To the best of our knowledge, our study is the first meta-analysis evaluating the efficacy of triplet chemotherapy plus anti-EGFR agent therapy. Nevertheless, reviews and studies of this topic were multiple, which implies the attention researchers have paid to this important issue. Our meta-analysis combined the outcome of 762 mCRC patients treated with triple chemotherapy plus anti-EGFR therapy from 14 studies, indicating that the treatment can improve the ORR to 82% and secondary R0 rate without increasing G3/4 adverse effect.
With a rich number of studies comparing the efficacy of different regimens in mCRC first-line treatment, optimal therapy is still controversial[26–27], especially on CLM or potentially resectable mCRC patients who may need more intensive treatment to achieve distinct tumor shrinkage to purchase the chance of secondary surgery.[28] The phase II OLIVIA study and phase III TRIBE study both confirmed that the addition of bevacizumab to mFOLFOXIRI were able to provide a higher ORR and R0RR.[7, 29] Because both FOLFOXIRI and anti-EGFR target therapy are considered to be related to more rapid tumor responses, we conducted this meta-analysis to assess the efficacy and safety of their combination.
In our meta-analysis, the pooled ORR of patients who accepted triplet chemotherapy plus anti-EGFR therapy was 82% (95%CI = 76%-88%,I2 = 76%), ranging from 60–100%. For comparison, we summarized several meta-analysis of important studies in mCRC treatment. Details were presented in Table 5. It is obviously that this result dominates the efficacy of all the other treatments. This indicates that triplet chemotherapy plus anti-EGFR therapy conspicuously increases the chance of secondary resection in mCRC patients. The same advantages can be observed in other outcomes as the R0RR of CLM, mPFS, and mOS. A possible explanation of such a remarkable improvement is discussed below.
Table 5
Summary of related meta-analysis for first-line treatment of mCRC.
Meta-analysis | Included studies | Treatment | ORR | mPFS(month) | mOS(month) | SRR | Grade3/4 toxicity |
Cremolini C[30] | CHARTA OLIVIA STEAM TRIBE TRIBE2 | bev + 3-CT vs bev + 2CT* | 64.5% vs 53.6% OR 1.57,p < 0.001 | 12.2 vs 9,9 HR:0.74, p < 0.001 | 28.9 vs 24.5 HR:0.81, p < 0.001 | 16.4% vs 11.8% OR 1.48, p = 0.007 | neutropenia 45.8% vs 21.5%; p < 0.001 FN 6.3% vs 3.7%; p = 0.019 Diarrhea 17.8% vs 8.4%; p < 0.001 |
C Bokemeyer[31] | CRYSTAL OPUS | Cet + 2CT Vs 2CT | 60.7% vs 40.9% OR 2.27, p < 0.0001 | 10.9 vs 7.7 HR 0.64,p < 0.0001 | 24.8 vs 21.1 HR 0.84,p = 0.0048 | NA | NA |
F Pietrantonio[32] | Valentino TRIBE TRIBE2 STEAM CHARTA. | pan + 3CT vs bev + 3CT | 73% vs 77% OR 0.79, p = 0.4 | 11.4vs 13.3 HR 0.83, p = 0.11 | 30.3 vs 33.1 HR 0.8, p = 0.14 | 22% vs 18% P = 0.51 | neutropenia 26% vs 48%; p = 0.001 Diarrhea 14% vs 6%; p = 0.82 febrile stomatitis 8% vs 6%; p = 0.67 |
G Tomasello[33] | 11 studies | Bev + 3CT | 69% (95%CI,65%−72%) | 12.4 (95%CI,10−14.3) | 30.2 (95%CI,26.5–33.7) | 36.6% (95%CI,24.6%−50.5%) | NA |
*,all KRAS and BRAF wild type; ORR, objective response rate; SRR, Secondary resection rate; mPFS, median progression free survival; mOS, median overall survival; 2CT, doublet chemotherapy; 3CT, triplet chemotherapy; FN, febrile neutropenia. |
4.1 RAS and BRAF mutation status.
KRAS codon 12 and 13, 61, HRAS, NRAS, and BRAF V600E mutations are the most important and widely studied biomarkers for anti-EGFR agents. Both RAS and BRAF mutations have been proved to be prognostic factors with worse survival.[34] Because anti-EGFR target therapy is not recommended in treating patients with RAS or BRAF mutations, there were notable differences in patient baseline gene status selection in the included studies. In our meta-analysis, among the 701 patients who accepted the RAS mutation test, 157 were KRAS wild-type, 390 were RAS wild, and three were KRAS mutants. Among the 311 patients accepted the BRAF mutation test, 291 were wild, and seven were mutant. It’s assumed that the impressive efficacy was associated with the large number of RAS and BRAF wid-type patients.
