You are here

A Highly Sensitive and Quantitative Test Platform for Detection of NSCLC EGFR Mutations in Urine and Plasma

Journal of Thoracic Oncology, Volume 11, Issue 10, October 2016, Pages 1690 - 1700



In approximately 60% of patients with NSCLC who are receiving EGFR tyrosine kinase inhibitors, resistance develops through the acquisition of EGFR T790M mutation. We aimed to demonstrate that a highly sensitive and quantitative next-generation sequencing analysis of EGFR mutations from urine and plasma specimens is feasible.


Short footprint mutation enrichment next-generation sequencing assays were used to interrogate EGFR activating mutations and the T790M resistance mutation in urine or plasma specimens from patients enrolled in TIGER-X (NCT01526928), a phase 1/2 clinical study of rociletinib in previously treated patients with EGFR mutant–positive advanced NSCLC.


Of 63 patients, 60 had evaluable tissue specimens. When the tissue result was used as a reference, the sensitivity of EGFR mutation detection in urine was 72% (34 of 47 specimens) for T790M, 75% (12 of 16) for L858R, and 67% (28 of 42) for exon 19 deletions. With specimens that met a recommended volume of 90 to 100 mL, the sensitivity was 93% (13 of 14 specimens) for T790M, 80% (four of five) for L858R, and 83% (10 of 12) for exon 19 deletions. A comparable sensitivity of EGFR mutation detection was observed in plasma: 93% (38 of 41 specimens) for T790M, 100% (17 of 17) for L858R, and 87% (34 of 39) for exon 19 deletions. Together, urine and plasma testing identified 12 additional T790M-positive cases that were either undetectable or inadequate by tissue test. In nine patients monitored while receiving treatment with rociletinib, a rapid decrease in urine T790M levels was observed by day 21.


DNA derived from NSCLC tumors can be detected with high sensitivity in urine and plasma, enabling diagnostic detection and monitoring of therapeutic response from these noninvasive “liquid biopsy” samples.

Keywords: Urine, Circulating tumor DNA, NSCLC, EGFR mutations, T790M.


A major challenge for assessing EGFR mutation status in advanced NSCLC is the availability of suitable biopsy tissue for molecular testing. Clinical studies suggest that 10% to 20% of all NSCLC biopsies are inadequate for molecular analysis because of a lack of either sufficient tumor cells or amplifiable DNA.1 and 2 Biopsies also pose an economic burden and health risk to patients, with biopsy-associated patient morbidity (e.g., pneumothorax) observed in 12% to 21% of image-guided transthoracic needle tissue biopsies.3 Moreover, despite guidelines recommending EGFR testing at diagnosis for guiding first-line treatment decisions,4 and 5 up to 25% of patients with lung cancer receive treatment before EGFR mutation assessment.6 Physicians cite tumor histologic features (i.e., squamous), insufficient tumor samples, poor health status of the patient, long turnaround times for tests, and patient’s desire to initiate therapy as reasons for failure to undergo timely molecular testing.6

In NSCLC patients receiving first-line tyrosine kinase inhibitors (TKIs) targeting EGFR mutation–positive tumors (i.e., erlotinib, gefitinib, and afatinib), resistance to therapy develops through the emergence of a second mutation in EGFR, T790M, in approximately 60% of cases.7 Performance of a repeat biopsy of these patients is still an emerging standard of care, and up to 25% of patients may be medically ineligible owing to comorbidities or the lack of an accessible lesion.8 False-negative results could occur with tissue biopsies, likely because of the underlying intratumoral and intertumoral heterogeneity often associated with resistance mechanisms such as T790M.9 and 10

Detection and monitoring of cancer-specific genomic alterations in blood, specifically through the assessment of circulating tumor DNA (ctDNA), is a minimally invasive alternative to a tissue biopsy that has shown promise in overcoming some of the challenges associated with sampling from tissue.11 However, ctDNA presents its own challenges for clinical diagnostics. It is highly fragmented, may be very rare (<0.01%) as a proportion of all circulating free nucleic acids in blood, and may be especially difficult to detect in certain cancers such as those localized in the central nervous system.12 and 13 Highly sensitive assays have been developed and continue to be improved upon to address these challenges.13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, and 24

ctDNA in the systemic circulation is eventually excreted into urine, where it is thought to undergo further degradation.25, 26, 27, and 28 Urine analysis is a truly noninvasive alternative to tissue biopsy that integrates DNA from multiple sites, thus potentially addressing the challenges posed by tumor heterogeneity. To date, only a limited number of published studies have examined the feasibility of ctDNA detection from urine.25, 26, 27, and 28 Although to our knowledge, none of these studies describe ctDNA detection from the urine of patients with NSCLC, patient-matched tissue and plasma and urine studies in colorectal cancer (KRAS) and histiocytic disorders (BRAF) indicate good concordance of DNA mutation status across all three biopsy specimens.29, 30, and 31

We report herein, for the first time, the development and characterization of an EGFR mutation detection platform that identifies L858R, exon 19 deletions, and T790M in ctDNA from both plasma and urine samples from patients with NSCLC. We evaluated the clinical performance of this platform in matched pretreatment urine and plasma samples and examined the feasibility of longitudinal monitoring of EGFR mutations from the urine of patients with NSCLC in the TIGER-X study, which is a phase 1/2 study of the third-generation EGFR-TKI rociletinib (CO-1686).

