Efficacy and Safety of Afatinib in the Treatment of Advanced Non-Small-Cell Lung Cancer with EGFR Mutations: A Meta-Analysis of Real-World Evidence

Introduction The purpose of this study was to explore the efficacy and safety of afatinib in advanced non-small-cell lung cancer (NSCLC) patients with epidermal growth factor receptor (EGFR) mutations based on real-world evidence. Materials and Methods Eligible real-world studies were identified from PubMed, Cochrane Library, and Embase. Cochrane guidelines were used to assess the quality of included studies. Cochran's Q test and I2 statistics were used for the heterogeneity analysis. Results Twenty-five studies were included in this meta-analysis; nine studies were included in the qualitative descriptive analysis. The summarized disease control rate (DCR) was 87.6% (81.5%, 92.7%), and the overall response rate (ORR) was 58.9% (48.8%, 68.7%). The pooled median progression-free survival (PFS) was 12.4 (10.3, 14.5) months, mean time to failure (TTF) was 15.4 (13.6, 17.2) months, and median overall survival (OS) was 31.6 (26.7, 36.5) months. The total incidences of adverse events (AEs) for skin rashes, diarrhea, paronychia, and mucositis were 71.4% (64.4%, 77.9%), 70.4% (60.1%, 79.8%), 52.1% (41.9, 62.3%), and 36.5% (29.5%, 43.8%), respectively. The incidences of severe adverse events (SAEs, Grade ≥3) for diarrhea, skin rashes, paronychia, and mucositis were 9.7% (6.8%, 13.1%), 5.8% (4.5%, 7.2%), 3.8% (2.0%, 6.2%), and 2.1% (1.0%, 3.6%), respectively. Differences in PFS and OS between the afatinib non-full-dose (<40 mg) and full-dose (>40 mg) groups were not significant (P > 0.05). However, the ORR in the full-dose group was 78.5% (66.7%, 88.4%), which was significantly higher than that in the non-full-dose group (67.8% [56.8%, 77.9%]). Conclusion The efficacy and safety of afatinib has been confirmed by real-world evidence in advanced NSCLC with EGFR mutation, consistent with randomized controlled trial results. In real-world setting, tolerability-guided dose adjustment might not affect the afatinib efficacy.


Introduction
Lung cancer is the leading cause of cancer-related deaths and a serious threat to human health [1]. Non-small-cell lung cancer (NSCLC) accounts for more than 80% of lung cancer [2]. Epidermal growth factor receptor (EGFR) mutations have been identified in approximately 50% of Asian and 10-15% of Caucasian lung adenocarcinoma patients [3,4]. Currently, tyrosine kinase inhibitors (TKIs), including erlotinib, gefitinib, dacomitinib, afatinib, and osimertinib, are the standard first-line treatment for advanced NSCLC patients with EGFR mutations [5].
Afatinib is an irreversible second-generation ErbB family blocker [6], which has been approved as a first-line treatment for NSCLC patients with EGFR exon 19 deletions or exon 21 L858R substitution mutations [7]. In 2013, afatinib was approved worldwide as a first-line treatment for patients with EGFR-mutant NSCLC [5,8].
e LUX-Lung 3/6/7 trials revealed that afatinib had obvious effects in the treatment of advanced EGFR-mutant NSCLC [9][10][11][12] and might provide a better curative effect than first-generation EGFR-TKIs [13]. Moreover, a meta-analysis based on randomized controlled trials (RCTs) has shown that afatinib prolonged progression-free survival (PFS), increased overall survival (OS), and the overall response rate (ORR) [14]. However, whether afatinib is effective in particular subgroups remains controversial due to the RCT exclusion criteria [14,15]. Furthermore, the adverse effects of afatinib limit its clinical application [16]. us, a dose-adjustment strategy guided by tolerability can yield clinical benefits based on RCT data [14,16], which need to be demonstrated through real-world evidence (RWE)-based data.
It is undeniable that the credibility of RCTs can be inferred from causality and is thus considered the gold standard of clinical research; however, RCTs may not reflect real-world practice due to strict inclusion criteria [17]. Patients in the real-world may differ in numerous characteristics from those in RCTs. RWE-based studies collect data of patients treated as a whole in clinical practice according to local government regulations, overcoming inherent limitations of RCTs, and can assess the efficacy and safety information of patients in the real world [18,19]. erefore, it is crucial to confirm the efficacy and safety of afatinib by using RWE data. Additionally, a previous study based on real-world data proved that afatinib dose adjustment decreases the intensity and frequency of adverse drug reactions without affecting the efficacy [20]. Although RWE contributes to the evaluation of certain clinical influential factors, the impact of dose adjustment, brain metastasis, mutation type, and Eastern Cooperative Oncology Group Performance Status (ECOG-PS) on the efficacy and safety of afatinib remains elusive.
Accordingly, we performed a meta-analysis based on real-world afatinib data in advanced NSCLC with EGFR mutations. To the best of our knowledge, this is the first meta-analysis based on RWE to explore the efficacy and safety of afatinib in advanced NSCLC with EGFR mutations. In this study, related RWE from Embase, PubMed, and Cochrane Library databases were analyzed. Herein, we aimed to comprehensively analyze the efficacy and safety of afatinib in NSCLC patients with advanced EGFR mutations based on real-world evidence. Furthermore, we explored the impact of tolerability-guided dose adjustment, brain metastasis, mutation type, and ECOG-PS.

