A nonparametric method of comparing the partial areas under two correlated receiver operating characteristic curves

A non-parametric method was developed and tested to compare the partial areas under two correlated Receiver Operating Characteristic curves. Based on the theory of generalized U-statistics the mathematical formulas have been derived for computing ROC area, and the variance and covariance between the portions of two ROC curves. A practical SAS application also has been developed to facilitate the calculations. The accuracy of the non-parametric method was evaluated by comparing it to other methods. By applying our method to the data from a published ROC analysis of CT image, our results are very close to theirs. A hypothetical example was used to demonstrate the effects of two crossed ROC curves. The two ROC areas are the same. However each portion of the area between two ROC curves were found to be significantly different by the partial ROC curve analysis. For computation of ROC curves with large scales, such as a logistic regression model, we applied our method to the breast cancer study with Medicare claims data. It yielded the same ROC area computation as the SAS Logistic procedure. Our method also provides an alternative to the global summary of ROC area comparison by directly comparing the true-positive rates for two regression models and by determining the range of false-positive values where the models differ. ^

IMMUNOGLOBULINS IN PERIODONTAL DISEASE

Periodontal disease is the major cause of tooth loss in man. The initial histological picture of the inflamed gingiva is characteristic of local inflammatory reaction involving polymorphonuclear leukocytes, vasculitis and localized tissue loss. Subsequent clinical stages of periodontal disease (mild gingivitis) show histological evidence of the involvement of the immune response with initial accumulation of macrophages, and lymphocytes devoid of surface staining immunoglobulins (presumably T cells). As the disease progresses, a predominance of surface and cytoplasmic staining lymphocytes and plasma cells are seen (severe gingivitis and periodontitis). Whether the occurrence of the immunoglobulin positive lymphocytes and the concurrent loss of collagen and resorption of alveolar bone seen in periodontitis is indicative of a direct cause and effect relationship has been a controversy.^ The majority of investigations in the periodontal field have involved the use of peripheral blood lymphocytes or serum. Blastogenic responses of peripheral blood lymphocytes and serum antibody titers from periodontal patients to a variety of oral bacteria have not shown any correlation between response and the severity of disease. The need to study the local immune response in inflamed gingiva is apparent. Since there are no baseline studies on the functional capabilities of the lymphoid cells present in gingiva from periodontitis patients, an in depth study involving the role of the immunoglobulin positive lymphocytes was investigated.^ Inflamed gingiva from four clinically defined periodontal disease states (mild gingivitis, severe gingivitis, periodontitis and severe periodontitis) were placed in gingival organ cultures. Class specific immunoglobulins were quantitated in gingival organ culture supernatants using an indirect sandwich technique. A significant difference in mean levels of IgA and IgG was seen between mild gingivitis and periodontitis (P < .00l, P = .001), as well as in IgG levels between periodontitis and severe periodontitis (P = .001). The predominance of IgG in gingival organ culture supernatants and the statistically significant findings that the overall mean levels of IgG between mild gingivitis and periodontitis (P = .014) and between severe periodontitis and periodontitis (P = .001) suggested a possible indicator of periodontal disease. The presence of IgG in gingival organ culture supernatants was shown to be a product of actively secreting plasma cells. The incorporation of radiolabelled amino acids into IgG was noted over a seven-day period with a peak response at day 4-5. The inhibition of IgG synthesis by cyclohexamide confirmed the contention that IgG was a product of de novo synthesis and not serum derived.^ The specificity of immunoglobulins derived from gingival organ cultures were studied using a whole bacterial agglutination test. Oral bacteria frequently cultured from periodontal patients were assessed for their ability to be agglutinated by gingival organ culture supernatants. A positive correlation of antibody titer and severity of disease was seen with five strains of Actinomyces viscosus, two of Actinomyces naeslundii and one Actinomyces israelii. The agglutination of bacteria was shown to be due to the specific interaction of immunoglobulin and cell-wall antigen. ^