4.2 Primary tumor side
It is now a consensus that primary tumor side of mCRC is biologically distinct.[35] The CALGB/SWOG 80405 study presented a significant longer mPFS and mOS of the left side than the right side (mOS 33.3 months vs. 19.4 months, p < 0,001).[36] Subgroup analysis showed that in left-side originated mCRC patients, compared with doublet chemotherapy plus bevacizumab, combination of doublet chemotherapy and cetuximab was associated with a significant longer mOS (36 vs.31.4 months, p = 0.018). This advantage was not reflected in the right-side originated mCRC patients; the mOS of cetuximab-arm and bevacizumab-arm were 16.7 vs. 24.2 months, p = 0.065.[37] The same tendency was found in the sub-analysis of trial FIRE-3 and PEAK.[9, 38] Based on this evidence, guidelines recommended treatment disparately according to the primary tumor side. Anti-EGFR target therapy is only recommended in left-side colon cancer combined with doublet chemotherapy including FOLFOX, FOLFIRI, and XELOX.[39] In our meta-analysis, tumor primary location was presented in 316 patients from five studies, 48 patients from the colon, 35 patients from the right colon, and 233 patients from the left colon or rectum among 762 patients, which may be one reason for our high ORR.
4.3 R0 resection rate and conversion therapy
The conversion therapy is a standard therapy for mCRC patients, especially for patients with CLM. Recent studies verified that CLM is a particular type of mCRC that is heterogeneous to other mCRC patients.[40] Jennie Engstrand et al. observed a 44% decrease in liver metastatic left-sided cancer and rectal cancer compared to right-sided cancer, and 5-year OS was 16.6% vs. 4.3%.[2] Despite that, convincing evidence proved that the R0RR of CLM can significantly increase 5-year survival rate and mOS.[41] This inspiring discovery reflect the importance of seeking the optimal treatment for the appropriate patients. The new treatment goal of CLM currently is to maximize the possibility of eradicating all LM lesions, which request a rapid and distinct tumor shrinkage.[42] In our meta-analysis, the pooled R0RR in CLM patients was 60% with a random effects model (95%CI = 49%-70%,I2 = 69%). Among the included 14 studies, five studies were designed specifically to evaluate the R0RR of CLM after treating by triplet chemotherapy plus anti-EGFR agent, and the R0RR ranged from 60–84%. Despite this encouraging results, there were an additional 14 patients from two studies that accepted R1 resection (2 patients) or R2 resection (12 patients). Other important indexes for conversion therapy include early tumor shrinkage and depth of response (DpR), which were verified as capable of predicting treatment outcomes and cetuximab efficacy.[43–44] Two studies used these indexes as end points. The FOCULM study demonstrated that, compared with mFOLFOXIRI alone, the addition of cetuximab improved the DpR from 44–56.1% (p = 0.012) and the overall resection rate was 55.2% and 29.4%, respectively.[23] Consistent results of R0RR were presented in the POCHER trial (60%, 95%CI 45.8–75.1) and another study conducted by E Samalin (57.4%).[12, 22] All the results mentioned above demonstrated that the use of triplet chemotherapy plus the anti-EGFR agent significantly increased the possibility of the R0 resection of CLM patients, indicating that this treatment is preferable for CLM patients. It is important to identify the possible molecular and genetic marks of the patients that may benefit most from the treatment. Because evidence has proven that anti-EFGR treatment was more effective in left-side colorectal cancer,[45] it is safe to hypothesize that RAS and BRAF wild-type patients with liver metastatic left-sided cancer may possibly be the selected group to benefit most from triplet chemotherapy plus anti-EGFR agent therapy. We should acknowledge the small sample size and low quality of the included studies. This assumption needs to be tested in large scale RCTs. The results of several ongoing RCTs, including DEEPER (mFOLFOXIRI + cetuximab vs. mFOLFOXIRI + bevacizumab,NCT02515734), TRIPLETE (mFOLFOXIRI + panitumumab vs. mFOLFOX6 + panitumumab,NCT03231722), and PANIRINOX (mFOLFIRINOX + panitumumab vs. mFOLFOX6 + panitumumab,NCT02980510) may be helpful in this regard.
Toxicity and dose adjustment
The most commonly observed adverse events in our study was diarrhea (174/648), followed by neutropenia (157/648) and skin toxicity(95/648). The toxicity profile was slightly different comparing with the triplet chemotherapy plus anti-VEGF therapy in which neutropenia was the most frequently observed grade 3/4 adverse event, with an incidence of 45.8%, and the incidence of diarrhea was only 17.8%.[30] In our meta-analysis, the pooled rates were 29% (95%CI = 20%-39%,I2 = 86%), 28% (95%CI = 20%-37%,I2 = 77%), and 17% (95%CI = 11%-24%,I2 = 66%) for diarrhea, neutropenia, and skin toxicity, respectively. Three studies reported a dose reduction after a high incidence of toxicity, mainly because of diarrhea and neutropenia. After a dose adjustment, all studies considered the side effects manageable through symptomatic measure. The detailed dosage of each drug is presented in Table 2. There were two studies that pointed out that dose reduction did not influence efficacy. The results should be interpreted discreetly for retrospective design and small sample size. It should also be noticed that nearly all the patients the patients included had an ECOG performance score of 0–1.
Limitation
This study has several limitations that should be recognized. First, the eligible studies and the sample sizes of the included studies were relatively small. Heterogeneity between studies was relatively high, which may bias the results. Most of the studies were case series studies or had a retrospective design, which may influence the accuracy of the results (especially in the assessment of adverse events). Second, a portion of the studies were only obtainable as conference abstracts. Despite efforts made to get in touch with authors for complete results, there remained crucial data uncollected. Third, the dosage of each study was not consistent; one of the included study used capecitabine instead of 5-fluorouracil, affecting the outcome of the study and toxicity evaluation.