Tao Shi

PURPOSE: The trial objectives were to identify the maximum-tolerated dose (MTD) of first-line gemcitabine plus nab-paclitaxel in metastatic pancreatic adenocarcinoma and to provide efficacy and safety data. Additional objectives were to evaluate positron emission tomography (PET) scan response, secreted protein acidic and rich in cysteine (SPARC), and CA19-9 levels in relation to efficacy. Subsequent preclinical studies investigated the changes involving the pancreatic stroma and drug uptake. PATIENTS AND METHODS: Patients with previously untreated advanced pancreatic cancer were treated with 100, 125, or 150 mg/m(2) nab-paclitaxel followed by gemcitabine 1,000 mg/m(2) on days 1, 8, and 15 every 28 days. In the preclinical study, mice were implanted with human pancreatic cancers and treated with study agents. RESULTS: A total of 20, 44, and three patients received nab-paclitaxel at 100, 125, and 150 mg/m(2), respectively. The MTD was 1,000 mg/m(2) of gemcitabine plus 125 mg/m(2) of nab-paclitaxel once a week for 3 weeks, every 28 days. Dose-limiting toxicities were sepsis and neutropenia. At the MTD, the response rate was 48%, with 12.2 median months of overall survival (OS) and 48% 1-year survival. Improved OS was observed in patients who had a complete metabolic response on [(18)F]fluorodeoxyglucose PET. Decreases in CA19-9 levels were correlated with increased response rate, progression-free survival, and OS. SPARC in the stroma, but not in the tumor, was correlated with improved survival. In mice with human pancreatic cancer xenografts, nab-paclitaxel alone and in combination with gemcitabine depleted the desmoplastic stroma. The intratumoral concentration of gemcitabine was increased by 2.8-fold in mice receiving nab-paclitaxel plus gemcitabine versus those receiving gemcitabine alone. CONCLUSION: The regimen of nab-paclitaxel plus gemcitabine has tolerable adverse effects with substantial antitumor activity, warranting phase III evaluation.

A random forest (RF) predictor is an ensemble of individual tree predictors. As part of their construction, RF predictors naturally lead to a dissimilarity measure between the observations. One can also define an RF dissimilarity measure between unlabeled data: the idea is to construct an RF predictor that distinguishes the “observed” data from suitably generated synthetic data. The observed data are the original unlabeled data and the synthetic data are drawn from a reference distribution. Here we describe the properties of the RF dissimilarity and make recommendations on how to use it in practice.An RF dissimilarity can be attractive because it handles mixed variable types well, is invariant to monotonic transformations of the input variables, and is robust to outlying observations. The RF dissimilarity easily deals with a large number of variables due to its intrinsic variable selection; for example, the Addcl 1 RF dissimilarity weighs the contribution of each variable according to how dependent it is on other variables.We find that the RF dissimilarity is useful for detecting tumor sample clusters on the basis of tumor marker expressions. In this application, biologically meaningful clusters can often be described with simple thresholding rules.