Deep Learning Models for Predicting Malignancy Risk in CT-Detected Pulmonary Nodules: A Systematic Review and Meta-analysis.

BACKGROUND: There has been growing interest in using artificial intelligence/deep learning (DL) to help diagnose prevalent diseases earlier. In this study we sought to survey the landscape of externally validated DL-based computer-aided diagnostic (CADx) models, and assess their diagnostic performance for predicting the risk of malignancy in computed tomography (CT)-detected pulmonary nodules. METHODS: An electronic search was performed in four databases (from inception to 10 August 2023). Studies were eligible if they were peer-reviewed experimental or observational articles comparing the diagnostic performance of externally validated DL-based CADx models with models widely used in clinical practice to predict the risk of malignancy. A bivariate random-effect approach for the meta-analysis on the included studies was used. RESULTS: Seventeen studies were included, comprising 8553 participants and 9884 nodules. Pooled analyses showed DL-based CADx models were 11.6% more sensitive than physician judgement alone, and 14.5% more than clinical risk models alone. They had a similar pooled specificity to physician judgement alone [0.77 (95% CI 0.68-0.84) v 0.81 (95% CI 0.71-0.88)], and were 7.4% more specific than clinical risk models alone. They had superior pooled areas under the receiver operating curve (AUC), with relative pooled AUCs of 1.03 (95% CI 1.00-1.07) and 1.10 (95% CI 1.07-1.13) versus physician judgement and clinical risk models alone, respectively. CONCLUSION: DL-based models are already used in clinical practice in certain settings for nodule management. Our results show their diagnostic performance potentially justifies wider, more routine deployment alongside experienced physician readers to help inform multidisciplinary team decision-making.

A survey of cancer genome signatures identifies genes connected to distinct chromosomal instability phenotypes.

Certain breast and ovarian cancers are characterised by high levels of chromosomal instability. We established a suite of eleven SNP array-based signatures of various forms of chromosomal instability. To understand what biological mechanisms might underpin these signatures, we generated and assembled genetic-feature data including allele-specific expression, fusion genes, gene expression, methylation, somatic coding mutations and protein expression. For each signature, we extracted a compendium of significantly associated genetic features using machine learning. We established an association between subchromosomal allelic imbalance-based measures and DNA repair genes. Numerical chromosomal instability and chromothripsis were found to have distinct genetic associations but shared a relationship to mitotic processes, while whole-genome doubling was characterised by TP53 mutation, and high AURKA and GINS1 expression. Furthermore, we identified two genetically distinct forms of tandem duplicator phenotypes. These findings identify potentially novel genomic targets for validation and drug development for specific subsets of breast and ovarian cancer.

Leukemic blasts program bone marrow adipocytes to generate a protumoral microenvironment.

Despite currently available therapies, most patients diagnosed with acute myeloid leukemia (AML) die of their disease. Tumor-host interactions are critical for the survival and proliferation of cancer cells; accordingly, we hypothesize that specific targeting of the tumor microenvironment may constitute an alternative or additional strategy to conventional tumor-directed chemotherapy. Because adipocytes have been shown to promote breast and prostate cancer proliferation, and because the bone marrow adipose tissue accounts for up to 70% of bone marrow volume in adult humans, we examined the adipocyte-leukemia cell interactions to determine if they are essential for the growth and survival of AML. Using in vivo and in vitro models of AML, we show that bone marrow adipocytes from the tumor microenvironment support the survival and proliferation of malignant cells from patients with AML. We show that AML blasts alter metabolic processes in adipocytes to induce phosphorylation of hormone-sensitive lipase and consequently activate lipolysis, which then enables the transfer of fatty acids from adipocytes to AML blasts. In addition, we report that fatty acid binding protein-4 (FABP4) messenger RNA is upregulated in adipocytes and AML when in coculture. FABP4 inhibition using FABP4 short hairpin RNA knockdown or a small molecule inhibitor prevents AML proliferation on adipocytes. Moreover, knockdown of FABP4 increases survival in Hoxa9/Meis1-driven AML model. Finally, knockdown of carnitine palmitoyltransferase IA in an AML patient-derived xenograft model improves survival. Here, we report the first description of AML programming bone marrow adipocytes to generate a protumoral microenvironment.

