Of the cases identified herein three known and novel FGFR
Of the 52 cases identified herein, three known and 11 novel FGFR fusion partners were identified. Notably, an oligomerization domain, including coiled-coil dimerization domains similar to that seen in echinoderm microtubule associated protein like 4 gene (EML4), which is the most common ALK partner gene, was present in eight of 10 unique 3' fusion partners. In the four cases with FGFR as the 3' gene, the 5' fusion partners included the BCL2 associated athanogene 4 antiapoptotic protein with retention of the BAG binding domain, the anoctamin 3 transmembrane protein, and nuclear receptor binding SET domain protein 1 and nuclear receptor binding SET domain protein 2 (WHSC1), which are SET domain–containing proteins with zinc finger domains. Tumors positive for FGFR3 fusions have been shown to be sensitive to FGFR inhibition. We have presented the case of a patient with lung adenocarcinoma harboring a novel FGFR2-LZTFL1 fusion who had a prolonged response to treatment with an FGFR inhibitor. This suggests that the FGFR2 fusion in this case was a driver alteration and conveyed sensitivity to FGFR inhibition. Furthermore, FGFR fusions may confer sensitivity to FGFR TKIs, possibly irrespective of the fusion partner, and this should be studied further. In our cohort, we also identified two patients who had FGFR3-TACC3 fusions in their tumor tissue obtained after treatment with EGFR TKIs for a primary EGFR-mutated NSCLC. In one case, comprehensive genomic testing on the tumor before Fluorescein TSA Fluorescence System Kit of an EGFR TKI was performed and no FGFR3 fusion was detected, confirming that the fusion was acquired after treatment. In the other case, only EGFR testing was performed before treatment, so we could not confirm that the fusion was not present initially. However, both of these cases, as well as similar cases in which the FGFR3 fusion was detected in tissue or circulating tumor DNA after administration of an EGFR TKI but was not found in the pretreatment sample, have been described previously. This adds to emerging data that support FGFR alterations not only as a primary driver mutation but also as a potential mechanism of resistance to other targeted agents. Terai et al. developed human NSCLC lines resistant to the EGFR TKI gefitinib and showed (after confirming the absence of the T790M mutation) that there was significantly increased expression of FGFR1 compared with that in the parent nonresistant cells. The addition of an FGFR inhibitor restored the sensitivity of these cells to gefitinib. However, in a phase I study (NCT0151969), the combination of an EGFR TKI (erlotinib) with dovitinib (a multireceptor TKI with activity against FGFR1 and FGFR3) led to intolerable toxicity. Therefore, it may be more feasible to consider combining EGFR inhibitors with more selective FGFR inhibitors currently in clinical development. Whether serial administration of these agents (i.e., an EGFR TKI followed by, or alternated with, a FGFR inhibitor when a FGFR alteration is detected) may be effective is also unknown. FGFR alterations as mechanisms of resistance should continue to be studied to further inform appropriate clinical care for these patients. Aside from the two cases with co-occurring primary EGFR exon 19 deletion mutations, only two of the 50 remaining FGFR fusion–positive cases in our data set had a co-occurring known targetable driver alteration (MET exon 14 splice and EGFR L861Q) included in the NSCLC National Comprehensive Cancer Network guidelines (version 1.2018). In the case with MET exon 14 splice and FGFR3-TACC3, the patient had not received any prior targeted therapy, and in the case with EGFR L861Q and ANO3-FGFR4 the treatment history is unknown. The lack of other known drivers in the remaining cases suggests that the fusion is likely to be the driving alteration. Further, in adenocarcinoma cases harboring FGFR fusions in the data set, the TMB was relatively low (median 5.2 mutations/Mb), which is consistent with that seen in tumors harboring other targetable driver alterations such as ALK fusions. This is also consistent with the median TMB being lower in patients with lung adenocarcinomas overall (median TMB 6.1 mutations/Mb in our data set), who tend to have a higher frequency of known driver alterations and a lower incidence of smoking history than do patients with lung SCCs (median TMB 8.7 mutations/Mb in our data set).