Effects of informative censoring on the hazard function: Application to the proportional hazards regression model

The standard analyses of survival data involve the assumption that survival and censoring are independent. When censoring and survival are related, the phenomenon is known as informative censoring. This paper examines the effects of an informative censoring assumption on the hazard function and the estimated hazard ratio provided by the Cox model.^ The limiting factor in all analyses of informative censoring is the problem of non-identifiability. Non-identifiability implies that it is impossible to distinguish a situation in which censoring and death are independent from one in which there is dependence. However, it is possible that informative censoring occurs. Examination of the literature indicates how others have approached the problem and covers the relevant theoretical background.^ Three models are examined in detail. The first model uses conditionally independent marginal hazards to obtain the unconditional survival function and hazards. The second model is based on the Gumbel Type A method for combining independent marginal distributions into bivariate distributions using a dependency parameter. Finally, a formulation based on a compartmental model is presented and its results described. For the latter two approaches, the resulting hazard is used in the Cox model in a simulation study.^ The unconditional survival distribution formed from the first model involves dependency, but the crude hazard resulting from this unconditional distribution is identical to the marginal hazard, and inferences based on the hazard are valid. The hazard ratios formed from two distributions following the Gumbel Type A model are biased by a factor dependent on the amount of censoring in the two populations and the strength of the dependency of death and censoring in the two populations. The Cox model estimates this biased hazard ratio. In general, the hazard resulting from the compartmental model is not constant, even if the individual marginal hazards are constant, unless censoring is non-informative. The hazard ratio tends to a specific limit.^ Methods of evaluating situations in which informative censoring is present are described, and the relative utility of the three models examined is discussed. ^

Survival analysis of longevity in siblings

Do siblings of centenarians tend to have longer life spans? To answer this question, life spans of 184 siblings for 42 centenarians have been evaluated. Two important questions have been addressed in analyzing the sibling data. First, a standard needs to be established, to which the life spans of 184 siblings are compared. In this report, an external reference population is constructed from the U.S. life tables. Its estimated mortality rates are treated as baseline hazards from which the relative mortality of the siblings are estimated. Second, the standard survival models which assume independent observations are invalid when correlation within family exists, underestimating the true variance. Methods that allow correlations are illustrated by three different methods. First, the cumulative relative excess mortality between siblings and their comparison group is calculated and used as an effective graphic tool, along with the Product Limit estimator of the survival function. The variance estimator of the cumulative relative excess mortality is adjusted for the potential within family correlation using Taylor linearization approach. Second, approaches that adjust for the inflated variance are examined. They are adjusted one-sample log-rank test using design effect originally proposed by Rao and Scott in the correlated binomial or Poisson distribution setting and the robust variance estimator derived from the log-likelihood function of a multiplicative model. Nether of these two approaches provide correlation estimate within families, but the comparison with the comparison with the standard remains valid under dependence. Last, using the frailty model concept, the multiplicative model, where the baseline hazards are known, is extended by adding a random frailty term that is based on the positive stable or the gamma distribution. Comparisons between the two frailty distributions are performed by simulation. Based on the results from various approaches, it is concluded that the siblings of centenarians had significant lower mortality rates as compared to their cohorts. The frailty models also indicate significant correlations between the life spans of the siblings. ^

Demographic analysis of Certified Industrial Hygienists (CIH) with the use of survival analysis techniques

Objective. The goal of this study is to characterize the current workforce of CIHs, the lengths of professional practice careers of the past and current CIHs.^ Methods. This is a secondary data analysis of data compiled from all of the nearly 50 annual roster listings of the American Board of Industrial Hygiene (ABIH) for Certified Industrial Hygienists active in each year since 1960. Survival analysis was performed as a technique to measure the primary outcome of interest. The technique which was involved in this study was the Kaplan-Meier method for estimating the survival function.^ Study subjects: The population to be studied is all Certified Industrial Hygienists (CIHs). A CIH is defined by the ABIH as an individual who has achieved the minimum requirements for education, working experience and through examination, has demonstrated a minimum level of knowledge and competency in the prevention of occupational illnesses. ^ Results. A Cox-proportional hazards model analysis was performed by different start-time cohorts of CIHs. In this model we chose cohort 1 as the reference cohort. The estimated relative risk of the event (defined as retirement, or absent from 5 consecutive years of listing) occurred for CIHs for cohorts 2,3,4,5 relative to cohort 1 is 0.385, 0.214, 0.234, 0.299 relatively. The result show that cohort 2 (CIHs issued from 1970-1980) has the lowest hazard ratio which indicates the lowest retirement rate.^ Conclusion. The manpower of CIHs (still actively practicing up to the end of 2009) increased tremendously starting in 1980 and grew into a plateau in recent decades. This indicates that the supply and demand of the profession may have reached equilibrium. More demographic information and variables are needed to actually predict the future number of CIHs needed. ^