Determinants of cancer screening awareness and participation among Indonesian women.

BACKGROUND: Cancer screening awareness and participation may be lower in low- and middle-income countries that lack established national screening programmes compared with those that do. We evaluated potential determinants of awareness about and participation in breast and cervical cancer screening, and breast self-examination (BSE) in women using survey data from Indonesia. METHODS: From the fifth Indonesian Family Life Survey (2014-2015), a total of 5397 women aged 40 and older without any history of cancer who responded to questionnaires concerning Pap smears, mammography, and BSE were included. Multilevel modelling was used to assess potential determinants in relation to awareness about Pap smears and mammography, and participation in Pap smears and BSE practice. Multivariable analyses were performed to identify independent predictors of cancer screening. RESULTS: Of the 5397 respondents, 1058 (20%) women were aware of Pap smears, of which 297 had never had the procedure. Only 251 (5%) participants were aware of mammography. A total of 605 (12%) of women reported they performed BSE. Higher education and household expenditure were consistently associated with higher odds of awareness about Pap smears and mammography (e.g. odds ratio [OR] of being aware of Pap smear and mammography: 7.82 (95% CI: 6.30-9.70) and 7.70 (6.19-9.58), respectively, for high school graduates compared to women with less educational attainment in the multivariable models), and participation in Pap smears and BSE. We also identified enabling factors linked with greater cancer screening awareness and participation, including health insurance, shorter distance to health services, and social participation. CONCLUSION: There are socioeconomic disparities in cancer screening awareness and participation among Indonesian women. Our findings may help inform targeted health promotion and screening for cancer in the presence of limited resources.

Circulating Prostate-Specific Antigen and Telomere Length in a Nationally Representative Sample of Men Without History of Prostate Cancer.

BACKGROUND: We investigated the association of prostate-specific antigen (PSA) with leukocyte telomere length, which may be altered in preclinical prostate malignancies. METHODS: This study was based on the 2001-2002 U.S. National Health and Nutrition Examination Survey (NHANES). A subsample of 1,127 men aged 40-85 years without prior history of prostate cancer who provided informed consent and blood samples were selected. Leukocyte telomere length (LTL) relative to standard DNA reference (T/S ratio) was quantified by polymerase chain reaction (PCR). Survey-weighted multivariable linear regression was performed to examine T/S ratio across quintiles of total and free PSA and free-to-total PSA ratio (%fPSA). A sensitivity analysis was performed by excluding men dying from prostate cancer during follow-up through to December 31, 2006. Stratification analyses were carried out to assess any effect modification by age group, race, body mass index (BMI), and levels of C-reactive protein (CRP), a marker of inflammation. RESULTS: Higher total PSA levels were associated to longer LTL, with approximately 8% increase in log-transformed T/S ratio (95% confidence interval [CI]: 2-13%) among men in the highest quintile of total PSA compared to the lowest in the fully adjusted model (Ptrend = 0.01). No significant association was found for free PSA or %fPSA, although nonlinearity between all PSA measures and T/S ratio was indicated. Similar results were found after excluding men who died from prostate cancer during follow-up. We also found the associations between total PSA and T/S ratio to be strongest among non-Hispanic blacks, non-obese men (BMI <30 kg/m2 ), and those with low CRP. However, a significant interaction was only found between total PSA and race/ethnicity (Pinteraction = 0.01). CONCLUSION: Total PSA levels were strongly associated to LTL, particularly among non-Hispanic blacks. Our findings support a potential link between PSA and specific mechanisms contributing to prostate cancer development. Prostate 77:22-32, 2017. © 2016 Wiley Periodicals, Inc.

Economic downturns, universal health coverage, and cancer mortality in high-income and middle-income countries, 1990-2010: a longitudinal analysis.

