Bayesian joint modeling of longitudinal and survival data

The joint modeling of longitudinal and survival data is a new approach to many applications such as HIV, cancer vaccine trials and quality of life studies. There are recent developments of the methodologies with respect to each of the components of the joint model as well as statistical processes that link them together. Among these, second order polynomial random effect models and linear mixed effects models are the most commonly used for the longitudinal trajectory function. In this study, we first relax the parametric constraints for polynomial random effect models by using Dirichlet process priors, then three longitudinal markers rather than only one marker are considered in one joint model. Second, we use a linear mixed effect model for the longitudinal process in a joint model analyzing the three markers. In this research these methods were applied to the Primary Biliary Cirrhosis sequential data, which were collected from a clinical trial of primary biliary cirrhosis (PBC) of the liver. This trial was conducted between 1974 and 1984 at the Mayo Clinic. The effects of three longitudinal markers (1) Total Serum Bilirubin, (2) Serum Albumin and (3) Serum Glutamic-Oxaloacetic transaminase (SGOT) on patients' survival were investigated. Proportion of treatment effect will also be studied using the proposed joint modeling approaches. ^ Based on the results, we conclude that the proposed modeling approaches yield better fit to the data and give less biased parameter estimates for these trajectory functions than previous methods. Model fit is also improved after considering three longitudinal markers instead of one marker only. The results from analysis of proportion of treatment effects from these joint models indicate same conclusion as that from the final model of Fleming and Harrington (1991), which is Bilirubin and Albumin together has stronger impact in predicting patients' survival and as a surrogate endpoints for treatment. ^

A urokinase receptor promoter element (-152/-135) bound with an AP-2-alpha-related protein and SP1 is required for the induction of gene expression by PMA and SRC

The urokinase-type plasminogen activator receptor (u-PAR) promotes extracellular matrix degradation, invasion and metastasis. A first objective of this dissertation was to identify cis-elements and trans-acting factors activating u-PAR gene expression through a previously footprinted (–148/–124) promoter region. Mobility shifting experiments on nuclear extracts of a high u-PAR-expressing colon cancer cell line (RKO) indicated Sp1, Sp3 and a factor similar to, but distinct from, AP-2α bound to an oligonucleotide spanning –152/–135. Mutations preventing the binding of the AP-2α-related factor reduced u-PAR promoter activity. In RKO, the expression of a dominant negative AP-2 (AP-2αB) diminished u-PAR promoter activity, protein and u-PAR mediated laminin degradation. Conversely, u-PAR promoter activity in low u-PAR-expressing GEO cells was increased by AP-2αA expression. PMA treatment, which induces u-PAR expression, caused an increased amount of the AP-2α-related factor-containing complex in GEO, and mutations preventing AP-2α-like and Sp1/Sp3 binding reduced the u-PAR promoter stimulation by PMA. In resected colon cancers, u-PAR protein amounts were related to the amount of the AP-2α-related factor-containing complex. In conclusion, constitutive and PMA- inducible u-PAR gene expression and -proteolysis are mediated partly through transactivation via a promoter sequence (–152/435) bound with an AP-2α-related factor and Sp1/Sp3. ^ A second interest of this dissertation was to determine if a constitutively active Src regulates the transcription of the u-PAR gene, since c-src expression increases invasion in colon cancer. Increased u-PAR protein and laminin degradation paralleling elevated Src activity was evident in SW480 colon cancer cells stably expressing a constitutively active Src (Y- c-src527F). Nuclear run-on experiments indicated that this was due largely to transcriptional activation. While transient transfection of SW480 cells with Y-c-src527F induced a u-PAR-CAT-reporter, mutations preventing Sp1-binding to promoter region –152/435 abolished this induction. Mobility shift assays revealed increased Sp1 binding to region –152/135 with nuclear extracts of Src-transfected SW480 cells. Finally, the amounts of endogenous u-PAR in resected colon cancers significantly correlated with Src-activity. These data suggest that u-PAR gene expression and proteolysis are regulated by Src, this requiring the promoter region (–152/–135) bound with Sp1, thus, demonstrating for the first time that transcription factor Sp1 is a downstream effector of Src. ^

