Optimal use of anti-EGFR monoclonal antibodies for patients with advanced colorectal cancer: a meta-analysis

Worldwide colorectal cancer is the second most common cancer in women and the third in men [1]. Irresectable, non-curable colorectal cancer can be treated with palliative chemotherapy to reduce cancer symptoms, improve quality of life, and overall survival. In 1963, Heidelberger et al. discovered 5-fluorouracil (5-FU) as the first systemic chemotherapy for colorectal cancer. To date, this is the most effective and widely used systemic treatment for colorectal adenocarcinoma. In the course of time, combinations of fluoropyrimidines, with oxaliplatin and irinotecan, as well as the use of anti-EGFR and anti-VEGF targeted agents have improved survival of patients with metastasized colorectal cancer (mCRC) to about 2.5 years [2].

Cetuximab, a chimeric human and mouse monoclonal antibody (mAb), and panitumumab, a fully human mAb, both bind the epidermal growth factor receptor (EGFR). This prevents activation of the intracellular EGFR tyrosine kinase, resulting in an inhibition of the associated downstream signaling pathways, such as the RAS-RAF-MAPK and the PI3K-PTEN-AKT axis [3]. Additionally, antibody-dependent cell-mediated cytotoxicity (ADCC) may play a role in the efficacy of anti-EGFR mAb.

In 2008, a retrospective analysis revealed that the presence of mutations in Kirsten rat sarcoma viral oncogene homolog (KRAS) exon 2 has a negative predictive value for benefit from anti-EGFR therapy [4, 5]. Recently, the same was demonstrated for mutations in KRAS exons 3 and 4, and for the rare mutations in neuroblastoma rat sarcoma (NRAS) viral oncogene homolog exon 2–4 [6]. Despite patient selection based on wild-type (WT) RAS status, approximately 30% will not have clinical benefit from anti-EGFR mAb treatment [7]. Therefore, additional predictive biomarkers are needed.

In multiple clinical trials, the efficacy of anti-EGFR mAb has been evaluated as monotherapy or combined with different types of chemotherapy in patients with mCRC. Yet, the optimal sequence and combination for the use of anti-EGFR therapy remains unclear. With this meta-analysis, we aim to get more insight in the optimal clinical strategy for the use of anti-EGFR therapy. All included randomized controlled clinical trials in a KRAS WT mCRC population compared the additional benefit of anti-EGFR mAb therapy to first- or second-line chemotherapy treatment or to best supportive care in third-line treatment. We pooled efficacy data to objectify and compare overall response rate (ORR), progression-free survival (PFS), and overall survival (OS) for each treatment line. With meta-regression, the influence of the chemotherapeutic backbone and type of anti-EGFR mAb were analyzed. Furthermore, we evaluated whether the addition of anti-EGFR mAb is superior to anti-VEGF mAb in first-line treatment.

With our literature search, 1856 records were obtained; 1803 records did not meet the inclusion criteria based on title and abstract (Fig. 1). Of the remaining 53 records, 37 records were excluded based on full text review. Reasons for exclusion were anti-EGFR mAb in both arms (19%), a misbalance between treatment arms (19%), non-randomized trials (17%), sub-analysis of an original article (17%), KRAS-mutated population (11%), the combination of anti-EGFR mAb with anti-VEGF mAb (8%), and no reported efficacy data (8%) (Fig. 1). The 17 included publications are summarized in Tables 1 and 2. Pooled analyses were done for six first-line studies (n = 2580 patients), two second-line studies (n = 1057), and two third-line studies (n = 444). Of these trials, four studies published RAS WT data (n = 1464 patients), which were pooled. Additionally, three first-line studies that compared anti-EGFR mAb with anti-VEGF mAb were pooled (n = 2014 patients). Risk of bias was assessed using the Cochrane collaboration’s for randomized trials; all studies had a fairly low risk for bias (Supplementary Fig. 1).

The aim of this meta-analysis was to define the optimal use of anti-EGFR mAb in the non-curative treatment of mCRC. To determine its optimal use, we evaluated benefit (e.g., ORR, PFS, and OS) and toxicity that resulted from anti-EGFR mAb addition to standard chemotherapy or as monotherapy. Additionally, we assessed the influence of the treatment line as well as the chemotherapeutic backbone on efficacy of anti-EGFR mAb.

