Psychosocial Sequelae of Stroke in Working-Age Adults: A Pilot Study.

OBJECTIVE: Few prospective studies have examined psychosocial symptoms in young adult stroke survivors. The purpose of this pilot study was to examine the feasibility of methods to assess and describe the presence of symptoms and their characteristics in young adult stroke survivors. METHODS: A prospective descriptive design using PROMIS and NeuroQoL patient-reported outcomes was used. Stroke survivors aged 18 to 65 years admitted for acute ischemic stroke completed baseline measures before hospital discharge; the same measures were collected by telephone interview 6 months later. RESULTS: Thirty patients were recruited, and 18 (60%) completed follow-up measures. There was wide variability in scores across patients for each patient-reported outcome that persisted at 6 months. NeuroQoL cognitive function was the only patient-reported outcome to have a statistically significant change from baseline to 6 months (P = .045). CONCLUSIONS: It is feasible to recruit and retain young adult stroke survivors in a prospective study. Wide variability in scores across patients suggests different trajectories of recovery in the first 6 months after stroke. Neuroscience nurses can assess for these symptoms and provide individualized interventions to aid in self-management of burdensome symptoms.

Standardizing the Frequency of Neurologic Assessment After Acute Stroke.

The National Institutes of Health Stroke Scale (NIHSS) is considered the standard for assessing neurologic status after an acute stroke. Currently, there is no guideline for when this assessment should be completed, nor is there consensus on how frequently or for how long.We initiated a quality improvement project that sought first to determine when and how often nurses at a variety of institutions in our multisite health care system completed the NIHSS assessment and then to identify the minimum frequency at which nurses should complete the assessment. After reviewing the literature and current practices and observing internal and external benchmarks, we set a new standard for all institutions in the system to follow.The new assessment frequency was based on patient condition and level of care, although that frequency would change if a patient showed new or worsening neurologic changes. The new standard was successfully implemented at all the primary and comprehensive certified stroke centers in our network. Ongoing monitoring confirmed that the frequency of NIHSS assessment met the needs of our patients and ensured staff adherence to the new practice.

Schlieren visualization of fluid dynamics effects in direct analysis in real time mass spectrometry.

RATIONALE: The success of ambient analysis using plasma-based ion sources depends heavily on fluid dynamics and mass transport efficiency in the sample region. To help characterize the influence of these determining factors, visualization of the gas flow profile for a Direct Analysis in Real Time (DART) ion source at the mass spectrometer atmospheric pressure (AP) interface was performed using the Schlieren technique. METHODS: The DART helium flow pattern was imaged in model systems incorporating different interface designs, i.e. skimmer or capillary inlet, and for sampling strategies using several types of traditional DART sample probes including a glass capillary, swab, and drug tablet. Notably, Schlieren experiments were conducted on instruments equipped with the gas-ion separator tube (GIST) adapter and Vapur® pump, and on setups featuring the transmission mode (TM) DART module used in standard practice. RESULTS: DART sources were seen to expel a collimated, highly laminar helium stream across interface distances up to ~8 cm. The helium stream was robust to the influence of gas temperature (50-500 °C) and flow rate (≤3.5 L min(-1) ), but considerable DART gas deflection or full disruption was observed in each sampling scenario. The severity of the flow disturbance depended on probe size and placement, the GIST/Vapur® settings, or counter-current gas movements present at the interface. CONCLUSIONS: The real-time Schlieren visualizations introduced in this work provide new insight on the fluid dynamics within the DART-MS sample gap while also helping to identify those experimental parameters requiring optimization for improved transmission.

Effect of premedication and other factors on the occurrence of acute transfusion reactions in dogs.