IMMUNOLOGICAL AND CYTOLOGICAL INVESTIGATION OF GAMMA INTERFERON PRODUCTION IN HUMAN T-LYMPHOCYTES INDUCED BY PMA AND PHA (IMMUNOGOLD, CELL CYCLE ANALYSIS)

The present study examined cellular mechanisms involved in the production and secretion of human (gamma)IFN. The hypothesis of this investigation was that (gamma)IFN is an export glycoprotein whose synthesis in human T lymphocytes is dependent on membrane stimulation, polypeptide synthesis in the rough endoplasmic reticulum, packaging in the Golgi complex, and release from the cell by exocytosis.^ The model system for this examination utilized T lymphocytes from normal donors and patients with chronic lymphocytic leukemia (CLL) induced in vitro with the tumor promoter, phorbol 12-myristate 13-acetate (PMA) and the lectin, phytohemagglutinin (PHA) to produce (gamma)IFN. This study reconfirmed the ability of PMA and PHA to synergistically induce (gamma)IFN production in normal T lymphocytes, as measured by viral inhibition assays and radio-immunoassays for (gamma)IFN. The leukemic T cells were demonstrated to produce (gamma)IFN in response to treatment with PHA. PMA treatment also induced (gamma)IFN production in the leukemic T cells, which was much greater than that observed in similarly treated normal T cells. In these same cells, however, combined treatment of the agents was shown to be ineffective at inducing (gamma)IFN production beyond the levels stimulated by the individual agents. In addition, the present study reiterated the synergistic effect of PMA/PHA on the stimulation of growth kinetics in normal T cells. The cell cycle of the leukemic T cells was also responsive to treatment with the agents, particularly with PMA treatment. A number of morphological alterations were attributed to PMA treatment including the acquisition of an elongated configuration, nuclear folds, and large cytoplasmic vacuoles. Many of the effects were observed to be reversible with dilution of the agents, and reversion to this state occurred more rapidly in the leukemic T cells. Most importantly, utilization of a thin section immuno-colloidal gold labelling technique for electron microscopy provided, for the first time, direct evidence of the cellular mechanism of (gamma)IFN production and secretion. The results of this latter study support the idea that (gamma)IFN is produced in the rough endoplasmic reticulum, transferred to the Golgi complex for accumulation and packaging, and released from the T cells by exocytosis. ^

VISUAL MEMORY PROCESSING BY PRIMATES AND AVIANS IN A RELATIONAL TASK

Three rhesus monkeys (Macaca mulatta) and four pigeons (Columba livia) were trained in a visual serial probe recognition (SPR) task. A list of visual stimuli (slides) was presented sequentially to the subjects. Following the list and after a delay interval, a probe stimulus was presented that could be either from the list (Same) or not from the list (Different). The monkeys readily acquired a variable list length SPR task, while pigeons showed acquisition only under constant list length condition. However, monkeys memorized the responses to the probes (absolute strategy) when overtrained with the same lists and probes, while pigeons compared the probe to the list in memory (relational strategy). Performance of the pigeon on 4-items constant list length was disrupted when blocks of trials of different list lengths were imbedded between the 4-items blocks. Serial position curves for recognition at variable probe delays showed better relative performance on the last items of the list at short delays (0-0.5 seconds) and better relative performance on the initial items of the list at long delays (6-10 seconds for the pigeons and 20-30 seconds for the monkeys and a human adolescent). The serial position curves also showed reliable primacy and recency effects at intermediate probe delays. The monkeys showed evidence of using a relational strategy in the variable probe delay task. The results are the first demonstration of relational serial probe recognition performance in an avian and suggest similar underlying dynamic recognition memory mechanisms in primates and avians. ^

