraction ≥ 1%. Bottom: inset showing individual diseases, with CNS liquid biopsy exhibiting extremely low rate of successfully delivering liquid biopsy results with TF > 1%.
Discussion
Here, we demonstrate that the real-world, pan-tumor CGP success rates for both tissue- and liquid-based samples with FDA-approved assays, F1CDx and F1LCDx are high (ie, >90%). In contrast, most of the previously published data on success rates of molecular testing are from retrospective studies of single centers or consortia and usually on smaller NGS panels and exclusively on tissue samples. In addition, previously reported success rates for clinical trials or real-world samples of select tumor types, such as prostate and lung cancer, were lower.2,8 A key difference is the prospective nature of these real-world samples that are used in direct patient management as opposed to archival clinical trial samples. Another difference is that the present study takes into account only CGP failure rates, whereas, if the unfeasibility of the tissue biopsy, procedural failures, and tissue inadequacy are also taken into account, tissue failure rates may be higher (ie, up to 43% for lung non–small cell carcinoma).8 Furthermore, recent optimization of our tissue handling for F1CDx, such as the implementation of precision needle-punch tumor enrichment for real-world samples and educating external pathology laboratories on how to process tissue for molecular testing, may have contributed to the high success rate of tissue CGP testing at Foundation Medicine.9
While overall CGP success rates were high, some tumor types may have better success rates depending on the sample type and assay platform. Additionally, tissue CGP delivered results at a higher rate than liquid CGP when only considering samples with high ctDNA TF. These data support the notion that both tissue- and liquid-based CGP as reasonable matrix types for molecular testing of routine real-world specimens. Additionally, liquid CGP may be an acceptable alternative when tissue CGP does not result in a successful report or when tissue is not available, especially in the context of an elevated TF result. Consistent with this notion, a previous study demonstrated that in patients with lung cancer with tissue CGP results, liquid CGP detected no additional patients with NCCN biomarkers, whereas liquid CGP added meaningfully in patients getting more limited, non-CGP tissue testing.10 In conclusion, the results of our study may guide informed decision to pursue tissue vs liquid testing of patients with cancer derived from both real-world and clinical trials and may represent a benchmark for future studies in this field.
Supplementary material
Supplementary material is available at The Oncologist online.
Conflicts of interest
All authors are full-time employees of Foundation Medicine, Inc., a whole subsidiary of Roche, and are stockholders of Roche.
Data Availability
The data underlying this article are available in the article and its online Supplementary Material.