STUDY OF THE EFFECTS OF TOTAL MODULATION TRANSFER FUNCTION CHANGES ON OBSERVER PERFORMANCE USING CLINICAL MAMMOGRAMS

The main goal of this study was to relate physical changes in image quality measured by Modulation Transfer Function (MTF) to diagnostic accuracy.^ One Hundred and Fifty Kodak Min-R screen/film combination conventional craniocaudal mammograms obtained with the Pfizer Microfocus Mammographic system were selected from the files of the Department of Radiology, at M.D. Anderson Hospital and Tumor Institute.^ The mammograms included 88 cases with a variety of benign diagnosis and 62 cases with a variety of malignant biopsy diagnosis. The average age of the patient population was 55 years old. 70 cases presented calcifications with 30 cases having calcifications smaller than 0.5mm. 46 cases presented irregular bordered masses larger than 1 cm. 30 cases presented smooth bordered masses with 20 larger than 1 cm.^ Four separated copies of the original images were made each having a different change in the MTF using a defocusing technique whereby copies of the original were obtained by light exposure through different thicknesses (spacing) of transparent film base.^ The mammograms were randomized, and evaluated by three experienced mammographers for the degree of visibility of various anatomical breast structures and pathological lesions (masses and calicifications), subjective image quality, and mammographic interpretation.^ 3,000 separate evaluations were anayzed by several statistical techniques including Receiver Operating Characteristic curve analysis, McNemar test for differences between proportions and the Landis et al. method of agreement weighted kappa for ordinal categorical data.^ Results from the statistical analysis show: (1) There were no statistical significant differences in the diagnostic accuracy of the observers when diagnosing from mammograms with the same MTF. (2) There were no statistically significant differences in diagnostic accuracy for each observer when diagnosing from mammograms with the different MTF's used in the study. (3) There statistical significant differences in detail visibility between the copies and the originals. Detail visibility was better in the originals. (4) Feature interpretations were not significantly different between the originals and the copies. (5) Perception of image quality did not affect image interpretation.^ Continuation and improvement of this research ca be accomplished by: using a case population more sensitive to MTF changes, i.e., asymptomatic women with minimum breast cancer, more observers (including less experienced radiologists and experienced technologists) must collaborate in the study, and using a minimum of 200 benign and 200 malignant cases.^

A LAPLACE TRANSFORM PAIR MODEL TO DETERMINE BREMSSTRAHLUNG SPECTRA FROM ATTENUATION DATA (RADIATION, X-RAY, SPECTRAL MODELING)

Every x-ray attenuation curve inherently contains all the information necessary to extract the complete energy spectrum of a beam. To date, attempts to obtain accurate spectral information from attenuation data have been inadequate.^ This investigation presents a mathematical pair model, grounded in physical reality by the Laplace Transformation, to describe the attenuation of a photon beam and the corresponding bremsstrahlung spectral distribution. In addition the Laplace model has been mathematically extended to include characteristic radiation in a physically meaningful way. A method to determine the fraction of characteristic radiation in any diagnostic x-ray beam was introduced for use with the extended model.^ This work has examined the reconstructive capability of the Laplace pair model for a photon beam range of from 50 kVp to 25 MV, using both theoretical and experimental methods.^ In the diagnostic region, excellent agreement between a wide variety of experimental spectra and those reconstructed with the Laplace model was obtained when the atomic composition of the attenuators was accurately known. The model successfully reproduced a 2 MV spectrum but demonstrated difficulty in accurately reconstructing orthovoltage and 6 MV spectra. The 25 MV spectrum was successfully reconstructed although poor agreement with the spectrum obtained by Levy was found.^ The analysis of errors, performed with diagnostic energy data, demonstrated the relative insensitivity of the model to typical experimental errors and confirmed that the model can be successfully used to theoretically derive accurate spectral information from experimental attenuation data. ^

STUDIES OF THE MOLECULAR MECHANISMS OF CHROMOSOME ABERRATION FORMATION (DNA REPAIR, CROSSLINKS, MITOMYCIN C)

