Rapid detection of mutations in CSF-cfTNA with the Genexus Integrated Sequencer

Abstract

Purpose

Genomic alterations are fundamental for molecular-guided therapy in patients with breast and lung cancer. However, the turn-around time of standard next-generation sequencing assays is a limiting factor in the timely delivery of genomic information for clinical decision-making.

Methods

In this study, we evaluated genomic alterations in 54 cerebrospinal fluid samples from 33 patients with metastatic lung cancer and metastatic breast cancer to the brain using the Oncomine Precision Assay on the Genexus sequencer. There were nine patients with samples collected at multiple time points.

Results

Cell-free total nucleic acids (cfTNA) were extracted from CSF (0.1–11.2 ng/μl). Median base coverage was 31,963× with cfDNA input ranging from 2 to 20 ng. Mutations were detected in 30/54 CSF samples. Nineteen (19/24) samples with no mutations detected had suboptimal DNA input (< 20 ng). The EGFR exon-19 deletion and PIK3CA mutations were detected in two patients with increasing mutant allele fraction over time, highlighting the potential of CSF-cfTNA analysis for monitoring patients. Moreover, the EGFR T790M mutation was detected in one patient with prior EGFR inhibitor treatment. Additionally, ESR1 D538G and ESR1::CCDC170 alterations, associated with endocrine therapy resistance, were detected in 2 mBC patients. The average TAT from cfTNA-to-results was < 24 h.

Conclusion

In summary, our results indicate that CSF-cfTNA analysis with the Genexus-OPA can provide clinically relevant information in patients with brain metastases with short TAT.

Introduction

Brain metastases most commonly arise in patients with lung cancer, breast cancer, and melanoma. Brain metastases are generally associated with poor quality of life and pose distinctive clinical challenges [1, 2]. Precision medicine and the use of targeted therapies in patients with brain metastases rely on the characterization of genomic alterations in the tumor in order to implement specific and precise therapeutic strategies. However, in patients with brain metastases, limitations to this approach include genetic heterogeneity between primary tumor and brain metastasis [3], challenges of obtaining tumor tissue from brain lesions, and the relatively slow turn-around time (TAT) of standard next-generation sequencing (NGS) analysis.

Circulating tumor DNA (ctDNA) is released into body fluids as a result of tumor cell content shedding [4]. Therefore, analysis of ctDNA can provide an inclusive genomic landscape addressing inter-tumor and intra-tumor heterogeneity which would be difficult to achieve with tissue analysis [4]. However, studies have shown that the detection of ctDNA from plasma for genetic characterization of central nervous system (CNS) tumors is challenging [5, 6]. Instead, cerebrospinal fluid (CSF) can present a better source of ctDNA for detecting mutations derived from CNS tumors due to its proximity to the brain parenchyma [7-12]. Moreover, previous studies suggest an increased sensitivity of CSF-ctDNA over CSF cytology for evaluating patients with CNS tumors [13-15].

At present, the advancement of NGS has made the simultaneous screening of a wide range of gene alterations in a single assay possible. However, the TAT of standard NGS assays can be a limiting factor in the timely delivery of genomic information for clinical decision-making [16]. The Genexus^™ Sequencer automates the NGS workflow delivering results in less than 24 h and the Genexus Oncomine Precision Assay (OPA) evaluates somatic mutations, copy number variations (CNV), and fusions in hotspot regions of 50 cancer-related genes (Supplementary Table 1). In this study, we evaluated genomic alterations in 54 CSF samples from patients with metastatic lung and breast cancer to the brain using Genexus OPA. We also evaluated the sensitivity of the assay and concordance rate of the genomic alterations detected in CSF with matched tumor tissue.

