Amphiregulin can predict treatment resistance to palliative first-line cetuximab plus FOLFIRI chemotherapy in patients with RAS wild-type metastatic colorectal cancer

Amphiregulin (AREG) is an epidermal growth factor receptor (EGFR) ligand. The aim of this study was to investigate the effects of baseline plasma AREG levels in KRAS, NRAS, and BRAF wild-type metastatic colorectal cancer (CRC) on treatment outcome with palliative first-line cetuximab + FOLFIRI chemotherapy. Chemotherapy outcomes were analyzed based on baseline plasma AREG levels. The clinical findings were further validated using an in vitro model of CRC. Among 35 patients, the progression-free survival (PFS) was significantly inferior in patients with high AREG than in those with low AREG levels: 10.9 vs. 24.2 months, respectively (p = 0.008). However, after failure of first-line chemotherapy, AREG levels were associated with neither PFS (4.8 vs. 11.6 months; p = 0.215) nor overall survival (8.4 vs. 13.3 months; p = 0.975). In SNU-C4 and Caco-2 cells which were relatively sensitive to cetuximab among the seven CRC cell lines tested, AREG significantly decreased the anti-proliferative effect of cetuximab (p < 0.05) via AKT and ERK activation. However, after acquiring cetuximab resistance with gradual exposure for more than 6 months, AREG neither increased colony formation nor activated AKT and ERK after cetuximab treatment. Our results suggest that plasma AREG is a potential biomarker to predict clinical outcomes after cetuximab-based chemotherapy.

www.nature.com/scientificreports/ may have worse treatment outcomes with anti-EGFR therapy. However, in some studies, conflicting results have been reported on the effects of AREG on anti-EGFR treatment outcomes [11][12][13] . Therefore, in this study, we aimed to investigate the effects of baseline plasma AREG levels in patients with colorectal cancer treated with palliative first-line cetuximab plus FOLFIRI regimen and subsequent second-line chemotherapy. In addition, we conducted an in vitro study to further define the molecular mechanisms of AREG in cetuximab-naïve and cetuximab-resistant colorectal cancer cell lines.

High baseline plasma AREG levels are associated with inferior progression-free survival. From
May 2015 to September 2019, a total of 35 patients were consecutively included in this study. All patients were treated with palliative first-line chemotherapy consisting of cetuximab plus FOLFIRI. Among them, 21 patients (60.0%) had left-sided colorectal cancer and most of the cancers (n = 29, 82.9%) showed moderate differentiation. All patients were evaluated for RAS mutation; no one had RAS mutation at the time of diagnosis. BRAF mutation was evaluated in 34 patients; 2 patients (5.9%) had BRAF mutant cancer ( Table 1).
The median follow-up time was 19.1 months (range 1.2-56.0 months). Among patients whose disease had target lesions defined by RECIST version 1.1 (n = 34), 21 patients (60.0%) showed partial response ( Table 2). In Table 1. Baseline characteristics of patients. ECOG PS Eastern Cooperative Oncology Group performance status, NOS not otherwise specified, MSI microsatellite instability, MSS microsatellite stable, MSI-L microsatellite instability-low, MSI-H microsatellite instability-high. a Percentage of RAS, BRAF mutation, and MSI status was calculated for patients whose data were available. the entire patient cohort, the median PFS and OS were 17.0 months (95% CI 15.0-51.1 months) and 33.0 months (95% confidence interval [CI] 8.7-25.3 months), respectively (Fig. 1A,B).
To determine the effect of cetuximab combined with AREG on the EGFR signaling pathways, Western blot analyses were conducted (Fig. 3B). AREG increased the phosphorylation of EGFR Y1068 (p-EGFR), which is the most sensitive site to EGFR ligand stimulation among p-EGFR residues, in Caco-2 and SNU-C4 cells 14,15 . In AREG-treated Caco-2 cells, the changes of EGFR downstream molecules involving phosphorylated AKT (p-AKT) and phosphorylated ERK1/2 (p-ERK1/2) were not apparent. In contrast, the phosphorylation of AKT and ERK1/2 strongly increased by adding AREG to SNU-C4 cells, indicating a more responsive phenotype against AREG in terms of signaling pathway activation compared with Caco-2 cells. Further, cetuximab monotherapy decreased the phosphorylation of EGFR, AKT, and ERK1/2 in both cell lines. To examine the effects of AREG on cetuximab response, AREG was administered prior to cetuximab treatment in Caco-2 and SNU-C4 cells. In www.nature.com/scientificreports/ both Caco-2 and SNU-C4 cells, AREG partially restored the cetuximab-induced inhibition of EGFR downstream molecules, AKT and ERK1/2. To assess the effect of cetuximab in the presence or absence of AREG on the cell cycle in Caco-2 and SNU-C4 cells, we conducted flow cytometry (Fig. 4A,B). We sought that cetuximab monotherapy significantly increased the G0/G1 phase but decreased the S and G2/M phase in Caco-2 and SNU-C4 cells. However, in the presence of AREG, cetuximab-mediated cell cycle alteration was partially reversed in Caco-2 and SNU-C4 cells. Particularly, in the presence of AREG, subG1 phase was significantly repressed with the cetuximab treatment in both cells, indicating that AREG might reduce the activity of cetuximab in Caco-2 and SNU-C4 cells.