Patients and Methods


A blinded, retrospective study was conducted on matched urine and plasma specimens collected from 63 patients with stage IIIB to IV disease who were enrolled in the TIGER-X trial (NCT01526928). Patients in TIGER-X were required to have histologically or cytologically confirmed NSCLC and documented evidence of one or more EGFR mutations. All patients signed an ethics committee/institutional review board–approved consent before any procedures. Further details regarding TIGER-X study design have been previously published.14 and 32

Sample Collection and Processing

Tissue biopsies were collected within 60 days of initiation of treatment with rociletinib. For all formalin-fixed, paraffin-embedded tissue specimens, tumor content was assessed by board-certified pathologists using hematoxylin and eosin–stained slides. Tumor specimens were considered evaluable if any tumor cells were identified. For seven cases, DNA was extracted from one 5-μm section and central laboratory tissue testing was performed with the cobas EGFR Mutation Test (Roche Molecular Systems, Pleasanton, CA). For 55 cases, DNA was extracted from two 5-μm sections and central laboratory tissue testing was performed with the therascreen EGFR RGQ Polymerase Chain Reaction (PCR) Kit (Qiagen, Valencia, CA). A local EGFR test result was used for one case in which tissue was not submitted to the central laboratory.

Blood and urine samples were obtained serially, before administration of the first dose, and with every 21-day cycle of rociletinib treatment. Blood samples were collected in K2 ethylenediaminetetraacetic acid BD Vacutainer tubes (BD Biosciences, San Jose, CA), processed into plasma within 30 minutes (1800 g for 10 minutes at 18°C–23°C), and stored at or below –70°C. For plasma DNA analysis, 1.5 to 4 mL of plasma was extracted using the QIAamp DNA Circulating Nucleic Acid Kit (Qiagen, Manchester, U.K.) according to the manufacturer’s instructions. Urine samples between 10 and 100 mL were collected in the clinic into 120-mL cups, supplemented with preservative, and stored at or below –70°C. For urinary DNA extraction, urine was concentrated to 4 mL using Vivacell 100 concentrators (Sartorius Corp., Bohemia, NY) and incubated with 700 μL of Q Sepharose Fast Flow quaternary ammonium resin (GE Healthcare, Pittsburgh, PA). Tubes were spun to collect Sepharose and bound DNA. The pellet was resuspended in a buffer containing guanidine hydrochloride and isopropanol, and the eluted DNA was collected as a flowthrough using polypropylene chromatography columns (BioRad Laboratories, Irvine, CA). The DNA was further purified using QiaQuick columns (Qiagen, Hilden, Germany). DNA in plasma and urine was quantitated using a droplet digital PCR (ddPCR) assay that amplifies a single-copy RNaseP reference gene (QX200 ddPCR system, Bio-Rad, CA) as described previously.30

Urine and Plasma EGFR Mutation Analysis

Quantitative analysis of the T790M resistance mutation and EGFR activating mutations (L858R and 69 deletion variants in exon 19) was performed using a mutation enrichment PCR coupled with next-generation sequencing (NGS) detection (Trovagene, San Diego, CA). Selective amplification of mutant fragments was accomplished by short amplicon (42–44 base pairs [bp]) kinetically driven PCR that amplifies the mutant fragments while suppressing the amplification of the wild-type (WT) sequence using a blocker oligonucleotide. PCR primers contained a 3′ gene-specific sequence and a 5′ common sequence that was used in the subsequent sample barcoding step. The PCR enrichment cycling conditions utilized the initial 98°C denaturation step followed by the assay-specific 5 to 15 cycles of preamplification PCR and 17 to 32 cycles of mutation enrichment PCR. Custom DNA sequencing libraries were constructed and indexed using the Access Array System for Illumina Sequencing Systems (Fluidigm Corp., San Francisco, CA). The indexed libraries were pooled, diluted to equimolar amounts with buffer and the PhiX Control library, and sequenced to 200,000× coverage on an Illumina MiSeq platform using 150-V3 sequencing kits (Illumina, San Diego, CA). Primary image analysis, secondary base calling, and data quality assessment were performed on the MiSeq instrument using RTAv1.18.54 and MiSeq Reporter v2.6.2.3 software (Illumina). The analysis output files (FASTQ) from the runs were processed using custom sequencing reads counting and variant calling algorithms to tally the sums of total target gene reads, WT, or mutant EGFR reads, which passed predetermined sequence quality criteria (q score ≥ 20). A custom quantification algorithm was developed to accurately determine the absolute number of mutant DNA molecules in the source ctDNA sample. To that end, each single multiplexed NGS run contained a set of standard curve samples in addition to clinical samples and controls. For each run the standard sample set was assayed in parallel with patient samples starting with PCR enrichment of mutant EGFR DNA followed by NGS. The number of mutant copies detected was determined by interpolation from a standard curve derived from the standard sample set. To account for time of residency in urine across serially collected samples, the results were standardized by normalizing the number of copies detected in the sample to a constant number of WT DNA genome equivalents (GEqs) observed in an average urine sample (i.e., 330 ng WT DNA = 100,000 GEqs).

Determination of ctDNA EGFR Mutation Detection Cutoffs in Urine and Plasma

Clinical EGFR mutation detection cutoffs for urine and plasma were determined for each assay by assessing the level of nonspecific signal present, if any, from urine and plasma DNA samples obtained from 54 to 64 unique healthy volunteers and metastatic patients with non-NSCLC cancers (∼50%:50%). Detection cutoffs were standardized to 100,000 WT GEqs yielding adjusted clinical detection cutoffs of 5.5, 5.5, and 12.6 copies per105 GEqs for exon 19 deletions, L858R, and T790M, respectively.