The objective of this research has been to study the molecular basis for chromosome aberration formation. Predicated on a variety of data, Mitomycin C (MMC)-induced DNA damage has been postulated to cause the formation of chromatid breaks (and gaps) by preventing the replication of regions of the genome prior to mitosis. The basic protocol for these experiments involved treating synchronized Hela cells in G(,1)-phase with a 1 (mu)g/ml dose of MMC for one hour. After removing the drug, cells were then allowed to progress to mitosis and were harvested for analysis by selective detachment. Utilizing the alkaline elution assay for DNA damage, evidence was obtained to support the conclusion that Hela cells can progress through S-phase into mitosis with intact DNA-DNA interstrand crosslinks. A higher level of crosslinking was observed in those cells remaining in interphase compared to those able to reach mitosis at the time of analysis. Dual radioisotope labeling experiments revealed that, at this dose, these crosslinks were associated to the same extent with both parental and newly replicated DNA. This finding was shown not to be the result of a two-step crosslink formation mechanism in which crosslink levels increase with time after drug treatment. It was also shown not to be an artefact of the double-labeling protocol. Using neutral CsCl density gradient ultracentrifugation of mitotic cells containing BrdU-labeled newly replicated DNA, control cells exhibited one major peak at a heavy/light density. However, MMC-treated cells had this same major peak at the heavy/light density, in addition to another minor peak at a density characteristic for light/light DNA. This was interpreted as indicating either: (1) that some parental DNA had not been replicated in the MMC treated sample or; (2) that a recombination repair mechanism was operational. To distinguish between these two possibilities, flow cytometric DNA fluorescence (i.e., DNA content) measurements of MMC-treated and control cells were made. These studies revealed that the mitotic cells that had been treated with MMC while in G(,1)-phase displayed a 10-20% lower DNA content than untreated control cells when measured under conditions that neutralize chromosome condensation effects (i.e., hypotonic treatment). These measurements were made under conditions in which the binding of the drug, MMC, was shown not to interfere with the stoichiometry of the ethidium bromide-mithramycin stain. At the chromosome level, differential staining techniques were used in an attempt to visualize unreplicated regions of the genome, but staining indicative of large unreplicated regions was not observed. These results are best explained by a recombinogenic mechanism. A model consistent with these results has been proposed.^

Childhood cancer survivor satisfaction with characteristics of clinical care

This study investigated the characteristics of a clinic that affect how satisfied survivors of childhood cancer are with their medical care. Questionnaire and interview data from the Passport for Care: Texas Implementation project collected between January 2011 to April 2012 were analyzed. Eleven clinics in Texas participated. Questionnaire respondents were childhood cancer survivor patients who had been off therapy for at least 2 years, or their parents. Interview respondents were clinical providers or research staff at the participating clinics. The outcomes evaluated were answers to a single question on satisfaction with care and a composite Percent Satisfaction Score created from seven other questionnaire items that were correlated (Spearman Rho >0.3) with the question on satisfaction. The following characteristics were also evaluated: sex, age, race, education, and type of cancer. The following clinic indicators were evaluated: type of clinic (general vs. dedicated cancer survivor clinics), number of providers, number of survivors, ratio of survivors/providers, distribution of handouts, distribution of treatment summaries, and use of Children's Oncology Group (COG) guidelines. ^ The only demographic characteristic that affected satisfaction was race. A Kruskal-Wallis test showed a statistically significant difference (Chi-square 6.129, 2 d.f., p = 0.0467). To analyze this further, Wilcoxon Rank Sum test of pairings of the three groups were performed. A Bonferroni correction for multiple testing was applied, with p = 0.017 indicating significance at alpha = 0.05. There was no significant difference between the White and Hispanic groups or between the Hispanic and "Other" groups. For the White and "Other" groups there was a significant difference for the satisfaction item (p = 0.0123) but not for the Percent Satisfaction Score (p = 0.0289). These results suggest that race may influence satisfaction and should be evaluated further in future studies. ^ None of the clinic indicators affected the Percent Satisfaction Score. Going to a clinic that distributed patient information handouts (Wilcoxon Rank Sum p = 0.048) and going to a clinic with >=100 survivors (Wilcoxon Rank Sum p = 0.021) were associated with increased satisfaction. The population of childhood cancer survivors is a growing group of individuals with special health needs. In the future survivors will likely seek medical care in a variety of clinical settings, so it is important to investigate features to improve patient satisfaction with clinical care.^

