PDF
Citation
Print
Introduction
Anti–epidermal growth factor receptor (EGFR) antibodies are important agents in the treatment of metastatic colorectal cancer (mCRC). Research on biomarkers for predicting the therapeutic response to anti-EGFR antibodies has been extensive. While RAS and BRAF mutations are associated with resistance to anti-EGFR antibodies, the therapeutic implications of EGFR amplification are less well understood. Various studies have aimed to elucidate the biologic significance of EGFR amplification in mCRC, but its prognostic and predictive impact remains unclear [1-3]. According to a recent international cohort study, EGFR amplification was observed in approximately 4% of mCRC cases, predominantly in left-sided colon and rectal cancers harboring wild-type RAS and BRAF [4]. Overall survival (OS) was anticipated to be more favorable among patients without EGFR amplification compared with those with EGFR amplification, although the use of anti-EGFR antibodies was not related to OS.
EGFR amplification is associated with resistance to anti-EGFR tyrosine kinase inhibitors (TKIs) in lung cancer [5,6]. Relatively little is known about its other clinical significance. Studies have shown that patients with EGFR amplification have a higher enrichment in other genetic mutations compared with those without amplification, suggesting a greater likelihood of developing resistance to conventional treatments by activating various signaling pathways from co-occurring mutations [1,7,8]. Therefore, it is considered that anti-EGFR antibody treatment alone may have limitations in patients with EGFR amplification. Combining anti-EGFR antibodies with anti-EGFR TKIs, MEK inhibitors, or immune checkpoint inhibitors is expected to yield better outcomes. Thus, understanding the clinical and pathologic features of patients with EGFR amplification will provide important foundation for developing further treatment options.
This study aims to compare clinicopathologic features and clinical outcome of mCRC based on EGFR amplification status in a single center.
Materials and Methods
1. Study design and participants
A total of 2,421 patients with mCRC who underwent next-generation sequencing (NGS) using a targeted 244-gene panel (AMC Oncopanel, designed through SureDesign at Asan Medical Center) from January 2016 to December 2021 were identified from the electronic medical records (Fig. 1).
EGFR copy numbers were processed using CNVkit [9]. The slides of cases with at least five EGFR copies from the pipeline were reviewed for the tumor purity, inferred from variant allelic fraction pattern, to adjust the copy number [2]. Patients with an adjusted copy number of 6 or greater were defined as EGFR-amplified (EGFR amp+), while those with a copy number fewer than 6 were defined as EGFR-nonamplified (EGFR amp–) (Fig. 2).
Clinicopathologic features, including mutation profiles, microsatellite instability (MSI) status, primary tumor location, age, sex, stage at diagnosis, metastatic sites, and histologic type, were collected.
2. Molecular analysis
Various genomic data other than EGFR amplification obtained from NGS were also analyzed. KRAS/NRAS/BRAF mutations were identified in all participants, and patients were further divided into subgroups according to the presence of the mutations. Immunohistochemistry (IHC) of EGFR was performed when available. MSI status was also identified in all patients according to the previously published criteria [10].
3. Statistical analysis
Clinicopathologic features among patients with or without EGFR amplification were compared using chi-square tests or Fisher’s exact tests for categorical variables, and t tests or Mann-Whitney U tests for continuous variables, as appropriate.
OS was calculated as the time from diagnosis of mCRC until death or final follow-up of the patient. Progression-free survival (PFS) was defined as the time from the first day of treatment with a specific chemotherapy regimen until death, disease progression (as evaluated by the Response Evaluation Criteria in Solid Tumors criteria ver. 1.1) [11], or the final visit of the patient. Both clinical outcomes were assessed using the Kaplan-Meier method, and Cox proportional hazards regression models were used to identify covariates associated with OS or PFS.
All tests were two-sided with an alpha level of 0.05. Statistical analyses were performed using SPSS ver. 23.0 (IBM Corp.), STATA IC ver. 15.1 (Stata Corp.), and R ver. 4.2.2 (R Foundation for Statistical Computing).