Methods

Patient samples

We evaluated genomic alterations in cell-free total nucleic acids (cfTNA) including cell-free DNA (cfDNA) and cell-free RNA (cfRNA) obtained from 54 CSF samples. A total of 32 CSF samples from 21 patients with metastatic lung cancer (mLC) and 22 CSF samples from 12 patients with metastatic breast cancer (mBC) to the brain were included in the study. The cohort included 19 females and 14 males. The patients’ ages ranged from 40 to 79 years. There were three patients with mLC and six patients with mBC with multiple samples collected at different time points. When available, the genomic profile of tumor tissues was compared with the genetic alterations detected in cfTNA.

Determination of LMD

The presence of leptomeningeal disease (LMD) was determined with contrasted brain and spine MR scans. Leptomeningeal disease was defined as metastatic disease progression into the leptomeninges, evidenced by localized or diffuse areas of nodular enhancement. The basal cisterns, cerebellar folia, cerebral sulci, cranial nerves, and spine were carefully evaluated to detect the presence of leptomeningeal disease.

CSF processing

CSF was collected in a sterile tube found in the lumbar puncture kit (Care Fusion, IL, USA). CSF was centrifuged at 4 °C, 1000×g for 10 min. The supernatant was transferred to a sterile Eppendorf tube. Subsequently, CSF was centrifuged a second time at 4 °C, 1000×g for 10 min. The supernatant was transferred to sterile cryotubes and stored at − 80 °C immediately until use.

Extraction of cfTNA

cfTNA was extracted using QIAamp Circulating Nucleic Acid (Qiagen following the manufacturer’s recommenda-tions. cfDNA was quantified using Qubit DNA HS kit (Life Technologies Corporation). The starting volume of CSF for extraction ranged from 0.3 to 6 ml. The cfDNA concentration ranged from 0.1 to 11.2 ng/μl.

Targeted next generation sequencing assay

The Genexus OPA was used to analyze CSF-cfTNA with the Ion Torrent^™ Genexus^™ Integrated Sequencer. Library preparation, templating, sequencing, and data analysis of cfTNA extracted from CSF were conducted in the instrument with default parameters for liquid biopsies. A maximum of four samples were loaded in the Genexus chip in one run with an expected 12–15 million reads per sample. The amount of cfDNA input ranged from 2 to 20 ng (input recommended by the manufacturer is 20 ng). The sequencing data and the results were mapped and analyzed using on-instrument Genexus software. The OPA evaluates somatic mutations, CNV, and fusions in hot-spot regions of 50 cancer-related genes. The system allows nucleic acid-to-result workflow in less than 24 h. The alterations detected by the OPA are listed in Supplementary Table 1. We used default quality control (QC) parameters set by the instrument which include the Median of the Absolute values of all Pairwise Differences (MAPD) between 0 and 0.4 and detection of at least two out of seven internal RNA control targets. Additional QC parameters monitored include average base coverage, uniformity of base coverage, molecular coverage, total reads, and mapped reads (Supplementary Table 2).

Results

Sample preparation for cfTNA analysis with Genexus OPA

cfDNA concentrations range from 0.1 to 11.2 (ng/μl). The average amount of DNA input used for Genexus OPA was 13.617 ng (range 2–20 ng). For the median cfTNA input of 15 ng, the median of total reads reported was 9,506,476 with median mapped reads of 9,079,925. The overall average and median base coverage depth were 29,599× and 31,963× respectively (range from 104.3 to 60,197×). The uniformity of base coverage showed an average of 96.94% with a median value of 99.39%. Due to low amount of cfTNA input, a subset of samples had molecular coverage < 500×. The molecular coverage ranged from 0 to 3701× with an overall average of 729× and a median value of 359×. Eleven samples had molecular coverage < 100×. Among these, no mutations were detected in 9/11 samples. The two samples for which variants were detected (2/11) were from the same patient #5 (samples 5-1 and 5-3 respectively; Table 1) with mBC. The mutations detected were 2 ESR1 (D538G) and 1 PIK3CA (E545K) with median molecular coverage of 81 and 62, respectively. Our results indicate that the mutation detection by the Genexus is significantly influenced by the cfDNA input. The summary of Genexus OPA parameters is shown in Supplementary Table 2.

Table 1.