Statistical Analysis

The correlation between input and output absolute EGFR mutant copies in the analytical spike-in experiments were examined using Spearman’s correlation, which is robust against nonlinearity. Analysis of trends observed in urine ctDNA EGFR signal upon patient treatment with rociletinib was assessed using a two-sided Wilcoxon’s paired two-sample test. p Values less than 0.05 were considered statistically significant. All statistical analyses were carried out using R v3.2.3 computer software (R Foundation for Statistical Computing, Vienna, Austria).


Development of EGFR Mutation Enrichment NGS Assays for ctDNA

A three-pronged approach was taken to overcome the inherent technical challenges of obtaining sensitive detection and robust quantification of ctDNA mutations in plasma and urine. First, ultrashort footprint PCR assays were developed to increase the likelihood of amplifying highly degraded ctDNA. Three ultrashort footprint assays were developed to detect the most common EGFR mutations: (1) a 42-bp EGFR exon 19 deletion assay recognizing 69 annotated deletions, (2) a 46-bp EGFR exon 21 L858R assay, and (3) a 44-bp EGFR exon 20 T790M assay. Secondly, mutant ctDNA fragments were enriched by a PCR-based method to maximize sensitivity for detecting ctDNA mutations having rare prevalence (i.e., ∼0.01%). Preferential PCR enrichment of mutant EGFR ctDNA was accomplished by using WT EGFR oligonucleotides that block the ability of PCR primers to anneal and amplify WT EGFR DNA, thereby increasing the likelihood of amplifying mutant EGFR templates (see Patients and Methods). Lastly, absolute ctDNA mutation copy numbers from a patient sample were quantitated by NGS methodology. This was achieved with the aid of a standard sample set spiked with known copies of mutant EGFR molecules.

Enrichment performance of EGFR mutant DNA was assessed by spiking five to 500 copies of mutant DNA into 18,181 GEqs of WT DNA (0.028% to 2.7%). Fold enrichment of EGFR mutant fragments increased as the proportion of mutant versus WT fragments decreased from 2.7% to 0.028% (Fig. 1A and B). The resulting sequencing libraries comprised 24% to 99.9% mutant reads, thus enabling sensitive mutation detection by NGS (Fig. 1A). For the three assays, 857- to 3214-fold enrichment of EGFR mutation signal was obtained for an input of five copies of mutant EGFR DNA within 60 ng (18,181 GEqs) of WT DNA (Fig. 1B).


Figure 1

Mutation (MT) enrichment next-generation sequencing. (A) Comparison between input ratio of mutant-to–wild-type (WT) EGFR copies and output ratio of mutant-to-WT EGFR sequencing reads for five to 500 input mutant copies of indicated EGFR variants diluted in 60 ng (∼18,180 genome equivalents) of WT DNA (mutation abundance of 0.028%–2.7%). Output sequencing reads are means of 18 replicates from six independent next-generation sequencing dilution series experiments performed on three different days by two operators on two MiSeq instruments. Percent mutant is calculated as a fraction of all copies or reads. (B) Data from (A) presented as mutation fold enrichment estimates. Fold enrichment is calculated as the percent of input mutant EGFR molecules divided by the percentage of output mutant EGFR sequencing reads.


Analytical Performance of ctDNA EGFR Mutation Assays

The lower limit of detection (LLoD) for the EGFR mutation assays was determined by using a statistical model based on the Poisson distribution of rare mutant DNA molecules within a series of highly diluted EGFR mutant DNA samples (Supplementary Digital Content). The observed frequency distribution of the number of copies detected within 80 replicates of samples with a mutant DNA spike-in level of one or two expected mutant DNA copies per replicate was within the 95% confidence intervals of expected frequency distribution within a Poisson model (Supplementary Table 1). These results indicated a LLoD of one copy in 18,181 WT GEqs (0.006%) for EGFR exon 19 deletion and L858R assays and a LLoD of two copies in 18,181 GEqs (0.01%) for the EGFR T790M assay.

The analytical accuracy and reproducibility of the EGFR mutation assays were determined by a dilution series of six replicates of 0, 5, 10, 50, 100, and 250 copies against a background of 18,181 WT DNA copies, spanning the linear range of the assays. The entire workflow was replicated six times with three dilution series replicates prepared by two different operators on three different days for a total of 18 measurements at each copy level; NGS analysis was performed on two different Illumina MiSeq instruments. The Spearman correlation between spiked-in absolute copy numbers (quantified by ddPCR) versus detected copy numbers (quantified by mutation enrichment NGS) ranged from 0.967 to 0.981 for the EGFR mutation assays (Fig. 2A). The mean coefficient of variation percentage was 34.5% across the reportable range of 5 to 250 copies for all three EGFR mutation assays (yielding an adjusted quantifiable range of 27.5 to 1375 copies per 100,000 GEqs), with the highest coefficient of variation percentage of 47.4% to 60.7% observed at the lowest input of five copies likely owing to Poisson limitations (Fig. 2B). Assay fold-discrimination performance was examined by comparing the expected copy ratio between two consecutive dilutions to the observed copy ratio (Fig. 2C). Known twofold and 2.5-fold differences within the dilution series were maintained and detected for all three EGFR mutation assays (Fig. 2C).