Results
1. Clinicopathologic characteristics
Among 2,421 patients with mCRC who underwent NGS from January 2016 to December 2021, 211 had measured copy number ≥ 5 processed through CNVkit, and the adjustment to the exact cellularity was made by examining tumor samples from these patients (Fig. 1). Among 211 cases, 27% (56/211) were identified to have copy numbers with 5 or higher after the adjustment. The corrected copy numbers are illustrated in Fig. 2, which indicates that most of them (n=176) had copy numbers fewer than 6, leaving 35 patients with EGFR amplification (copy number ≥ 6). Thus, the overall prevalence of EGFR amplification in this population was 1.4% (35/2,421). The correlation between measured and corrected EGFR copy number is presented in Fig. 3, suggesting that most patients had marginal values (between 5 and 10) of copy numbers, and corrected copy numbers were generally lower than measured ones. To further explore the relationship between EGFR overexpression and amplification, an analysis using EGFR IHC was also conducted, which revealed no significant correlation with EGFR amplification (S1 Table).
As shown in Table 1, most EGFR amp+ cases were RAS/BRAF wild type, except for two patients with KRAS mutations. Among the 2,386 EGFR amp– patients, 1,143 (48%) were RAS/BRAF wild-type. EGFR amp+ tended to be associated with a predilection for the rectum without statistical significance (48.6% vs. 43.8%, p=0.58). All 35 EGFR amp+ patients were microsatellite-stable (MSS), while 78 out of 2,386 EGFR amp– patients (3.3%) showed microsatellite instability. EGFR amp+ patients tended to have fewer peritoneal seedings at presentation (8.6% vs. 21.8%, p=0.059). The proportion of patients who received anti-EGFR agents throughout the course was 45.7% vs. 23.3%, respectively, in EGFR amp+ and EGFR amp– groups. Among these cases, seven EGFR amp+ patients received anti-EGFR agents as first-line chemotherapy (20.0%), compared with 346 patients treated with anti-EGFR agents as first-line chemotherapy among EGFR amp– patients (14.5%). The combination of 5-fluorouracil, folinic acid, and oxaliplatin or irinotecan doublet was used in almost all patients receiving first-line anti-EGFR agents (100% vs. 99.7%).
2. Clinicogenomic characteristics
Co-alterations in patients with EGFR-amplified colorectal cancer were analyzed and summarized in an OncoPrint (Fig. 4). The most common alteration found was APC mutation, followed by TP53 mutations. The sidedness of tumor was also described, indicating a prominent distribution in the left colon.
3. Clinical outcomes of EGFR amplification in metastatic colorectal cancer
The median follow-up was 22.0 months (range, 1 to 207 months). OS tended to be better in EGFR amp+ patients (median OS, 76 months; 95% confidence interval [CI], 20.6 to 131.4) than in EGFR amp– patients (median OS, 37 months; 95% CI, 34.9 to 39.1) but the difference did not reach statistical significance (p=0.145) (Fig. 5A). Among the 572 patients who received anti-EGFR antibody–based chemotherapy during the course of their disease, the median OS was better in 16 EGFR amp+ patients at 79 months (95% CI, 37.7 to 120.3) compared with at 39 months (95% CI, 36.0 to 42.1) in 556 EGFR amp– patients (p=0.048) (Fig. 6). Univariate analysis of survival outcomes in patients treated with anti-EGFR therapy showed a trend between the absence of EGFR amplification and a higher hazard ratio, though it did not reach statistical significance (Table 2). The OS of patients with EGFR amplification with and without anti-EGFR treatment also did not show significant difference (79 months vs. not available, p=0.192) (Fig. 5B). Survival analysis based on EGFR overexpression (IHC 0-2+ vs. 3+) revealed no statistically significant differences (S2 Fig.).
Cox regression analysis for OS of all patients indicated sidedness, number of metastatic sites, and resection as clinically significant factors (S3 Table). However, when adjusting the survival outcomes according to EGFR amplification and other clinical factors including primary location of tumor, number of metastatic sites, stage at diagnosis, primary tumor resection, and metastasectomy, EGFR amplification did not have an impact on survival outcomes in those who received anti-EGFR antibodies (Table 3). A further stratified analysis was performed to evaluate the impact of factors that affect anti-EGFR treatment efficacy, including tumor sidedness and RAS/RAF wild-type mutations; no significant effect of sidedness or mutation was observed on survival (S4 Table). Only seven out of 35 EGFR amp+ patients were given front-line anti-EGFR chemotherapy, and their PFS did not differ from the PFS of EGFR amp– patients treated with first-line anti-EGFR chemotherapy (median PFS, 20 vs. 14 months; p=0.344) (Fig. 7).