TIME SERIES ANALYSIS AS INPUT FOR PREDICTIVE MODELING: PREDICTING CARDIAC ARREST IN A PEDIATRIC INTENSIVE CARE UNIT

The first manuscript, entitled "Time-Series Analysis as Input for Clinical Predictive Modeling: Modeling Cardiac Arrest in a Pediatric ICU" lays out the theoretical background for the project. There are several core concepts presented in this paper. First, traditional multivariate models (where each variable is represented by only one value) provide single point-in-time snapshots of patient status: they are incapable of characterizing deterioration. Since deterioration is consistently identified as a precursor to cardiac arrests, we maintain that the traditional multivariate paradigm is insufficient for predicting arrests. We identify time series analysis as a method capable of characterizing deterioration in an objective, mathematical fashion, and describe how to build a general foundation for predictive modeling using time series analysis results as latent variables. Building a solid foundation for any given modeling task involves addressing a number of issues during the design phase. These include selecting the proper candidate features on which to base the model, and selecting the most appropriate tool to measure them. We also identified several unique design issues that are introduced when time series data elements are added to the set of candidate features. One such issue is in defining the duration and resolution of time series elements required to sufficiently characterize the time series phenomena being considered as candidate features for the predictive model. Once the duration and resolution are established, there must also be explicit mathematical or statistical operations that produce the time series analysis result to be used as a latent candidate feature. In synthesizing the comprehensive framework for building a predictive model based on time series data elements, we identified at least four classes of data that can be used in the model design. The first two classes are shared with traditional multivariate models: multivariate data and clinical latent features. Multivariate data is represented by the standard one value per variable paradigm and is widely employed in a host of clinical models and tools. These are often represented by a number present in a given cell of a table. Clinical latent features derived, rather than directly measured, data elements that more accurately represent a particular clinical phenomenon than any of the directly measured data elements in isolation. The second two classes are unique to the time series data elements. The first of these is the raw data elements. These are represented by multiple values per variable, and constitute the measured observations that are typically available to end users when they review time series data. These are often represented as dots on a graph. The final class of data results from performing time series analysis. This class of data represents the fundamental concept on which our hypothesis is based. The specific statistical or mathematical operations are up to the modeler to determine, but we generally recommend that a variety of analyses be performed in order to maximize the likelihood that a representation of the time series data elements is produced that is able to distinguish between two or more classes of outcomes. The second manuscript, entitled "Building Clinical Prediction Models Using Time Series Data: Modeling Cardiac Arrest in a Pediatric ICU" provides a detailed description, start to finish, of the methods required to prepare the data, build, and validate a predictive model that uses the time series data elements determined in the first paper. One of the fundamental tenets of the second paper is that manual implementations of time series based models are unfeasible due to the relatively large number of data elements and the complexity of preprocessing that must occur before data can be presented to the model. Each of the seventeen steps is analyzed from the perspective of how it may be automated, when necessary. We identify the general objectives and available strategies of each of the steps, and we present our rationale for choosing a specific strategy for each step in the case of predicting cardiac arrest in a pediatric intensive care unit. Another issue brought to light by the second paper is that the individual steps required to use time series data for predictive modeling are more numerous and more complex than those used for modeling with traditional multivariate data. Even after complexities attributable to the design phase (addressed in our first paper) have been accounted for, the management and manipulation of the time series elements (the preprocessing steps in particular) are issues that are not present in a traditional multivariate modeling paradigm. In our methods, we present the issues that arise from the time series data elements: defining a reference time; imputing and reducing time series data in order to conform to a predefined structure that was specified during the design phase; and normalizing variable families rather than individual variable instances. The final manuscript, entitled: "Using Time-Series Analysis to Predict Cardiac Arrest in a Pediatric Intensive Care Unit" presents the results that were obtained by applying the theoretical construct and its associated methods (detailed in the first two papers) to the case of cardiac arrest prediction in a pediatric intensive care unit. Our results showed that utilizing the trend analysis from the time series data elements reduced the number of classification errors by 73%. The area under the Receiver Operating Characteristic curve increased from a baseline of 87% to 98% by including the trend analysis. In addition to the performance measures, we were also able to demonstrate that adding raw time series data elements without their associated trend analyses improved classification accuracy as compared to the baseline multivariate model, but diminished classification accuracy as compared to when just the trend analysis features were added (ie, without adding the raw time series data elements). We believe this phenomenon was largely attributable to overfitting, which is known to increase as the ratio of candidate features to class examples rises. Furthermore, although we employed several feature reduction strategies to counteract the overfitting problem, they failed to improve the performance beyond that which was achieved by exclusion of the raw time series elements. Finally, our data demonstrated that pulse oximetry and systolic blood pressure readings tend to start diminishing about 10-20 minutes before an arrest, whereas heart rates tend to diminish rapidly less than 5 minutes before an arrest.