Figure 2

Quantification of EGFR mutant and wild-type DNA blends by mutation enrichment next-generation sequencing. (A) Analysis of dilution series of indicated mutant EGFR variants spiked into 60 ng (∼18,180 GEq) of WT DNA. An analysis algorithm was applied to transform the mutant EGFR sequencing reads into the absolute mutant copies detected. The box-and-whisker plots show the median (center line), 25th, and 75th percentiles (box) with the connecting “whiskers” extending from the first quartile minus 1.5 of the interquartile range (the third quartile minus the first quartile) and the third quartile plus 1.5 of the interquartile range. (B) Interrun reproducibility of the EGFR exon 19 deletions, L858R, and T790M mutation enrichment next-generation sequencing assays for the dilution series shown in (A). Coefficient of variation percent (CV%) is presented. (C) Quantification of twofold or 2.5-fold differences between subsequent mutant copy input levels. The expected fold ratio was calculated as the ratio of the two input mutant copy levels. The observed fold ratio was calculated as the ratio of two means for two measured mutant copy levels with a 95% confidence interval (CI).


Clinical Performance of ctDNA EGFR Mutation Assays in Urine

Matched baseline tumor tissue biopsies and urine samples were obtained from the first 63 consecutive patients in TIGER-X who agreed to optional pretreatment urine sample collection. TIGER-X was a phase 1/2 trial of rociletinib in patients with stage IIIB to IV NSCLC who had been treated with at least one prior EGFR inhibitor and had an EGFR activating mutation in their medical record. The clinical characteristics of the 63 patients are shown in Table 1. Tumor tissue was processed by a central laboratory for EGFR mutation testing. Of the 63 tumor tissue biopsy samples, 60 were adequate for analysis, with 47 positive for T790M, 16 positive for L858R, and 42 positive for exon 19 deletion mutations. Thirteen of 60 evaluable cases were negative for T790M, and two of 60 cases were negative for either L858R or exon 19 deletion mutation by central laboratory testing. Urine volumes ranged from 10 to 100 mL, with 19 of 63 samples meeting the prespecified criteria for the recommended urine volume of 90 to 100 mL. Tumor tissue testing was used as a reference standard. For all samples (volumes 10–100 mL), the sensitivity of the urine assays for EGFR mutation detection was 72% (34 of 47 samples) for T790M, 75% (12 of 16) for L858R, and 67% (28 of 42) for exon 19 deletion mutations (Table 2). For samples with the recommended urine volume of 90 to 100 mL, the sensitivity of EGFR mutation detection was 93% (13 of 14) for T790M, 80% (four of five) for L858R, and 83% (10 of 12) for exon 19 deletions (see Table 2). For urine samples with lower volumes (10–89 mL), the sensitivity for EGFR mutation detection was 64% (21 of 33) for T790M, 73% (eight of 11) for L858R, and 60% (18 of 30) for exon 19 deletions (see Table 2).

Table 1

Patient Demographic and Baseline Characteristics


Characteristics Patients in TIGER-X with Pretreatment Urine Samples (N = 63)
Age, y
 Median 64
 Range 40–85
Sex, n (%)
 Female 45 (71.4)
 Male 18 (28.6)
Race, n (%)
 Asian 17 (27.0)
 Black/African American 1 (1.6)
 White 44 (69.8)
 Other 0
 Missing 1 (1.6)
Histologic diagnosis, n (%)
 Adenocarcinoma 63 (100)
ECOG at study entry, n (%0)
 0 18 (28.6)
 1 45 (71.4)
No. previous therapies at baseline, n (%)
 0 0
 1 21 (33.3)
 2 11 (17.5)
 3 7 (11.1)
 4 9 (14.3)
 5+ 15 (23.8)
No. previous EGFR TKIs at baseline, n (%)
 1 34 (54.0)
 2 17 (27.0)
 3 12 (19.0)
Smoking history, n (%)
 Current smoker 1 (1.6)
 Former smoker 22 (34.9)
 Never-smoker 40 (63.5)

TIGER-X, Study to Evaluate Safety, Pharmacokinetics, and Efficacy of Rociletinib (CO-1686) in Previously Treated Mutant EGFR in NSCLC Patients; ECOG, Eastern Cooperative Oncology Group; TKI, tyrosine kinase inhibitor.

Table 2

Performance of Mutation Enrichment NGS Assays for Detection of EGFR Mutations in Urine and Plasma


Characteristics Specimen Type Value, n %
Sensitivity All urine volumes, 10–100 mL 72% (34 of 47)
Urine volumes, 90–100 mLa 93% (13 of 14)
Urine volumes, 10–89 mL 64% (21 of 33)
Plasma 93% (38 of 41)
Specificity Urine 96% (54 of 56)
Plasma 94% (60 of 64)
Sensitivity All urine volumes, 10–100 mL 75% (12 of 16)
Urine volumes, 90–100 mLa 80% (4 of 5)
Urine volumes, 10–89 mL 73% (8 of 11)
Plasma 100% (17 of 17)
Specificity Urine 100% (50 of 50)
Plasma 100% (48 of 48)
Exon 19 deletions
Sensitivity All urine volumes, 10–100 mL 67% (28 of 42)
Urine, volumes 90–100 mLa 83% (10 of 12)
Urine, volumes 10–89 mL 60% (18 of 30)
Plasma 87% (34 of 39)
Specificity Urine 94% (47 of 50)
Plasma 96% (47 of 49)

a Recommended urine volume: ≥90 mL.

NGS, next-generation sequencing.

The specificity of the EGFR urine assays was determined using urine samples obtained from healthy donors and patients with non-NSCLC metastatic cancers (see Patients and Methods) and was 96% for T790M, 100% for L858R, and 94% for the exon 19 deletion mutations (see Table 2).

Clinical Performance of ctDNA EGFR Mutation Assays in Plasma

Plasma was available for 60 of the 63 patients. When tumor tissue testing results were used as a reference standard, the detection sensitivity of the assays in plasma was 93% (38 of 41 samples [three of 44 available plasma samples failed NGS]) for T790M, 100% (17 of 17 samples) for L858R, and 87% (34 of 39 samples) for exon 19 deletions (see Table 2). The specificity of the EGFR plasma tests was determined using plasma samples obtained from healthy donors and patients with non-NSCLC metastatic cancers and was 94% for T790M, 100% for L858R, and 96% for exon 19 deletion mutations (see Table 2).