OBJECTIVE: To evaluate the effect of premedication on transfusion reactions (TRs) within 24 hours after blood product transfusions in dogs. DESIGN: Retrospective study between 2008 and 2011. SETTING: Private veterinary referral hospital. ANIMALS: Nine hundred and thirty-five transfusion events in 558 dogs. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Medical records of dogs receiving blood product transfusions were reviewed. Information collected included signalment, weight, transfusion product type, reason for transfusion, first or subsequent transfusion, whether an acute reaction occurred, type of reaction, whether the reaction was treated, premedication prior to the transfusion and the premedication used, other medications the animal was given, whether the animal had an immune-mediated process, and whether the transfusion was administered in the perioperative period. A total of 144 (15%) acute TRs were documented in 136 dogs. The most common TRs were fever alone (77/144 [53%]) and vomiting alone (26/144 [18%]). Six dogs died due to the TR (4%). TR was not associated with age (P = 0.257), sex (P = 0.754), weight (P = 0.829), or premedication (P = 0.312). The type of blood product transfused (P < 0.001) was significantly associated with TRs, with packed RBCs most likely associated with a TR, and plasma least likely. Immune disease (P = 0.015) was significantly associated with occurrence of a TR. Significantly fewer reactions were documented following transfusions given in the perioperative period (P = 0.023). CONCLUSIONS: While most TRs were mild, there were some serious reactions observed including hemolysis, dyspnea, and 6 deaths. Immune-mediated disease was associated with development of a TR, while transfusion during the perioperative period was associated with lower likelihood of reaction. Packed RBC transfusions were associated with development of acute TRs. Overall occurrence of TR was not significantly altered with premedication; however, when evaluated alone, antihistamines decreased the incidence of acute allergic reactions.

Plasma-spray ionization (PLASI): a multimodal atmospheric pressure ion source for liquid stream analysis.

A new ion generation method, named plasma-spray ionization (PLASI) for direct analysis of liquid streams, such as in continuous infusion experiments or liquid chromatography (LC), is reported. PLASI addresses many of the analytical limitations of electrospray ionization (ESI) and has potential for real time process stream analysis and reaction monitoring under atmospheric conditions in non-ESI friendly scenarios. In PLASI-mass spectrometry (MS), the liquid stream is pneumatically nebulized and partially charged at low voltages; the resultant aerosol is thus entrained with a gaseous plasma plume from a distal glow discharge prior to MS detection. PLASI-MS not only overcomes ESI-MS limitations but also generates simpler mass spectra with minimal adduct and cluster formation. PLASI utilizes the atomization capabilities of an ESI sprayer operated below the ESI threshold to generate gas-phase aerosols that are then ionized by the plasma stream. When operated at or above the ESI threshold, ionization by traditional ESI mechanisms is achieved. The multimodal nature of the technique enables readily switching between plasma and ESI operation. It is expected that PLASI will enable analyzing a wide range of analytes in complex matrices and less-restricted solvent systems, providing more flexibility than that achievable by ESI alone.

Head group-functionalized poly(ethyleneglycol)-lipid (PEG-lipid) surface modification for highly selective analyte extractions on capillary-channeled polymer (C-CP) fibers.

Polypropylene (PP) capillary-channeled polymer (C-CP) fibers were modified by adsorption of a head group-functionalized lipid to generate analyte-specific surfaces for application as a stationary phase in high performance liquid chromatography (HPLC) or solid phase extraction (SPE). The aliphatic moiety of the lipid adsorbs strongly to the hydrophobic PP surface, with the hydrophilic active head groups orienting themselves toward the more polar mobile phase, thus allowing for interactions with the desired solutes. Initial proof-of-concept was achieved by adsorbing a biotin-poly(ethylene glycol)-functionalized lipid to the surface of the PP C-CP fibers. Surface modification and uniformity was evaluated by binding streptavidin labeled with Texas Red (SAv-TR) to the biotin moiety. Isolation of SAv-TR from a mixture in neat buffer and in cleared lysate demonstrated the capability of the modified fibers to extract an analyte of interest from a complex viscous mixture. It is believed that this surface modification approach is generally applicable to a diversity of selective protein immobilization applications, including clinical diagnostics and preparative scale HPLC on C-CP fibers as well as to other hydrophobic supports.

Sonoclot evaluation of whole blood coagulation in healthy adult dogs.