Patient information, QC status, and detected mutations of the mLC and mBC samples

Sample ID	Sample type	Sex	Age	CSF volume (ml)	DNA input (ng)	QC Status	Mutations detected in CSF- cfTNA (CNV or MAF)	LMD status	PM status
1-1	mLC	Male	64	0.3	2	RNA-fail, DNA-pass	No mutations detected	Yes	Yes
1-2	mLC	Male	64	6	15.96	Pass	PTEN CNV deletion (0.52) CDKN2A CNV deletion (0.31)	Yes	Yes
1-3	mLC	Male	64	6	20	Pass	PTEN CNV deletion (0.39) CDKN2A CNV deletion (0.16)	Yes	Yes
1-4	mLC	Male	64	6	20	Pass	PTEN CNV deletion (0.5) CDKN2A CNV deletion (0.39)	Yes	Yes
1-5	mLC	Male	64	4.5	20	Pass	No mutations detected	Yes	Yes
1-6	mLC	Male	64	1.7	8.17	Pass	No mutations detected	Yes	Yes
2-1	mLC	Female	53	1.8	10.14	Pass	No mutations detected	Yes	No
2-2	mLC	Female	53	1.7	13.14	Pass	No mutations detected	Yes	No
2-3	mLC	Female	53	4	20	Pass	No mutations detected	Yes	No
2-4	mLC	Female	53	5	4.29	Pass	No mutations detected	Yes	No
3-1	mLC	Male	40	5	20	Pass	EGFR E746_A750del (24.7%)	Yes	No
3-2	mLC	Male	40	5	20	Pass	EGFR E746_A750del (30.7%)	Yes	No
3-3	mLC	Male	40	4.2	20	Pass	EGFR E746_A750del (33.7%)	Yes	No
3-4	mLC	Male	40	5	20	Pass	EGFR E746_A750del (47.8%)	Yes	No
4	mLC	Female	67	2	2.2	RNA-pass, DNA-fail	No mutations detected	Yes	Yes
5	mLC	Male	74	1.8	2.8	Fail	No mutations detected	No	No
6	mLC	Male	63	4.3	4.5	RNA-pass, DNA-fail	No mutation detected	No	Yes
7	mLC	Male	61	1.7	7.03	RNA-pass DNA-fail	No mutations detected	No	Yes
8	mLC	Female	59	4.4	8.5	Fail	TP53 R273L (26.3%)	Yes	Yes
9	mLC	Male	65	4.3	10.26	Pass	FGFR3 F384L (86.4%) EGFR E746_A750del (46.7%) EGFR T790M (13.7%) CTNNB1 S37C (8.6%) EGFR C797S (8%)	No	Yes
10	mLC	Male	71	1.7	11.088	Pass	No mutations detected	No	Yes
11	mLC	Female	74	5	13.59	Pass	ERBB2 Y772_A775dup (12%) TP53 D281H (4.2%)	No	Yes
12	mLC	Male	57	5	13.9	Fail	No mutations detected	No	Yes
13	mLC	Male	57	2	15.6	RNA-fail, DNA-pass	KRAS G12D (2.7%)	Yes	Yes
14	mLC	Male	60	1.6	15.75	Pass	TP53 C135F (37%) ERBB2 Y772_A775dup (32.4%) TP53 C135S (1.3%)	No	Yes
