Abstract Background Correct identification of the EGFR c.2369C>T p.(Thr790Met) variant is key to decide on a targeted therapeutic strategy for patients with acquired EGFR TKI resistance in non-small cell lung cancer. The aim of this study was to evaluate the correct detection of this variant in 12 tumor tissue specimens tested by 324 laboratories participating in External Quality Assessment (EQA) schemes. Methods Data from EQA schemes were evaluated between 2013 and 2018 from cell lines (6) and resections (6) containing the EGFR c.2369C>T p.(Thr790Met) mutation. Adequate performance was defined as the percentage of tests for which an outcome was available and correct. Additional data on the used test method were collected from the participants. Chi-squared tests on contingency tables and a biserial rank correlation were applied by IBM SPSS Statistics version 25 (IBM, Armonk, NY, USA). Results In 26 of the 1190 tests (2.2%) a technical failure occurred. For the remaining 1164 results, 1008 (86.6%) were correct, 151 (12.9%) were false-negative and 5 (0.4%) included incorrect mutations. Correct p.(Thr790Met) detection improved over time and for repeated scheme participations. In-house non-next-generation sequencing (NGS) techniques performed worse (81.1%, n = 293) compared to non-NGS commercial kits (85.2%, n = 656) and NGS (97.0%, n = 239). Over time there was an increase in the users of NGS. Resection specimens performed worse (82.6%, n = 610 tests) compared to cell line material (90.9%, n = 578 tests), except for NGS (96.3%, n = 344 for resections and 98.6%, n = 312 for cell lines). Samples with multiple mutations were more difficult compared to samples with the single p.(Thr790Met) variant. A change of the test method was shown beneficial to reduce errors but introduced additional analysis failures. Conclusions A significant number of laboratories that offer p.(Thr790Met) testing did not detect this relevant mutation compared to the other EQA participants. However, correct identification of this variant is improving over time and was higher for NGS users. Revising the methodology might be useful to resolve errors, especially for resection specimens with low frequency or multiple variants. EQA providers should include challenging resections in the scheme.
