The generated prognostic model was able to powerfully stratify patients into 3 classes [54]. Recently, a large retrospective analysis from the SEER database showed that the increasing number of resected positive nodes and a higher ratio between metastatic and overall resected nodes have an independent negative prognostic

impact for overall survival in N1 patients [55, 56]. Although selleck chemicals llc a prospective validation is mandatory, these results suggest the inclusion in the next TNM of other nodal descriptors than site and status to improve the prognostic power. Molecular factors Several molecular prognostic (and predictive) models have been published in the attempt to improve the clinical decision process [57–62]. Although promising, their effectiveness and clinical utility were undermined by several limitations: the inability to account for comorbidities and other clinical factors (which affect prognosis) [63], methodological and statistical biases, lack of a solid and reproducible internal and external validation [64, 65]. The proposal of a new prognostic model should always be supported by reliable validations. A pivotal example is represented by the recent retraction by Potti et al of their NU7026 ic50 promising metagene expression model (which led to stop the CALBG 30506 trial and to remove the metagene analysis from the study

design) [66]. The analysis of the data from the available randomized trials exploring

the role of eventual predictors for either prognosis or treatment efficacy hides many drawbacks given its retrospective nature. This is further complicated by the extremely learn more relevant impact of the attrition rate oxyclozanide for the analysis of biological samples [67]. Nevertheless, many investigators involved in those trials exploring the benefit of adjuvant chemotherapy planned and conducted intriguing analyses to generate working hypotheses for future biomarker-driven randomized trials, as follows: IALT-BIO : the low IHC expression of the excision repair cross complementation group 1 (ERCC1) represents a marker of better outcome in patients receiving cisplatinum in ACT (HR = 0.65; p = .0002 vs HR = 1.14; p = .4 for high expression; interaction test p = .0009); conversely, high ERCC1 expression correlates with longer OS in the control group (HR = 0.66) [68]. In the context of cell cycle regulators, p27, while having a predictive role for patients treated with ACT, does not affect prognosis (p27 negative HR 0.66; p = .006; p27 positive HR 1.09; p = .54; interaction test p = .02)[69]. Similarly to ERCC1, the low expression of MutS homologue 2 (MSH2) was predictive of benefit from platinum based -ACT (low MSH2 HR = 0.76; p = .03 vs high MSH 2 HR = 1.12; p = .48; interaction test p = .06). MSH2 high expression was also prognostic for longer survival in untreated patients (HR = 0.66; p = .01)[70].