Radiomics and Machine Learning for prediction of two-year disease-specific mortality and KRAS mutation status in metastatic colorectal cancer.

PURPOSE: Colorectal cancer is usually accompanied by liver metastases. The prediction of patient evolution is essential for the choice of the appropriate therapy. The aim of this study is to develop and evaluate machine learning models to predict KRAS gene mutations and 2-year disease-specific mortality from medical images. METHODS: Clinical and follow-up information was collected from patients with metastatic colorectal cancer who had undergone computed tomography prior to liver resection. The dominant liver lesion was segmented in each scan and radiomic features were extracted from the volumes of interest. The 65% of the cases were employed to perform feature selection and to train machine learning algorithms through cross-validation. The best performing models were assembled and evaluated in the remaining cases of the cohort. RESULTS: For the mortality model development, 101 cases were used as training set (64 alive, 37 deceased) and 35 as test set (22 alive, 13 deceased); while for KRAS mutation models, 55 cases were used for training (31 wild-type, 24 mutated) and 30 for testing (17 wild-type, 13 mutated). The ensemble of top performing models resulted in an area under the receiver operating characteristic curve of 0.878 for mortality and 0.905 for KRAS prediction. CONCLUSIONS: Predicting the prognosis of patients with metastatic colorectal cancer is useful for making timely decisions about the best treatment options. This study presents a noninvasive method based on quantitative analysis of baseline images to identify factors influencing patient outcomes, with the aim of incorporating these tools as support systems.

Severe toxicity in adult patients with lung cancer under treatment with pemetrexed: a prospective cohort study.

Pemetrexed is an antimetabolite approved for treatment of non-small cell lung cancer. Harbouring interindividual variability in both the pharmacokinetic and pharmacogenetic profiles may lead to life-threatening toxicities. A prospective cohort study of adult patients initiating treatment with pemetrexed in combination with platinum between 2013 and 2015 were follow up. Primary exposure were the methylenetetrahydrofolate reductase (MTHFR) single base polymorphisms in exon 4 and 7 and 5'-UTR- thymidylate synthase (TYMS) VNTR genotypes, in addition to baseline clinical and demographic variables. We used a Cox regression model to evaluate patient's survival and toxicity experience and its association with both baseline characteristics, and a-priori determined genetic polymorphisms. Seventy two patients were included, 52.7% developed severe hematologic toxicity during follow-up. None of the tested genotypes were significantly associated with the main outcome on multivariate analysis, nor other basal clinical variables. Overall survival between patients experiencing the outcome was not different from those without it, but hospital admissions were more frequent. MTHFR and 5'-UTR-TYMS genotypes were not useful for predicting high grade toxicity events in patients under treatment with pemetrexed.

Universal determination of microsatellite instability using BAT26 as a single marker in an Argentine colorectal cancer cohort.

Microsatellite instability (MSI) is a hallmark tool for Lynch syndrome (LS) screening and a prognostic marker for sporadic colorectal cancer (CRC). In regions with limited resources and scarce CRC molecular characterization as South America, the implementation of universal MSI screening is under debate for both its purposes. We sought to estimate the frequency of BAT26 in colorectal adenocarcinomas and to determine associated clinical and histological features. Consecutive patients from a CRC registry were included. BAT26 determination was performed in all cases; if instability was found, immunohistochemistry (IHC) and BRAF mutation analyses were done, as appropriate. Differences were assessed by chi-squared or Fisher's exact test, or by T test or Mann-Whitney. Multiple logistic regression was used to identify factors independently associated with BAT26-unstable tumors. We included 155 patients; mean age was 65.6 (SD 14.4) and 56.1% were male. The frequency of BAT26-unstable tumors was 22% (95% CI 15.7-29.3). Factors independently associated with BAT26-unstable tumors were right colon localization (OR 3.4, 95% CI 1.3-8.7), histological MSI features (OR 5.1, 95% CI 1.9-13.6) and Amsterdam criteria (OR 23.2, 95% CI 1.9-286.7). IHC was altered in 85.3% BAT26-unstable tumors and 70.6% lacked MLH1 expression; 47.8% of these harbored BRAF V600E mutation. We provide evidence to link the frequency of BAT26 to an increased diagnostic yield (up to 1.4-folds) of suspected LS cases in comparison to the revised Bethesda guidelines alone. In regions with limited resources, clinical and histological features associated with BAT26-unstable status could be useful to direct MSI screening in sporadic CRCs and may help guide clinical care and future research.