Urine and Plasma Identify Additional EGFR T790M–Positive Cases Undetectable by Tissue Biopsy

Contingency tables for the EGFR mutation detection in urine, plasma, and tissue are presented in Figure 3. For all urine sample volumes, there were 11 cases that were urine T790M–positive but tumor tissue T790M–negative or tissue sample–inadequate (Fig. 3A). Of these 11 cases, 10 were also T790M positive in plasma (one sample was T790M negative in plasma). Similarly, of the 11 discordant cases that were plasma T790M–positive but tissue T790M–negative or tissue sample–inadequate, 10 were also positive by urine T790M testing (one sample was T790M negative in urine). Together, urine and plasma T790M testing identified a higher proportion of positive cases (89% [56 of 63) than did tissue testing alone (75% [47 of 63]).


Figure 3

Contingency tables for the analysis of EGFR T790M mutation in matched tumor, urine, and plasma specimens from patients enrolled in the TIGER-X clinical trial. (A) Urine versus tumor analysis of T790M in 63 matched tumor and urine specimens. (B) Plasma versus tumor analysis of T790M in 60 matched tumor and plasma specimens. (C) Urine versus plasma analysis of T790M in 60 matched urine and plasma specimens. (D) Venn diagram showing T790M-positive status of 60 cases with available matched tumor, plasma, and urine specimens. Four cases not identified as T790M-positive by tumor, plasma, or urine are not depicted in the diagram. FFPE, formalin-fixed, paraffin-embedded.


When 60 cases with all three patient-matched specimen types were considered, urine analysis identified seven T790M cases not detected or failed in plasma and 11 cases were not detected by tissue or found to be tissue sample–inadequate (Fig. 3D). Analysis of urine and blood combined detected T790M in 93% of patients (56 of 60). Tissue analysis alone detected T790M in 73% of patients (44 of 60). Four of 60 cases were negative according to testing of tissue, plasma, and urine samples.

Association of EGFR T790M Levels in Urine with Patient Response to Rociletinib

Recent studies have suggested that the extent to which plasma EGFR mutation levels drop after introduction of EGFR TKI therapy may predict depth of response.14 and 16 To assess this relationship in urine, longitudinal urine samples were obtained from 15 patients treated with therapeutic doses of rociletinib (500, 625, 750, or 1000 mg twice daily [hydrobromide formulation]). Of these 15 patients, five had progressive disease as their best overall confirmed response and 10 patients had either stable disease or a partial response (PR) as their overall confirmed response. Among these, nine patients were identified with quantifiable levels of baseline T790M in urine (>27.5 copies per 100,000 GEqs), with seven experiencing PR or stable disease as best overall confirmed response and two with progressive disease as best overall confirmed response. For all nine patients, there was a significant decrease in T790M levels in urine after cycle 1 relative to baseline irrespective of best overall confirmed response (range −51% to −100%; p = 0.0091, two-sided Wilcoxon test; p < 0.0001, two-sided t-test) (Fig. 4A and B). However, in the two patients with progressive disease, there was an observed attenuated decrease (−51% and −70%) compared with in patients with PR and stable disease as best overall confirmed response (range −83% to −100%).


Figure 4

Longitudinal dynamics of EGFR T790M signal in urine of patients treated with rociletinib. (A) A rapid decrease in urinary T790M signal from baseline was observed after one cycle of treatment (21 days) in all patients treated with rociletinib. (B) Clinical outcomes of patients in the longitudinal analysis. Best overall confirmed response was assessed by investigator. For patient 2, percent change in urine T790M from baseline was estimated at day 42. PR, partial response; SD, stable disease.



Herein we have described a highly sensitive method for detection of actionable EGFR mutations in the urine and plasma of patients with advanced NSCLC. In our study cohort of relapsed patients, the sensitivity of EGFR mutation detection in urine with tumor as a reference was 72% (34 of 47) for T790M, 67% (28 of 42) for exon 19 deletions, and 75% (12 of 16) for L858R mutations (all urine volumes). A higher sensitivity was achieved with urine samples at the recommended volume of 90 to 100 mL as compared with urine volumes less than 90 mL: 93% (13 of 14) for T790M, 71% (five of seven) for L858R, and 83% (10 of 12) for exon 19 deletions (90–100 mL) versus 64% (21 of 33) for T790M, 72% (eight of 11) for L858R, and 60% (18 of 30) for the exon 19 deletion mutations (volumes of 10–89 mL). The specificity of the urine EGFR assays in healthy volunteers or patients without NSCLC was 96% for T790M, 100% for L858R, and 94% for exon 19 deletion mutations. To our knowledge, this study represents the first successful demonstration of EGFR mutation detection in the urine of patients with metastatic NSCLC.

The transrenal clearance of systemically derived DNA was first demonstrated in 2000 by Botezatu et al. who detected male-specific sequences in the urine of women transfused with male blood or pregnant with male fetuses.25 In addition, this pioneering work demonstrated that circulating nucleic acid of tumor origin could be identified in the urine of patients with cancer, specifically those with colorectal or pancreatic cancer.25 Further studies suggested that the size of the systemically derived DNA fragments in urine can range from approximately 35 to 250 bp.27, 33, 34, and 35 Subsequently, an anion exchange–based urinary DNA isolation technique coupled with ultrashort amplicon PCR detection was developed to maximize the detection of systemically derived DNA.27, 33, and 34 The method described in the present study builds on this work, incorporating an anion exchange–based method to preferentially isolate low-molecular-weight DNA from urine, short amplicon PCR with WT DNA suppression for mutation enrichment, and ultradeep sequencing to further enhance the identification of rare mutations. Using this approach, we show single-copy detection with spiked sample material.