BACKGROUND: The global economic crisis has been associated with increased unemployment and reduced public-sector expenditure on health care (PEH). We estimated the effects of changes in unemployment and PEH on cancer mortality, and identified how universal health coverage (UHC) affected these relationships. METHODS: For this longitudinal analysis, we obtained data from the World Bank and WHO (1990-2010). We aggregated mortality data for breast cancer in women, prostate cancer in men, and colorectal cancers in men and women, which are associated with survival rates that exceed 50%, into a treatable cancer class. We likewise aggregated data for lung and pancreatic cancers, which have 5 year survival rates of less than 10%, into an untreatable cancer class. We used multivariable regression analysis, controlling for country-specific demographics and infrastructure, with time-lag analyses and robustness checks to investigate the relationship between unemployment, PEH, and cancer mortality, with and without UHC. We used trend analysis to project mortality rates, on the basis of trends before the sharp unemployment rise that occurred in many countries from 2008 to 2010, and compared them with observed rates. RESULTS: Data were available for 75 countries, representing 2.106 billion people, for the unemployment analysis and for 79 countries, representing 2.156 billion people, for the PEH analysis. Unemployment rises were significantly associated with an increase in all-cancer mortality and all specific cancers except lung cancer in women. By contrast, untreatable cancer mortality was not significantly linked with changes in unemployment. Lag analyses showed significant associations remained 5 years after unemployment increases for the treatable cancer class. Rerunning analyses, while accounting for UHC status, removed the significant associations. All-cancer, treatable cancer, and specific cancer mortalities significantly decreased as PEH increased. Time-series analysis provided an estimate of more than 40,000 excess deaths due to a subset of treatable cancers from 2008 to 2010, on the basis of 2000-07 trends. Most of these deaths were in non-UHC countries. INTERPRETATION: Unemployment increases are associated with rises in cancer mortality; UHC seems to protect against this effect. PEH increases are associated with reduced cancer mortality. Access to health care could underlie these associations. We estimate that the 2008-10 economic crisis was associated with about 260,000 excess cancer-related deaths in the Organisation for Economic Co-operation and Development alone. FUNDING: None.

Cigarette smoking and telomere length: A systematic review of 84 studies and meta-analysis.

BACKGROUND: Cigarette smoking is a risk factor for ageing-related disease, but its association with biological ageing, indicated by telomere length, is unclear. METHODS: We systematically reviewed evidence evaluating association between smoking status and telomere length. Searches were performed in MEDLINE (Ovid) and EMBASE (Ovid) databases, combining variation of keywords "smoking" and "telomere". Data was extracted for study characteristics and estimates for association between smoking and telomere length. Quality of studies was assessed with a risk of bias score, and publication bias was assessed with a funnel plot. I2 test was used to observe heterogeneity. Meta-analysis was carried out to compare mean difference in telomere length by smoking status, and a dose-response approach was carried out for pack-years of smoking and telomere length. A sensitivity analysis was carried out to examine sources of heterogeneity. RESULTS: A total of 84 studies were included in the review, and 30 among them were included in our meta-analysis. Potential bias was addressed in half of included studies, and there was little evidence of small study bias. Telomere length was shorter among ever smokers compared to never smokers (summary standard mean difference [SMD]: -0.11 (95% CI -0.16 to -0.07)). Similarly, shorter telomere length was found among smokers compared to non-smokers, and among current smokers compared to never or former smokers. Dose-response meta-analysis suggested an inverse trend between pack-years of smoking and telomere length. However, heterogeneity among some analyses was observed. CONCLUSION: Shorter telomeres among ever smokers compared to those who never smoked may imply mechanisms linking tobacco smoke exposure to ageing-related disease.

SoftPanel: a website for grouping diseases and related disorders for generation of customized panels.

