Serial Circulating Tumor DNA Mutational Status in Patients with KRAS-Mutant Metastatic Colorectal Cancer from the Phase 3 AIO KRK0207 Trial

Abstract

Background

We assessed the usefulness of circulating tumor DNA (ctDNA) pre- or post-treatment initiation for outcome prediction and treatment monitoring in metastatic colorectal cancer (mCRC).

Methods

Droplet digital PCR was used to measure absolute mutant V-Ki-ras2 Kirsten rat sarcoma viral oncogene (^(mut)KRAS) ctDNA concentrations in 214 healthy controls (plasma and sera) and in 151 tissue-based ^mutKRAS positive patients with mCRC from the prospective multicenter phase 3 trial AIO KRK0207. Serial ^mutKRAS ctDNA was analyzed prior to and 2–3 weeks after first-line chemotherapy initiation with fluoropyrimidine, oxaliplatin, and bevacizumab in patients with mCRC and correlated with clinical parameters.

Results

^mutKRAS ctDNA was detected in 74.8% (113/151) of patients at baseline and in 59.6% (90/151) at follow-up. ^mutKRAS ctDNA at baseline and follow-up was associated with poor overall survival (OS) (hazard ratio [HR] =1.88, 95% confidence interval [CI] 1.20–2.95; HR = 2.15, 95% CI 1.47–3.15) and progression-free survival (PFS) (HR = 2.53, 95% CI 1.44–4.46; HR = 1.90, 95% CI 1.23–2.95), respectively. ^mutKRAS ctDNA clearance at follow-up conferred better disease control (P = 0.0075), better OS (log-rank P = 0.0018), and PFS (log-rank P = 0.0018). Measurable positive ^mutKRAS ctDNA at follow-up was the strongest and most significant independent prognostic factor on OS in multivariable analysis (HR = 2.31, 95% CI 1.40–3.25).

Conclusions

Serial analysis of circulating ^mutKRAS concentrations in mCRC has prognostic value. Post treatment ^mutKRAS concentrations 2 weeks after treatment initiation were associated with therapeutic response in multivariable analysis and may be an early response predictor in patients receiving first-line combination chemotherapy.

Clinicaltrialsgov Identifier

NCT00973609.

Colorectal cancer (CRC) is a leading cause of mortality and the second most common malignancy in Europe (1, 2). While chemo- and targeted-therapies improve survival in patients with metastatic colorectal cancer (mCRC), benefits are limited and survival rates remain dissatisfying (1, 3). First-line treatment of v-Ki-ras2 Kirsten rat sarcoma viral oncogene (KRAS) mutant (^mutKRAS) mCRC currently relies on standard combination chemotherapy regimens of fluoropyrimidine with oxaliplatin or irinotecan. Combination of bevacizumab and fluorouracil-based therapy improves response rates and prolongs survival (4, 5), however, a lack of efficacy in patient subsets and cumulative toxicities (e.g., oxaliplatin-induced neurotoxicity) are a constant challenge in clinical care of patients with CRC. Thus, there is an unmet need for better biomarkers for predicting patient outcome and early treatment response.

The plasma concentrations of carcinoembryonic antigen (CEA) are often used for disease and treatment monitoring in colorectal cancer (6–9). Although prediagnostic concentrations of CEA have strong clinical value, its usefulness as an independent screening tool remains limited (10). Furthermore, CEA concentrations are unstable during chemotherapy and consequently response monitoring is not clinical routine (11). Thus, treatment response monitoring relies on radiologic morphological assessments precluding real-time monitoring.