Our methodology may have also contributed to the high clinical sensitivity that we observed for detecting EGFR mutations in plasma (range 87%–100%), which compares favorably with the published performance of real-time and ddPCR platforms. For example, T790M detection sensitivity in patients who experienced a relapse while receiving first-line EGFR TKIs was shown to range between 64% and 73% for the cobas test (a test platform based on real-time PCR), 73% to 81% for BEAMing (Beads, Emulsion, Amplification, and Magnetics, a technology based on digital PCR), and 77% for the ddPCR platform.13, 18, and 24 In the same studies, the sensitivity for detection of activating EGFR mutations was 73% to 84% for the cobas test, 82% to 84% for the BEAMing assay, and 74% to 82% for the ddPCR platform.14, 18, and 24 In five recent trials in patients with previously untreated NSCLC (NCT01203917, FASTACT-2, TRIGGER, EUROTAC, and NCT02279004), the sensitivity for detection of EGFR exon 19 deletions and L858R mutations in plasma was 78% for a real-time peptide nucleic acid clamp test, 69% to 86% for a ddPCR test, 62% to 68% for the therascreen EGFR RGQ PCR test, and 62% to 100% for the cobas test.2, 15, 16, 22, and 24 The specificities of our plasma assays in healthy volunteers and patients without NSCLC were 94% to 100%, which are similar to that found in urine. Limitations of this study include a small sample size for urine with volumes of 90 to 100 mL and a less than 100% assay specificity. Analysis of a larger cohort of patients is under way. An improved error correction bioinformatics algorithm is being developed to further increase assay specificity.

Assay sensitivity in the present study was calculated using tumor as the reference sample type. This method has limitations, particularly when applied to resistance mutations such as T790M that will have a significant false-negative rate in analysis of biopsy samples owing to tumor heterogeneity and low tumor cellularity.9, 10, 14, and 24 In our study, the combination of urine and plasma testing identified 12 T790M-positive cases that were undetectable by central laboratory testing of tumor tissue. Although 10 of the 12 cases were positive by both ctDNA specimen types, one case each was unique to plasma and urine. Urine may therefore provide complementary information about a patient’s mutational status that is not captured by plasma or tissue tests. These results indicate for the first time that either urine or plasma T790M testing may be considered as an alternative to tissue biopsy testing. Urine testing can be the preferable option because it represents a truly noninvasive alternative that can be collected in a patient’s own home, and the results of both diagnostic and monitoring testing can be made available without scheduling visits to the clinic.

Given the ease of sample collection, urine holds promise for the serial monitoring of patients. Studies in plasma have already shown that early changes in ctDNA may predict response to targeted therapies and that emergence of resistance mutations can be identified before radiographic progression.12, 14, 15, 19, and 23 Reported here, T790M levels in urine rapidly decreased to a fraction of their pretreatment levels in patients treated with rociletinib, regardless of RECIST response status. These data are consistent with previous findings in plasma and suggest that rociletinib reduces proliferation and consequently turnover of T790M-positive clones even in patients with primary resistance.9 and 14 It is an intriguing observation that there was an attenuated decrease in T790M levels for the two patients with progressive disease as best overall response. Further urine analysis of a substantially larger cohort of patients from the TIGER-X study is ongoing and should further inform the utility of longitudinal monitoring in this patient population.

In conclusion, our data demonstrate that urine testing using the mutation enrichment NGS method successfully identifies EGFR mutations in patients with metastatic NSCLC and has high concordance with tumor and plasma, suggesting that EGFR mutation detection from urine or plasma should be considered as a viable approach for assessing EGFR mutation status.


The authors thank patients for their participation in the study. We thank the study staff, who encouraged the optional collection of urine specimens; Onkar Dha and David Gustafson for coordinating biospecimen collection; Latifa Hassain for assistance with urine analyses; and Peter Morello, Sandeep Pingle, and Victoria Raymond for assistance with manuscript preparation.