15	mLC	Male	68	3.5	16	Pass	No mutations detected	No	Yes
16	mLC	Female	68	2.8	16.72	Pass	No mutations detected	No	Yes
17	mLC	Female	76	1.7	17.44	Pass	TP53 V216M (3.2%)	No	Yes
18	mLC	Male	65	1.8	18.64	RNA-pass, DNA-fail	No mutations detected	No	Yes
19	mLC	Male	79	3.4	20	Pass	CDK4 R24L (68.6%)	Yes	Yes
20	mLC	Male	62	4.25	20	Pass	No mutations detected	Yes	No
21	mLC	Female	70	2.8	20	Pass	TP53 R280I (29.8%)	Yes	Yes
1-1	mBC	Female	63	3.6	20	Pass	No mutations detected	Yes	No
1-2	mBC	Female	63	3.9	5.7	Pass	No mutations detected	Yes	No
2-1	mBC	Female	60	1.7	20	Pass	PIK3CA H1047R (84.2%) TP53 R280T (92.7%) ERBB2 CNV Amp (12.12), PIK3CA CNV Amp (5.22)	Yes	No
2-2	mBC	Female	60	3	8	Pass	PIK3CA H1047R (83.1%) TP53 R280T (71.6%) ERBB2 CNV Amp (9.44), PIK3CA CNV Amp (4.22)	Yes	No
2-3	mBC	Female	60	5	9.5	Pass	PIK3CA H1047R (34.7%) TP53 R280T (14.9%) ERBB2 CNV Amp (3.81)	Yes	No
3-1	mBC	Female	48	1.3	18.7	DNA-pass RNA-fail	No mutations detected	Yes	No
3-2	mBC	Female	48	5	8	Pass	No mutations detected	Yes	No
4-1	mBC	Female	56	1.8	10.4	Pass	No mutations detected	Yes	Yes
4-2	mBC	Female	56	6	20	RNA-pass DNA-fail	No mutations detected	Yes	Yes
5-1	mBC	Female	45	0.8	9.8	Pass	ESR1 D538G (10.1%) PIK3CA E545K (10.7%)	Yes	Yes
5-2	mBC	Female	45	1.7	16	Pass	No mutations detected	Yes	Yes
5-3	mBC	Female	45	6	15	Pass	ESR1 D538G (11.8%)	Yes	Yes
6-1	mBC	Female	60	5	10	Pass	PIK3CA E545K (1.9%)	Yes	No
6-2	mBC	Female	60	5	6	Pass	PIK3CA E545K (2.5%)	Yes	No
6-3	mBC	Female	60	4.9	4.4	Pass	PIK3CA E545K (3.9%)	Yes	No
6-4	mBC	Female	60	5	8	Pass	PIK3CA E545K (4.5%)	Yes	No
7	mBC	Female	60	1.1	20	DNA-pass RNA-fail	TP53 R175H (93.5%) ERBB2 CNV Amp (26.7)	Yes	Yes
8	mBC	Female	48	4	20	Pass	TP53 R273H (28.6%)	No	Yes
9	mBC	Female	50	5	13.6	Pass	ESR1::CCDC170. E2C8.1	Yes	No
10	mBC	Female	54	0.35	15	Pass	AKT1 E17K (62.7%)	Yes	Yes
11	mBC	Female	43	4.4	20	Pass	PIK3CA E542K (34.8%) PIK3CA H1047Y (25.2%)	Yes	No
12	mBC	Female	47	2.9	9.5	Pass	TP53 R196* (44.1%)	Yes	No