An alternative to the use of CEA and other blood-based proteins is the use of circulating tumor-derived DNA (ctDNA) to monitor treatment response and predict patient outcome. Several oncogenic alterations including mutations in the adenomatous polyposis coli (APC), tumor protein p53 (TP53), KRAS, v-raf murine sarcoma viral oncogene homolog B1 (BRAF), phosphatase and tensin homolog (PTEN), and phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA) are frequently encountered in colorectal cancer cases (12). KRAS mutations are found in about 35%–45% of all cases of colorectal cancer, with hotspot mutations in codon 12 and 13 representing more than 95% of these mutations (13, 14). Among patients with mCRC, higher KRAS mutation rates (53%) have been reported (13). While several studies have described the utility of ^mutKRAS measurement prior to therapy initiation, the utility of serial ^mutKRAS measurement for treatment response monitoring in mCRC has barely been explored. In this translational study within the framework of the randomized multicenter phase 3 AIO KRK0207 trial, we used droplet digital PCR (ddPCR) to measure the circulating concentrations of ^mutKRAS in therapy-naive patients with mCRC prior to therapy initiation and 2–3 weeks post treatment initiation. We correlated absolute baseline concentrations of ^mutKRAS, follow-up ^mutKRAS load as measured after treatment initiation (2–3 weeks), or ^mutKRAS kinetics between baseline and follow-up with radiological response status at the end of induction treatment, progression-free survival (PFS), and overall survival (OS).

Materials and Methods

Study Design and Patient Population

Patients analyzed in this study were recruited within the framework of the clinical study AIO KRK0207. The study is a randomized 3-arm phase 3 trial with different maintenance strategies following a 24-week period of combination chemotherapy, consisting of treatment with a fluoropyrimidine, oxaliplatin, and bevacizumab.

All patients provided written informed consent for biomedical research approved by the institutional ethics committee for the AIO KRK0207 trial (ClinicalTrials. gov identifier: NCT00973609) to perform molecular analysis on tissue and plasma collections.

Tumor response to therapy was assessed by means of computed tomography (CT) and magnetic resonance imaging (MRI) scans at week 24 according to RECIST version 1.0, as previously described (15). Blood samples were collected at baseline (prior to treatment initiation) and 15 to 22 days after treatment commencement (follow-up), corresponding to the first administration of fluorouracil, leucovorin, and oxaliplatin, or capecitabine and oxaliplatin regimes, respectively. Plasma was collected for the isolation of circulating free DNA (cfDNA) and serum samples were prepared for the measurement of CEA. In total, 825 patients (recruited from over 160 centers around Germany) were included in the study, of which plasma samples from 151 patients bearing codon 12 and codon 13 hotspot mutations in the KRAS gene were analyzed.

Healthy Controls

To estimate the upper limit of ^mutKRAS load in serum and plasma samples from healthy blood donors, a total of 110 sera and 104 plasma samples from healthy donors were analyzed. All donors provided written informed consent for biomedical research. Approval was granted by the institutional ethics committee (14-5961-BO). Of the healthy donor group, gender and age data was available for 79 plasma and 22 serum samples.

Sample Processing

Preparation of plasma and serum from blood samples was performed as previously described (16). The supernatant was collected, stored at -80°C and later used for cfDNA isolation. CEA was quantified using a microplate immune-enzymometric assay (AxSYM, Abbott Laboratories).

DNA Isolation and ^mutKRAS Measurement

In order to avoid DNA sample cross contamination, cfDNA isolation was performed in batches of 16 samples followed by UV-sterilization in a fully automated process. Samples were thawed for 1 hour at room temperature and then 1 mL of either plasma or serum was loaded on a cartridge of the Maxwell RSC^® ccfDNA plasma kit (Promega Corporation). DNA was isolated following the manufacturer’s instructions and eluted in 50 µL of elution buffer. The DNA concentration was measured using a Quantus fluorometer (Promega). Circulating concentrations of ^mutKRAS were measured by means of ddPCR using the ddPCR™ KRAS screening multiplex kit (Bio-Rad) covering all codon 12 and codon 13 hotspot mutations (G12A, G12C, G12D, G12V, G12R, G12S, G13D). All reactions were performed in duplicates using 5 µL of cfDNA from each sample. The reaction components were constituted following manufacturers instruction to a final volume of 22 µL, of which 20 µL were used for droplet generation in a QX100™/QX200™ droplet generator (Bio-Rad). PCR reactions were performed in a C1000 Touch™ thermocycler (Bio-Rad) and droplets were read in a QX100™/QX200™ droplet reader (Bio-Rad). The raw ^mutKRAS concentration (copies/20 µL reaction) was used to determine the absolute ^mutKRAS load per mL of plasma by multiplying the concentration in 20 µL reaction by a factor of 10 to cover the entire elution volume from 1 ml of plasma.

