Predicting clinical trial results based on announcements of interim analyses.

BACKGROUND: Announcements of interim analyses of a clinical trial convey information about the results beyond the trial's Data Safety Monitoring Board (DSMB). The amount of information conveyed may be minimal, but the fact that none of the trial's stopping boundaries has been crossed implies that the experimental therapy is neither extremely effective nor hopeless. Predicting success of the ongoing trial is of interest to the trial's sponsor, the medical community, pharmaceutical companies, and investors. We determine the probability of trial success by quantifying only the publicly available information from interim analyses of an ongoing trial. We illustrate our method in the context of the National Surgical Adjuvant Breast and Bowel (NSABP) trial, C-08. METHODS: We simulated trials based on the specifics of the NSABP C-08 protocol that were publicly available. We quantified the uncertainty around the treatment effect using prior weights for the various possibilities in light of other colon cancer studies and other studies of the investigational agent, bevacizumab. We considered alternative prior distributions. RESULTS: Subsequent to the trial's third interim analysis, our predictive probabilities were: that the trial would eventually be successful, 48.0%; would stop for futility, 7.4%; and would continue to completion without statistical significance, 44.5%. The actual trial continued to completion without statistical significance. CONCLUSIONS: Announcements of interim analyses provide information outside the DSMB's sphere of confidentiality. This information is potentially helpful to clinical trial prognosticators. 'Information leakage' from standard interim analyses such as in NSABP C-08 is conventionally viewed as acceptable even though it may be quite revealing. Whether leakage from more aggressive types of adaptations is acceptable should be assessed at the design stage.

Neutrophil gelatinase-associated lipocalin: a novel marker of contrast nephropathy risk.

BACKGROUND: Neutrophil gelatinase-associated lipocalin (NGAL, siderocalin) is a protein secreted by the kidney in the setting of acute kidney injury in an attempt to regulate and bind the release of catalytic iron from injured cells. We sought to evaluate the relationships between baseline NGAL, renal filtration function, and the degree of injury reflected by further increases in NGAL. METHODS: This study was a prospective, blinded assessment of blood samples taken from patients with estimated glomerular filtration rate (eGFR) <75 ml/min/1.73 m(2) undergoing non-urgent coronary angiography and intervention using iodinated contrast. Renal transplant recipients, dialysis patients, and administration of iodinated contrast in the prior 30 days were exclusion criteria. Plasma NGAL was measured using the Alere™ assay. Serum creatinine (Cr) was measured using calibrated methods at a core laboratory. Samples were obtained at baseline, 1, 2, 4, 6, 12, 24, and 48 h after contrast administration. RESULTS: A total of 63 subjects were enrolled with a mean age of 69.4 ± 9.1 years, 73% male, 35% with diabetes, and a mean eGFR of 47.82 ± 15.46 ml/min/1.73 m(2). The correlation between eGFR and NGAL was r = -0.61, 95% CI -0.74 to -0.44, p < 0.001. When stratified by baseline NGAL tertile, the peak NGAL observed for each group occurred at 29.0 ± 22.2 h and there was a twofold increase in the mean and peak change in NGAL across the tertiles. NGAL began to rise 6 h after contrast exposure and followed a similar course to serum Cr and at 48 h the overall mean NGAL was still rising. Only 2 patients sustained a rise in Cr of >25% or ≥0.5 mg/dl. Multivariate regression revealed that baseline NGAL (p < 0.001) and not eGFR (p = 0.95) was independently associated with the NGAL value at 48 h. CONCLUSIONS: Baseline NGAL is strongly correlated with eGFR in patients with reduced renal filtration function undergoing coronary angiography. The magnitude of rise in NGAL is positively associated with the baseline value and is analogous to the time course of Cr in blood after contrast exposure. NGAL and not eGFR is an independent predictor of changes in the post-procedure NGAL. A baseline NGAL level is necessary for the interpretation of NGAL levels in the evaluation of acute kidney injury.

Estrogen-regulated genes in rat testes and their relationship to recovery of spermatogenesis after irradiation.