Materials and Methods
is meta-analysis was performed following the guidelines of the meta-analyses (PRISMA) statement [21].

Data Sources.
Relevant studies were searched and identified in electronic databases, including Embase, PubMed, and Cochrane Library (updated to December 31, 2020). e main search keywords included "afatinib," "nonsmall cell lung cancer," NSCLC, "lung adenocarcinoma," and "adenosquamous carcinoma" e search was performed based on a combination of subject and free words. Additionally, a manual search of references in identified literature was performed to obtain additional information regarding the procedure. No language restrictions were applied to the current meta-analysis.

Inclusion and Exclusion Criteria.
In the present study, the inclusion criteria employed were as follows: (i) the subjects were patients with advanced EGFR-mutated NSCLC diagnosed by histology and cytology; (ii) the study reported the efficacy (ORR, disease control rate (DCR), OS, PFS, and time to failure (TTF)) or safety (adverse reactions/ serious adverse reactions) of afatinib in the treatment of NSCLC; (iii) the study reported the difference in efficacy and safety based on different groups, including different afatinib doses (full dose (40 mg/day for 6 months or more) vs. nonfull dose (<40 mg/day for 6 months or 40 mg/day for a reduction in the first 6 months)), mutation type (exon 19 deletion vs. uncommon, exon 21 L858R vs. uncommon, exon 19 deletion vs. exon 21 L858R), ECOG-PS (0-1 vs. 2-4), and brain metastases; (iv) the research type was RWE. e exclusion criteria were as follows: (i) nontreatment studies, including letters, reviews, and comments; (ii) the study lacked parameters required for quantitative analysis (including mean, sample size, standard deviation of the experimental group and the control group required for continuous variable meta-analysis) and could not be obtained through the conversion of other data in the study (enrolled as the qualitative description study); (iii) the study lacked baseline information such as gender, age, tumor classification, and ECOG-PS, reporting only the efficacy (PFS and TTF) of afatinib (enrolled as the qualitative description study); (iv) the research type was RCT. Furthermore, only one study (containing complete information) was extracted if more than one study was published using the same data.

Data Extraction and Quality Assessment.
In the present study, two independent researchers participated in data extraction. e available data included first author name, year of publication, gender, sample size, history, country, age, stage, ORR, DCR, OS, PFS, TTF, and adverse reactions. e extraction tables were exchanged after both researchers had completed the above data extraction work. Any inconsistencies in extracted results were resolved through discussion.

Statistical
Analysis. STATA software (version 11.0) was used for statistical analysis. e incidence rate (IR) and 95% confidence interval (CI) were used as the effect size to evaluate the incidence of ORR, DCR, AEs, and SAEs in patients with NSCLC administering afatinib. e median (95% CI) was used to assess the months of PFS, OS, and TTF, while weighted mean difference (WMD) and 95% CI were utilized as combined indices of effect quantity. e relative risk (RR) and 95% CI were used as the combined indices of effect quantity to comprehensively evaluate differences in ORR, DCR, and incidence of adverse reactions between the two groups. Moreover, the risk of OS/PFS was analyzed using hazard risk (HR) and 95% CI. If the study did not directly report the HR (95% CI) of OS and PFS but reported the median survival and log-rank P-value in non-full dose vs. full dose of afatinib, the HR (95% CI) could be converted using the method described before [22]. 2 Journal of Oncology Cochran's Q test and I 2 statistics were used to perform the heterogeneity analysis [23]. A random-effects model was used if heterogeneity was obtained (P < 0.05, I 2 > 50%); conversely, a fixed-effects model was employed. Moreover, to assess the effect of the above factors on the combined results of PFS, a subgroup analysis was conducted on PFS based on the timing of afatinib administration and data sources (direct reports in study or obtained by conversion). Furthermore, a sensitivity analysis was undertaken to assess the effect of combined results by analyzing the relevant studies individually. Finally, the Egger test [24] was used to analyze publication bias between the two groups.