To verify whether AREG disturbs cetuximab-mediated apoptosis, the caspase-3/7-positive apoptotic cells were measured in Caco-2 and SNU-C4 cells with AREG in the presence or absence of cetuximab (Fig. 4C,D). AREG alone did not alter the number of apoptotic rates. Notably, in the presence of AREG, the cetuximabmediated apoptotic cells were significantly suppressed in both Caco-2 and SNU-C4 cells compared with AREGfree condition.
In addition, the colony-forming assays were carried out to analyze the prolonged effect of cetuximab in the presence or absence of AREG on the proliferation of Caco-2 and SNU-C4 cells (Fig. 4E). We observed that cetuximab monotherapy significantly decreased the growth of colonies compared with the control in Caco-2 and SNU-C4 cells. In the absence of cetuximab treatment, high AREG did not significantly increase cell proliferation in Caco-2 cells, whereas in SNU-C4 cells, the proliferation was significantly increased in the presence of AREG relative to that in the control. The colony-forming assay results appeared to be correlated with the western blot data (Fig. 3B): SNU-C4 cells showed higher susceptibility to the effect of AREG on EGFR downstream molecules, AKT and ERK1/2, possibly resulting in higher proliferation, compared with Caco-2 cells. Notably, in the presence of AREG, the anti-proliferative effects of cetuximab were significantly decreased in both Caco-2 and SNU-C4 cells compared to AREG-free environment.
We examined the influence of AREG on the EGFR signaling pathways by western blot analysis in Caco-2_R and SNU-C4_R cells (Fig. 5C). In line with our western blot data in the previous section (Fig. 3B), AREG increased the phosphorylation of EGFR, AKT, and ERK1/2 in SNU-C4_P cells, whereas SNU-C4_R cells did not respond to AREG. In contrast, in Caco-2_P cells, high AREG did not cause AKT phosphorylation, but led www.nature.com/scientificreports/ to ERK1/2 phosphorylation. High AREG slightly increased the expression of p-ERK1/2 in Caco-2_R1 cells but had no such effect in Caco-2_R2. In Caco-2_R1 and Caco-2_R2 cells, p-AKT slightly decreased in the presence of AREG ( Supplementary Fig. S1). Further, in cetuximab resistant cells, EGFR and its downstream signaling molecules were examined after cetuximab treatment in the presence or absence of AREG (Fig. 5D). Based on the anti-proliferation assay data (Fig. 5B), relatively more resistant clones, Caco-2_R1 and SNU-C4_R2, were selected for the study. In these cetuximab-resistant cells, additional western blot analysis showed that, after cetuximab treatment, the presence or absence of AREG did not elicit apparent differences in terms of EGFR downstream signaling (Fig. 5D).
Next, we carried out colony-forming assays to determine the prolonged effect of AREG on proliferation in cetuximab-resistant cells (Fig. 5E). Colony-forming assays of Caco-2_P and SNU-C4_P cells showed similar www.nature.com/scientificreports/ The cells were also exposed to the indicated concentrations of AREG for 96 h. Apoptosis was measured using the Caspase-Glo ® 3/7 assay kit. Results indicate mean ± standard deviation. The values were compared using the Student's t-test. *p < 0.05; **p < 0.01; ***p < 0.001. (E) Colony-forming assays were conducted to evaluate the long-term effect of AREG. AREG 50 ng/mL and cetuximab (10 μg/mL for Caco-2 and 20 μg/mL for SNU-C4) were added to the cells for 3 to 4 weeks. The colonies were quantified for optical density at 595 nm. www.nature.com/scientificreports/ www.nature.com/scientificreports/ results as in cetuximab-naïve cells (Fig. 4E). However, in Caco-2_R1 and SNU-C4_R2 cells, the anti-proliferative effects of cetuximab were not significant both in the presence and absence of AREG.