To determine optimal treatment for patients with non-curative mCRC, OS is the most important clinical outcome measure. However, benefit in OS is difficult to objectify due to the influence of subsequently applied systemic or, in some cases, local treatment strategies. Additionally, subsequent treatment data are often incompletely collected and reported. In the first- and second-line studies, there was no clear benefit in OS for the KRAS WT cohorts receiving anti-EGFR mAb therapy (HR 0.89, p = 0.13; HR 0.96, p = 0.54, respectively), but retrospective analyses in the RAS WT population indicated that the addition of anti-EGFR mAb to first-line chemotherapy significantly improved OS (HR 0.77, CI 0.67–0.89). In the third line, only the CO.17 trial provided a correct representation of benefit in OS with a HR of 0.55 (p < 0.001) compared to BSC. Benefit in PFS due to the addition of anti-EGFR mAb was comparable in first and second line (HR 0.79 and 0.80, respectively). In third line, the HR for PFS was 0.43. Yet, this greater effect of anti-EGFR mAb treatment is partly caused by the fact that it is compared to BSC, as the added median time to PFS with the addition of anti-EGFR mAb was comparable between all treatment lines. The last outcome measure, ORR, is often not paramount in a non-curative setting. An exception is the intent to convert unresectable to resectable disease, frequently a point of discussion for patients with unresectable liver-limited CRC metastases. In literature, the addition of anti-EGFR mAb to neo-adjuvant treatment has given contradictory results [23, 24]. Currently, a prospective multicenter RCT is including patients with liver-limited CRC metastases to further investigate the role of anti-EGFR mAb to convert irresectable to resectable disease [25]. In our study, pooled efficacy data revealed a significantly higher ORR in all treatment lines with the addition of an anti-EGFR mAb. However, based on the evaluated data, no obvious differences in added gain of OS, PFS, and ORR were demonstrated with the addition of anti-EGFR mAb in first-, second-, and third-line treatment to guide its optimal clinical use.

It has been suggested that the addition of an anti-EGFR mAb to irinotecan-based regimen has a synergetic effect, opposed to oxaliplatin-based regimens [26, 27]. With meta-regression, we have shown that anti-EGFR mAb rendered a significantly better ORR in the irinotecan-based regimen compared to the oxaliplatin-based regimen (p = 0.04). However, this superior effect was not confirmed for PFS and OS (p = 0.09 and p = 0.06, respectively). These results are in concordance with multiple randomized trials, which demonstrated no differences in treatment benefit for irinotecan and anti-EGFR mAb compared to oxaliplatin and anti-EGFR [8, 28‐30].

The addition of either anti-EGFR mAb or anti-VEGF mAb to first-line non-curative chemotherapy is a much-debated current clinical issue. Pooled data of all available first-line studies, which compared the addition of both antibodies, demonstrated a comparable ORR and PFS, while OS was significantly longer in the anti-EGFR arm (HR of 0.8, CI 0.65–0.97). The benefit in OS was unexpected, as it is not in line with the other outcome measurements such as ORR and PFS. The significant benefit in OS was observed in two studies, the FIRE-3 [22] and the PEAK [21]. In the FIRE-3, ORR (the primary endpoint) as well as the PFS were similar in the cetuximab arm compared to the bevacizumab arm (62 versus 58% and 10.0 versus 10.3 months), while OS was 28.7 versus 25.0 months (p = 0.017, respectively) [22]. Although a recent post hoc analysis on centrally reviewed CT images reported a significant improvement with the addition of anti-EGFR mAb (72 versus 56%; p = 0.003), PFS remained similar between the anti-EGFR and the anti-VEGF mAb arms (8.4 versus 9.7 months; p = 0.53) [31]. The observed discrepancy between PFS and OS is most likely caused by small imbalances in subsequent treatments, such as more use of oxaliplatin and fluoropyrimidine in the anti-EGFR mAb arm [32]. Additionally, only 52% of the patients in the anti-VEGF mAb arm received anti-EGFR mAb treatment, compared to all patients in the anti-EGFR arm [32]. In fact, the PFS for the second-line therapy in the anti-EGFR mAb arm was significantly longer (HR 0.68, p < 0.001) [32], indicating that indeed the differences in OS were not caused by first-line treatment but by differences in later treatment. Perhaps, data of the patient distribution of left versus right sidedness in both arms might provide more insight in this discrepancy between ORR, PFS, and OS. For the smaller PEAK trial, no data concerning the subsequent treatment were published [21]. In the largest trial, which evaluated anti-EGFR mAb versus anti-VEGF mAb as addition to first-line treatment, the CALGB/SWOG 80405 trial, 1137 patients were treated with FOLFIRI and cetuximab or bevacizumab. No significant differences in PFS or OS were found between the two arms. Unfortunately, no full paper of this study has been published yet, precluding detailed analysis. In the second line, the SPIRITT trial is the only study that compared anti-EGFR with anti-VEGF mAb in addition to combination chemotherapy (FOLFIRI), after progression on first-line, oxaliplatin-based treatment with bevacizumab. As there are no other studies comparing these two agents in second line, we could not pool data. We did not pool this study with the three first-line studies, because of the difference in treatment setting. In the SPIRITT phase 2 trial, PFS and OS were similar for the two regimens [33]. It is remarkable that no difference was observed, because patients received bevacizumab in first line, probably not all patients were truly resistant to the first-line treatment regimen. Oxaliplatin toxicity may have influenced the switch to second-line treatment. Nevertheless, these data suggest that the addition of either mAb to this treatment setting improves clinical outcome.