Downstream targets of the Rhox5 homeobox gene

The X-linked mouse Rhox gene cluster contains over 30 homeobox genes that are candidates to regulate multiple steps in male and female gametogenesis. The founding member of the Rhox gene cluster, Rhox5, is an androgen-dependent gene expressed in Sertoli cells that promotes the survival and differentiation of the adjacent male germ cells. To decipher downstream signaling pathways of Rhox5, I used in vivo and in vitro microarray profiling to identify and characterize downstream targets of Rhox5 in the testis. This led to the identification of many Rhox5 -regulated genes, two of which I focused on in more detail. One of them, Unc5c, encodes a pro-apoptotic receptor with tumor suppressor activity that I found is negatively regulated by Rhox5 through a Rhox5-response element in the Unc5c 5' untranslated region (5' UTR). Examination of other mouse Rhox family members revealed that Rhox2 and Rhox3 also have the ability to downregulate Unc5c expression. The human RHOX protein RHOXF2 also had this ability, indicating that Unc5c repression is a conserved Rhox-dependent response. The repression of Unc5c expression by Rhox5 may, in part, mediate Rhox5's pro-survival function in the testis, as I found that Unc5c mutant mice have decreased germ cell apoptosis in the testis. This along with my other data leads me to propose a model in which Rhox5 is a negative regulator upstream of Unc5c in a Sertoli-cell pathway that promotes germ-cell survival. The other Rhox5-regulated gene that I studied in detail is insulin II (Ins2). Several lines of evidence, including electrophoretic mobility shift anaylsis, promoter mutagenesis, and chromatin immuoprecipitation analysis indicated that Ins2 is a direct target of Rhox5. Structure-function analysis identified homeodomain residues and the RHOX5 amino-terminal domain crucial for conferring Ins2 inducibility. Rhox5 regulates not only the Ins2 gene but also genes encoding other secreted proteins regulating metabolism (adiponectin and resistin), the rate-liming enzyme for monosaturated fatty acid biosynthesis (SCD-1), and transcription factors crucial for regulating metabolism (the nuclear hormone receptor PPARγ). I propose that the regulation of some or all of these molecules in Sertoli cells is responsible for the Rhox5-dependent survival of the adjacent germ cells. ^

NOVEL THERAPEUTIC TARGETS IDENTIFIED BY HIGH-THROUGHPUT TECHNOLOGIES FOR TRIPLE-NEGATIVE BREAST CANCER

Triple-negative breast cancers (TNBC) are characterized by the lack of or reduced expression of the estrogen and progesterone receptors, and normal expression of the human epidermal growth factor receptor 2. The lack of a well-characterized target for treatment leaves only systemic chemotherapy as the mainstay of treatment. Approximately 60-70% of patients are chemosensitive, while the remaining majority does not respond. Targeted therapies that take advantage of the unique molecular perturbations found in triple-negative breast cancer are needed. The genes that are frequently amplified or overexpressed represent potential therapeutic targets for triple-negative breast cancer. The purpose of this study was to identify and validate novel therapeutic targets for triple-negative breast cancers. 681 genes showed consistent and highly significant overexpression in TNBC compared to receptor-positive cancers in 2 data sets. For two genes, 3 of the 4 siRNAs showed preferential growth inhibition in TNBC cells. These two genes were the low density lipoprotein receptor-related protein 8 (LRP8) and very low-density lipoprotein receptor (VLDLR). Exposure to their cognate ligands, reelin and apolipoprotein E isoform 4 (ApoE4), stimulated the growth of TNBC cells in vitro. Suppression of the expression of either LRP8 or VLDLR or exposure to RAP (an inhibitor of ligand binding to LRP8 and VLDLR) abolished this ligand-induced proliferation. High-throughput protein and metabolic arrays revealed that ApoE4 stimulation rescued TNBC cells from serum-starvation induced up-regulation of genes involved in lipid biosynthesis, increased protein expression of oncogenes involved in the MAPK/ERK and DNA repair pathways, and reduced the serum-starvation induction of biochemicals involved in oxidative stress response and glycolytic metabolism. shLRP8 MDA-MB-231 xenografts had reduced tumor volume, in comparison to parental and shCON xenografts. These results indicate that LRP8-APOE signaling confers survival advantages to TNBC tumors under reduced nutrient conditions and during cellular environmental stress. We revealed that the LRP8-APOE receptor-ligand system is overexpressed in human TNBC. We also demonstrated that this receptor system mediates a strong growth promoting and survival function in TNBC cells in vitro and helps to sustain the growth of MDA-MD-231 xenografts. We propose that inhibitors of LRP8-APOE signaling may be clinically useful therapeutic agents for triple-negative breast cancer.