BACKGROUND: Targeted next-generation sequencing is playing an increasingly important role in biological research and clinical diagnosis by allowing researchers to sequence high priority genes at much higher depths and at a fraction of the cost of whole genome or exome sequencing. However, in designing the panel of genes to be sequenced, investigators need to consider the tradeoff between the better sensitivity of a broad panel and the higher specificity of a potentially more relevant panel. Although tools to prioritize candidate disease genes have been developed, the great majority of these require prior knowledge and a set of seed genes as input, which is only possible for diseases with a known genetic etiology. RESULTS: To meet the demands of both researchers and clinicians, we have developed a user-friendly website called SoftPanel. This website is intended to serve users by allowing them to input a single disorder or a disorder group and generate a panel of genes predicted to underlie the disorder of interest. Various methods of retrieval including a keyword search, browsing of an arborized list of International Classification of Diseases, 10th revision (ICD-10) codes or using disorder phenotypic similarities can be combined to define a group of disorders and the genes known to be associated with them. Moreover, SoftPanel enables users to expand or refine a gene list by utilizing several biological data resources. In addition to providing users with the facility to create a "hard" panel that contains an exact gene list for targeted sequencing, SoftPanel also enables generation of a "soft" panel of genes, which may be used to further filter a significantly altered set of genes identified through whole genome or whole exome sequencing. The service and data provided by SoftPanel can be accessed at http://www.isb.pku.edu.cn/SoftPanel/ . A tutorial page is included for trying out sample data and interpreting results. CONCLUSION: SoftPanel provides a convenient and powerful tool for creating a targeted panel of potential disease genes while supporting different forms of input. SoftPanel may be utilized in both genomics research and personalized medicine.

Irinotecan chemotherapy combined with fluoropyrimidines versus irinotecan alone for overall survival and progression-free survival in patients with advanced and/or metastatic colorectal cancer.

BACKGROUND: Chemotherapy is the treatment of choice in patients with advanced or metastatic colorectal cancer (CRC) where surgical resection of metastases is not an option. Both irinotecan (IRI) and fluoropyrimidines are often included in first- or second- line chemotherapy treatment regimens in such patients. However, it is not clear whether combining these agents is superior to irinotecan alone. OBJECTIVES: To compare the efficacy and safety of two chemotherapeutic regimens, irinotecan monotherapy or irinotecan in combination with fluoropyrimidines, for patients with advanced CRC when administered in the first or second-line settings. SEARCH METHODS: We searched the following electronic databases to identify randomized controlled trials: Cochrane Colorectal Cancer Group Specialised Register (January 13, 2016), Cochrane Central Register of Controlled Trials (CENTRAL)(The Cochrane Library Issue 12, 2016), Ovid MEDLINE (1950 to January 13, 2016), Ovid EMBASE (1974 to January 13, 2016), registers of controlled trials in progress, references cited in relevant publications and conference proceedings in related fields (BioMed Central and Medscape's Conference). The key authors or investigators of all eligible studies, and professionals in the field were contacted when necessary. The search from January 2016 identified one eligible study, an ongoing trial currently presented as an abstract, to be considered in an update of this review. SELECTION CRITERIA: Randomized controlled trials (RCTs) investigating the efficacy and safety of IRI chemotherapy combined with fluoropyrimidine compared with IRI alone for the treatment of patients with advanced CRC, regardless of treatment line settings. DATA COLLECTION AND ANALYSIS: Study eligibility and methodological quality were assessed independently by the two authors, and any disagreement was solved by a third author. The data collected from the studies were reviewed qualitatively and quantitatively using the Cochrane Collaboration statistical software RevMan 5.3. MAIN RESULTS: Five studies were included in this review with a total of 1,726 patients. The top-up search resulted in an additional ongoing trial, the results of which have not been incorporated in this review. Among five included studies, no reduction in all-cause mortality was observed in the combination arm, with a summary hazard ratio (HR) of 0.91 (95% CI: 0.81-1.02). Longer progression-free survival was observed in those treated with the combination chemotherapy (HR: 0.68, 95% CI: 0.53-0.87), however, this result may have been driven by findings from the single first-line treatment setting study.The quality of evidence for overall survival was low and for progression-free survival was moderate, mainly due to study limitation from the lack of information on randomisation methods and allocation concealment.There were higher risks of toxicity outcomes grade 3 or 4 diarrhoea and grade 1 or 2 alopecia, and a lower risk of grade 3 or 4 neutropenia in controls compared to the invervention group. Evidence for toxicity has been assessed to be low to moderate quality. AUTHORS' CONCLUSIONS: There was no overall survival benefit of the irinotecan and fluoropyrimidine treatment over irinotecan alone, thus both regimens remain reasonable options in treating patients with advanced or metastatic CRC. Given the low and moderate quality of the evidence, future studies with sufficient numbers of patients in each treatment arms are needed to clarify the benefit observed in progression-free survival with combination irinotecan and fluoropyrimidines.