Despite numerous observations of the effects of estrogens on spermatogenesis, identification of estrogen-regulated genes in the testis is limited. Using rats in which irradiation had completely blocked spermatogonial differentiation, we previously showed that testosterone suppression with gonadotropin-releasing hormone-antagonist acyline and the antiandrogen flutamide stimulated spermatogenic recovery and that addition of estradiol (E2) to this regimen accelerated this recovery. We report here the global changes in testicular cell gene expression induced by the E2 treatment. By minimizing the changes in other hormones and using concurrent data on regulation of the genes by these hormones, we were able to dissect the effects of estrogen on gene expression, independent of gonadotropin or testosterone changes. Expression of 20 genes, largely in somatic cells, was up- or downregulated between 2- and 5-fold by E2. The unexpected and striking enrichment of transcripts not corresponding to known genes among the E2-downregulated probes suggested that these might represent noncoding mRNAs; indeed, we have identified several as miRNAs and their potential target genes in this system. We propose that genes for which expression levels are altered in one direction by irradiation and in the opposite direction by both testosterone suppression and E2 treatment are candidates for controlling the block in differentiation. Several genes, including insulin-like 3 (Insl3), satisfied those criteria. If they are indeed involved in the inhibition of spermatogonial differentiation, they may be candidate targets for treatments to enhance recovery of spermatogenesis following gonadotoxic exposures, such as those resulting from cancer therapy.

Changes in gene expression in somatic cells of rat testes resulting from hormonal modulation and radiation-induced germ cell depletion.

Although gonadotropins and androgen are required for normal spermatogenesis and both testosterone and follicle-stimulating hormone (FSH) are responsible for the inhibition of spermatogonial differentiation that occurs in irradiated rats, it has been difficult to identify the specific genes involved. To study specific hormonally regulated changes in somatic cell gene expression in the testis that may be involved in these processes, without the complication of changing populations of germ cells, we used irradiated LBNF(1) rats, the testes of which contain almost exclusively somatic cells except for a few type A spermatogonia. Three different groups of these rats were treated with various combinations of gonadotropin-releasing hormone antagonist, an androgen receptor antagonist (flutamide), testosterone, and FSH, and we compared the gene expression levels 2 wk later to those of irradiated-only rats by microarray analysis. By dividing the gene expression patterns into three major patterns and 11 subpatterns, we successfully distinguished, in a single study, the genes that were specifically regulated by testosterone, by luteinizing hormone (LH), and by FSH from the large number of genes that were not hormonally regulated in the testis. We found that hormones produced more dramatic upregulation than downregulation of gene expression: Testosterone had the strongest upregulatory effect, LH had a modest but appreciable upregulatory effect, and FSH had a minor upregulatory effect. We also separately identified the somatic cell genes that were chronically upregulated by irradiation. Thus, the present study identified gene expression changes that may be responsible for hormonal action on somatic cells to support normal spermatogenesis and the hormone-mediated block in spermatogonial development after irradiation.

RNA-binding specificity of Y-box protein 1.

Y-box protein 1 (YB-1) is a multifunctional DNA/RNA-binding protein that regulates transcription and translation. The specificity of YB-1's RNA binding and its consequences are unknown. Because expression and subcellular localization of YB-1 have been reported to be important in breast cancer, we determined the specificity and functional impact of YB-1 mRNA-binding in MCF7 breast cancer cells. We used YB-1 antibodies to immunoprecipitate YB-1 and microarray profiling to compare YB-1-bound and total poly(A) RNA. We demonstrated that YB-1 mRNA-binding was preferential. Transcript sequences significantly associated with this binding had high GC content. Selected YB-1 mRNA-binding targets were confirmed by QRT-PCR. However, downregulation of YB-1 levels by siRNA did not affect their RNA or protein expression. Thus, YB-1 has RNA-binding specificity; however, YB-1 binding does not necessarily regulate the stability or translation of its mRNA targets. Further study is needed to determine the functional consequences of selective YB-1 mRNA binding.

Comparative analysis of genes regulated in acute myelomonocytic leukemia with and without inv(16)(p13q22) using microarray techniques, real-time PCR, immunohistochemistry, and flow cytometry immunophenotyping.

