Study quality and characteristics
In total, 1,511 articles were identified in the literature search. After the removal of duplicates, the titles and abstracts of 1,148 articles were screened. Of these, 934 articles were excluded, and the full texts of the remaining 214 articles were assessed for eligibility. A further 154 articles were excluded because of insufficient data, overlapped data, or because they were unsuitable based on the inclusion criteria. The remaining 60 studies were included in this meta-analysis. The study selection process was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [15].
The characteristics of the 60 studies are listed in Table 1. Patients with stage IIIB and IV NSCLC were enrolled in the included 61 studies and only one study was performed in patients with IIB and IIIA NSCLC [58]. Using the Jadad score, 37 studies were classified as low quality (a score of 2 and or less), whereas 23 studies were classified as high quality (a score of 3 or greater) (Table 1).
Incidence and relative risk of erlotinib-induced eye disorders
Forty-four of the 60 studies were assessed for the incidence of eye disorders in 20,964 participants treated with erlotinib (Table 1). In total, 611 eye-related adverse events of any grades were reported, and thirty three of the 611 adverse events were classified as grade 3–4. The incidence of eye disorders in each study was between 0.00% and 45.7%. The index of eye disorders mainly included dry eye, conjunctivitis, blurry vision, and other ocular adverse events.
The overall incidence (event rate) of eye disorders for any grade was 3.30% (95% confidence interval [CI] 2.20%–5.00%) using the random-effects model (Table 2).
Four studies were assessed for the specific contribution of erlotinib to the development of eye disorders by comparing erlotinib-treatment groups and control groups (placebo or other treatment). The risk ratios (RR) and 95% CI of the comparison between the two groups were 2.91 and 1.70–4.98, respectively (Figure 1). Reanalysis using a random-effects model revealed the significant differences (RR = 3.34; 95% CI 1.32–8.45). This result indicated that patients who received erlotinib had significantly increased the risk of ocular toxicities.
Incidence and relative risk of erlotinib-induced hepatobiliary disorders
Fifty-two of the 60 studies were assessed for the incidence of hepatobiliary disorders in 21,339 participants treated with erlotinib (Table 1). The index of hepatobiliary disorders mainly included alanine transaminase (ALT) or aspartate transaminase (AST) elevations, alkaline phosphatase elevation, hyperbilirubinemia, and other hepatobiliary adverse events. In detail, 751 ALT increases, 456 bilirubin increases, and 1,025 other adverse events of any grade were reported and 301 of them were classified as grade 3–4. The incidence in each study of the ALT increase ranged from 0.00% to 50.9% and the incidence of bilirubin increase ranged from 0.00% to 38.9%.
The overall incidence (event rate) of ALT and bilirubin increases were 6.40% (95% CI 3.90–10.4) and 3.8% (95% CI 2.30%–6.10%), respectively, using the random-effects model (Table 2). The overall incidence of other adverse events except ALT and bilirubin increases were 1.00 % (95% CI 0.60%–1.80%). The incidence of any hepatobiliary disorders of grade 3–4 was 2.20% (95% CI 1.50%–3.10%) (Table 2).
Five studies were included to compare the liver toxicity of erlotinib, representing as ALT elevation, between erlotinib-treatment groups and control groups (placebo or other treatment). The RR and 95% CI of the comparison between the two groups were 1.319 and 0.913–1.904, respectively, using the fixed-effects model (Figure 2). Reanalysis using a random-effects model revealed no significant differences. This result indicated that patients who received erlotinib had no significantly increased risk of liver-related toxicities.
Incidence of erlotinib-induced renal disorders
Forty-three studies were assessed for erlotinib-induced renal disorders in 10,367 participants treated with erlotinib (Table 1). In total, 218 renal adverse events of any grade were reported and nine of the 218 adverse events were classified as grade 3–4. The incidence of renal disorders in each study ranged from 0.00% to 25.4%. The index of renal disorders mainly included elevated serum creatinine, proteinuria, renal failure, and other renal adverse events.
The overall incidence (event rate) of renal disorders was 3.10% (95% CI 1.90%–5.00%), using the random-effects model (Table 2). The incidence of renal disorder of grade 3–4 was 1.10% (95% CI 0.70%–1.60%), using the fixed-effects model (Table 2). Reanalysis using a random-effects model showed the same result.
Sensitivity analysis and publication bias
As the results of sensitivity analysis, no significant differences were observed (data available on request). The results of publication bias through the Begg’s rank correlation test and Egger’s regression test are shown in Table 2 and 3.