Spatiotemporal organization of Aurora-B by APC/CCdh1 after mitosis coordinates cell spreading through FHOD1.

Spatiotemporal regulation of mitotic kinase activity underlies the extensive rearrangement of cellular components required for cell division. One highly dynamic mitotic kinase is Aurora-B (AurB), which has multiple roles defined by the changing localisation of the chromosome passenger complex (CPC) as cells progress through mitosis, including regulation of cytokinesis and abscission. Like other mitotic kinases, AurB is a target of the anaphase-promoting complex (APC/C) ubiquitin ligase during mitotic exit, but it is not known if APC/C-mediated destruction plays any specific role in controlling AurB activity. We have examined the contribution of the Cdh1 coactivator-associated APC/C(Cdh1) to the organization of AurB activity as cells exit mitosis and re-enter interphase. We report that APC/C(Cdh1)-dependent proteolysis restricts a cell-cortex-associated pool of active AurB in space and time. In early G1 phase this pool of AurB is found at protrusions associated with cell spreading. AurB retention at the cortex depends on a formin, FHOD1, critically required to organize the cytoskeleton after division. We identify AurB phosphorylation sites in FHOD1 and show that phosphomutant FHOD1 is impaired in post-mitotic assembly of oriented actin cables. We propose that Cdh1 contributes to spatiotemporal organization of AurB activity and that organization of FHOD1 activity by AurB contributes to daughter cell spreading after mitosis.

Digital imaging in the immunohistochemical evaluation of the proliferation markers Ki67, MCM2 and Geminin, in early breast cancer, and their putative prognostic value.

BACKGROUND: Immunohistochemical assessment of proliferation may provide additional prognostic information in early breast cancer. However, due to a lack of methodological standards proliferation markers are still not routinely used for determining therapy. Even for Ki67, one of the most widely-studied markers, disagreements over the optimal cutoff exist. Improvements in digital microscopy may provide new avenues to standardise and make data more reproducible. METHODS: We studied the immunohistochemical expression of three markers of proliferation: Ki67, Mini-Chromosome Maintenance protein 2 and Geminin, by conventional light microscope and digital imaging on triplicate TMAs from 309 consecutive cases of primary breast cancers. Differences between the average and the maximum percentage reactivity in tumour cell nuclei from the three TMA cores were investigated to assess the validity of the approach. Time-dependent Receiver Operating Characteristic curves were utilized to obtain optimal expression level cut-offs, which were then correlated with clinico-pathological features and survival. RESULTS: High concordance between conventional and digital scores was observed for all 3 markers (Ki67: rs = 0.87, P < 0.001; MCM2: rs = 0.94, P < 0.001; and Geminin: rs = 0.86, P < 0.001; Spearman's rank). There was no significant difference according to the number of TMA cores included for either Ki67 or MCM2; analysis of two or three cores produced comparable results. Higher levels of all three proliferation markers were significantly associated with higher grade (P < 0.001) and ER-negativity (P < 0.001). Optimal prognostic cut-offs for percentage expression in the tumour were 8 %, 12 and 2.33 % for Ki67, MCM2 and Geminin respectively. All 3 proliferation marker cutoffs were predictive of 15-year breast cancer-specific survival in univariable Cox regression analyses. In multivariable analysis only lymph node status (HR = 3.9, 95 % CI = 1.79-8.5, P = 0.0006) and histological grade (HR = 1.84, 95 % CI = 1-3.38, P = 0.05) remained significantly prognostic. CONCLUSIONS: Here we show that. MCM2 is a more sensitive marker of proliferation than Ki67 and should be examined in future studies, especially in the lymph node-negative, hormone receptor-positive subgroup. Further, digital microscopy can be used effectively as a high-throughput method to evaluate immunohistochemical expression.