Acute myeloid leukemia with inv(16)(p13q22), also known as M4Eo, is a distinct type of leukemia with characteristic clinicopathologic and cytogenetic features. Patients with M4Eo have monocytosis, high blast counts, and abnormal bone marrow eosinophils that contain large basophilic granules. The inv(16)(p13q22) or, less commonly, the t(16;16)(p13;q22) causes fusion of the CBFbeta gene at 16q22 and the MYH11 gene at 16p13, creating the novel chimeric protein CBFbeta-MYH11. To understand the underlying molecular mechanisms unique to M4Eo biology, we determined the gene expression profile of M4Eo cases by using cDNA and long oligonucleotide microarrays. Cases of acute myelomonocytic leukemia without CBFbeta-MYH11 (M4) acted as our control. We found that in the gene expression profile of M4Eo, NF-kappaB activators and inhibitors were upregulated and downregulated, respectively, suggesting that the NF-kappaB signaling pathway is activated at a higher level in M4Eo than in acute myelomonocytic leukemia M4. In addition, the gene expression profile of M4Eo indicates high cell proliferation and low apoptosis. We used real-time PCR, immunohistochemistry, and flow cytometry immunophenotyping to confirm some of our microarray data. These findings most likely represent the functional consequences of the abnormal chimeric protein CBFbeta-MYH11, which is unique to this disease, and suggest that NF-kappaB is a potential therapeutic target for treating M4Eo patients.

Molecular signatures associated with clinical outcome in patients with high-risk head-and-neck squamous cell carcinoma treated by surgery and radiation.

PURPOSE: The local-regional control rate for advanced head-and-neck squamous cell carcinoma (HNSCC) remains poor and is unpredictable for a given individual. This study examined whether gene expression patterns developed from tumors from surgicopathologic, criteria-defined, high-risk HNSCC patients could be correlated with clinical outcomes, namely, metastasis or nonrecurrent disease. METHODS AND MATERIALS: Fifteen primary tumors from patients treated with a consistent protocol of surgery followed by radiotherapy were examined. Seven of these tumors were from high-risk patients who developed distant metastasis (DM), and eight tumors were from patients with no recurrence (NR) (median follow-up, 59 months). RESULTS: Unsupervised clustering of gene expression did not separate the two groups from one another, but when supervised methodologies were applied, 205 genes discriminated the two groups. Within the DM group, genes associated with cell growth and proliferation; DNA replication, recombination, and repair; antiapoptotic pathways; cell adhesion; and angiogenesis were identified. For NR samples, discriminatory genes were associated with the onset of apoptosis. CONCLUSIONS: Our data suggest that gene expression analysis of surgically excised HNSCC tumors from patients considered at high risk for recurrence has the potential to identify individuals susceptible to metastasis on the basis of distinct gene-expression patterns. These patients would be ideal candidates for testing systemic therapy.

Classification analysis of the transcriptosome of nonlesional cultured dermal fibroblasts from systemic sclerosis patients with early disease.

OBJECTIVE: To compare the transcriptosome of early-passage nonlesional dermal fibroblasts from systemic sclerosis (SSc) patients with diffuse disease and matched normal controls in order to gain further understanding of the gene activation patterns that occur in early disease. METHODS: Total RNA was isolated from early-passage fibroblasts obtained from nonlesional skin biopsy specimens from 21 patients with diffuse SSc (disease duration <5 years in all but 1) and 18 healthy controls who were matched to the cases by age (+/-5 years), sex, and race. Array experiments were performed on a 16,659-oligonucleotide microarray utilizing a reference experimental design. Supervised methods were used to select differentially expressed genes. Quantitative polymerase chain reaction (PCR) was used to independently validate the array results. RESULTS: Of the 8,324 genes that passed filtering criteria, classification analysis revealed that <5% were differentially expressed between SSc and normal fibroblasts. Individually, differentially expressed genes included COL7A1, COL18A1 (endostatin), DAF, COMP, and VEGFB. Using the panel of genes discovered through classification analysis, a set of model predictors that achieved reasonably high predictive accuracy was developed. Analysis of 1,297 gene ontology (GO) classes revealed 35 classes that were significantly dysregulated in SSc fibroblasts. These GO classes included anchoring collagen (30934), extracellular matrix structural constituent (5201), and complement activation (6958, 6956). Validation by quantitative PCR demonstrated that 7 of 7 genes selected were concordant with the array results. CONCLUSION: Fibroblasts cultured from nonlesional skin of patients with SSc already have detectable abnormalities in a variety of genes and cellular processes, including those involved in extracellular matrix formation, fibrillogenesis, complement activation, and angiogenesis.