Discussion
In this study, we elucidated the clinical features and significance of EGFR-amplified mCRC. EGFR amplification was more prevalent in MSS tumors with wild-type RAS/BRAF and was associated with a lower incidence of peritoneal seeding. While a favorable prognostic impact of EGFR amplification was observed in patients treated with anti-EGFR chemotherapy, the OS and PFS did not reach statistical significance in this study.
EGFR amplification has been defined using diverse methods and heterogeneous cut-off values. For instance, when analyzed using fluorescence in-situ hybridization (FISH), the usual cut-off values for the ratio of EGFR gene to chromosome 7 enumeration probe (CEP7) range from 2 to 3 [12,13]. However, the cut-off value for NGS has yet to be determined. In our analysis, we adopted a cut-off value of at least six copies from a multinational cohort study utilizing NGS, published in 2021 [4]. This cut-off value is also supported by another study on glioblastoma, which showed levels exceeding five EGFR gene copies in NGS were correlated with positive cases identified by FISH (EGFR/CEP7 ratio ≥ 2 in 15% of recorded cells) [14].
However, estimating copy number variation using data from targeted NGS panels is challenging. Most bioinformatic tools for detecting copy number variation are tailored for whole-genome or exome sequencing data, which limits their applicability to targeted gene panels. CNVkit, a bioinformatic pipeline for detecting copy numbers from targeted DNA sequencing, uses targeted and off-target reads to infer copy numbers across the genome, achieving exon-level resolution in targeted regions and sufficient resolution in intronic and intergenic regions [9]. In this study, copy numbers were estimated using CNVkit. However, they are still subject to biases due to gaps in coverage between enriched regions and can be substantially affected by tumor cell content. Copy numbers are often overestimated in samples with low tumor purity due to normal cell contamination. To address this limitation, we screened and identified 211 cases with copy numbers of 5 or higher using CNVkit. After adjusting for exact tumor cellularity, only 27% (56/211) of these cases had EGFR copy numbers of 5 or higher, as shown in Fig. 2. Notably, 90% of the cases had 5 to 10 copy numbers, as shown in Fig. 3B. This was consistent with a “shoulder” in EGFR copy number variation between 5 to 10 gene copies observed in glioblastoma cases [14]. These low copy number cases showed relatively few EGFR-amplified nuclei in FISH analysis compared with cases with higher copy numbers in this study, underscoring the importance of applying accurate tumor cellularity adjustments to estimate copy numbers accurately.
Previous studies for 30 to 500 patients with CRC reported the prevalence of increased EGFR copy numbers ranging from 6.9% to 88.9%. However, the concept of “increased EGFR copy number” included both amplification and high polysomy, defined as gene copies per nucleus greater than 3-4. Amplification detected by FISH also varied from 0% to 30%, due to the small number of study patients [15]. With the introduction of high-throughput analysis methods such as NGS into clinical practice, we can screen a larger number of patients for EGFR amplification, enabling more accurate prevalence estimates. In this study, the prevalence of EGFR amplification was 1.4%, using the criteria of copy number ≥ 6. A multicenter consortium study detected 62 EGFR-amplified cases out of 5,685 CRC samples (1.1%) [4]. Another study using a circulating tumor DNA-based NGS panel (Guardant 360) reported a prevalence of 16.3% (458 of 2,807) in colorectal cancer [1]; however, this higher prevalence could be attributed to the use of a lower cut-off value, with a median amplification level of 2.55. Further investigation is warranted to determine the optimal level of cut-off value of EGFR amplification that correlates with clinical outcomes from EGFR-targeted strategies, especially considering the correlation with diverse co-alterations in hyper-amplified cases, which play a crucial role in developing resistance to anti-EGFR therapy.
The clinical features of EGFR-amplified colorectal cancer in this study align with previous studies, demonstrating a left-side dominant distribution, rare peritoneal seeding, and mutually exclusive RAS/BRAF mutations [4,16]. Patients with EGFR-amplified tumors exhibited diverse co-alterations other than RAS/BRAF mutations, with APC mutations being the most common, followed by TP53 mutations. This suggests that EGFR-amplified colorectal cancer may be classified as an extreme type of consensus molecular subtype 2 (CMS2) of colorectal cancer [17], which is also prominent in left colon cancer. The CMS2 subtype is characterized by higher somatic copy number alterations, enriched copy number gains of oncogenes, losses in tumor suppressor genes, upregulated MYC downstream signaling, and frequent APC mutations [18]. The predictive and prognostic value of CMS subtypes was assessed in a phase III trial comparing cetuximab with bevacizumab (CALGB/SWOG 80405), showing that the CMS2 subtype benefited most from cetuximab. This underscores the need to evaluate EGFR-amplified colorectal cancer in relation to the CMS2 subtype and their reliability as prognostic markers [19].