TV = total volume of cfDNA used in the ddPCR reaction;

PV = total plasma volume used for isolation of cfDNA

For graphical representation and statistical analysis, a value of 0.5 copies/20 µL reaction was added to all samples with or without detectable ^mutKRAS as previously described (17). Samples with less than 10 000 droplets were reanalyzed and only included if they had more than 10 000 droplets in both replicates.

Statistical Analysis

Overall survival was defined from enrollment to event time (death). Progression-free survival was defined as being the time from randomization [n = 95] into the maintenance strategies to either detection of first progression or death. Overall and progression-free survival were analyzed as Kaplan-Meier estimates and groups were compared with log-rank test (univariable) and Cox proportional hazards model (multivariable). The latter was restricted to the respective populations with a full set of covariates available. All P-values are 2-sided, and considered explorative, without any adjustment for multiple testing.

ctDNA conversion groups were defined as follows: “negative-positive” were cases with a negative ^mutKRAS load at baseline but became positive at the follow-up time point; “Positive-negative,” were cases with a positive ^mutKRAS load at baseline but became negative at the follow-up time point; stable were cases whose baseline and follow-up ^mutKRAS loads remained negative or positive at both time points, even if they increased or decreased.

Results

Patient Characteristics and ^mutKRAS ctDNA Concentrations

Oncogenic mutations are present on otherwise healthy subjects (18–20). To determine the highest achievable circulating ^mutKRAS (^mutKRAS) load in healthy blood donors, we first isolated cfDNA from 104 plasma and 110 serum samples of healthy donors and analyzed them by ddPCR for ^mutKRAS. ^mutKRAS was detected in a total of 11/110 (10%) serum samples and in 14/104 (13.5%) plasma samples from healthy donors. We determined the upper limit of ^mutKRAS load in healthy donors to be ∼32 copies/mL in plasma (Supplemental Fig. 1) as previously described (18, 19). Based on the predetermined ^mutKRAS cut-off in healthy controls, we defined a patient group with positive circulating ^mutKRAS (> 32 copies/mL) and another group considered ^mutKRAS negative (≤ 32 copies/mL of plasma). Of 104 plasma samples from healthy donors, gender and age (18-67 years, median 35) information was available for 79 of them. Overall, healthy donors were younger than patients from the cancer patient study cohort and we could match 10 healthy donors to cancer patients. Paired comparison (10 vs 10) between age-matched patients and healthy donors revealed a significantly higher ^mutKRAS load in patients than controls (Supplemental Fig. 2). Since this was a large study involving multiple and small study sites in a real-world clinical setting, we investigated potential variations in ^mutKRAS load across different sites. As shown in Supplemental Fig. 3, no significant differences could be observed, although the case number per site was limited.

