EGFR-TKIs versus taxanes agents in therapy for nonsmall-cell lung cancer patients

Supplemental Digital Content is available in the text


Introduction
Carcinoma of lungs have become the first killer among all cancers, and is the leading cause of cancer-related mortality across the world, with a 5-year survival of less than 15%. [1,2] Nonsmall-cell lung cancer (NSCLC) accounts for nearly 80% to 85% among all cases of lung cancers, locally advanced NSCLC 25% to 30% of all cases and metastatic diseases 40% to 50% of all cases. [3,4] In the last decade, the therapeutic method for these patients consisted of 1st-line chemotherapy that can significantly improve the curative effects, such as gemcitabine, taxanes combining with platinum, and pemetrexed. However, the response rates are modest and after standard 1st-line therapy, several patients relapse of the malady, hence patients with NSCLC demand 2nd-line chemotherapy after 1st-line chemotherapy. [5] Currently, it is safe to say that NSCLC patients benefit from taxanes agents, such as paclitaxel and docetaxel, which can be seen as representative of the new generation of anticarcinogen with a unique mechanism: they play a role in the microtubule and tubulin system, combine with free tubulin and promote tubulin assembly into stable microtubules, and inhibit their depolymerization. Therefore, they prevent the division of a large number of cells, leading to cell death. [6,7] Among paclitaxel plus carboplatin as 1st-line treatment in advanced NSCLC. [8] Apart from paclitaxel, docetaxel is paclitaxel in the process of structural transformation synthesized paclitaxel derivatives, which has high bioavailability and small side effects. Docetaxel is approved as 1st-line therapy in combination with cisplatin, as single-agent 2nd-line therapy, or as single-agent maintenance therapy for patients with advanced NSCLC in numerous countries. [9,10] To date, epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKIs) as molecular targeted therapeutical drugs have aroused people's attention, therein gefitinib and erlotinib have secured approvals for the treatment of advanced NSCLC, especially for those with sensitizing EGFR mutations. [11] Nevertheless, different mutations may result in different structural changes, thereby affecting subsequent clinical outcomes. [12] EGFR-TKIs play a role in tumor suppression by blocking the signal transduction of tumor cells, including inhibition of tumor cell proliferation, acceleration of apoptosis, and antiangiogenesis. [13] Compared with other traditional medicines, gefitinib and erlotinib can prolong progression-free survival (PFS) in EGFR mutation-positive patients, and can be administered easily (orally). A phase 2B open-label randomized controlled trial has shown that gefitinib exhibits good tolerability and antitumor activities in NSCLC. [14] Based on these, we performed a meta-analysis and metaregression to explore the efficacy and safety of these medications for NSCLC patients, which could dedicate to make evidencebased clinical decisions for the treatment of pulmonary cancer.

Method
This review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement. [15] The project was prospectively registered with PROSPERO database of systematic reviews, number CRD42016038700. [16]

Data sources and search strategy
We systematically searched 3 search engines: PubMed, EMbase, and the Cochrane library from inception to April 2016. The search strategy included keywords and MeSH terms related to therapy using EGFR-TKIs and taxanes. Clinical trials were in any languages with patients presenting with NSCLC (see details in Table S1, http://links.lww.com/MD/B469). We also scrutinized the reference list of relevant publications for additional studies.

Inclusion criteria
The relevant literature was selected carefully had to meet the following 4 criteria: all patients were diagnosed with NSCLC; the treatment arm were given EGFR-TKIs for therapy; while the control arm were given taxanes for cure; measurable outcomes were reported; and randomized controlled trials (RCTs). Ethical approval was not necessary, because of meta-analysis is a type of secondary statistics study, not directly associated with the subjects.