OBJECTIVE: To establish a standard protocol for analysis of canine whole blood and generate reference intervals for healthy dogs using the Sonoclot analyzer, and to compare Sonoclot values to standard and viscoelastic coagulation tests. DESIGN: Prospective study. SETTING: Veterinary University research facility and teaching hospital. ANIMALS: Twelve healthy random source dogs and 52 healthy dogs from the general veterinary school population. INTERVENTIONS: Blood sampling for viscoelastic coagulation testing. MEASUREMENTS AND MAIN RESULTS: Blood was collected from 12 healthy adult dogs by jugular venipuncture. After a rest period at room temperature of 30, 60, or 120 minutes, 340 µL of citrated blood was added to 20 µL of 0.2 M CaCl(2) in 1 of 2 cuvette types warmed to 37° C. Cuvettes contained a magnetic stir-bar with glass beads (gbACT+) or only a magnetic stir-bar (nonACT). Reference interval samples were collected from 52 healthy adult dogs and analyzed in duplicate. The ACT, CR, and PF were not affected by duration of rest period for either cuvette type. ACT variability was decreased when using gbACT+ cuvettes (P < 0.05). In normal dogs reference intervals (mean ± 2 SD) using gbACT+ cuvettes were: ACT 56.0-154.0 seconds, CR 14.85-46.0, and PF 2.1-4.05. ACT correlated to TEG R-time, K-time, and angle, while CR correlated with all TEG parameters. Fibrinogen correlated with ACT, CR, and PF. Sonoclot did not correlate with other common coagulation tests. CONCLUSIONS: Sonoclot provides viscoelastic evaluation of canine whole blood coagulation and correlated to several TEG parameters and fibrinogen. A standard protocol and reference intervals were established.

Effects of rest temperature, contact activation, and sample technique on canine thrombelastography.

OBJECTIVE: To determine the effects of rest temperature, contact activation (CA), and sample collection technique on thrombelastography (TEG) using canine whole blood. DESIGN: Prospective, experimental study. SETTING: University-based research facility. ANIMALS: Twelve healthy, adult, mixed-breed dogs. INTERVENTIONS: Blood was collected by jugular venipuncture. Tubes containing 3.2% sodium citrate, with and without 75 µg/mL corn trypsin inhibitor (CTI), were filled by vacuum. Samples rested for 30 minutes at 3 temperatures: 37°C, room temperature (RT, 20-22°C), or warmed to 37°C 5 minutes prior to analysis (prewarmed). Samples were analyzed at 37°C. CTI-treated samples were analyzed with and without 1:50,000 tissue factor (TF) as activator. Six dogs were also tested similarly using a needle/syringe collection technique. MEASUREMENT AND MAIN RESULTS: Prewarmed samples exhibited greater MA compared to RT (55.5 ± 7.2 mm vs. 53.5 ± 6.0, P< 0.05), while 37°C samples exhibited a steeper angle (56.7 ± 10.4°C vs. 52.4 ± 8.6°C) and greater MA (55.9 ± 7.5 mm vs. 53.5 ± 6.0 mm) than RT samples (both P< 0.05). CTI-treated samples were hypocoagulable (R time 45 min [7.5-56.8 min], angle 8.2°C [5.1-42.5°C], MA 29.2 ± 9.7 mm, P< 0.001), with TF activation returning all but the angle (42.5 ± 7.6°C) to values similar to citrated samples (angle = 56.7 ± 10.4°C, P = 0.017). Collection using a syringe/needle method revealed a shorter R time for prewarmed samples only (R time 4.7 ± 0.7 min, vs. 5.6 ± 0.8 min for vacuum-collected samples, P = 0.008). CONCLUSIONS: Even in the absence of exogenous activators, CA has an impact on canine TEG results. The effects of rest temperatures and sample collection technique on TEG appear to be minimal.

Capillary-channeled polymer (C-CP) films as processing platforms for protein analysis by matrix-assisted laser/desorption ionization mass spectrometry (MALDI-MS).

Polypropylene (PP) capillary-channeled polymer (C-CP) films have parallel, µm-sized channels that induce solution wicking via capillary action. Efficient mass transport from the solution phase to the channel surface leads to adsorption of hydrophobic protein solutes. The basic premise by which C-CP films can be used as media to manipulate analyte solutions (e.g., proteins in buffer), for the purpose of desalting or chromatographic separation prior to MALDI-MS analysis is presented here. Cytochrome c and myoglobin prepared in a Tris-HCl buffer, and ribonuclease A, lysozyme, and transferrin prepared in phosphate buffered saline (PBS), are used as the test solutions to demonstrate the desalting concept. Protein analysis is performed after deposition on a C-CP film with and without a water washing step, followed by spray deposition of a typical sinapinic acid matrix. Extracted MALDI mass spectra exhibit much improved signal-to-noise characteristics after water washing. A mixture of cytochrome c and myoglobin (2 µL of 2.5 µM each in Tris-HCl buffer) was applied, washed with water and spatially separated via simple capillary action (wicking) using a reversed-phase solvent composition of 0.1% trifluoroacetic acid (TFA) in 50:50 acetonitrile (ACN):H(2)O. Subsequent application of sinapinic acid followed by imaging of the film using MALDI-MS reveals that as the protein solution is wicked down the film, separation occurs.