Discussion
This study was conducted to investigate the effects of AREG on cetuximab response in patients with metastatic colorectal cancer and determine the mode of action via the EGFR signaling pathway using in vitro experiments with both cetuximab-naïve and resistant colorectal cancer cell lines. Our results indicated that high baseline plasma AREG levels could predict PFS after first-line cetuximab plus FOLFIRI chemotherapy in patients with RAS/BRAF wild-type colorectal cancer. Our in vitro study also indicated that AREG significantly enhanced the growth of cetuximab-sensitive RAS/BRAF wild-type colorectal cancer cell lines with increased phosphorylation of AKT and ERK1/2. However, after progression of first-line chemotherapy, the plasma AREG levels could not predict treatment outcomes of second-line non-EGFR-directed treatment, suggesting that AREG may not be a prognostic biomarker in these patients. In line with this clinical finding, in cetuximab-resistant cell lines, AREG neither impeded the signal transduction inhibition by cetuximab nor was associated with the cetuximab-induced anti-proliferative effects. It is still controversial whether AREG protein or mRNA expression levels are prognostic factors in patients with colorectal cancer who did not undergo cetuximab-or panitumumab-based chemotherapy. For example, Ohchi et al. showed that higher AREG expression by IHC was significantly associated with worse relapse-free survival and OS 16 . In contrast, Stahler et al. reported that AREG mRNA expression in tumor tissue was not prognostic 17 . Besides, several studies demonstrated that higher AREG expression levels by tissue immunohistochemistry (IHC) and serum ELISA were associated with worse pathologic parameters [18][19][20] .
Other studies evaluated the predictive role of mRNA expression of AREG in the patients receiving cetuximab-or panitumumab-based treatment. In most studies, mRNA expression of AREG predicted better response to anti-EGFR treatment. However, mRNA-based analyses are less likely to be reproducible and thus difficult to adopt in clinical practice. Yonesaka et al. evaluated plasma AREG levels by ELISA in colorectal cancer patients treated with cetuximab in combination with cytotoxic chemotherapy 21 . However, treatment in the patients were heterogeneous in terms of combined cytotoxic chemotherapy and prior lines of treatment. In addition, this study evaluated the role of plasma AREG levels in combination with heregulin, therefore, further validation is required to determine whether plasma AREG levels by ELISA could be a predictive biomarker for cetuximabbased treatment.
Furthermore, our study group previously conducted a biomarker study of serum AREG in patients with HER2-positive breast cancer treated with trastuzumab plus taxane and found that high AREG levels were associated with poor clinical outcomes 22 . EGFR and HER2 belong to the epidermal growth factor family of receptor tyrosine kinases (ErbBs), which includes EGFR, HER2, HER3, and HER4 23 . HER2 is a receptor which has no identified ligand but shares the same downstream pathway with EGFR. AREG is a ligand of EGFR, which can form a heterodimer with HER2, leading to down-stream signaling activation 24 . Therefore, our interpretation in our previous study was that AREG indirectly confers trastuzumab resistance via EGFR activation in HER2positive breast cancer. In contrast, in the present study, AREG-induced cetuximab resistance may be attributed to direct EGFR activation in RAS/BRAF wild-type colorectal cancer.
We further analyzed the biological mechanisms of AREG using in vitro experiments. In our clinical data, plasma AREG was a negative predictive biomarker of cetuximab-based treatment, which was validated in our in vitro study as well. However, it was no longer predictive after patients acquired cetuximab resistance. Thus, it is speculated that AREG interferes with the inhibitory effect of cetuximab on EGFR. To prove the hypothesis, we generated cells with acquired cetuximab resistance using cetuximab-sensitive Caco-2 and SNU-C4 cell lines. As a result, AREG could not activate the EGFR down-stream signaling in cetuximab-resistant Caco-2 and SNU-C4 cells.
Based on these findings, it is hypothesized that AREG inhibition might have a therapeutic potential in patients receiving anti-EGFR treatment. Several preclinical studies have investigated the effect of AREG inhibition using anti-AREG antibodies or AREG shRNA in various human cancer cell lines [25][26][27] . However, there is still no clinical data on AREG inhibition in patients with high plasma AREG levels receiving anti-EGFR treatment. Our results may provide evidence of an AREG-targeting strategy in patients with RAS/BRAF wild-type colorectal cancer receiving cetuximab-based chemotherapy.
One of the limitations of our study is a small patient population and lack of validation cohort. Hence, it should be further validated in a larger patient cohort whether amphiregulin is an independent biomarker in this patient population. In addition, our experiment did not include various mechanisms of action of cetuximab including the antibody-dependent cell mediated cytotoxicity 28 . In this regard, further research is warranted.