There were 21 patients with mLC. Sample ID 1-1 to 1-6, 2-1 to 2-4, and 3-1 to 3-4 represent the three different patients’ whose CSF was collected at different time points. Patient one had six samples, and patients two and three had four CSF cfTNA samples each from the same patient used for the study. There were 12 patients with mBC. Sample ID 1-1 to 1-2, 2-1 to 2-3, 3-1 to 3-2, 4-1 to 4-2, 5-1 to 5-3, and 6-1 to 6-4 represent the six different patients’ whose CSF was collected at different time points. Patients one, three, and four had two samples, patients two and five had three samples, and patient six had four CSF-cfTNA samples each from the same patient used for the study

LMD leptomeningeal disease, PM status parenchymal metastasis status

Quality control parameters of cfTNA samples

Mutations were detected in 30/54 (55.5%) samples. The great majority of samples (19/24, 79.1%) with no mutations detected had less than the recommended amount of cfDNA input (< 20 ng). Approximately 1/4 (12/54, 22.2%) of the samples failed the default QC parameters set by the instrument. Around 10/12 (83.3%) of the samples with cfDNA input below the recommended amount failed the QC parameters. Among the samples that failed QC parameters, three samples failed both DNA and RNA parameters, five samples failed DNA only parameters and four samples failed RNA only QC parameters. In 3/12 samples that failed the QC parameters, we detected mutations in TP53 and KRAS mutations in two patients with mLC (sample #8 and #13, respectively; Table 1) and TP53 and ERBB2 in one patient (sample #7, Table 1) with mBC. Our results indicate that the mutation detection is significantly influenced by whether or not a sample meets the QC parameters of the OPA assay, which is largely influenced by the cfTNA input (Fig. 1). The detailed patient information, QC status, and the detected mutations of the samples analyzed from patients with mLC and mBC are shown in Table 1.

Fig. 1.

QC parameters for all samples analyzed. A QC status of all the samples included in the study. Forty-two (42/54) samples passed QC and twelve 12/54 samples failed to meet the QC parameters: 3/12 failed both RNA and DNA QC; 5/12 failed DNA QC only, 4/12 failed RNA QC only. B Dot plot showing the correlation between the amount of input DNA and QC. Ten (10/12, 83.3%) samples that failed QC had low input DNA (< 20 ng) and 2 (2/12, 16.7%) samples with input DNA of 20ng failed to meet the QC parameters

Genomic alterations detected in CSF cfTNA by Genexus OPA

The detected mutations in patients with mLC include single nucleotide variants (SNV) in TP53, EGFR, ERBB2, FGFR3, CDK4, CTNNB1 genes (Fig. 2A) and copy number variants (CNV) in PTEN and CDKN2A (Fig. 2B). For patients with mBC, the variants detected include SNV in PIK3CA, ESR1, TP53, AKT1, and CNVs in ERBB2 and PIK3CA (Fig. 3A, B). Mutations with a mutant allele frequency (MAF) of 1% more were reported. For copy number variants, 6 copies and 0.5 copies were used as cut off for amplification and deletion, respectively. The genetic alterations detected in CSF-cfTNA using Genexus OPA are shown in Table 1.

Fig. 2.

The key variants detected within the gene in patients with mLC are shown. A Bar graph showing the mutations detected from CSF cfTNA by Genexus OPA. B and C Data from two patients with multiple samples collected at different time points. B Bar graph showing PTEN and CDKN2A copy number loss in three samples from patient 1 (shown in Table 2). The x-axis depicts the three Samples whose CSF was collected at three different time intervals. Sample 1-3 was collected 8 days after sample 1-2, and sample 1-4 were collected 4 days after sample 1-3. C Line graph representing the MAF of the EGFR p.E746_A750del mutation detected in four Samples from patient #3 (shown in Table 2). The x-axis depicts the four samples whose CSF was collected at four different time intervals. Sample 3-2 was collected 21 days after sample 3-1, sample 3-3 was collected 3 days after sample 3-2 and sample 3-4 was collected 4 days after sample 3-3

Fig. 3.

Summary of the mutations detected in patients with mBC from CSF cfTNA by Genexus OPA. A Bar graph showing the number of point mutations detected. B Bar graph showing the number of Samples with copy number variants detected. C MAF for the PIK3CA E545K mutation was detected in four samples from patient #6 (Table 1). The x-axis depicts the four CSF samples collected at four different time intervals. Sample 6-2 was collected 21 days after Sample 6-1, sample 6-3 was collected 16 days after sample 6-2 and Sample 6-4 was collected 56 days after sample 6-3. D Line graph showing PIK3CA H1047R and TP53 R280T mutations in three samples from patient #2 (Table 1). The x-axis represents the three samples collected at different time points. Sample 2-2 was collected 32 days after Sample 2-1 and sample 2-3 was collected 28 days after sample 2-2. E Bar graph showing ERBB2 and PIK3CA copy numbers in three samples from patient #2 (Table 1). F Bar graph showing ESR1 D538G and PIK3CA E545K mutations in multiple samples from patient #5. The x-axis represents the three samples collected at different time points. Sample 5-2 was collected 4 days after sample 5-1 and sample 5-3 was collected 7 days after sample 5-2