Cyclin D1 mediates epithelial proliferation and links {\it ras\/} activation to the cell cycle

The studies presented in this thesis focus on two aspects of the involvement of cyclin D1 in epithelial proliferation. Since cyclin D1 has been identified as a target for genetic alterations and deregulation in a variety of human cancers, we studied cyclin D1 expression in two experimental models of epithelial carcinogenesis. These studies provided evidence that cyclin D1 was a potential target of the activating mutation of the Ha-ras gene characteristic of the experimental protocol. In addition, evidence from two independent in vitro models suggested that cyclin D1 was indeed part of the primary cellular response to activated ras, and at least partly responsible for the increase in proliferation observed in ras-transformed cells.^ Cyclin D1 has also been described as a key regulator of the passage through the G1 phase of the cell cycle. Cyclin D1 is induced in response to mitogens in a variety of cell lines, and cells engineered to overexpress cyclin D1 show accelerated G1 transit. In order to study the involvement of cyclin D1 in epithelial cell growth and differentiation, we generated transgenic mice that constitutively overexpress cyclin D1 in stratified epithelia. These mice developed thymic hyperplasia and skin hyperproliferation, providing in vivo evidence of the potential of cyclin D1 to regulate growth of epithelial cells. ^

The function of the murine complement *C3a and *C5a anaphylatoxin receptors in endotoxemia, bacteremia and septic shock

The complement system functions as a major effector for both the innate and adaptive immune response. Activation of the complement cascade by either the classical, alternative, or lectin pathway promotes the proteolysis of C3 and C5 thereby generating C3a and C5a. Referred to as anaphylatoxins, the C3a and C5a peptides mediate biological effects upon binding to their respective receptors; C3a binds to the C3a receptor (C3aR) while C5a binds to the C5a receptor (C5aR, CD88). Both C3a and C5a are known for their broad proinflammatory effects. Elevated levels of both peptides have been isolated from patients with a variety of inflammatory diseases such as COPD, asthma, RA, SLE, and sepsis. Recent studies suggest that C5a is a critical component in the acquired neutrophil dysfunction, coagulopathy, and progressive multi-organ dysfunction characteristic of sepsis. The primary hypothesis of this dissertation was that preventing C3a-C3aR and C5a-C5aR mediated pro-inflammatory effects would improve survival in endotoxic, bacteremic and septic shock. To test this hypothesis, the murine C3aR and C5aR genes were disrupted. Following disruption of both the C3aR and C5aR genes, no abnormalities were identified other than the absence of their respective mRNA and protein. In models of both endotoxic and bacteremic shock, C3aR deficient mice suffered increased mortality when compared to their wild type littermates. C3aR deficient mice also had elevated circulating IL-1β levels. Using a model of sepsis, C3aR deficient mice had a higher circulating concentration of IL-6 and decreased peritoneal inflammatory infiltration. While these results were unexpected, they support an emerging role for C3a in immunomodulation. In contrast, following endotoxic or bacteremic shock, C5aR deficient mice experienced increased survival, less hemoconcentration and less thrombocytopenia. It was later determined that C5a mediated histamine release significantly contributes to host morbidity and mortality in bacteremic shock. These studies provide evidence that C5a functions primarily as a proinflammatory molecule in models of endotoxic and bacteremic shock. In the same models, C3a-C3aR interactions suppress the inflammatory response and protect the host. Collectively, these results present in vivo evidence that C3a and C5a have divergent biological functions. ^