Previous studies on the benefits of anti-EGFR agents in EGFR-amplified patients have yielded varying results. A phase II trial did not find a correlation between EGFR copy number and the efficacy of anti-EGFR agents [3], while a single-center retrospective study indicated potential responsiveness to cetuximab only in EGFR-negative tumors [20]. Although not reaching statistical significance, this study suggested improved OS among patients with EGFR amplification. Patients treated with anti-EGFR agents had a lower hazard ratio. Our results suggest that EGFR-amplified colorectal cancer is a subtype that can benefit from anti-EGFR agents. This finding aligns with a meta-analysis of 19 studies on the predictive value of EGFR gene copy number for anti-EGFR monoclonal antibody treatments, where EGFR amplification was associated with improved outcomes [15]. One explanation for this result is that EGFR amplification is related to high reactive oxygen species levels, leading to increased chemosensitivity [21]. A Finnish study showed that anti-EGFR treatment yielded better outcomes for EGFR-amplified patients, significantly suppressing downstream signaling in EGFR-amplified, KRAS wild-type tumors compared with unamplified, wild-type tumors through cell line studies [22]. This suggests that EGFR signaling is abundant in distal colon cancer, potentially leading to better expectations for EGFR-targeted treatment [16]. However, further investigation is needed regarding the efficacy of therapy.
In lung cancer, EGFR amplification is associated with resistance mechanisms to EGFR TKIs. Amplification of the EGFR wild-type allele occurs in TKI-resistant EGFR-mutated cells, which might contribute to resistance [6]. Combining TKIs with cetuximab restores sensitivity in resistant cells, which suggests that anti-EGFR antibodies could help overcome EGFR amplification–mediated resistance. This notion is consistent with our study; unlike EGFR TKI, EGFR amplification may be a therapeutic target rather than a resistant marker for anti-EGFR antibodies. This highlights the importance of interpreting biomarkers within the specific context of each cancer type, genomic profile, and type of treatment and suggests that our results could help inform treatment strategies for mCRC.
Recently, the potential activity of a combination of bevacizumab plus erlotinib has been proposed, as resistance to EGFR blockade might be attributed to vascular endothelial growth factor, which shares a common downstream pathway with EGFR [23-25]. Trials in non-small cell lung cancer have shown improved PFS with combination therapy compared with erlotinib alone. Preliminary data from the GERCOR DREAM trial in colorectal cancer suggested that the combination of bevacizumab and erlotinib as maintenance therapy might be superior in terms of PFS and OS [26,27]. Notably, the anti-tumor activity of erlotinib did not depend on KRAS mutations in this trial, unlike other anti-EGFR agents such as cetuximab or panitumumab. However, the same combination in lung cancer showed significantly better outcomes in cases with the L858R mutation [28], suggesting the potential for maximum efficacy in EGFR signaling in colorectal cancer and patients with EGFR amplification.
The strength of this study lies in the substantial dataset of over two thousand mCRC cases, which enabled the inclusion of as many EGFR-amplified cases as possible, despite the low prevalence of EGFR amplification. However, this study has several limitations. First, as a retrospective study based on medical records from a single center, there is some inevitable information loss. Selection bias may also exist due to the characteristics of our center and the severity of the cases it handles. Nevertheless, the inclusion of over 2,000 patients with diverse features helps mitigate this bias. Second, long-term data were often not available due to patients being lost to follow-up when referred to local centers for later-line treatments and supportive care, which could have influenced OS and PFS.
In conclusion, this single-center analysis elucidates the clinical features and significance of EGFR-amplified metastatic colorectal cancer. Unlike lung cancer, EGFR amplification does not confer resistance to anti-EGFR antibodies in metastatic colorectal cancer, and it suggests favorable prognostic impact in patients treated with anti-EGFR chemotherapy.
XML Download