We next analyzed the ^mutKRAS load in plasma samples by ddPCR from patients with mCRC with tissue-confirmed ^mutKRAS recruited within the framework of the phase 3 AIO-KRK0207 clinical trial (15). Of the 825 cases included in the study, 151 cases with known KRAS mutations in tumor tissue had sufficient plasma material at the baseline and follow-up time points, as well as complete clinical information (Table 1 and Fig. 1) and were eligible for analysis. Importantly, since sample processing and collection in this large multicenter trial was performed in many small-volume centers, we evaluated potential bias with regard to recruiting study centers. However, there was no statistically significant difference in the ^mutKRAS concentrations in samples collected from 5 different sampling sites (Supplemental Fig. 3). Of the 151 cases included in the study, a positive ^mutKRAS load was detected in 113/151 (74.5%) at baseline. At the follow-up time point, 90/151 cases were positive for ^mutKRAS (59.6%) (Table 1, Supplemental Table 1, and Supplemental Fig. 4A). Of the 38 cases with negative ^mutKRAS load at baseline, 22 cases were negative for ^mutKRAS and 16 cases had ^mutKRAS loads ≤ 32 copies/mL. Among these cases, 10/38 (26%) baseline negative cases converted to positive at the follow-up time point (Supplemental Fig. 4B) and the others remained negative at both time points (Supplemental Fig. 4C).

Clinical Outcome and Radiological Response according to ^mutKRAS ctDNA

For all cases, clinicopathological parameters data were available at both time points except CEA, for which data was available in only 138 cases. As shown in Table 1, positive ^mutKRAS (in both ctDNAb and ctDNAf) was associated with unfavorable clinical features. A positive ^mutKRAS load was more common among elderly patients (≥ 65 years) and associated with multiple metastatic sites (N° metastatic sites >1), high CEA (>20 ng/mL) serum concentrations, and a primary tumor located at the right side (Table 1).

Radiological response data was available for 150/151 eligible cases. Supplemental Table 2 presents the prognostic impact of ctDNA on treatment outcome. There was no significant association between baseline ctDNA (ctDNAb) and disease control rate, which included complete response, partial response, and stable disease based on radiological assessment) (Supplemental Table 2, P = 1.0). However, when ctDNA at follow-up (ctDNAf) was analyzed, a significantly higher fraction of nonresponders (based on radiological staging about 6 months later) had a positive ^mutKRAS ctDNA at this time point (P = 0.0075).

Prognostic Impact of ^mutKRAS ctDNAb, ctDNAf, and ctDNA Kinetics on Survival

We evaluated the prognostic relevance of ^mutKRAS ctDNA and ^mutKRAS ctDNA dynamics on PFS and OS. As shown in Fig. 2A and B, both ctDNAb and ctDNAf were strong prognostic factors for progression-free survival (hazard ratio [HR]=2.53, 95% confidence interval [CI]: 1.44–4.46, P = 0.00079; HR = 1.90, 95% CI: 1.23–2.95, P = 0.0036). ^mutKRAS ctDNA kinetics were also prognostic for PFS. When cases were categorized into conversion and nonconversion categories, a survival difference was observed. Cases with positive ^mutKRAS ctDNAb and negative ctDNAf (Supplemental Fig. 4E) showed significantly better progression-free survival than cases with positive ^mutKRAS ctDNA at both time points (Fig. 3A and B and Supplemental Fig. 4F). Conversely, cases with negative to positive ^mutKRAS ctDNA conversion between baseline and follow-up had lower survival compared with those that remained negative at both time points (Supplemental Fig. 4D). This exploratory analysis, however, was limited by a low number of cases.

Both ctDNAb and ctDNAf were both strong prognostic factors for OS (HR = 1.88, 95% CI: 1.20–2.95, P = 0.0054; HR= 2.15, 95% CI: 1.47–3.15, P = 0.000060, respectively) (Fig. 2C and D). As with PFS, ^mutKRAS ctDNA kinetics between ctDNAb and ctDNAf also had a prognostic impact on OS (Fig. 3B).