Analyzing left-truncated right-censored data with uncertain onset time with parametric models

Prevalent sampling is an efficient and focused approach to the study of the natural history of disease. Right-censored time-to-event data observed from prospective prevalent cohort studies are often subject to left-truncated sampling. Left-truncated samples are not randomly selected from the population of interest and have a selection bias. Extensive studies have focused on estimating the unbiased distribution given left-truncated samples. However, in many applications, the exact date of disease onset was not observed. For example, in an HIV infection study, the exact HIV infection time is not observable. However, it is known that the HIV infection date occurred between two observable dates. Meeting these challenges motivated our study. We propose parametric models to estimate the unbiased distribution of left-truncated, right-censored time-to-event data with uncertain onset times. We first consider data from a length-biased sampling, a specific case in left-truncated samplings. Then we extend the proposed method to general left-truncated sampling. With a parametric model, we construct the full likelihood, given a biased sample with unobservable onset of disease. The parameters are estimated through the maximization of the constructed likelihood by adjusting the selection bias and unobservable exact onset. Simulations are conducted to evaluate the finite sample performance of the proposed methods. We apply the proposed method to an HIV infection study, estimating the unbiased survival function and covariance coefficients. ^

Joint effects of pulmonary function and aerobic power on survival in a cohort of healthy adults

Objective measurements of physical fitness and pulmonary function are related individually to long-term survival, both in healthy people and in those who are ill. These factors are furthermore known to be related to one another physiologically in people with pulmonary disease, because advanced pulmonary disease causes ventilatory limitation to exercise. Healthy people do not have ventilatory limitation to exercise, but rather have ventilatory reserve. The relationship between pulmonary function and exercise performance in healthy people is minimal. Exercise performance has been shown to modify the effect of pulmonary function on mortality in people with chronic obstructive pulmonary disease, but the relationship between these factors in healthy people has not been studied and is not known. The purpose of this study is to quantify the joint effects of pulmonary function and exercise performance as these bear on mortality in a cohort of healthy adults. This investigation is an historical cohort study over 20 years of follow-up of 29,624 adults who had complete preventive medicine, spirometry and treadmill stress examinations at the Cooper Clinic in Dallas, Texas.^ In 20 years of follow-up, there were 738 evaluable deaths. Forced expiratory volume in one second (FEV$\sb1$) percent of predicted, treadmill time in minutes percent of predicted, age, gender, body mass index, baseline smoking status, serum glucose and serum total cholesterol were all significant, independent predictors of mortality risk. There were no frank interactions, although age had an important increasing effect on the risk associated with smoking when other covariates were controlled for in a proportional-hazards model. There was no confounding effect of exercise performance on pulmonary function. In agreement with the pertinent literature on independent effects, each unit increase in FEV$\sb1$ percent predicted was associated with about eight tenths of a percent reduction in adjusted mortality rate. The concept of physiologic reserve is useful in interpretation of the findings. Since pulmonary function does not limit exercise tolerance in healthy adults, it is reasonable to expect that exercise tolerance would not modify the effect of pulmonary function on mortality. Epidemiologic techniques are useful for elucidating physiological correlates of mortality risk. ^

Constitutive NF-kappaB and NFAT activation leads to stimulation of the BLyS survival pathway in aggressive B-cell lymphomas.