There were three patients with mLC and six patients with mBC with multiple samples collected at different time points. Among these three patients with mLC, we detected PTEN and CDKN2A CNV loss in three samples from one patient, and EGFR exon 19 deletion in four samples from another patient. Six CSF samples collected at different time points were analyzed for patient #1 (Table 1, mLC). However, the CNVs (PTEN and CDKN2A loss) were detected in 3/6 samples (Fig. 2B). Among 3/6 samples with no mutations detected, two had a low amount of input DNA. The MAF of the exon 19 deletion detected in four CSF cfTNA samples from patient #3 (Table 1, mLC) increased over time from 24.7% to 47.8%. The time interval between the four samples collected (samples 3-1 to 3-4) were 21, 3, and 4 days, respectively (Fig. 2C). Also, multiple mutations were detected in Sample #9 (Table 1, mLC) including mutations in FGFR3 (86.4%), EGFR p.E746_A750del (46.7%), EGFR p.T790M (13.7%), CTNNB1 p.S37C (8.6%) and EGFR p.C797S (8%).

With regards to mBC, there were six patients with multiple samples collected at different time points. Among these six patients, we detected mutations in three patients. Among these six patients, we detected a PIK3CA (with increased MAF from 1.9% to 4.5%) in four samples from patient #6 (Table 1, mBC). The time interval between the four samples (Sample 6-1 to 6-4) were 21, 16, and 56 days, respectively (Fig. 3C). For patient #2, we detected PIK3CA and T53 mutations, and ERBB2 and PIK3CA amplification in 3/3 CSF samples. The mutations detected showed decreasing MAF over time (Fig. 3D); The change in the copy number in ERBB2 and PIK3CA from the same patient is shown in Fig. 3E. The time interval between the three samples (Sample 2-1 to 2-3) were 32 and 28 days, respectively. Patient #5 (Table 1, mBC) had three samples collected at multiple time points with mutations in ESR1 and PIK3CA (Fig. 3F). The ESR1 D538G mutation was detected in CSF cfTNA samples 5-1 and 5-3 with increasing MAF from 10.1% to 11.8%. The time interval between the three samples (Sample 5-1 to 5-3) were 4 and 7 days, respectively. However, the mutation was not detected in sample 5-2 with input DNA of 16 ng.

The concordance rate of genetic alterations between tissue biopsy and CSF liquid biopsy

We were able to analyze the concordance between mutations detected in tumor tissue and CSF-cfTNA for 17 patients. Among these, 57.1% (4/7) (2TP53, 3PTEN loss, 3CDKN2A loss, 4EGFR) of the hot spot actionable mutations detected in tumor tissue of mLC patients showed concordance to those detected in CSF-cfTNA. Among patients with mBC, the tumor tissue was analyzed by immunohistochemistry for estrogen receptors (ER), progesterone receptors (PR) and HER2 status but sequencing analysis was not performed as a part of routine clinical care. In 2/5 patients, HER2 amplification was detected in both tumor tissue and CSF-cfTNA. In 16/33 patients’ mutation information from tumor tissue was not available for comparison. The summary of the key variants detected in tumor tissue and CSF cfTNA is shown in Fig. 4.

Fig. 4.

A Summary of mutations detected in tumor tissue and CSF-cfTNA from patients with mLC. B Summary of mutations detected in tumor tissue and CSF-cfTNA from patients with mBC. For mBC, the tumor tissue was analyzed by immunohistochemistry for estrogen receptors (ER), progesterone receptors (PR) and HER2 status but sequencing analysis was not performed as a part of routine clinical care

Overall survival (OS) of patients with positive vs negative CSF-ctDNA results

Kaplan–Meier analysis showed no significant differences in OS based on CSF-ctDNA detection in our patient cohort (Supplementary Fig. 1). However, it is important to note that the analysis is limited by the relatively small sample size. For the patients with mLC, there were 10 patients with CSF ctDNA positive and 11 patients with CSF ctDNA negative results and the median survival was 37 months and 45 months for CSF ctDNA positive and CSF ctDNA negative patients, respectively. For the patients with mBC, there were nine patients with CSF ctDNA positive and three with CSF ctDNA negative results. The median survival was undefined for both CSF ctDNA positive and CSF ctDNA negative mBC patients.