The effect of unfractionated heparin on thrombelastographic analysis in healthy dogs.

OBJECTIVE: To determine the effect of single and multiple doses of SQ heparin (200 U/kg) on the thrombelastogram of healthy dogs. DESIGN: Prospective study. SETTING: University research facility. ANIMALS: Six random-source female dogs. INTERVENTIONS: Baseline parameters, including a CBC with platelet count, prothrombin time, activated partial thromboplastin time (aPTT), and antithrombin were performed. Thrombelastography (TEG) and aPTT were performed hourly for 12 hours after unfractionated heparin dosing (200 U/kg, SQ). Anti-Xa activity was assayed at 0, 3, 6, and 8 hours. Heparin was then administered every 8 hours for 3 days. The sampling protocol on Day 4 was identical to Day 1. MEASUREMENTS AND MAIN RESULTS: On Day 1, percentage change from baseline for TEG parameter R, as well as absolute values of K, angle, and maximum amplitude (MA) were evaluated. Statistically significant (P<0.01) prolongation of the R time and a decrease in angle and MA was seen in all dogs by hour 3. R and MA were unmeasurable for most dogs between 3 and 5 hours. All TEG tracings returned to baseline by 12 hours. Day 4 TEG tracings mimicked those on Day 1. Only 1 dog achieved aPTT values outside the reference interval on both days. Anti-Xa activity levels increased on Day 4 but not on Day 1. Based on post hoc in vitro analysis, prolongation of R time occurred at plasma heparin levels as low as 0.075 U/mL, well below the lower limit of detection of the anti-Xa activity level assay. CONCLUSIONS: Administration of SQ heparin results in progressive changes in the TEG tracing, with maximal change occurring 3-5 hours after dosing. The extensive prolongation of the R time also indicates that TEG may be too sensitive and limits its utility as a monitoring tool for unfractionated heparin therapy.

Registered nurse job satisfaction and collective bargaining unit membership status.

OBJECTIVE: To examine differences in job satisfaction levels between registered nurses who were or were not members of a nursing collective bargaining unit. BACKGROUND: The nursing shortage could lead to decreased quality of patient care, heavier workloads, and mandatory overtime, resulting in decreased job satisfaction and increased intent to leave. Nursing collective bargaining units use contracts to help decrease patient workload, decrease and eliminate mandatory overtime, increase pay and benefits, and increase job security. Exploring differences in job satisfaction between nurses who are and are not members of collective bargaining units is pertinent to understanding the retention and recruitment of nurses. METHODS: A descriptive secondary analysis using a survey database from the Minnesota Department of Health. The survey, which included a job satisfaction section largely based on the Index of Work Satisfaction, was sent in 2002 to 3,645 registered nurses in Minnesota. RESULTS: Members of collective bargaining units had higher satisfaction with wages. Nonmembers had higher satisfaction with nursing supervision, patient care, work setting, professional relationships, and overall job satisfaction. CONCLUSIONS: There is a need for interventions in institutions with collective bargaining units to improve job satisfaction, nurse retention, and job recruitment.

Information Conversion, Effective Samples, and Parameter Size.

Consider the relative entropy between a posterior density for a parameter given a sample and a second posterior density for the same parameter, based on a different model and a different data set. Then the relative entropy can be minimized over the second sample to get a virtual sample that would make the second posterior as close as possible to the first in an informational sense. If the first posterior is based on a dependent dataset and the second posterior uses an independence model, the effective inferential power of the dependent sample is transferred into the independent sample by the optimization. Examples of this optimization are presented for models with nuisance parameters, finite mixture models, and models for correlated data. Our approach is also used to choose the effective parameter size in a Bayesian hierarchical model.

Genomic prediction of locoregional recurrence after mastectomy in breast cancer.