B-lymphocyte stimulator (BLyS), a relatively recently recognized member of the tumor necrosis factor ligand family (TNF), is a potent cell-survival factor expressed in many hematopoietic cells. BLyS binds to 3 TNF-R receptors, TACI, BCMA, BAFF-R, to regulate B-cell survival, differentiation, and proliferation. The mechanisms involved in BLYS gene expression and regulation are still incompletely understood. In this study, we examined BLYS gene expression, function, and regulation in B-cell non-Hodgkin lymphoma (NHL-B) cells. Our studies indicate that BLyS is constitutively expressed in aggressive NHL-B cells, including large B-cell lymphoma (LBCL) and mantle cell lymphoma (MCL), playing an important role in the survival and proliferation of malignant B cells. We found that 2 important transcription factors, NF-kappaB and NFAT, are involved in regulating BLyS expression through at least one NF-kappaB and 2 NFAT binding sites in the BLYS promoter. We also provide evidence suggesting that the constitutive activation of NF-kappaB and BLyS in NHL-B cells forms a positive feedback loop associated with lymphoma cell survival and proliferation. Our findings indicate that constitutive NF-kappaB and NFAT activations are crucial transcriptional regulators of the BLyS survival pathway in malignant B cells that could be therapeutic targets in aggressive NHL-B.

Hepatocellular carcinoma survival in uninsured and underinsured patients.

BACKGROUND: The incidence of hepatitis C virus (HCV) and hepatocellular carcinoma (HCC) is increasing. The purpose of this study is to establish baseline survival in a medically-underserved population and to evaluate the effect of HCV seropositivity on our patient population. MATERIALS AND METHODS: We reviewed clinicopathologic parameters from a prospective tumor registry and medical records from the Harris County Hospital District (HCHD). Outcomes were compared using Kaplan-Meier survival analysis and log-rank tests. RESULTS: A total of 298 HCC patients were identified. The median survival for the entire cohort was 3.4 mo. There was no difference in survival between the HCV seropositive and the HCV seronegative groups (3.6 mo versus 2.6 mo, P = 0.7). Patients with a survival <1 mo had a significant increase in>αfetoprotein (AFP), international normalized ratio (INR), model for end-stage liver disease (MELD) score, and total bilirubin and decrease in albumin compared with patients with a survival ≥ 1 mo. CONCLUSIONS: Survival for HCC patients in the HCHD is extremely poor compared with an anticipated median survival of 7 mo reported in other studies. HCV seropositive patients have no survival advantage over HCV seronegative patients. Poorer liver function at diagnosis appears to be related to shorter survival. Further analysis into variables contributing to decreased survival is needed.

FUNCTION OF ZNF668 IN CANCER DEVELOPMENT

Human cancer develops as a result of accumulation of mutations in oncogenes and tumor suppressor genes. Zinc finger protein 668 (ZNF668) has recently been identified and validated as one of the highly mutated genes in breast cancer, but its function is entirely unknown. Here, we report two major functions of ZNF668 in cancer development. (1) ZNF668 functions as a tumor suppressor by regulating p53 protein stability and function. We demonstrate that ZNF668 is a nucleolar protein that physically interacts with both MDM2 and p53. By binding to MDM2, ZNF668 regulates MDM2 autoubiquitination and prevents MDM2-mediated p53 ubiquitination and degradation; ZNF668 deficiency impairs DNA damage-induced p53 stabilization. Notably, ZNF668 effectively suppresses breast cancer cell proliferation and transformation in vitro and tumorigenicity in vivo. Consistently, ZNF668 knockdown readily transforms normal mammary epithelial cells. Together, our studies identify ZNF668 as a novel breast tumor suppressor gene that acts at least in part by regulating the stability and function of p53. (2) ZNF668 functions as a DNA repair protein by regulating histone acetylation. DNA repair proteins need to access the chromatin by chromatin modification or remodeling to use DNA template within chromatin. Dynamic posttranslational modifications of histones are critical for cells to relax chromatin in DNA repair. However, the precise underlying mechanism mediating enzymes responsible for these modifications and their recruitment to DNA lesions remains poorly understood. We observed ZNF668 depletion causes impaired chromatin relaxation as a result of impaired DNA-damage induced histone H2AX hyper-acetylation. This results in the decreased recruitment of repair proteins to DNA lesions, defective homologous recombination (HR) repair and impaired cell survival after DNA damage, albeit with the presence of a functional ATM/ATR dependent DNA-damage signaling cascade. Importantly, the impaired loading of repair proteins and the defect in DNA repair in ZNF668-deficient cells can be counteracted by chromatin relaxation, indicating that the DNA-repair defect that was observed in the absence of ZNF668 is due to impeded chromatin accessibility at sites of DNA breaks. Our findings therefore identify ZNF668 as a key molecule that links chromatin relaxation with response to DNA damage in the control of DNA repair.