Data extraction
Two investigators (AN and ZYS) read the references and extracted the data independently. If we had any disagreements, we asked the 3rd investigator (LQ or ZQC). Every eligible study included the following information: the first author, publication year, trial design, sample size, age, gender, performance status, clinical phase, EGFR status, disease status, intervention, outcome assessment time, the main outcomes, risk of bias, and Newcastle-Ottawa Scale (NOS) score. The main outcomes were as follows: PFS, progression-free survival rate (PFSR), overall survival (OS), overall survival rate (OSR), objective response rate (ORR), disease control rate (DCR), quality of life (QoL), and adverse events (AEs).

Study quality assessment
The Cochrane Collaboration tool was used to evaluate the quality of these studies. [17] We assessed the following 7 items of risk of bias: random sequence (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), and other bias. Low risk, high risk, and unclear risk were classified in all studies. The NOS [18,19] as recommended by the Cochrane Non-Randomized Studies Methods Working Group, our meta-analysis also used NOS to assess the quality of the included RCTs. NOS using the following criteria label as "yes," or "no": Is the case definition adequate? Representativeness of the cases? Selection of Controls? Definition of Controls? Comparability of cases and controls? (0-2) Ascertainment of exposure? (0-2) Same method of ascertainment for cases and controls? and nonresponse rate? We excluded some studies which score less than 5 score.

Statistical analysis
The hazard ratio (HR), risk ratio (RR), and odds ratio (OR) with 95% confidence interval (CI) were used in eligible study. If studies did not have significant heterogeneity (P > 0.10, I 2 < 50%), fixedeffects model was calculated. By contrast, random-effects model was employed. Furthermore, using subgroup analysis for the possible causes of sources of heterogeneity factors. The results of the meta-regression with the P value less than 0.05 means that the factors could cause significant impact to overall. Funnel plot was produced to assess publication bias. All statistical analyses were conducted with Review Manager 5.3.5 statistical software (Cochrane Collaboration) and STATA 13.0 software (StataCorp, College Station).

Article selection and risks of bias
After searching the PubMed, EMbase, and the Cochrane library, we identified 633 articles, based on title and abstract screening, and obtained as full texts records. A total of 26 studies were included (Fig. 1).
We evaluated the risks of bias of all articles by the Cochrane Collaboration's tool and NOS scale, the required data can be evaluated as acceptable quality. The detail of quality assessment was shown in Table 1, Table S2, http://links.lww.com/MD/B469 and Fig. S1, http://links.lww.com/MD/B469.

Subgroup analyses and meta-regression
We also did meta-regression to find the source of heterogeneity. We found that grouping by platinum in control arm revealed differences in outcomes of PFS and DCR with P value less than 0.05. Moreover, EGFR status might have influenced heterogeneity in PFS (P = 0.039). Besides, grouping by clinical phase of trials, differences could be found in OS (P = 0.036).

Publication bias
We did the funnel plot according to PFS, OS, ORR, and DCR was shown in Fig. 6. The funnel plot showed asymmetry among our included studies, which proved the existence of publication bias.

