Small Molecular TKIs Versus Chemotherapy In EGFR Mutation-Negative, Previously Treated Non-Small-Cell Lung Cancer-meta-analysis

Lung cancer was the most commonly diagnosed cancer as well as the leading cause of cancer death in males and the second leading cause of cancer death in females in 2008 globally. It also accounts for the first leading cause of cancer death in China. Lung cancer included non small cell lung cancer and small cell lung cancer according to its pathological type and treatment methods. Most of patients are NSCLC, which can be mainly treated by surgery, radiotherapy, chemotherapy and molecule targeting agents. But many NSCLC patients can not be treated by surgery because they are advanced when diagnosed (70%-80%).The standard first line treatment for advanced non small cell lung cancer is cisplatin-based or carboplatin-based chemotherapy (paclitaxel, gemcitabine, docetaxel, pemetrexed, vinorelbine). Although many new active drugs are available, the reported response rate to second-line systematic therapy have generally been less than 10%. Many patients are treated by single agent because they are intolerance to two agents after progressed for first line treatment. Pemetrexed, docetaxel and gemcitabine have been approved for the treatment in previously treated NSCLC patients.At the beginning of twenty-first Century, drug therapy except for chemotherapy, molecular targeted drugs in the tumor therapy have shown astonishing effects so far. Molecular targets of lung cancer include:EGFR, PIK3CA, MEK1, BRAF, ERBB2, KRAS, MET amplification, fusion gene (ALK, ROS, RET), FGFR amplification, PTEN, DDR2 etc. Among them, EGFR (epidermal growth factor receptor) gene targeted drugs plays an important role in the therapy of lung cancer. Mutation of EGFR include:.Exon 19 deletion and exon 21 (L858R,L861), exon 18(G719X,G719), exon 20 (S768I) mutations.Small molecular EGFR tyrosine kinase inhibitors (EGFR TKIs), including erlotinib (Tarceva; OSI Pharmaceuticals Inc, Melville, NY) and gefitinib (Iressa; AstraZeneca, Macclesfield, United Kingdom), have demonstrated promising effects in some NSCLC patients. The drug sensitivity to gefitinib and erlotinib is associated with EGFR mutations, which occur in about 50% of Asian patients with NSCLC and about 10% of patients with NSCLC in Caucasian patients. Asian women, lung adenocarcinoma and non-smokers are existing high gene mutation rate, so the curative effect is better. Their main side effects of these agents are rash and diarrhea,which is mild comparing with chemotherapy.Chemotherapy or molecular targeted drug therapy can be used for the two or three line treatment of advanced non small cell lung cancer. Whether the EGFR mutation of gefitinib or erlotinib or icotinib or not is closely related to the treatment effect. And the treatment effect is good for the EGFR mutation positive patients. But EGFR TKIs compared with chemotherapy, which is the better treatment agents are still controversial for mutation-negative patients. Some RCTs have been performed to test the role of eroltinib or gefitinib versus single agent chemotherapy after progression to first-line therapy in EGFR mutation negative advanced NSCLC patients in recent years. Considering the inconsistent results above, a systematic review of the literature was conducted.ObjectsWe performed this meta-analysis to compare the efficacy of small molecular EGFR TKIs with commonly used chemotherapy in EGFR mutation-negative, previously treated NSCLC patients. We hope it can help clinician make decisions and help patients choose the most optimal agents.MethodsWe Searched the PubMed database, CENTRAL library, EMBASE database, China National Knowledge Infrastructure (CNKI) database and the ASCO Annual Meeting abstracts, we made slightly different search strategy according to different database, but in principle, all search strategies are identical as far as possible as follows:Patient, Intervention, Comparison and Outcome. We accepted all RCTs (randomized controlled trials) comparing small molecule EGFR TKIs with chemotherapy of NSCLC treatment progress after first line chemotherapy for EGFR mutation negative patients, and searching the relevant references, if the information is still lack, then contacting the researchers to get more information, and seek help from the authors for original data. Collecting data of literature including the first author, published year, race, nationality, tumor type with histological or physiological evidence of non small cell lung cancer, research type, sample numbers of control group and case group. The results included PFS (progress free survival) and OS (overall survival), their HR (hazard ratio) and 95% CI (95% confidence interval). PFS is defined as the period from random to progress, and OS is defined as from the project beginning to patients death.HR is the result comparing TKIs to chemotherapy.We set up the strict and slightly different including and excluding criteria satisfied with different databases. Two researchers both used standard data extraction forms to collect all possible relevant titles and abstracts. If the two researchers had different opinions, we find the full text to evaluate. An oncologist will assess it if there were still inconsistent after we evaluated full text. We used the Cochrane Collaboration risk assessment tool review manager 5.1.0 to score the quality of the studies. Statistical analysis was performed using stataSE12.0 (Stata Corporation, College station, Texas, USA). Heterogeneity was tested with the χ2-based Q statistic and considered statistically significant when I2> 50% or p-value less than 0.1. When p-value more than 0.1, we could think these similar