Unemployment, public-sector health-care spending and breast cancer mortality in the European Union: 1990-2009.

BACKGROUND: The global economic crisis has been associated with increased unemployment, reduced health-care spending and adverse health outcomes. Insights into the impact of economic variations on cancer mortality, however, remain limited. METHODS: We used multivariate regression analysis to assess how changes in unemployment and public-sector expenditure on health care (PSEH) varied with female breast cancer mortality in the 27 European Union member states from 1990 to 2009. We then determined how the association with unemployment was modified by PSEH. Country-specific differences in infrastructure and demographic structure were controlled for, and 1-, 3-, 5- and 10-year lag analyses were conducted. Several robustness checks were also implemented. RESULTS: Unemployment was associated with an increase in breast cancer mortality [P < 0.0001, coefficient (R) = 0.1829, 95% confidence interval (CI) 0.0978-0.2680]. Lag analysis showed a continued increase in breast cancer mortality at 1, 3, 5 and 10 years after unemployment rises (P < 0.05). Controlling for PSEH removed this association (P = 0.063, R = 0.080, 95% CI -0.004 to 0.163). PSEH increases were associated with significant decreases in breast cancer mortality (P < 0.0001, R = -1.28, 95% CI -1.67 to -0.877). The association between unemployment and breast cancer mortality remained in all robustness checks. CONCLUSION: Rises in unemployment are associated with significant short- and long-term increases in breast cancer mortality, while increases in PSEH are associated with reductions in breast cancer mortality. Initiatives that bolster employment and maintain total health-care expenditure may help minimize increases in breast cancer mortality during economic crises.

Genomic scars as biomarkers of homologous recombination deficiency and drug response in breast and ovarian cancers.

Poly (ADP-ribose) polymerase (PARP) inhibitors and platinum-based chemotherapies have been found to be particularly effective in tumors that harbor deleterious germline or somatic mutations in the BRCA1 or BRCA2 genes, the products of which contribute to the conservative homologous recombination repair of DNA double-strand breaks. Nonetheless, several setbacks in clinical trial settings have highlighted some of the issues surrounding the investigation of PARP inhibitors, especially the identification of patients who stand to benefit from such drugs. One potential approach to finding this patient subpopulation is to examine the tumor DNA for evidence of a homologous recombination defect. However, although the genomes of many breast and ovarian cancers are replete with aberrations, the presence of numerous factors able to shape the genomic landscape means that only some of the observed DNA abnormalities are the outcome of a cancer cell's inability to faithfully repair DNA double-strand breaks. Consequently, recently developed methods for comprehensively capturing the diverse ways in which homologous recombination deficiencies may arise beyond BRCA1/2 mutation have used DNA microarray and sequencing data to account for potentially confounding features in the genome. Scores capturing telomeric allelic imbalance, loss of heterozygosity (LOH) and large scale transition score, as well as the total number of coding mutations are measures that summarize the total burden of certain forms of genomic abnormality. By contrast, other studies have comprehensively catalogued different types of mutational pattern and their relative contributions to a given tumor sample. Although at least one study to explore the use of the LOH scar in a prospective clinical trial of a PARP inhibitor in ovarian cancer is under way, limitations that result in a relatively low positive predictive value for these biomarkers remain. Tumors whose genome has undergone one or more events that restore high-fidelity homologous recombination are likely to be misclassified as double-strand break repair-deficient and thereby sensitive to PARP inhibitors and DNA damaging chemotherapies as a result of prior repair deficiency and its genomic scarring. Therefore, we propose that integration of a genomic scar-based biomarker with a marker of resistance in a high genomic scarring burden context may improve the performance of any companion diagnostic for PARP inhibitors.