The relationship between rheumatoid arthritis and reproductive function

There is growing support for the theory that an interaction between the immune and reproductive/endocrine systems underlies the pathogenesis of autoimmune rheumatic diseases. Most of the recent evidence derives from studies of sex hormones and pregnancy in women with systemic lupus. Other than an ameliorative effect of pregnancy, little is known about reproductive factors in relation to rheumatoid arthritis. To elucidate the relationship, a population-based retrospective study was undertaken. Included were 378 female residents of Olmsted County, Minnesota diagnosed with rheumatoid arthritis between 1950 and 1982 (cases) and 325 arthritis-free, married female controls matched to the 324 married cases on birth-year, age at first marriage, and duration of Olmsted County residency. Information of reproductive factors was extracted from the medical records system maintained by the Mayo Clinic.^ Cases had lower fertility rates compared with the female population of Minnesota (rate ratio = 0.86, 95% confidence interval (CI)= 0.80-0.92). Fertility was significantly reduced even prior to the onset of rheumatoid factor positive arthritis. Restricting the comparison to married Olmsted County residents did not alter the results. Further adjustments for time not at risk of conception using survival analysis and proportional hazards modeling only intensified the fertility reduction in the married cases compared with controls. Nulligravidity was more common among cases than controls (odds ratio = 3.16, CI = 1.61-6.20). Independent of fertility, pregnancy had a protective effect against rheumatoid arthritis (odds ratio = 0.31, CI = 0.11-0.89), which was dramatically reversed in the 12 months postpartum (odds ratio = 4.67, CI = 1.50-14.47). Cases were younger at menopause than controls (p $<$ 0.01).^ Small but statistically insignificant associations were observed between rheumatoid arthritis and the following factors: increased frequency of complaints to a physician of infertility; increased frequency of spontaneous abortion, premature birth, and congenital malformations following arthritis onset; and increased prevalence of menopause at arthritis onset. Cases did not differ from controls on age at menarche, duration of pregnancy, or birth weight.^ The findings provide further support for the involvement of the reproductive/endocrine systems in the pathogenesis of autoimmune rheumatic disease. The search for biological mechanisms should be intensified. ^

p75 receptor-mediated neurotrophism: A new mechanism of melanoma cell survival and invasion

Trophism as a "clonal dominance" support mechanism for tumor cells is an unexplored area of tumor progression. This report presents evidence that the human melanoma low-affinity neurotrophin receptor (p75) can signal independently of its high-affinity tyrosine kinase counterparts, the TRK family of kinases. Signaling may be accomplished by a p75-associated purine-analog-sensitive kinase and results in enhanced invasion into a reconstituted basement membrane with a corresponding stimulation of matrix metalloproteinase-2 expression. Additionally, a "stress culture" survival assay was developed to mimic the growth limiting conditions encountered by melanoma cells in a rapidly growing primary tumor or metastatic deposit prior to neoangiogenesis. Under these conditions, p75, promotes the survival of high p75 expressing brain-colonizing melanoma cells. Extensive 70W melanoma cell-cell contact, which downregulates p75, immediately precedes the induction of cell death associated with diminished production of two key cell survival factors, bcl-2 and the p85 subunit of phosphoinositol-3-kinase, and an elevation in apoptosis promoting intracellular reactive oxygen species (ROSs). Since one function of bcl-2 may be to control the generation of ROSs via the antioxidant pathway, these cells may receive a apoptosis-prompting "double hit". 70W melanoma cell death occurred by an apoptotic mechanism displaying classical morphological changes including plasma membrane blebbing, loss of microvilli and redistribution of ribosomes. 70W apoptosis could be pharmacologically triggered following anti-p75 monoclonal antibody-mediated clustering of p75 receptors. 70W cells fluorescently sorted for high-p75 expression (p75$\sp{\rm H}$ cells) exhibited an augmented survival potential and a predilection to sort with the S + G2/M growth phase, relative to their low p75 expressing, p75$\sp{\rm L}$ counterparts. Apoptosis is significantly delayed by p75$\sp{\rm H}$ cells, whereas p75$\sp{\rm L}$ cells are exquisitely prone to initiate apoptosis. Importantly, the p75$\sp{\rm L}$ cells that survive apoptosis, highly re-expressed p75 and were remarkably responsive to exogenous NGF.^ These are the first data to implicate p75-mediated neurotrophism as an invasion and survival support mechanism employed by brain-metastatic cells. In particular, these results may have implications in little understood phenomena of tumor progression, such as the emergence of "clonal dominance" and tumor dormancy. ^

Molecular mechanisms of signal transducer and activator of transcription (STAT) protein function in hematopoiesis