Comparison of clinical/imaging information with the CSF ctDNA results

We also analyzed the CSF ctDNA results in the context of the available MRI information for the subset of patients with multiple CSF samples. Given the retrospective nature of the study the MRIs and CSF collection were performed on different dates, however, the information is useful in providing clinical context for the CSF ctDNA results. The MRI results and CSF ctDNA results for patients with multiple CSF samples are provided as Supplementary Fig. 2A-E.

Discussion

CSF liquid biopsy analysis has the potential to improve the management of patients with brain metastases. However, some limitations of CSF-cfTNA analysis by NGS include slow TAT and limited sensitivity. In this study, we evaluated 54 CSF cfTNA samples from 33 patients with brain mLC and mBC, with a small subset of patients having multiple CSF samples collected at different time points.

Our results show the feasibility of performing rapid CSF cfTNA analysis using the Genexus instrument. One of the main advantages of the assay is the fast TAT. The faster TAT of the Genexus OPA assay (< 24 h) facilitates using CSF cfTNA for screening actionable hot-spot mutations without the need for tissue analysis or while waiting for the results of tumor tissue sequencing, which can take up to several weeks. In addition, the platform automates the entire NGS library preparation workflow from cfDNA to results, with minimal hands-on-time required from the laboratory staff. In this study, we detected mutations in 55.5% of CSF cfTNA samples with 100% specificity (no false positive mutations detected in CSF compared to tissue). Although the recommended input for the OPA-Genexus assay of cfDNA is 20 ng, we identified mutations with as low as 6 ng of cfDNA input. The mutant allele frequency cut-off utilized for reporting a mutation was 1%. It is possible that mutations with a MAF below 1% were missed, this is a limitation of the assay, but the MAF cut-off is important to reduce the chances of false positive results. The rate of mutation detection at ~ 55% is consistent with the rate of mutation detection in previous studies analyzing CSF cfDNA [13]. Even in samples with less than the recommended 20 ng input of cfDNA, we were able to identify mutations in CSF cfTNA. However, the majority of the samples with no mutations detected in CSF cfTNA (19/24 samples) had less than the recommended amount of cfDNA input (< 20 ng). This highlights that one of the limitations of using CSF cfTNA analysis for patient care is that the small amount of cfDNA obtained from some CSF samples precludes the successful detection of mutations in some cases. Nonetheless, our results show that in ~ 55% of CSF samples, clinically relevant mutations can be detected in cfDNA with a TAT of less than 24 h.

The small number of samples with matched tissue sequencing is a limitation due to the retrospective nature of the study. Nonetheless, we observed concordance for the majority of the samples for which we were able to make the comparison. Despite this limitation the results of the study are informative to the neuro-oncology community and demonstrate the feasibility of rapid NGS analysis of CSF-ctDNA, which could potentially benefit patients with CNS tumors. From 17 patients with available brain tumor tissue, the CSF-cfTNA demonstrated great specificity as all the mutations detected in the CSF were also present in tissue. This could be due to the strict criteria utilized in this study for the detection of a mutation to avoid false positive results. This specificity could provide crucial information in the treatment of brain metastases, as the detection on cfTNA mutations like EGFR in mLC or HER2 in mBC could be enough to justify the change of targeted therapy as it has been documented that ~ 20% of brain mBC have discord-ant HER2 status compared to their primary tumor [17, 18]. In addition, no mutations were identified in our previous analysis of CSF cfDNA from patients with no history of cancer, supporting the specificity of CSF-cfTNA analysis by NGS [15].

