The wide uptake of "liquid biopsy" diagnostics in the care of advanced cancer patients highlights the desire for improved access to tumor allowing accurate tumor genotyping (1). Genotyping of plasma cfDNA is now routine for detection of EGFR driver mutations at diagnosis of NSCLC, or for detection of the EGFR T790M mutation after TKI resistance, and is an emerging approach for the detection of other drivers (HER2 or BRAF mutations, ALK or ROS1 fusions…) (2) or the estimation of tumor mutation burden (TMB) (3). However, the most sensitive plasma genotyping platforms still have a sensitivity of only 70%-80%, such that a negative result requires tissue biopsy confirmation.

The wide uptake of "liquid biopsy" diagnostics in the care of advanced cancer patients highlights the desire for improved access to tumor allowing accurate tumor genotyping (1). Genotyping of plasma cfDNA is now routine for detection of EGFR driver mutations at diagnosis of NSCLC, or for detection of the EGFR T790M mutation after TKI resistance, and is an emerging approach for the detection of other drivers (HER2 or BRAF mutations, ALK or ROS1 fusions…) (2) or the estimation of tumor mutation burden (TMB) (3). However, the most sensitive plasma genotyping platforms still have a sensitivity of only 70%-80%, such that a negative result requires tissue biopsy confirmation. This poses a clinical challenge because negative plasma genotyping is correlated with more limited metastatic spread and lower tumor burden, such that biopsy of these patients may be even more challenging. Because invasive biopsy remains an integral part of the diagnostic strategy, methods are needed for maximizing the yield from these biopsy procedures.

There is a current paradox between the need for large amounts of tissue for multiplex analysis of an increasing number of targetable drivers and markers of response to immune therapy (PD-L1, TMB) and the development of minimally invasive biopsy procedures that results in limited specimens. Up to 25% of patients are thus treated without knowledge of the molecular profile of their tumor (4). In particular, 20% of endobronchial ultrasonography transbronchial needle aspiration (EBUS-TBNA) are rejected from genotyping due to lack of tissue (5) after time and tissue consuming diagnostics steps that are sometimes not required (resistance setting). Circulating tumor DNA is an emerging approach for cancer genotyping but sensitivity is limited to 70-80% (6) by inconsistent tumor shed and low DNA concentrations, so that tissue biopsy is still routine. Also, feasibility of TMB assessment on tissue is only 60% (likely much less on EBUS-TBNA specimens) (7) and approximately 80% in plasma (blood TMB, bTMB) (3).

The presence of cfDNA in several biological fluids and the feasibility of detecting mutations of interest (usually targeting only EGFR) in these fluids (urine, pleural fluid, CSF) have been clearly demonstrated (8-12), while blood is the most widely studied liquid biopsy substrate in advanced NSCLC.

Furthermore, we showed in a proof of concept study, investigating various FNA specimens in a limited numbers of patients that cytology samples' supernatant (usually discarded) is a rich source of DNA. Our results suggest that supernatant free DNA (sfDNA) can be used for baseline and resistance genotyping (13).

Inclusion Criteria:

• Age > 18 years-old

• Patients planned for an EBUS-TBNA for

  1. Suspicion of stage IV lung cancer (PET+ mediastinal node(s)) (Cohort 1)
  2. Stage IV NSCLC with an EGFR, BRAF, HER2, MET mutation or ALK, RET or ROS1 rearranged NSCLC and acquired resistance to targeted therapy (Cohort 2)

• Performance status 0-3

• Informed consent

Exclusion Criteria:

• Refusal to participate
• Patient under legal tutelage

Patients will be recruited in our Department in Toulouse University Hospital, during appointment needed for information and planning of the EBUS-TBNA procedure.