Rossi, Sabrina, Gasparotto, Daniela, Miceli, Rosalba, Toffolatti, Luisa, Gallina, Giovanna, Scaramel, Enrico, Marzotto, Alessandra, Boscato, Elena, Messerini, Luca, Bearzi, Italo, Mazzoleni, Guido, Capella, Carlo, Arrigoni, Gianluigi, Sonzogni, Aurelio, Sidoni, Angelo, Mariani, Luigi, Amore, Paola, Gronchi, Alessandro, Casali, Paolo G., Maestro, Roberta, and Dei Tos, Angelo P.
Supplemental Digital Content is available in the text.The mutation status of KITor PDGFRAnotoriously affects the response of advanced gastrointestinal stromal tumors (GISTs) to tyrosine kinase inhibitors. Conversely, it is currently still unclear whether mutation status impinges on the prognosis of localized, untreated GISTs. Hence, at present, this variable is not included in decision making for adjuvant therapy. A series of 451 primary localized GISTs were analyzed for KIT, PDGFRA, and BRAFmutations. Univariable and multivariable analyses and a backward selection procedure were used to assess the impact of mutation status on overall survival and to identify prognostically homogenous groups. Mutation was a significant prognostic indicator of overall survival in naive, localized GISTs (P<0.001): KIT-mutated patients had a worse outcome than PDGFRA-mutated or triple-negative (KIT, PDGFRA, BRAFwild-type) cases. Multivariable Cox regression models allowed us to identify 3 molecular risk groups: group I exhibited the best outcome and included PDGFRAexon 12, BRAF, and KITexon 13-mutated cases; group II, of intermediate clinical phenotype (HR=3.06), included triple-negative, KITexon 17, PDGFRAexon 18 D842V, and PDGFRAexon 14-mutated cases; group III displayed the worst outcome (hazard ratio=4.52), and comprised KITexon 9 and exon 11 and PDGFRAexon 18 mutations apart from D842V. This study highlights the prognostic impact of mutation status on the natural course of GIST and suggests that the molecular prognostic grouping may complement the conventional clinicopathologic risk stratification criteria in decision making for adjuvant therapy.