Literature Review and Characteristics of Included Studies.
In total, 994, 990, and 281 studies were explored in the Embase, PubMed, and Cochrane Library databases, respectively ( Figure 1). After eliminating duplicate studies, a total of 1525 studies were obtained, with 1479 studies further excluded as these failed to meet the inclusion criteria after assessing the abstract and title. Among the remaining 46 studies, 12 relevant studies were identified after reading the full text. Manual retrieval did not detect any study that could be included in the current analysis. Finally, a total of 25 studies [20,[25][26][27][28][29][30][31][32][33] with sufficient data were included (Table 1). A total of nine studies [34][35][36][37][38][39][40][41] were included in the current qualitative descriptive analysis ( Table 2).

Study Characteristics.
e 25 enrolled studies were published between 2014 and 2020.
e study areas were primarily in Asia. Moreover, 19, 1, and 5 studies were enrolled in first-line, second-line, and mixed population groups, respectively.

Effect of Tolerability-Guided Afatinib Dose Adjustment.
Compared with the full dose of 40 mg/day, 57.8% (1917/ 3319) patients administered non-full dose of afatinib (Table 1). Seven studies (8 sets of research data) reported PFS between the non-full-dose and full-dose groups ( Figure 6(a)). Among them, HRs (95% CI) in four studies (five sets of data) were obtained by conversion. e heterogeneity for these data was not significant (I 2 < 0.0%, P > 0.05). For the combined results, the HR was 1.2 (95% CI [0.9, 1.5]; P > 0.05) (Table 3, Figure 6(a)), which indicated that dose reduction does not impact the therapeutic efficacy in terms of PFS. e PFS of the non-full-dose group was 5.0-14.2 months, while the PFS in the full-dose group was 3.0-15.7 months. For the combined results, the WMD was −1.6 months (95% CI [−5.7, 2.5]) ( Figure 6(b)). Two studies reported the difference in OS risk between the non-full-dose and full-dose groups ( Figure 6(c)). e result of the  (Figure 7(b)). Although the difference in PFS and OS was not significant, the ORR was significantly higher in the full-dose group compared with the non-full-dose group.
Two studies reported differences in total AEs based on the full-dose and non-full-dose groups. Significant heterogeneity in the total AEs was observed in diarrhea and skin rashes (I 2 > 50.0%; P < 0.05) but was not observed in mucositis and paronychia (I 2 > 50.0%, P < 0.05) (Figure 8(a)). e heterogeneity of diarrhea, skin rashes, and paronychia (Grade ≥3) was not significant (I 2 > 50.0%, P < 0.05) (Figure 8(b)). e incidence of total AEs and SAEs, including diarrhea, skin rash, mucositis, paronychia, dry skin, and pruritus, showed a decreased trend in the non-full-dose group (Table 4), which indicated the tolerability-guided dose adjustment alleviated afatinib-related adverse effects.

Effect of Subgroup on Efficacy and Safety of Afatinib.
Two studies reported differences in the DCR and ORR in brain metastases. e DCRs in the without metastases group and the brain metastases group were 94.6% (91.3%, 97.3%) and 89.1% (82.2%, 94.6%), respectively. e ORRs in the Articles excluded with reasons (n = 9) 9 without available data for meta-analyses.            Figure 4: e meta-analysis results for various incidence rates after afatinib treatment in advanced NSCLC with EGFR mutation: (a) the incidence rate of diarrhea; (b) the incidence rate of mucositis; (c) the incidence rate of skin rashes; and (d) the incidence rate of paronychia.  (Figure 9(c)). e ORRs in the exon 19 (Figure 9(d)).
Moreover, two studies reported PFS in patients with brain metastases based on the full-dose and non-full-dose groups (Supplementary Figure 3A). For the combined results, the HR was 2.4 (95% CI [0.9, 5.9]; P > 0.05). Furthermore, the combined results showed that the PFS in the exon 19 deletion group was undoubtedly higher than that in the uncommon mutation group (HR: 0.2, 95% CI [0.1, 0.4]; P < 0.05), and PFS in patients without brain metastasis was significantly lower than that in patients with brain metastasis  Figure 3B). However, the heterogeneity between the two studies was significant (I 2 � 59.4%) (Figure 6(c)).

Sensitivity Analysis and Publication Bias.
Publication bias and sensitivity analyses were performed on the study outcomes. e sensitivity analysis results revealed that none of the included studies had a noticeable influence on the combined result of PFS between the full-dose and the nonfull-dose groups. e combined results of PFS ranged from HR: 1 (d) Figure 5: e meta-analysis results for serious adverse reaction incidence rates after afatinib treatment in advanced NSCLC with EGFR mutation: (a) the adverse reaction incidence rate of diarrhea; (b) the adverse reaction incidence rate of mucositis; (c) the adverse reaction incidence rate of skin rashes; and (d) the adverse reaction incidence rate of paronychia.