Discussion
We carried out this meta-analysis to compare PFS, PFSR, OS, OSR, ORR, DCR, QoL, and AEs between EGFR-TKIs and taxanes. EGFR-TKIs can significantly prolong PFS and PFSR after therapy. The therapeutic effects of EGFR-TKIs were similar to taxanes in OS. Furthermore, taxanes were inferior to EGFR-TKIs in ORR. There was no significant difference between EGFR-TKIs and taxanes in DCR, while taxanes had a tendency to improve DCR. We found whether in FACT-L, LCS, or TOI, the results showed EGFR-TKIs surpassed taxanes in QoL with NSCLC patients. We found that comparing taxanes, NSCLC patients with EGFR mutation, EGFR mutation-positive, and unknown EGFR mutation can benefit from EGFR-TKIs on PFS, OS, and ORR. However, they cannot get beneficial treatment, who with EGFR wild-type and EGFR mutation-negative. There was no significant difference in EGFR copy number and EGFR protein expression. Thus, EGFR-TKIs are more suitable for patients with EGFR mutations and EGFR mutation-positive. Li et al [46] made a relevant study, they also found that EGFR-TKIs were more efficient in EGFR mutations patients.
As per the analysis of heterogeneity, we did meta-regression and detail subgroup with EGFR status, platinum in control arm, clinical phase of trials, and trial design. First after EGFR status subgroup analyses, although the results showed PFS and ORR were different to drug groups, they had same tendency with drug groups. EGFR status on the effect of OS had difference on drug groups; hence, EGFR status might cause heterogeneity. Besides, we used platinum in control arm, clinical phase of trials, and trial design were operated in different groups, the results of PFS, OS, ORR, and DCR were equivalent to drug groups. However, analysis by meta-regression was found that platinum in control arm and clinical phase of trials were same to lead to heterogeneity, the detail results were showed in Table 3. Subsequent analysis will be confirmed. In fact, small sample size may be one of the causes of heterogeneity.
There are also published meta-analysis, such as that by Zhao et al [3] who compared the therapeutic values of gefitinib versus docetaxel, while we included more studies and outcomes. In fact, incorporating more clinical outcomes will bring more strong evidences in evaluation of efficacy and safety with NSCLC patients. Furthermore, we expanded the sample size and obtained consistent with their conclusions. To date, there has been no published meta-analysis comparing the efficacy and safety between EGFR-TKIs and taxanes. Pilkington et al [47] published a systematic review of the clinical effectiveness of 1stline chemotherapy, and they mentioned that, compared with paclitaxel and platinum, gefitinib had a statistically significant improvement in PFS, which was also consistent with the findings of our study, proving above-described results were trustworthy. EGFR-TKIs could prolong PFS, improve ORR and QoL, yet they have many side-effects, such as rash and diarrhea. [48,49] Taxanes also have several AEs: gastrointestinal reaction, alopecia, and hematological toxicity, particularly grade 3/4 leukopenia and neutropenia which tended to be more frequent after treatment with taxanes. [50,51] In the meta-analysis of this study, except for diarrhea and rash, there was a slightly worse trend toward EGFR-TKIs compared with taxanes. EGFR-TKIs were superior to taxanes in rates of many AEs, such as all   hematologic toxicity, myalgia, and pyrexia, etc. All of the data were listed in Table 2. It illustrated that the risk of AE rates was not increased when EGFR-TKIs instead of taxanes were applied for the treatment of NSCLC. Holistic nursing care can improve the curable effects and significantly reduce adverse effects in the treatment of patients with hematological system disorders using high-dose dexamethasone pulse, and it deserves to be promoted to clinic. [52] Auricular acupressure can significantly reduce the gastrointestinal side effects in lung cancer patients after chemotherapy, and be without any adverse reaction and high compliance. [53] Meanwhile pantoprazole joint granisetron and methoxychlor Puan and dexamethasone prevent chemotherapy-induced gastrointestinal reactions with better efficacy, adverse reactions are mild, worthy of clinical application. [54] Certainly this meta-analysis had several limitations need to be addressed. First most of included trials were allocated in Asian region (Table 1), which may cause the geographical limitations. Besides, due to limited or missing data about current trials, details such as gender, age, smoking, and cancer stage were unable to be analyzed. Moreover, not all of the patients in this study were serious, especially the performance status ≦2, which may be proved that the basic level was mixed. Although anticancer drugs have been used widely in NSCLC, related randomized clinical trials appear to be limited. Furthermore, the different outcome assessment times could lead to the existence of publication bias. Positive results are easy to be published, negative results with several AEs are not likely to be viewed. Finally the quality of included studies were variable, although most of them with acceptable quality, high-quality, well-level, and large-scale double-blind RCTs are needed for further research. Considering the limitations above, further studies were warranted to complete the information and the results of this research must be interpreted with caution.
In terms of PFS, PFSR, ORR, QoL, and AEs, EGFR-TKIs were superior to taxanes in NSCLC patients from the present metaanalysis study, particularly who were with EGFR mutationpositive. There were no differences between EGFR-TKIs and