Hematopoietic growth factors play important roles in regulating blood cell growth and development in vivo. In this work, we investigated the signaling mechanisms of two growth factors with clinical significance, erythropoietin (Epo) and granulocyte colony-stimulating factor (G-CSF). Epo is essential for the survival, proliferation and differentiation of red blood cell progenitors, while G-CSF plays an important role in controlling mature neutrophil production. To identify which amino acid(s) and/or motif in EpoR is responsible for cell survival, wild type or mutant EpoR isoforms were transfected into the growth factor-dependent 32D cell line. Proliferation and apoptosis assays demonstrated that an EpoR isoform that lacks intracellular tyrosine residues and is truncated after 321 amino acids in the cytoplasmic tail (EpoR 1-321) mediates Epo-dependent cell survival. Furthermore, in absence of fetal calf serum (FCS), Epo signaling through wild type or mutant receptors supported anti-apoptosis, but not proliferation during 72 hours in response to Epo. To investigate the signaling pathway by which EpoR regulates cell survival, a dominant negative Stat5b (dnStat5b) isoform was generated and coexpressed with EpoR in stable cell lines. Expression of dnStat5b causes a significant induction of apoptosis in the presence of Epo in cells expressing EpoR 1-321, indicating that Stat5 is essential for survival signaling through tyrosine independent sequences in the EpoR. In a second project to investigate G-CSF signaling, we studied mechanisms by which G-CSF regulates the expression of PU.1, an important transcription factor in myeloid and B cell development. We demonstrated, by immunoblot and real time RT-PCR, that PU.1 is induced by G-CSF ex vivo as well as in vivo. To test whether G-CSF signaling through Stat3 is required for PU.1 regulation, the upstream region of the PU.1 gene was analyzed for potential Stat3 binding motifs. Four potential sites were identified; chromatin immunoprecipitations demonstrated that G-CSF activated Stat3 binds to 3 of the 4 binding motifs. In addition, PU.1 induction by G-CSF was completely abrogated in bone marrow from hematopoietic conditional Stat3 knockout mice. These results indicate an important role for Stat3 in G-CSF-dependent PU.1 gene regulation. Collectively, our works demonstrate that Stat protein play important and diverse roles in hematopoietic growth factor signaling. ^

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. ^

Hypoxia-induced transcriptional regulation through chromatin modifications and NC2 function

In response to tumor hypoxia, specific genes that promote angiogenesis, proliferation, and survival are induced. Globally, I find that hypoxia induces a mixed pattern of histone modifications that are typically associated with either transcriptional activation or repression. Furthermore, I find that selective activation of hypoxia-inducible genes occurs simultaneously with widespread repression of transcription. I analyzed histone modifications at the core promoters of hypoxia-repressed and -activated genes and find that distinct patterns of histone modifications correlate with transcriptional activity. Additionally, I discovered that trimethylated H3-K4, a modification generally associated with transcriptional activation, is induced at both hypoxia-activated and repressed genes, suggesting a novel pattern of histone modifications induced during hypoxia. ^ In order to determine the mechanism of hypoxia-induced widespread repression of transcription, I focused my studies on negative cofactor 2 (NC2). Previously, we found that hypoxia-induced repression of the alpha-fetoprotein (AFP) gene occurs during preinitiation complex (PIC) assembly, and I find that NC2, an inhibitor of PIC assembly, is induced during hypoxia. Moreover, I find that the beta subunit of NC2 is essential for hypoxia-mediated repression of AFP, as well as the widespread repression of transcription observed during hypoxia. Previous data in Drosophila and S. cerevisiae indicate that NC2 functions as either an activator or a repressor of transcription. The mechanism of NC2-mediated activation remains unclear; although, Drosophila NC2 function correlates with specific core promoter elements. I tested if NC2 activates transcription in mammalian cells using this core promoter-specific model as a guide. Utilizing site-specific mutagenesis, I find that NC2 function in mammalian cells is not dependent upon specific core promoter elements; however, I do find that mammalian NC2 does function in a gene-specific manner as either an activator or repressor of transcription during hypoxia. Furthermore, I find that binding of the alpha subunit of NC2 specifically correlates with NC2-mediated transcriptional activation. NC2α and NC2β are both required for NC2-mediated transcriptional activation; whereas, NC2β alone is required for hypoxia-induced transcriptional repression. Together, these data indicate that hypoxia mediates changes in gene expression through both chromatin modifications and NC2 function. ^