Discussion
Based on RCT data, a meta-analysis has revealed that in the first-line treatment of EGFR-mutated NSCLC, there is no conclusive evidence that afatinib is more effective than gefitinib or erlotinib [15]. Meanwhile, Wang et al. have performed a meta-analysis of RCTs in advanced NSCLC to assess the safety and efficacy of afatinib when compared with chemotherapy and first-generation EGFR-TKIs. eir results revealed that compared with control groups, afatinib treatment apparently increased ORR   2.1% (1.0%, 3.6%), respectively. Furthermore, in the present study, the efficacy of afatinib in the first-line-only group was significantly superior to that in the second-line treatment. erefore, the efficacy and safety of afatinib has been confirmed by RWE.
e efficacy of tolerability-guided dose adjustment remains controversial. Previously, it has been suggested that 40 mg was the recommended afatinib dose for first-line therapy [10]. A recent study has revealed that the PFS of the non-full-dose group was 12.8 months, while the PFS was    [11,12,44]. 40 mg afatinib daily presented a significantly higher incidence of Grade 3 skin rash (16% vs. 0%) and diarrhea (100% vs. 41%) than 30 mg daily afatinib [45]. In the present study, the frequency and severity of adverse events (including diarrhea, skin rash, mucositis, paronychia, and pruritus) was higher in patients who administered 40 mg afatinib daily than in those who administered 30 mg afatinib daily. However, the differences of adverse reactions in the two groups of tolerability-guided afatinib dose adjustment were not significant. Moreover, compared with the 40 mg/day dose, 57.8% (1917/3319) of patients received a lower afatinib dose, with only 0.5% (18/3319) of patients receiving a higher afatinib dose; this could partly explain the lower tolerability and higher toxicities associated with afatinib 40 mg daily. However, it should be noted that the anticancer efficacy ORR of afatinib 30 mg daily did not surpass that of the 40 mg daily dose. Besides, the incidence and severity of adverse reactions showed a decreased trend in patients receiving non-full dose, which indicated the tolerability-guided dose adjustment alleviated afatinib-related adverse effects.
us, the realworld data support that dose adjustment can be guided according to tolerance once adverse reactions occur.
In addition to the afatinib dose, clinical factors such as brain metastases can influence the results of patients with advanced EGFR-mutant NSCLC [46]. In the LUX-Lung 6 trial, the median PFS of patients with brain metastases treated with afatinib was lower than that of patients without brain metastases [47], which was in accordance with our results, suggesting that brain metastases is an influence factor of patients with advanced EGFR-mutant NSCLC based on afatinib dose. e current study was the first RWE-based meta-analysis to explore the efficacy and adverse reactions in patients with advanced EGFR-mutated NSCLC. However, some limitations persist in the current study: (1) the small sample size of some included studies influenced certain outcome indicators  of meta-analysis; (2) it was not possible to assess the methodological quality of included studies and the impact of quality on the results in this RWE study owing to a lack of suitable quality evaluation tools; (3) subgroup analysis was not performed on first-generation or second-generation EGFR-TKIs for comparing afatinib with erlotinib, dacomitinib, and gefitinib.
In conclusion, afatinib is a safe and effective first-line treatment in patients with EGFR-mutated NSCLC, and tolerability-guided afatinib dose adjustment might not affect the PFS of these patients. is study was performed based on real-world data, reflecting information on curative effects in real-world patients and fully compensates for disadvantages of RCTs.

CI:
Confidence interval DCR: Disease control rate ECOG-PS: Eastern Cooperative Oncology Group Performance Status EGFR: Epidermal growth factor receptor HR: Hazard risk IR: Incidence rate NSCLC: Non-small-cell lung cancer ORR: Overall response rate OS: Overall survival PFS: Progression-free survival RCT: Randomized clinical trial RR: Relative risk RWE: Real-world evidence TTF: Time to failure TKIs: Tyrosine kinase inhibitors WMD: Weighted mean difference AE: Adverse event SAE: Severe adverse event.
Data Availability e raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

Disclosure
An earlier version of abstract of this manuscript has been presented as meeting abstract in 2021 ASCO.

Conflicts of Interest
e authors declare that they have no conflicts of interests.

Authors' Contributions
Zhang Lemeng, Luo Yongzhong, and Chen Jianhua carried out the conception and design of the research, and Cheng Tianli and Yang Hua participated in the acquisition of data. Pan Changqie, Li Haitao, and Jiang Zhou carried out the analysis and interpretation of data. Zhang Lemeng and Luo Yongzhong participated in the design of the study and prepared and revised the manuscript. All authors read and approved the final manuscript.