Supplementary Data

Download file

Supplementary Digital Content

Download file

Supplementary Table 1


  • 1 M.E. Arcila, G.R. Oxnard, K. Nafa, et al. Rebiopsy of lung cancer patients with acquired resistance to EGFR inhibitors and enhanced detection of the T790M mutation using a locked nucleic acid-based assay. Clin Cancer Res. 2011;17:1169-1180 Crossref
  • 2 J.Y. Douillard, G. Ostoros, M. Cobo, et al. Gefitinib treatment in EGFR mutated Caucasian NSCLC: circulating-free tumor DNA as a surrogate for determination of EGFR status. J Thorac Oncol. 2014;9:1345-1353 Crossref
  • 3 M.K. Accordino, J.D. Wright, D. Buono, et al. Trends in use and safety of image-guided transthoracic needle biopsies in patients with cancer. J Oncol Pract. 2015;11:e351-359
  • 4 B.P. Levy, M.D. Chioda, D. Herndon, et al. Molecular testing for treatment of metastatic non-small cell lung cancer: how to implement evidence-based recommendations. Oncologist. 2015;20:1175-1181
  • 5 N.I. Lindeman, P.T. Cagle, M.B. Beasley, et al. Molecular testing guideline for selection of lung cancer patients for EGFR and ALK tyrosine kinase inhibitors: guideline from the College of American Pathologists, International Association for the Study of Lung Cancer, and Association for Molecular Pathology. J Thorac Oncol. 2013;8:823-859 Crossref
  • 6 J. Spicer, B. Tischer, M. Peters. EGFR mutation testing and oncologist treatment choice in advanced NSCLC: global trends and differences. Ann Oncol. 2015;26(suppl):i57-i61
  • 7 H.A. Yu, M.E. Arcila, N. Rekhtman, et al. Analysis of tumor specimens at the time of acquired resistance to EGFR-TKI therapy in 155 patients with EGFR-mutant lung cancers. Clin Cancer Res. 2013;19:2240-2247 Crossref
  • 8 H.J. Yoon, H.Y. Lee, K.S. Lee, et al. Repeat biopsy for mutational analysis of non-small cell lung cancers resistant to previous chemotherapy: adequacy and complications. Radiology. 2012;265:939-948 Crossref
  • 9 Z. Piotrowska, M.J. Niederst, C.A. Karlovich, et al. Heterogeneity underlies the emergence of EGFRT790 wild-type clones following treatment of T790M-positive cancers with a third-generation EGFR inhibitor. Cancer Discov. 2015;5:713-722 Crossref
  • 10 K. Suda, I. Murakami, T. Katayama, et al. Reciprocal and complementary role of MET amplification and EGFR T790M mutation in acquired resistance to kinase inhibitors in lung cancer. Clin Cancer Res. 2010;16:5489-5498 Crossref
  • 11 L.A. Diaz Jr., A. Bardelli. Liquid biopsies: genotyping circulating tumor DNA. J Clin Oncol. 2014;32:579-586 Crossref
  • 12 C. Bettegowda, M. Sausen, R.J. Leary, et al. Detection of circulating tumor DNA in early- and late-stage human malignancies. Sci Transl Med. 2014;6 224ra224
  • 13 F. Diehl, K. Schmidt, M.A. Choti, et al. Circulating mutant DNA to assess tumor dynamics. Nat Med. 2007;14:985-990
  • 14 C. Karlovich, J.W. Goldman, J.M. Sun, et al. Assessment of EGFR mutation status in matched plasma and tumor tissue of NSCLC patients from a phase I study of rociletinib (CO-1686). Clin Cancer Res. 2016;22:2386-2395
  • 15 A. Marchetti, J.F. Palma, L. Felicioni, et al. Early prediction of response to tyrosine kinase inhibitors by quantification of EGFR mutations in plasma of NSCLC patients. J Thorac Oncol. 2015;10:1437-1443
  • 16 T. Mok, Y.L. Wu, J.S. Lee, et al. Detection and dynamic changes of EGFR mutations from circulating tumor DNA as a predictor of survival outcomes in NSCLC patients treated with first-line intercalated erlotinib and chemotherapy. Clin Cancer Res. 2015;21:3196-3203 Crossref
  • 17 M. Sausen, J. Phallen, V. Adleff, et al. Clinical implications of genomic alterations in the tumour and circulation of pancreatic cancer patients. Nat Commun. 2015;6:7686 Crossref
  • 18 K.S. Thress, R. Brant, T.H. Carr, et al. EGFR mutation detection in ctDNA from NSCLC patient plasma: a cross-platform comparison of leading technologies to support the clinical development of AZD9291. Lung Cancer. 2015;90:509-515
  • 19 G. Siravegna, B. Mussolin, M. Buscarino, et al. Clonal evolution and resistance to EGFR blockade in the blood of colorectal cancer patients. Nat Med. 2015;21:795-801 Crossref
  • 20 S. Couraud, F. Vaca-Paniagua, S. Villar, et al. Noninvasive diagnosis of actionable mutations by deep sequencing of circulating free DNA in lung cancer from never-smokers: a proof-of-concept study from BioCAST/IFCT-1002. Clin Cancer Res. 2014;20:4613-4624 Crossref
  • 21 S.J. Dawson, D.W. Tsui, M. Murtaza, et al. Analysis of circulating tumor DNA to monitor metastatic breast cancer. N Engl J Med. 2013;368:1199-1209 Crossref
  • 22 N. Karachaliou, C. Mayo-de las Casas, C. Queralt, et al. Association of EGFR L858R mutation in circulating free DNA with survival in the EURTAC trial. JAMA Oncol. 2015;1:149-157 Crossref
  • 23 G.R. Oxnard, C.P. Paweletz, Y. Kuang, et al. Noninvasive detection of response and resistance in EGFR-mutant lung cancer using quantitative next-generation genotyping of cell-free plasma DNA. Clin Cancer Res. 2014;20:1698-1705 Crossref
  • 24 A.G. Sacher, C. Paweletz, S.E. Dahlberg, et al. Prospective validation of rapid plasma genotyping for the detection of EGFR and KRAS mutations in advanced lung cancer. JAMA Oncol. 2016;2:1014-1022
  • 25 I. Botezatu, O. Serdyuk, G. Potapova, et al. Genetic analysis of DNA excreted in urine: a new approach for detecting specific genomic DNA sequences from cells dying in an organism. Clin Chem. 2000;46:1078-1084
  • 26 A.V. Lichtenstein, H.S. Melkonyan, L.D. Tomei, et al. Circulating nucleic acids and apoptosis. Ann N Y Acad Sci. 2001;945:239-249
  • 27 Y.H. Su, M. Wang, D.E. Brenner, et al. Human urine contains small, 150 to 250 nucleotide-sized, soluble DNA derived from the circulation and may be useful in the detection of colorectal cancer. J Mol Diagn. 2004;6:101-107 Crossref
  • 28 Y.H. Su, M. Wang, B. Aiamkitsumrit, et al. Detection of a K-ras mutation in urine of patients with colorectal cancer. Cancer Biomark. 2005;1:177-182
  • 29 D.M. Hyman, E.L. Diamond, C.R. Vibat, et al. Prospective blinded study of BRAFV600E mutation detection in cell-free DNA of patients with systemic histiocytic disorders. Cancer Discov. 2015;5:64-71 Crossref
  • 30 F. Janku, C.R. Vibat, K. Kosco, et al. BRAF V600E mutations in urine and plasma cell-free DNA from patients with Erdheim-Chester disease. Oncotarget. 2014;5:3607-3610
  • 31 Y.-H. Su, J. Song, Z. Wang, et al. Removal of high-molecular-weight DNA by carboxylated magnetic beads enhances the detection of mutated K-ras DNA in urine. Ann N Y Acad Sci. 2008;1137:82-91 Crossref
  • 32 L.V. Sequist, J.C. Soria, J.W. Goldman, et al. Rociletinib in EGFR-mutated non-small-cell lung cancer. N Engl J Med. 2015;372:1700-1709 Crossref
  • 33 K.C. Chan, S.F. Leung, S.W. Yeung, et al. Quantitative analysis of the transrenal excretion of circulating EBV DNA in nasopharyngeal carcinoma patients. Clin Cancer Res. 2008;14:4809-4813 Crossref
  • 34 H.S. Melkonyan, W.J. Feaver, E. Meyer, et al. Transrenal nucleic acids: from proof of principle to clinical tests. Ann N Y Acad Sci. 2008;1137:73-81 Crossref
  • 35 N.B. Tsui, P. Jiang, K.C. Chow, et al. High resolution size analysis of fetal DNA in the urine of pregnant women by paired-end massively parallel sequencing. PLoS One. 2012;7:e48319 Crossref