Determining the impact of an artificial intelligence tool on the management of pulmonary nodules detected incidentally on CT (DOLCE) study protocol: a prospective, non-interventional multicentre UK study.

INTRODUCTION: In a small percentage of patients, pulmonary nodules found on CT scans are early lung cancers. Lung cancer detected at an early stage has a much better prognosis. The British Thoracic Society guideline on managing pulmonary nodules recommends using multivariable malignancy risk prediction models to assist in management. While these guidelines seem to be effective in clinical practice, recent data suggest that artificial intelligence (AI)-based malignant-nodule prediction solutions might outperform existing models. METHODS AND ANALYSIS: This study is a prospective, observational multicentre study to assess the clinical utility of an AI-assisted CT-based lung cancer prediction tool (LCP) for managing incidental solid and part solid pulmonary nodule patients vs standard care. Two thousand patients will be recruited from 12 different UK hospitals. The primary outcome is the difference between standard care and LCP-guided care in terms of the rate of benign nodules and patients with cancer discharged straight after the assessment of the baseline CT scan. Secondary outcomes investigate adherence to clinical guidelines, other measures of changes to clinical management, patient outcomes and cost-effectiveness. ETHICS AND DISSEMINATION: This study has been reviewed and given a favourable opinion by the South Central-Oxford C Research Ethics Committee in UK (REC reference number: 22/SC/0142).Study results will be available publicly following peer-reviewed publication in open-access journals. A patient and public involvement group workshop is planned before the study results are available to discuss best methods to disseminate the results. Study results will also be fed back to participating organisations to inform training and procurement activities. TRIAL REGISTRATION NUMBER: NCT05389774.

Bayesian clinical classification from high-dimensional data: Signatures versus variability.

When data exhibit imbalance between a large number d of covariates and a small number n of samples, clinical outcome prediction is impaired by overfitting and prohibitive computation demands. Here we study two simple Bayesian prediction protocols that can be applied to data of any dimension and any number of outcome classes. Calculating Bayesian integrals and optimal hyperparameters analytically leaves only a small number of numerical integrations, and CPU demands scale as O(nd). We compare their performance on synthetic and genomic data to the mclustDA method of Fraley and Raftery. For small d they perform as well as mclustDA or better. For d = 10,000 or more mclustDA breaks down computationally, while the Bayesian methods remain efficient. This allows us to explore phenomena typical of classification in high-dimensional spaces, such as overfitting and the reduced discriminative effectiveness of signatures compared to intra-class variability.

Profile of the breast cancer susceptibility marker rs4245739 identifies a role for miRNAs.

OBJECTIVE: To determine the influence of the single nucleotide polymorphism (SNP) rs4245739 on the binding and expression of microRNAs and subsequent MDM4 expression and the correlation of these factors with clinical determinants of ER-negative breast cancers. METHODS: FindTar and miRanda were used to detect the manner in which potential microRNAs are affected by the SNP rs4245739-flanking sequence. RNA sequencing data for ER-negative breast cancer from The Cancer Genome Atlas (TCGA) were used to compare the expression of miR-184, miR-191, miR-193a, miR-378, and MDM4 in different rs4245739 genotypes. RESULTS: Comparison of ER-negative cancer patients with and without the expression of miR-191 as well as profile microRNAs (miR-184, miR-191, miR-193a and miR-378 altogether) can differentiate the expression of MDM4 among different rs4245739 genotypes. Although simple genotyping alone did not reveal significant clinical relationships, the combination of genotyping and microRNA profiles was able to significantly differentiate individuals with larger tumor size and lower number of involved lymph nodes (P < 0.05) in the risk group (A allele). CONCLUSIONS: We present two novel methods to analyze SNPs within 3'UTRs that use: (i) a single miRNA marker expression and (ii) an expression profile of miRNAs predicted to bind to the SNP region. We demonstrate that the application of these two methods, in particular the miRNA profile approach, permits detection of new molecular and clinical features related to the rs4245739 variant in ER-negative breast cancer.