H2-Ea deficiency is a risk factor for bleomycin-induced lung fibrosis in mice.

Pulmonary fibrosis is a significant complication in cancer patients when treated by radiation, e.g., thoracic malignant diseases, or chemotherapeutic agents. Bleomycin is one of the primary drugs used to treat testicular cancer, but the incidence of significant pulmonary fibrosis limits the dose. It is known that susceptibility to bleomycin-induced pulmonary fibrosis is a heritable trait controlled by multiple genes, none of which, however, are yet known. In this study, we used expression profiling and genetic analysis in mouse models of bleomycin-induced pulmonary fibrosis and identified MHC class II antigen Ealpha (H2-Ea) as a risk factor for this disease. We found that a loss-of-function deletion in the H2-Ea gene was linked to susceptibility. A functional test of H2-Ea in transgenic mice showed 100% survival in the transgenic mice compared with 53% in C57BL/10J mice and significantly decreased pulmonary fibrosis from 16.42% (C57BL/10J) to 5.76% (transgenic; P = 1.20e(-8)). These results show that H2-Ea expression protects mice from bleomycin-induced pulmonary fibrosis, which implicates H2-Ea as a candidate susceptibility gene for pulmonary fibrosis.

Global gene expression changes during neoadjuvant chemotherapy for human breast cancer.

PURPOSE: The purpose of this study was to analyze global gene expression changes in serial tumor core biopsy specimens taken during neoadjuvant chemotherapy for primary breast cancer. PATIENTS AND METHODS: Core biopsy specimens from tumors were obtained before treatment and 24 and/or 48 hours after treatment from 21 women who were beginning chemotherapy for breast cancer. RNA was extracted, and radiolabeled complementary DNA was synthesized. The complementary DNA probes were hybridized to high-density microarray membranes that contained more than 25,000 human sequence clones. Hierarchical cluster analysis was used to compare the degree of similarity between expression profiles. RESULTS: Twenty-five (45%) of the 56 available core specimens yielded sufficient quantity and quality RNA for microarray analysis. Microarray profiles were performed only on samples from patients with pretreatment and posttreatment specimens, resulting in serial data sets for five patients (14 specimens). The serial samples from individual patients clustered more closely than the samples taken from different patients. Analyses of the variance of individual gene expression showed that there were significantly fewer genes with fivefold differences in expression in an individual tumor at different times (average, 359 genes) versus pretreatment samples of different tumors (average, 732 genes). Patients with a good pathological response to treatment had gene patterns that clustered distinctly from those of poor responders. Significant transcriptional response occurred in all patients during therapy. Surprisingly, all patients had different genes change after chemotherapy, with no single gene having a significant expression change in all five patients. DISCUSSION: This is the first report to show global gene expression changes during chemotherapy in a human solid tumor. Comprehensive gene expression profiles of more than 25,000 genes can be obtained from core biopsy specimens. A remarkable diversity in transcriptional response was observed for individual cases. Further data are needed to determine whether gene profiling can predict response to chemotherapy.

Identifying and quantifying sources of variation in microarray data using high-density cDNA membrane arrays.

Microarray experiments involve many steps, including spotting cDNA, extracting RNA, labeling targets, hybridizing, scanning, and analyzing images. Each step introduces variability, confounding our ability to obtain accurate estimates of the biological differences between samples. We ran repeated experiments using high-density cDNA microarray membranes (Research Genetics Human GeneFilters Microarrays Version I) and 33P-labeled targets. Total RNA was extracted from a Burkitt lymphoma cell line (GA-10). We estimated the components of variation coming from: (1) image analysis, (2) exposure time to PhosphorImager screens, (3) differences in membranes, (4) reuse of membranes, and (5) differences in targets prepared from two independent RNA extractions. Variation was assessed qualitatively using a clustering algorithm and quantitatively using a version of ANOVA adapted to multivariate microarray data. The largest contribution to variation came from reusing membranes, which contributed 38% of the total variation. Differences in membranes and in exposure time each contributed about 10%. Differences in target preparations contributed less than 5%. The effect of image quantification was negligible. Much of the effect from reusing membranes was attributable to increasing levels of background radiation and can be reduced by using membranes at most four times. The effects of exposure time, which were partly attributable to variation in the scanning process, can be minimized by using the same exposure time for all experiments.