Multivariable Analysis of ^mutKRAS ctDNA

Using the 5 most relevant independent prognostic factors derived from the prognostic model on OS from the complete study population (825 cases) as determined previously (21), a multivariable Cox regression was performed. Table 2 shows the results of the multivariable Cox model on PFS. Only the number of metastatic sites (HR = 3.03, 95% CI: 1.67–5.47, P = 0.00025) and ^mutKRAS ctDNAb (HR = 2.08, 95% CI: 1.05–24.13, P = 0.037) were multivariably significant prognosticators. Similarly, when ctDNAf was considered, only the number of metastatic sites (HR = 3.18, 95% CI: 1.74–5.82, P = 0.00017) and ctDNAf (HR = 1.96, 95% CI: 1.14–3.38, P = 0.015) showed a significant independent prognostic impact on PFS (Table 2).

An analogous Cox model was performed for OS based on 126 patients with a full parameter set. As shown in Table 3, although ctDNAb showed a prognostic impact numerically comparable to most of the other relevant factors, it failed to reach statistical significance because of limited sample size. As with PFS, the number of metastatic sites (HR = 1.83, 95% CI: 1.18–2.83, P = 0.0066) and the Eastern Cooperative Oncology Group (ECOG) performance status (HR = 1.52, 95% CI: 1.01–2.28, P = 0.044) were the only independently significant multivariable prognosticators for OS.

As expected from the univariable analyses, ctDNAf was the most pronounced and significant independent prognosticator for OS. In the Cox model on OS, in addition to ctDNAf (HR = 2.31, 95% CI: 1.40–3.25, P = 0.00047), the ECOG performance status (HR = 1.57, 95% CI: 1.24—3.03, P = 0.033) and the number of metastatic sites (HR = 1.94, 95% CI: 1.24—3.-03, P = 0.0037) were significant.

Discussion

We analyzed serial ^mutKRAS ctDNA profiles in patients with mCRC recruited within the framework of the AIO KRK0207 phase 3 trial to determine whether baseline (ctDNAb), early post-treatment initiation (ctDNAf) circulating tumor DNA concentrations, or ctDNA conversion could provide value for therapy surveillance and disease prognosis. The present study was a large study in mCRC from a randomized multicenter phase 3 trial to measure ctDNA at baseline and as early as 2–3 weeks after treatment initiation. We compared the performance of ctDNAb and ctDNAf to conventional protein biomarkers such as CEA and other clinically relevant factors (e.g., primary tumor location) in a large mCRC patient cohort.

Since sample collection in this trial was likely heterogenous in quality due to the large amount of participating small volume centers and due to the limited available sample volumes, we first analyzed 214 healthy serum and plasma samples of limited volume from healthy donors for rigorous analysis of our DNA isolation and ^mutKRAS detection method by digital droplet PCR to determine background cut-off of ^mutKRAS. We indeed found a difference in ^mutKRAS detection between plasma and serum (Supplemental Fig. 2). This result supports the importance of using the appropriate sample type for such studies.

The identification and frequency of low-level KRAS mutations in healthy blood donors observed in this study has also been reported in ctDNA analyses elsewhere (22–24). A recent study reported low level KRAS mutations in 30% of plasma samples from healthy controls using a further developed dd-PCR approach (24). Our rate of 13.5% of low level ^mutKRAS detection in healthy subjects was lower using a standard ddPCR approach. The young median age in the control group and the low amounts of plasma volume tested may be additional factors that affect the observed ^mutKRAS concentrations in healthy donors in our study. Another caveat is the high clinical specificity of the ddPCR-based analysis, thus allowing for mutation monitoring in patients at advanced disease stages, but sensitivity issues limiting mutation detection in healthy individuals as recently described (25). Given the lack of further information from healthy blood donors in our study, we cannot exclude the contribution of additional factors contributing to the ^mutKRAS concentrations.

