BACKGROUND Despite current therapies, diffuse cutaneous systemic sclerosis (scleroderma) often has a devastating outcome. We compared myeloablative CD34+ selected autologous hematopoietic stem-cell transplantation with immunosuppression by means of 12 monthly infusions of cyclophosphamide in patients with scleroderma. METHODS We randomly assigned adults (18 to 69 years of age) with severe scleroderma to undergo myeloablative autologous stem-cell transplantation (36 participants) or to receive cyclophosphamide (39 participants). The primary end point was a global rank composite score comparing participants with each other on the basis of a hierarchy of disease features assessed at 54 months: death, event-free survival (survival without respiratory, renal, or cardiac failure), forced vital capacity, the score on the Disability Index of the Health Assessment Questionnaire, and the modified Rodnan skin score. RESULTS In the intention-to-treat population, global rank composite scores at 54 months showed the superiority of transplantation (67% of 1404 pairwise comparisons favored transplantation and 33% favored cyclophosphamide, P = 0.01). In the per-protocol population (participants who received a transplant or completed ≥9 doses of cyclophosphamide), the rate of event-free survival at 54 months was 79% in the transplantation group and 50% in the cyclophosphamide group (P = 0.02). At 72 months, Kaplan–Meier estimates of event-free survival (74% vs. 47%) and overall survival (86% vs. 51%) also favored transplantation (P = 0.03 and 0.02, respectively). A total of 9% of the participants in the transplantation group had initiated disease-modifying antirheumatic drugs (DMARDs) by 54 months, as compared with 44% of those in the cyclophosphamide group (P = 0.001). Treatment-related mortality in the transplantation group was 3% at 54 months and 6% at 72 months, as compared with 0% in the cyclophosphamide group. CONCLUSIONS Myeloablative autologous hematopoietic stem-cell transplantation achieved long-term benefits in patients with scleroderma, including improved event-free and overall survival, at a cost of increased expected toxicity. Rates of treatment-related death and post-transplantation use of DMARDs were lower than those in previous reports of nonmyeloablative transplantation. (Funded by the National Institute of Allergy and Infectious Diseases and the National Institutes of Health; ClinicalTrials.gov number, NCT00114530.)

In the phase I/II pediatric hydroxyurea safety trial (HUG-KIDS), school-aged children with sickle cell anemia receiving hydroxyurea at the maximally tolerated dose (MTD) had variable increases in the percentage of fetal hemoglobin (%HbF). To identify predictors of the HbF response to hydroxyurea therapy, baseline clinical and laboratory values (age, sex, hemoglobin concentration, %HbF, reticulocytes, white blood cell [WBC], platelets, and serum chemistries), as well as treatment variables (number of toxicities, noncompliance, MTD dose, and MTD blood counts) were analyzed in 53 HUG-KIDS children who achieved MTD. Baseline %HbF values (P ‫؍‬ .001), baseline hemoglobin concentration (P ‫؍‬ .01), MTD dose (P ‫؍‬ .02), and compliance (P ‫؍‬ .02) were significantly associated with a higher %HbF at MTD; in contrast, age, sex, number of toxicities, and other baseline hematologic parameters were not. After adjusting for variations in baseline %HbF, the baseline reticulocyte count (P ‫؍‬ .05) and baseline WBC count (P ‫؍‬ .05) were also significantly associated with a higher %HbF at MTD. Hydroxyurea-induced increases in the hemoglobin concentration and mean corpuscular volume (both higher absolute values at MTD and larger positive changes from baseline values), as well as hydroxyurea-induced decreases in reticulocytes and WBC count, were significantly associated with a higher %HbF at MTD. These data suggest that selected baseline laboratory parameters, a higher MTD dose with attention to compliance, and greater therapy-related changes in blood counts may predict the HbF response to hydroxyurea therapy for children with sickle cell anemia. The HbF response to hydroxyurea is variable and complex, however, and even children with IntroductionThe level of fetal hemoglobin (HbF) expression is one of the most important modifiers of disease expression for patients with sickle cell anemia. 1 The percentage of HbF (%HbF) influences both laboratory values and clinical features of children and adults with sickle cell anemia. For example, an elevated %HbF has been significantly associated with fewer painful vasoocclusive events, 2 fewer episodes of acute chest syndrome, 3 and reduced early mortality. 4,5 Pharmacologic enhancement of HbF expression has been accomplished by using myelosuppressive agents, cytokines, and short-chain fatty acids. [6][7][8] To date, however, the prototypic agent for increasing HbF expression is hydroxyurea, based on its ease of oral administration, modest toxicity profile, and rarity of serious side effects. A phase I/II trial of hydroxyurea for adults with sickle cell anemia demonstrated a significant increase in HbF expression, accompanied by a significant increase in hemoglobin concentration and significant decreases in reticulocytes, neutrophils, and platelets. 9 The Multicenter Study of Hydroxyurea (MSH), a doubleblinded, placebo-controlled phase III trial for adults with sickle cell anemia, demonstrated that hydroxyurea therapy significantly reduced the number of painful events, episodes of acute che...

BackgroundRecursive partitioning is a non-parametric modeling technique, widely used in regression and classification problems. Model-based recursive partitioning is used to identify groups of observations with similar values of parameters of the model of interest. The mob() function in the party package in R implements model-based recursive partitioning method. This method produces predictions based on single tree models. Predictions obtained through single tree models are very sensitive to small changes to the learning sample. We extend the model-based recursive partition method to produce predictions based on multiple tree models constructed on random samples achieved either through bootstrapping (random sampling with replacement) or subsampling (random sampling without replacement) on learning data.ResultsHere we present an R package called “mobForest” that implements bagging and random forests methodology for model-based recursive partitioning. The mobForest package constructs large number of model-based trees and the predictions are aggregated across these trees resulting in more stable predictions. The package also includes functions for computing predictive accuracy estimates and plots, residuals plot, and variable importance plot.ConclusionThe mobForest package implements a random forest type approach for model-based recursive partitioning. The R package along with it source code is available at http://CRAN.R-project.org/package=mobForest.