To the Editor Arrieta and colleagues 1 recently published an openlabel randomized clinical trial (RCT) evaluating the addition of metformin to tyrosine kinase inhibitor (TKI) treatment in nondiabetic patients with non-small cell lung cancer carrying epidermal growth factor receptor (EGFR) variants. The authors reported significantly improved survival benefits without additional toxic effects. However, there are some major issues about this work that need to be addressed, especially considering that our larger-scale placebo-controlled doubleblind RCT 2 attempting to answer the same question had negative results.According to their article 1 and a poster that Arrieta et al submitted to the 2018 American Society of Clinical Oncology annual meeting, 139 patients were randomly assigned in a 1:1 ratio to metformin plus a TKI (n = 69) or a TKI alone (n = 70). However, in their original abstract submitted approximately 2 months earlier, 3 only 49 patients were enrolled in the combination arm and 67 in the TKI arm. Therefore, 23 patients were recruited between the 2 submissions, but oddly, 20 of them ended up in the metformin group while only 3 were placed in the control arm, so obviously allocation randomization/ concealment, the cornerstone of RCTs, was disrupted.Patient enrollment also seemed biased because loss of LKB1 was extraordinarily enriched (33.3%) in the subset evaluated, 1 while LKB1 variants, which predominantly predict protein loss, are mutually exclusive with EGFR variants as confirmed by the 1.45% rate of LKB1-EGFR covariants in The Cancer Genome Atlas non-small cell lung cancer cohort. 4 Baseline characteristics were not balanced, either, because all 5 patients with exon 18 G719X variants, who are supposedly less sensitive to firstgeneration TKIs than to afatinib, were allocated to the combination group. 5 The heterogeneous selection of TKIs represents another source of bias, with more than 40% of the patients choosing afatinib over gefitinib/erlotinib, for which the rationale was never provided. The uneven distribution of TKIs between treatment arms further complicated the picture, especially without stratification by types of variants and TKIs.Indeed, the monotherapy arm exhibited remarkably lower overall survival and objective response rate 1 than historical data on first-line TKIs, which they unconvincingly attributed to a high incidence of brain metastases and adoption of chemotherapy beyond progression; however, the control group's progression-free survival looked reasonable, and switching to chemotherapy on dropout/completion was common in pivotal studies. Moreover, the intolerable toxic effects of metformin, as inferred from discontinuation occurring in 16 of 69 patients (23.2%) and dosage reduction occurring in 24 of 69 patients (34.8%), 1 also contradict their claim that the 2 groups had similar safety profiles. Metformin plus gefitinib was shown to cause a higher frequency of diarrhea than gefitinib alone. 2 Given the issues enumerated above, we advise extra caution before any conclusions are...