Notably, ctDNA concentrations from healthy donors associate with donor age (26). We found a correlation of donor age and ^mutKRAS in healthy blood donors (Supplemental Fig. 5A) but not in our patient population (Supplemental Fig. 5B), although this analysis is limited by low case numbers. KRAS mutations have been reported in about 20% of nonpatient subjects, albeit with substantially lower mutation loads (22, 27). Notably, in our patient cohort ctDNA positivity was significantly associated with elderly patients. In addition to age, exposure to potential carcinogens (such as benzo-a-pyrene in smokers) may affect ^mutKRAS concentrations in noncancer subjects (27). Furthermore, healthy donors primed for cancer development may have ^mutKRAS in the circulation years before cancer development (28).

ctDNA analysis for mutation detection in patients with CRC provide the potential as a biomarker for prognosis, tailored therapeutic decision making, and for early prediction of treatment response. We observed that, although positive measures of both ^mutKRAS ctDNAb and ctDNAf were associated with unfavorable clinical phenotype and conventional prognostic factors such as CEA, only ^mutKRAS ctDNAf was predictive of better treatment outcome. Two recent studies in colorectal cancer reported a prognostic and predictive value of ctDNA concentrations. In one study, patients with lower ctDNA at 8 weeks after treatment showed a longer OS and PFS (29), while a ctDNA reduction before cycle 2 predicted response after 4 cycles of chemotherapy (30). In pancreatic cancer, a cancer with high frequency of KRAS mutations, patients with detectable KRAS mutations in ctDNA pre- or postsurgery showed inferior recurrence-free as well as overall survival (31). As ctDNAf concentrations were evaluated as early as 15-22 days post treatment initiation, this result supports the value of ctDNA measurements for early response prediction. Our data are in line with a recent report in advanced pancreatic cancer, where serial measurement of ^mutKRAS ctDNA was reported as a promising tool for early response prediction and monitoring tool for therapy response (32). A few studies have evaluated the use of baseline ctDNA as treatment response predictor in mCRC [overview in (33)]. One study of 20 patients evaluated ctDNA for response prediction to kinase inhibitors and reported worse OS and PFS for high ctDNAb and increased ctDNAf (34). Another study that included 115 patients with mCRC described baseline ctDNAb as an early response predictor (35). Similar observations were also reported in a study of 15 cases of non-small cell lung cancer, uveal melanoma, and colorectal cancer, in which ctDNA was measured at baseline and 8 weeks of treatment (36).

In line with previous reports in CRC (34, 35, 37) including a recent meta-analysis (38), our study confirms the prognostic significance of ^mutKRAS ctDNA. However, unlike previous studies, our study reveals ctDNAf as the numerically strongest and most significant prognostic factor for both overall and progression-free survival, outcompeting conventional protein-based biomarkers such as CEA or other important variables such as primary tumor location. The absolute ^mutKRAS load in ctDNAb showed a prognostic impact comparable with other factors but this did not reach statistical significance due to limited sample size.

Another finding in our study is the prognostic impact of ^mutKRAS ctDNA conversion. Our results reveal that positive to negative ^mutKRAS ctDNA conversion has a positive effect on both overall and progression-free survival, whereas a negative-positive ^mutKRAS ctDNA profile between baseline and follow-up has a negative impact on survival. A positive to negative ctDNAf conversion before start of the second therapy cycle was an early indicator of radiological disease control at the end of induction treatment with superior performance over CEA. ^mutKRAS ctDNA clearance before the second cycle was prognostic for higher overall and progression-free survival. Such early predictors of treatment and patient outcome are of profound clinical utility, as they could help to design adapted treatment strategies and spare some patient the toxic effects of some chemotherapeutic agent that will likely procure no clinical benefit.

Our study has several limitations. Collection and processing of blood samples was performed at multiple sites including small medical practices. Consequently, operating procedures and quality assessments are not as stringent and standardized as when performed in single high-volume institutions. Therefore, the setting represents clinical daily practice conditions and as such, our results reflect a real-life scenario. Furthermore, we were only able to obtain and process a maximum of 1 mL of blood from each sample, limiting the amount of overall ctDNA. Using an independent sample cohort for comparison of 4 different assay kits for isolation of ctDNA (data not shown), we found the method performed in this study to produce the most sensitive and consistent results in comparison to other assay kits. Given these limitations, ^mutKRAS ctDNA was detected beyond background concentrations in 113/153 (74.5%) cases with confirmed tissue KRAS mutations. This detection is in line with many previous reports and supports the principal realization of such an approach in a real-world setting.