PURPOSE: This study aims to explore gene expression profiles that are associated with locoregional (LR) recurrence in breast cancer after mastectomy. PATIENTS AND METHODS: A total of 94 breast cancer patients who underwent mastectomy between 1990 and 2001 and had DNA microarray study on the primary tumor tissues were chosen for this study. Eligible patient should have no evidence of LR recurrence without postmastectomy radiotherapy (PMRT) after a minimum of 3-year follow-up (n = 67) and any LR recurrence (n = 27). They were randomly split into training and validation sets. Statistical classification tree analysis and proportional hazards models were developed to identify and validate gene expression profiles that relate to LR recurrence. RESULTS: Our study demonstrates two sets of gene expression profiles (one with 258 genes and the other 34 genes) to be of predictive value with respect to LR recurrence. The overall accuracy of the prediction tree model in validation sets is estimated 75% to 78%. Of patients in validation data set, the 3-year LR control rate with predictive index more than 0.8 derived from 34-gene prediction models is 91%, and predictive index 0.8 or less is 40% (P = .008). Multivariate analysis of all patients reveals that estrogen receptor and genomic predictive index are independent prognostic factors that affect LR control. CONCLUSION: Using gene expression profiles to develop prediction tree models effectively identifies breast cancer patients who are at higher risk for LR recurrence. This gene expression-based predictive index can be used to select patients for PMRT.

Patterns of gene expression that characterize long-term survival in advanced stage serous ovarian cancers.

PURPOSE: A better understanding of the underlying biology of invasive serous ovarian cancer is critical for the development of early detection strategies and new therapeutics. The objective of this study was to define gene expression patterns associated with favorable survival. EXPERIMENTAL DESIGN: RNA from 65 serous ovarian cancers was analyzed using Affymetrix U133A microarrays. This included 54 stage III/IV cases (30 short-term survivors who lived <3 years and 24 long-term survivors who lived >7 years) and 11 stage I/II cases. Genes were screened on the basis of their level of and variability in expression, leaving 7,821 for use in developing a predictive model for survival. A composite predictive model was developed that combines Bayesian classification tree and multivariate discriminant models. Leave-one-out cross-validation was used to select and evaluate models. RESULTS: Patterns of genes were identified that distinguish short-term and long-term ovarian cancer survivors. The expression model developed for advanced stage disease classified all 11 early-stage ovarian cancers as long-term survivors. The MAL gene, which has been shown to confer resistance to cancer therapy, was most highly overexpressed in short-term survivors (3-fold compared with long-term survivors, and 29-fold compared with early-stage cases). These results suggest that gene expression patterns underlie differences in outcome, and an examination of the genes that provide this discrimination reveals that many are implicated in processes that define the malignant phenotype. CONCLUSIONS: Differences in survival of advanced ovarian cancers are reflected by distinct patterns of gene expression. This biological distinction is further emphasized by the finding that early-stage cancers share expression patterns with the advanced stage long-term survivors, suggesting a shared favorable biology.

High insulin-like growth factor-2 (IGF-2) gene expression is an independent predictor of poor survival for patients with advanced stage serous epithelial ovarian cancer.

OBJECTIVE: Epithelial ovarian cancer is the deadliest gynecologic malignancy, yet its molecular etiology remains poorly understood. Evidence is accumulating to support a role for the insulin-like growth factor family in human carcinogenesis, and recently using microarray expression analysis, we demonstrated over-expression of the insulin-like growth factor-2 (IGF-2) gene in advanced stage epithelial ovarian cancers. The purpose of the current study is to further elucidate the role of the IGF-2 gene in ovarian cancer development and progression. METHODS: Relative expression of IGF-2 was measured in 109 epithelial ovarian cancers and eight normal ovarian surface epithelial (NOSE) samples, using quantitative real-time polymerase chain reaction. Associations with clinicopathological parameters were examined. RESULTS: Expression of the IGF-2 gene was more than 300-fold higher in ovarian cancers compared with normal ovarian surface epithelium samples (P <0.001). High IGF-2 expression was associated with advanced stage disease at diagnosis (P <0.001), high-grade cancers (P <0.05) and sub-optimal surgical cytoreduction (P = 0.08). In multivariate analysis, relative IGF-2 expression was an independent predictor of poor survival. CONCLUSIONS: Expression of the IGF-2 gene is significantly higher in ovarian cancers relative to normal ovarian surface epithelium. Further, high IGF-2 gene expression is associated with high grade, advanced stage disease, and is an independent predictor of poor survival in patients with epithelial ovarian cancer. As such, IGF-2 is a molecular marker and potential therapeutic target for the most aggressive epithelial ovarian cancers.

Bayesian analysis of binary prediction tree models for retrospectively sampled outcomes.