Among the clinically relevant alterations detected in CSF-cfTNA, we identified HER2 amplification, EGFR exon19 deletion, EGFR T790M, ESR1 D538G, and ESR1::CCDC170 fusion. There were nine samples from patients with mLC with information about tissue mutations and enough cfDNA to meet the 20 ng input recommended for the OPA. We obtained concordant results in seven of these nine samples, with no mutations identified in the remaining two samples. There were 11 mBC samples from patients with HER2 amplification in tumor tissue. HER2 amplification was detected in CSF in 4/11 (36.3%) samples from patients with HER2-amplified tumors. The contributing factor to the apparent discrepancy in HER2 amplification status between tissue and CSF results in seven samples agrees with the discordant HER2 status reported with previous studies [17, 18]. In addition, 7/11 samples did not meet the recommended 20 ng of cfDNA input, which could also contribute to the limited sensitivity of the OPA for detecting HER2 amplification in CSF. Nonetheless, we were able to identify HER2 amplification in ~ 1/3 of CSF samples with no false positive results.

A small number of patients had CSF samples collected at multiple time points. For example, the EGFR exon 19 deletion was detected in four CSF samples from one patient, with increasing MAF in CSF over time. Similarly, PIK3CA and TP53 mutations were detected in multiple samples from a patient with mBC, with decreasing MAF over time. Given that matching brain MRI results are not available, we cannot evaluate the concordance between changes in MAF in CSF and tumor burden. Nonetheless, this result highlights the potential of rapid CSF-cfDNA analysis for monitoring patients, in agreement with previous studies [19, 20].

Endocrine therapy is routinely used in the treatment of patients with estrogen receptor (ER) positive breast cancer [21]. However, the effectiveness of the treatment is limited by the development of resistance [21]. We detected ESR1 D538G and ESR1::CCDC170 alterations, associated with endocrine therapy resistance, in the CSF of two patients with mBC. Similarly, we detected the EGFR T790M mutation, associated with resistance to EGFR inhibitors, in a patient (sample ID #9) with mLC. Our results highlight the potential utility of CSF cfTNA in detecting ESR1 alterations that can influence response to treatments.

All the patients in the study had metastatic lung cancer or metastatic breast cancer to the brain with a subset of them showing evidence of leptomeningeal disease (LMD) as described in Fig. 4. Additional information documenting which patients had parenchymal brain metastases is shown in Table 1. However, in patients with both parenchymal lesions and LMD, it is unclear if the ctDNA in CSF originates from the parenchymal tumor, the LMD or both. There were 9/21 mLC patients diagnosed with (LMD) based on imaging findings. Among the mLC patients with LMD, 6/9 showed mutations in CSF and mutations were detected in 4/12 patients without LMD. In patients with mBC (n = 12), mutations were detected in CSF in 8/11 patients with LMD and 1/1 patient without LMD. This data shows that mutations in CSF-cfTNA can be identified in samples obtained from patients without radiologic evidence of LMD. Kaplan–Meier analysis for the OS of patients with positive and negative CSF-ctDNA results showed no significant dif-ferences in our patient cohort. However, the analysis is limited by the relatively small sample size. It is also possible that some CSF-ctDNA negative patients had potential driver mutations not covered by OPA.

To the best of our knowledge, this is the first study evaluating the utility of the Genexus OPA assay for analysis of CSF-cfTNA in patients with lung and breast cancer metastasis to the brain. This is also the first study demonstrating the feasibility of rapid NGS analysis of CSF-cfTNA. One of the main advantages of the Genexus OPA is the short TAT. Genexus OPA required less than 20 h to complete the nucleic acid-to-result workflow. Therefore, the Genexus system facilitates successful cfTNA analysis from CSF with short TAT, which could assist clinicians in making treatment decisions faster than standard NGS assays. In summary, our results indicate that CSF-cfTNA analysis with the OPA in the Genexus instrument can provide clinically relevant information in patients with brain metastases with a short TAT highlighting the potential clinical utility of liquid biopsy for improving the management of patients with lung and breast cancer brain metastases.

Data availability

All data generated or analyzed during this study are included in this published article.