In conclusion, our data from a large phase 3 trial suggest that serial ^mutKRAS ctDNA measurement is a strong prognostic factor in patients with ^mutKRAS mCRC. Evaluation of dynamics in ctDNA concentrations identified ^mutKRAS ctDNA clearance between baseline and follow-up with a positive impact on OS and PF, while positive ctDNA conversion (negative to positive) had a negative impact on survival. Notably, ^mutKRAS ctDNAf, which was analyzed 2 weeks after treatment initiation, was found to be a relevant early response prediction marker in multivariable analysis. These results may help guide future strategies exploring or validating liquid biopsy approaches.

Supplemental Material

Supplemental material is available at Clinical Chemistry online.

Prior Presentation: Presented as online publication at ASCO 2018 Congress, Chicago, IL, USA, June 1-5, 2018.

Author Contributions

All authors confirmed they have contributed to the intellectual content of this paper and have met the following 4 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved.

A. Herbst, provision of study material or patients; S.-T. Liffers, statistical analysis, administrative support, provision of study material or patients; P.A. Horn, provision of study material or patients; A. Reinacher-Schick, administrative support, provision of study material or patients; A. Hinke, statistical analysis; S. Hegewisch-Becker, provision of study material or patients; F.T. Kolligs, administrative support, provision of study material or patients.

Authors’ Disclosures or Potential Conflicts of Interest

Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:

Employment or Leadership

None declared.

Consultant or Advisory Role

A. Reinacher-Schick, Amgen, Baxalta, BMS, Celgene, Merck Serono, MSD, Onkowissen.de, Pfizer, Roche, Sanofi-Aventis, Servier; J.T. Siveke, AstraZeneca, Baxalta, BMS, Celgene, Novartis, Roche, Servier.

Stock Ownership

None declared.

Honoraria

S. S. Lueong, Merck advisory on liquid biospies; A. Reinacher-Schick, Amgen, Baxalta, BMS, Celgene, Lilly, Merck Serono, MSD, Pfizer, Roche, Sanofi-Aventis, Servier, Shire; A. Hinke, Roche Pharma AG, Germany; J.T. Siveke, AstraZeneca, Baxalta, BMS, Celgene, Novartis, Servier, Roche.

Research Funding

J.T. Siveke, the German Cancer Consortium (DKTK) and the B. Braun Foundation.

Expert Testimony

None declared.

Patents

None declared.

Other Remuneration

A. Reinacher-Schick, Amgen, Celgene, Ipsen, onkowissen.de, Roche, Servier.

Role of Sponsor

The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, preparation of manuscript, or final approval of manuscript.

Acknowledgment

We thank Simon Schaefer for excellent technical support. We thank Prof. Dr. Nils von Neuhoff for infrastructural support.

References

Nonstandard Abbreviations:

 
• CRC

  colorectal cancer

 
• mCRC

  metastatic colorectal cancer

 
• KRAS

  v-Ki-ras2 Kirsten rat sarcoma viral oncogene

 
• ^mutKRAS

  mutantKRAS

 
• CEA

  carcinoembryonic antigen

 
• ctDNA

  circulating tumor-derived DNA

 
• cfDNA

  circulating free DNA

 
• ctDNAf

  ctDNA at follow-up

 
• APC

  adenomatous polyposis coli

 
• TP53

  tumor protein p53

 
• PTEN

  phosphatase and tensin homolog

 
• PIK3CA

  phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit alpha

 
• BRAF

  v-raf murine sarcoma viral oncogene homolog B1

 
• ddPCR

  droplet digital PCR

 
• PFS

  progression-free survival

 
• OS

  overall survival