Classification tree models are flexible analysis tools which have the ability to evaluate interactions among predictors as well as generate predictions for responses of interest. We describe Bayesian analysis of a specific class of tree models in which binary response data arise from a retrospective case-control design. We are also particularly interested in problems with potentially very many candidate predictors. This scenario is common in studies concerning gene expression data, which is a key motivating example context. Innovations here include the introduction of tree models that explicitly address and incorporate the retrospective design, and the use of nonparametric Bayesian models involving Dirichlet process priors on the distributions of predictor variables. The model specification influences the generation of trees through Bayes' factor based tests of association that determine significant binary partitions of nodes during a process of forward generation of trees. We describe this constructive process and discuss questions of generating and combining multiple trees via Bayesian model averaging for prediction. Additional discussion of parameter selection and sensitivity is given in the context of an example which concerns prediction of breast tumour status utilizing high-dimensional gene expression data; the example demonstrates the exploratory/explanatory uses of such models as well as their primary utility in prediction. Shortcomings of the approach and comparison with alternative tree modelling algorithms are also discussed, as are issues of modelling and computational extensions.

Gene expression phenotypes of atherosclerosis.

OBJECTIVE: Fulfilling the promise of personalized medicine by developing individualized diagnostic and therapeutic strategies for atherosclerosis will depend on a detailed understanding of the genes and gene variants that contribute to disease susceptibility and progression. To that end, our group has developed a nonbiased approach congruent with the multigenic concept of complex diseases by identifying gene expression patterns highly associated with disease states in human target tissues. METHODS AND RESULTS: We have analyzed a collection of human aorta samples with varying degrees of atherosclerosis to identify gene expression patterns that predict a disease state or potential susceptibility. We find gene expression signatures that relate to each of these disease measures and are reliable and robust in predicting the classification for new samples with >93% in each analysis. The genes that provide the predictive power include many previously suspected to play a role in atherosclerosis and additional genes without prior association with atherosclerosis. CONCLUSIONS: Hence, we are reporting a novel method for generating a molecular phenotype of disease and then identifying genes whose discriminatory capability strongly implicates their potential roles in human atherosclerosis.

Integrated modeling of clinical and gene expression information for personalized prediction of disease outcomes.

We describe a comprehensive modeling approach to combining genomic and clinical data for personalized prediction in disease outcome studies. This integrated clinicogenomic modeling framework is based on statistical classification tree models that evaluate the contributions of multiple forms of data, both clinical and genomic, to define interactions of multiple risk factors that associate with the clinical outcome and derive predictions customized to the individual patient level. Gene expression data from DNA microarrays is represented by multiple, summary measures that we term metagenes; each metagene characterizes the dominant common expression pattern within a cluster of genes. A case study of primary breast cancer recurrence demonstrates that models using multiple metagenes combined with traditional clinical risk factors improve prediction accuracy at the individual patient level, delivering predictions more accurate than those made by using a single genomic predictor or clinical data alone. The analysis also highlights issues of communicating uncertainty in prediction and identifies combinations of clinical and genomic risk factors playing predictive roles. Implicated metagenes identify gene subsets with the potential to aid biological interpretation. This framework will extend to incorporate any form of data, including emerging forms of genomic data, and provides a platform for development of models for personalized prognosis.

Towards integrated clinico-genomic models for personalized medicine: combining gene expression signatures and clinical factors in breast cancer outcomes prediction.

Genomic data, particularly genome-scale measures of gene expression derived from DNA microarray studies, has the potential for adding enormous information to the analysis of biological phenotypes. Perhaps the most successful application of this data has been in the characterization of human cancers, including the ability to predict clinical outcomes. Nevertheless, most analyses have used gene expression profiles to define broad group distinctions, similar to the use of traditional clinical risk factors. As a result, there remains considerable heterogeneity within the broadly defined groups and thus predictions fall short of providing accurate predictions for individual patients. One strategy to resolve this heterogeneity is to make use of multiple gene expression patterns that are more powerful in defining individual characteristics and predicting outcomes than any single gene expression pattern. Statistical tree-based classification systems provide a framework for assessing multiple patterns, that we term metagenes, selecting those that are most capable of resolving the biological heterogeneity. Moreover, this framework provides a mechanism to combine multiple forms of data, both genomic and clinical, to most effectively characterize individual patients and achieve the goal of personalized predictions of clinical outcomes.