a City of Hope Comprehensive Cancer Center, Duarte, California

b Trovagene, Inc., San Diego, California

c Clovis Oncology, Inc., Boulder, Colorado

d Massachusetts General Hospital, Boston, Massachusetts

e University of Colorado, Denver, Colorado

f Stanford University Medical Center, Stanford, California

g Hospices Civils de Lyon, Lyon, France

h Dana Farber Cancer Institute, Boston, Massachusetts

i Beverly Hills Cancer Center, Beverly Hills, CA; City of Hope, Duarte, California

j Barbara Karmanos Cancer Institute, Detroit, Michigan

Corresponding author. Address for correspondence: Karen L. Reckamp, MD, City of Hope Comprehensive Cancer Center, 1500 E Duarte Rd, Duarte, CA 91010.

Drs. Reckamp, Melnikova, and Karlovich equally contributed to this work.

Disclosure: Dr. Reckamp reports personal fees from Trovagene and institutional clinical trial support from Clovis during the conduct of the study. Drs. Melnikova, Kosco, Chroucher, Mr. Samuelsz, Drs. Vibat, Geiss, and Erlander are employees of Trovagene. Dr. Karlovich, Ms. Mann, Mr. Rolfe, and Dr. Raponi are employees of Clovis Oncology and hold stock in Clovis Oncology. Dr. Sequist reports institutional trial support from Novartis during the conduct of the study; noncompensated consulting for Boehringer Ingelheim, Clovis Oncology, Novartis, Merrimack Pharmaceuticals, and Taiho; and compensated consulting for AstraZeneca and Ariad. Dr. Camidge reports personal fees from Clovis outside the submitted work. Dr. Wakelee reports personal fees from Peregrine and ACEA; grants from Novartis, Bristol-Myers Squibb, XCovery, Celgene, Roche/Genentech, MedImmune, Gilead, AstraZeneca, and Lilly; and grants and personal fees from Pfizer outside the submitted work. Dr. Perol reports personal fees from Clovis Oncology, Astra-Zeneca, Roche, and Boehringer-Ingelheim outside the submitted work. Dr. Oxnard reports personal fees from AstraZeneca, Clovis, Sysmex, Boehringer-Ingelheim, and Inivata outside the submitted work; in addition, Dr. Oxnard has a patent development and application of NSCLC plasma genotyping using digital polymerase chain reaction pending. Dr. Gadgeel reports personal fees from Roche/Genentech and AstraZeneca outside the submitted work. The remaining author declares no conflict of interest.

Commentary by Tom Stinchcombe

Approximately 60% of patients treated with an EGFR TKI develop a T790M acquired resistance mutation, and these patients are candidates for treatment with a third generation EGFR TKI (e.g. osimertinib). Historically patients have undergone a repeat tumor biopsy to detect the T790M. This study investigated the use of plasma and urine testing. In this phase I/II study plasma, urine, and tumor samples were collected, and 60 patients had evaluable tumor specimens. When the tumor sample was used as a reference the sensitivity for plasma for T790M was 93%, for exon 21 L858R 100% and for exon 19 deletions was 87% When tumor samples were used as a reference the sensitivity of urine for T790M was 72%, for exon 21 L858R was 75%, and for exon 19 deletions was 67%. The urine and plasma together identified an additional 12 patients with a T790M mutation that were undetectable on tumor testing or the sample inadequate tumor tissue. This is a small study and the number of patients in each mutation subset is limited. The potential use of urine will facilitate the detection of additional patients with a T790M resistance mutation, and may assistance in the serial measurement of T790M levels.  Importantly if the original EGFR mutation, exon 19 deletion or exon 21 L858R, is not detected on these test this tests cannot be accurately interpreted.