In conclusion, high baseline plasma AREG levels predicted worse PFS in patients with RAS/BRAF wild-type metastatic colorectal cancer treated with palliative first-line cetuximab plus FOLFIRI chemotherapy, but not with non-EGFR-directed second-line treatment. AREG reversed the inhibitory effects of cetuximab in RAS/BRAF wild-type cetuximab-naïve colorectal cancer cells, but not in cetuximab-resistance cells in vitro. Our results suggest that plasma AREG is a potential biomarker to predict clinical outcomes after cetuximab-based chemotherapy. www.nature.com/scientificreports/ m 2 bolus, and then 5-fluorouracil 2400 mg/m 2 infusion over 46 h every 2 weeks. Before the first dose of cetuximab + FOLFIRI treatment, blood samples were collected after obtaining informed consent during two preceding prospective studies: pharmacogenomic study in patients with advanced solid tumors receiving palliative chemotherapy (IRB registration No. B-1603/340-305) and biomarker study in patients with metastatic colorectal cancer (IRB registration No. B-1211/180-007). The present study was conducted using these consecutively collected blood samples. Retrospective medical record reviews were conducted for these patients. The study was approved by the institutional review board of Seoul National University Bundang Hospital (IRB registration No

Establishment of acquired cetuximab-resistant colorectal cancer cell lines. Cetuximab-resistant
Caco-2 and SNU-C4 cell lines were generated by continuous exposure to increasing concentrations of cetuximab up to 400 μg/mL for 3 months, and then maintained in cetuximab 400 μg/mL-containing media for more than 6 months (Fig. 3A). Cell cycle analysis. The Caco-2 and SNU-C4 cells were plated in 60-mm culture dishes at a density of 1 × 10 6 cells per dish for 72 h, and then treated with AREG and/or cetuximab. Next, the cells were harvested at 72 h by trypsinization, centrifugation, and fixation with 70% ethanol. A staining solution containing 0.05 μg/mL of propidium iodide (PI) and 0.2 mg/mL of RNase was used for DNA staining. Attune ® NxT cytometer (Thermo Fisher Scientific) was used for cell cycle analysis. The flow cytometry data was processed using FlowJo software (BD Biosciences, San Jose, CA, USA). www.nature.com/scientificreports/ Apoptosis assay. Apoptosis assays were performed according to manufacturer's recommendations using the Caspase-Glo ® 3/7 assay kit (Promega). Significant differences between values obtained in DMSO control groups and different treatment groups were determined using the Student's t-test.
Colony-forming assay. Cells were seeded in 6-well plates at 1000 cells/well (30006, SPL Life Science, Pocheon, Korea). The treatments, 10 and 20 μg/mL of cetuximab and 50 ng/mL of AREG, were added three times weekly for 3 or 4 weeks to SNU-C4 or Caco-2, respectively. The cell colonies were stained with Coomassie brilliant blue R-250 solution (1610436, BioRad) for 2 h at room temperature. The representative colony images were photographed by ChemiDoc Touch Imaging System. The procedure of colony de-staining was performed using a 1% SDS solution to analyze the relative colony area. The relative colony area was determined at 595 nm using a Synergy H1 microplate reader. Protein expression was analyzed by ImageJ software (NIH, Baltimore, MD, USA).
Statistical analysis. The Eastern Cooperative Oncology Group (ECOG) scale was used to evaluate performance status (PS) of patients. KRAS, NRAS, and BRAF mutation status was evaluated using polymerase chain reaction (PCR) or next generation sequencing as a part of clinical practice. Microsatellite instability was determined according to the revised Bethesda guidelines (14). Disease status was evaluated using computed tomography (CT) scans or magnetic resonance imaging (MRI). Treatment response was assessed in accordance with the response evaluation criteria in solid tumor (RECIST) version 1.1 29 .
To identify the optimal cut-off value for plasma AREG levels, Maxstat R package (a maximal chi-square method) was used in R 4.0.3 (R Development Core Team, Vienna, Austria). PFS was defined from the start of systemic chemotherapy to the date of documented disease progression, relapse, or death of any cause. The second PFS (PFS2) was defined from the start of second-line anticancer treatment to the date of documented disease progression, relapse, or death of any cause. OS was measured from the start day of systemic chemotherapy to the date of death of any cause. The second OS (OS2) was defined from the start of the second-line anticancer treatment to the date of death of any cause. The survival outcomes were calculated using the Kaplan-Meier method and compared with the log-rank test.
All in vitro experiments were conducted at least three times independently. The differences between groups were evaluated by Student's t-test using GraphPad Prism 5 software (San Diego, CA, USA). Error bars represent mean ± standard deviation. Two-sided p-values < 0.05 were considered statistically significant (*p < 0.05, **p < 0.01, and ***p < 0.001).

Data availability
The datasets generated during and/or analysis during the current study are available from the corresponding author on reasonable request.