REGULATION OF BRAIN NORADRENERGIC-RECEPTOR FUNCTION BY ADRENOCORTICOTROPHIN AND ANTIDEPRESSANT DRUGS (ADRENERGIC RECEPTOR, CYCLIC-AMP, ADENYLATE CYCLASE, IMIPRAMINE, STRESS)

Human behavior appears to be regulated in part by noradrenergic transmission since antidepressant drugs modify the number and function of (beta)-adrenergic receptors in the central nervous system. Affective illness is also known to be associated with the endocrine system, particularly the hypothalamic-pituitary-adrenal axis. The aim of the present study was to determine whether hormones, in particular adrencorticotrophin (ACTH) and corticosterone, may influence behavior by regulating brain noradrenergic receptor function.^ Chronic treatment with ACTH accelerated the increase or decrease in rat brain (beta)-adrenergic receptor number induced by a lesion of the dorsal noradrenergic bundle or treatment with the antidepressant imipramine. Chronic administration of ACTH alone had no effect on (beta)-receptor number although it reduced norepinephrine stimulated cyclic AMP accumulation in brain slices. Treatment with imipramine also reduced the cyclic AMP response to norepinephrine but was accompanied by a decrease in (beta)-adrenergic receptor number. Both the imipramine and ACTH treatments reduced the affinity of (beta)-adrenergic receptors for norepinephrine, but only the antidepressant modified the potency of the neurotransmitter to stimulate second messenger production. Neither ACTH nor imipramine treatment altered Gpp(NH)p- or fluoride-stimulated adenylate cyclase, cyclic AMP, cyclic GMP, or cyclic GMP-stimulated cyclic AMP phosphodiesterase, or the activity of the guanine nucleotide binding protein (Gs). These findings suggested that post-receptor components of the cyclic nucleotide generating system are not influenced by the hormone or antidepressant. This conclusion was verified by the finding that neither treatment altered adenosine-stimulated cyclic AMP accumulation in brain tissue.^ A detailed examination of the (alpha)- and (beta)-adrenergic receptor components of norepinephrine-stimulated cyclic AMP production revealed that ACTH, but not imipramine, administration reduced the contribution of the (alpha)-receptor mediated response. Like ACTH treatment, corticosterone diminished the (alpha)-adrenergic component indicating that adrenal steroids probably mediate the neurochemical responses to ACTH administration. The data indicate that adrenal steroids and antidepressants decrease noradrenergic receptor function by selectively modifying the (alpha)- and (beta)-receptor components. The functional similarity in the action of the steroid and antidepressants suggests that adrenal hormones normally contribute to the maintenance of receptor systems which regulate affective behavior in man. ^

LONGITUDINAL GROWTH PARAMETERS AND THEIR ECOLOGICAL DETERMINANTS (SEASONALITY)

The growth patterns of weight from birth through the first twelve months of life among rural Taiwanese infants were investigated with the following objectives: (i) compare each of the parameters of the Count model estimated for infants who were nutritionally at risk with those for a reference population from the United States; and (ii) within the Taiwanese infants, account for the variance in the growth patterns in the first and second six months of life on the basis of selected ecological factors.^ The significance between group differences were observed in the patterns of the weight growth in both linear growth and in the timing and the direction of velocity changes. A significant decline in growth velocity was observed among Taiwanese infants at about the fourth month of life. The decline is in keeping with a recent proposal made by J. C. Waterlow regarding the timing of change in growth velocity among nutritionally at risk populations in developing countries. The growth course of a nutritionally at risk infant during the first three months is apparently protected by the nurturance of the mother and innate biological properties of the infant.^ A highly significant portion of the growth variance in the second six months of life was accounted for by exogenous factors and biological factors related to the infant. Conversely, none of the growth variance in the first six months of life was accounted for by predictor variables. The most potent determinant of growth in the second six months of life was seasonality which represents a multiple environmental event.^ The model parameters estimated from the Count model represent different aspect of physical growth; yet the correlation coefficients between parameters b and c are high (r > .80). Clearly, the biological interpretation of the model parameters requires analysis of the whole function in the specific context of a given age period. ^