studies are homogeneity, so we can use fixed-effects model. But when p-value less than or equal to 0.1, we can use random effects model if the heterogeneity was still exist after analysis the cause and dealing with it. A statistical test with a p-value less than 0.1 was considered significant. HR>1 indicated more deaths or progression in EGFR-TKIs than chemotherapy. So we used random effect model and fixed-effects model to analyses PFS and OS, respectively. Results report included PFS and OS and their HR and 95%. HR more than 1 suggest TKIs had more death and progress than chemotherapy. The integrity of the included studies is the most important factor to influence the results and conclusions of meta-analysis, which is mainly measured by reporting bias. It can be divided into the following categories:publication bias, lag bias, repeated publication bias, publication location bias, citation bias, language bias, results reporting bias. The most important and frequently studied bias among these is publication bias among these biases. Funnel plot and Egger test were used to detect publication bias. A symmetric inverted funnel and P-value for Egger test value more than 0.01 suggested there are no publication bias existed.ResultsIdentification of studies and study characteristicsA total of 799 publications were identified which met our requirement with the search strategy. After initial screening, this number reduced to 26 possible relevant trials, of which 19 articles were removed by reading abstract to satisfy the inclusion criteria of the present study. At last,7 trials were confirmed eligible for inclusion. At last, gefitinib and erlotinib were included in EGFR TKIs, single agent docetaxed and single agent pemetrexed were included in chemotherapy. For all trials included in this meta-analysis, direct sequencing was applied to detect EGFR mutation type. Rare types (7 types excepted deletion in exon 19 and the exon 21 mutation L858R) and wild type were considered as negative for EGFR mutation in one trial, others were all EGFR wild type. Patients enrolled in 3 trials of these 7 trials were all of EGFR negative mutation, while the other 4 trials included both EGFR mutation negative and positive patients. Data were extracted from subgroup of negative mutation in each trial.In these 7 trials,1 trial was carried out in East Asia,4 trials were in only one country, the other 3 were international. However, all of them were multicentric. The analysis contained a combined total of 1126 participants. The sample size of per trial ranged from 38 to 253. Participants in two arms for each trials received small molecular TKIs or chemotherapy. Risk of biasData were extracted from full text published at PubMed database, Cochrane library and Embase database except for two articles which were presented at ASCO in abstract form. The bias of risk from all included trials fell into three types:low risk, high risk, unclear (undescribed in available data). Only one trial introduced a bias concerning the efficacy of pemetrexed in histologically unselected NSCLC patients because patients who had squamous cell lung cancer had been included between January 2006 and July 2008, while the concrete proportion of patients with squamous disease wasn’t introduced. These 7 RCTs had similar designs that compared the efficacy of small molecular TKIs versus chemotherapy. We all agreed to perform the meta-analysis on them. OutcomesNo significant heterogeneity was found in OS (I2=10.1%, p=0.348), so the fixed effect model was applied to continue the following analysis. However, significant heterogeneity was found in PFS (I2=55.2%, p=0.048), so the random effect model was applied to continue the following analysis.The pooled analysis on PFS showed a statistically significant inferiority in the study arm treated with small molecular TKIs (HR=1.34,95% CI=1.08-1.65; P= 0.007). Meta-analysis on OS showed no difference between two arms (HR= 1.07, 95% CI=0.92-1.24; P= 0.395).Because of the small number and EGFR rare mutation type of patients enrolled in one trial and treatment with pemetrexed for squamous cell lung cancer before NCCN and ESMO guideline recommended pemetrexed was better to treat non- squamous NSCLC patients in another trial, we respectively removed the two trials to reduce heterogeneity. No significant heterogeneity was found for both OS (I2=28.2%, p= 0.243) and PFS (I2=13.3%, p= 0.329) in the left trials, so the fixed effect model was applied to continue the pool analysis. The outcomes of TKIs were still significant inferior to chemotherapy in PFS and still not significantly different in OS after removing the two trials (HR= 1.05,95% CI=0.90-1.23; P= 0.536).According to the OS of TAILOR and HORG, subgroup analysis classify as the patient’s gender, pathological type and smoking history, respectively. Using randomized effect model, the results show that every subgroup between erlotinib and chemotherapy have no statistically significant difference.Funnel plot and Egger’s test were applied to detect publication bias on pool data for PFS analysis and OS analysis. Funnel plots all showed inversion symmetry, bottom has no obvious big gap. Egger’s test showed no significant publication bias both in PFS before (P=0.220 for EGFR-TKIs vs. chemotherapy) or after (P=0.399 for EGFR-TKIs vs. chemotherapy) removing a trial, and in OS before (P=0.998 for EGFR-TKIs vs. chemotherapy) or after removing a trial (P=0.732 for EGFR-TKIs vs. chemotherapy).ConclusionsFor EGFR mutation negative second line and third line advanced NSCLC patients, chemotherapy is the best choices to whom can undergo chemotherapy. However, TKIs can be treated to such patients with poor performance status.