It might be argued that first-line treatment is the most important treatment line, and responsible for the main gain in OS. This is the key rationale to add anti-EGFR mAb to first-line treatment. However, in the last decade, OS improved with 10 months, whereas PFS of the first-line treatment remained the same [34], indicating that the gain in OS most likely results from improved care for patients by using the total arsenal of available systemic treatments. In addition, multiple studies showed that survival benefit of combination therapy is comparable to sequential therapy, whereas combination therapy is significantly more toxic [35‐40]. Indeed, in our meta-analysis, the occurrence of any ≥3 grade adverse events in first-line combination treatment was 82% compared to 64% with standard chemotherapy. Based on these data, using anti-EGFR mAb as third-line monotherapy is a sensible and tolerable treatment option after progression on standard (combination) chemotherapy.

In order to improve efficacy of anti-EGFR treatment, predictive biomarkers are urgently needed. An obvious biomarker, EGFR expression on tumor tissue, did not correlate with treatment benefit [8, 27‐29]. A well-established biomarker that predicts primary resistance to anti-EGFR mAb are RAS mutations (KRAS and NRAS exon 2–4). This meta-analysis only included studies excluding patients with KRAS exon 2 mutated tumors, as these were the first and most common mutations known to induce resistance. Thus, our results are most likely an underestimation of the efficacy of anti-EGFR mAb treatment for the RAS wild-type cohort. Retrospective analyses of these additional RAS mutations were performed in four of the included studies [36]. Additional RAS mutations were confirmed in 14–26% of patients, resulting in improved efficacy data upon exclusion of these patients. A potential explanation for primary resistant patients with RAS wild-type mCRC is intralesional and interlesional heterogeneity in RAS mutation [41], making the RAS mutation determination on a single needle biopsy or old resection material prone for sampling errors. A promising novel approach is to evaluate these mutations in circulating cell-free DNA [42] or circulating tumor cells [43].

Recently, right-sided location of the primary tumor has been reported to negatively influence treatment benefit of anti-EGFR mAb [34, 44]. In the meta-analysis of Arnold et al., single patient data of six trials, which were also included in this meta-analysis, were pooled to evaluate the prognostic and predictive value of sidedness [45]. Indeed, they demonstrated that the addition of anti-EGFR mAb in patients with left-sided tumors significantly improved OS (HR = 0.75, CI 0.67–0.84) and PFS (HR = 0.78, CI 0.70–0.87), in contrast to patients with right-sided tumors (HR = 1.12, CI 0.87–1.45; HR = 1.12, CI 0.87–1.44, for OS and PFS, respectively). Further research is needed to evaluate the clinical utility of this biomarker and understand the underlying mechanisms of resistance.

Based on our meta-analysis, we conclude that the anti-EGFR treatment significantly improves response and survival outcome of patients with (K)RAS wild-type mCRC, regardless of treatment line or chemotherapeutic backbone. It is a sensible treatment strategy to save anti-EGFR mAb as third-line monotherapy for patients with mCRC in a true non-curative setting, as combination therapy is more toxic and has no clinically significant benefit compared to sequential therapy. For patients with limited disease, first-line combination therapy with anti-EGFR mAb can be considered, if local radical treatment may still be an option upon downstaging. As sound data to support this last consideration are lacking, further research is necessary.