INFANT MORTALITY IN THE MEXICAN AMERICAN COMMUNITY: AN ANALYSIS OF DIFFERENTIALS IN INFANT, NEONATAL, AND POSTNEONATAL MORTALITY BY ETHNICITY AND PARENTAL NATIVITY, 1974-75 SINGLE LIVE BIRTH COHORT, HARRIS COUNTY, TEXAS

The paradoxically low infant mortality rates for Mexican Americans in Texas have been attributed to inaccuracies in vital registration and idiosyncracies in Mexican migration in rural areas along the U.S.-Mexico border. This study examined infant (IMR), neonatal (NMR), and postneonatal (PNMR) mortality rates of Mexican Americans in an urban, non-border setting, using linked birth and death records of the 1974-75 single live birth cohort (N = 68,584) in Harris County, Texas, which includes the city of Houston and is reported to have nearly complete birth and death registration. The use of parental nativity with the traditional Spanish surname criterion made it possible to distinguish infants of Mexican-born immigrants from those of Blacks, Anglos, other Hispanics, and later-generation, more Anglicized Mexican Americans. Mortality rates were analyzed by ethnicity, parental nativity, and cause of death, with respect to birth weight, birth order, maternal age, legitimacy status, and time of first prenatal care.^ While overall IMRs showed Spanish surname rates slightly higher than Anglo rates, infants of Mexican-born immigrants had much lower NMRs than did Anglos, even for moderately low birth weight infants. However, among infants under 1500 grams, presumably unable to be discharged home in the neonatal period, Mexican Americans had the highest NMR. The inconsistency suggested unreported deaths for Mexican American low birth weight infants after hospital discharge. The PNMR of infants of Mexican immigrants was also lower than for Anglos, and the usual mortality differentials were reversed: high-risk categories of high birth order, high maternal age, and late/no prenatal care had the lowest PNMRs. Since these groups' characteristics are congruent with those of low-income migrants, the data suggested the possibility of migration losses. Cause of death analysis suggested that prematurity and birth injuries are greater problems than heretofore recognized among Mexican Americans, and that home births and "shoebox burials" may be unrecorded even in an urban setting.^ Caution is advised in the interpretation of infant mortality rates for a Spanish surname population of Mexican origin, even in an urban, non-border area with reportedly excellent birth and death registration. ^

A cohort study of body fatness and coronary mortality in middle -aged men

Data from the Chicago Western Electric Study were used to investigate whether central fat distribution, as estimated by the ratio of subscapular-to-triceps skinfold, was associated with 25-year risk of death from coronary heart disease in a cohort of 1,945 middle-aged employed men. Subscapular-triceps skinfold ratio was found positively and significantly associated with risk of coronary death after adjustment for age and body mass index. The age-adjusted proportional hazards regression coefficient was 0.2078 with 95% confidence interval of 0.0087 to 0.4069. A difference of 1.1 in the subscapular-triceps skinfold ratio (the difference between the mean of the fifth quintile and of the first and second quintiles combined) was associated with a relative risk of 1.31 with 95% confidence interval of 1.06 to 1.62. The coefficient was decreased to 0.1961 (95% confidence interval of ($-$0.0028 to 0.3950) after adjustment for diastolic blood pressure, serum cholesterol and cigarette smoking as well as age and body mass index. At least some of the effect of central fat on coronary risk is probably mediated by blood pressure and serum lipids, but whether all of the effect can be accounted for blood pressure and serum lipids is uncertain.^ This study supports the concept that central fat distribution is a risk factor for 25-year risk of coronary death in middle-aged men. ^