Effects of economic downturns on child mortality: a global economic analysis, 1981-2010.

OBJECTIVES: To analyse how economic downturns affect child mortality both globally and among subgroups of countries of variable income levels. DESIGN: Retrospective observational study using economic data from the World Bank's Development Indicators and Global Development Finance (2013 edition). Child mortality data were sourced from the Institute for Health Metrics and Evaluation. SETTING: Global. PARTICIPANTS: 204 countries between 1981 and 2010. MAIN OUTCOME MEASURES: Child mortality, controlling for country-specific differences in political, healthcare, cultural, structural, educational and economic factors. RESULTS: 197 countries experienced at least 1 economic downturn between 1981 and 2010, with a mean of 7.97 downturns per country (range 0-21; SD 0.45). At the global level, downturns were associated with significant (p<0.0001) deteriorations in each child mortality measure, in comparison with non-downturn years: neonatal (coefficient: 1.11, 95% CI 0.855 to 1.37), postneonatal (2.00, 95% CI 1.61 to 2.38), child (2.93, 95% CI 2.26 to 3.60) and under 5 years of age (5.44, 95% CI 4.31 to 6.58) mortality rates. Stronger (larger falls in the growth rate of gross domestic product/capita) and longer (lasting 2 years rather than 1) downturns were associated with larger significant deteriorations (p<0.001). During economic downturns, countries in the poorest quartile experienced ∼1½ times greater deterioration in neonatal mortality, compared with their own baseline; a 3-fold deterioration in postneonatal mortality; a 9-fold deterioration in child mortality and a 3-fold deterioration in under-5 mortality, than countries in the wealthiest quartile (p<0.0005). For 1-5 years after downturns ended, each mortality measure continued to display significant deteriorations (p<0.0001). CONCLUSIONS: Economic downturns occur frequently and are associated with significant deteriorations in child mortality, with worse declines in lower income countries.

Inflammatory Differences in Plaque Erosion and Rupture in Patients With ST-Segment Elevation Myocardial Infarction.

BACKGROUND: Plaque erosion causes 30% of ST-segment elevation myocardial infarctions, but the underlying cause is unknown. Inflammatory infiltrates are less abundant in erosion compared with rupture in autopsy studies. We hypothesized that erosion and rupture are associated with significant differences in intracoronary cytokines in vivo. METHODS AND RESULTS: Forty ST-segment elevation myocardial infarction patients with <6 hours of chest pain were classified as ruptured fibrous cap (RFC) or intact fibrous cap (IFC) using optical coherence tomography. Plasma samples from the infarct-related artery and a peripheral artery were analyzed for expression of 102 cytokines using arrays; results were confirmed with ELISA. Thrombectomy samples were analyzed for differential mRNA expression using quantitative real-time polymerase chain reaction. Twenty-three lesions were classified as RFC (58%), 15 as IFC (38%), and 2 were undefined (4%). In addition, 12% (12 of 102) of cytokines were differentially expressed in both coronary and peripheral plasma. I-TAC was preferentially expressed in RFC (significance analysis of microarrays adjusted P<0.001; ELISA IFC 10.2 versus RFC 10.8 log2 pg/mL; P=0.042). IFC was associated with preferential expression of epidermal growth factor (significance analysis of microarrays adjusted P<0.001; ELISA IFC 7.42 versus RFC 6.63 log2 pg/mL, P=0.036) and thrombospondin 1 (significance analysis of microarrays adjusted P=0.03; ELISA IFC 10.4 versus RFC 8.65 log2 ng/mL, P=0.0041). Thrombectomy mRNA showed elevated I-TAC in RFC (P=0.0007) epidermal growth factor expression in IFC (P=0.0264) but no differences in expression of thrombospondin 1. CONCLUSIONS: These results demonstrate differential intracoronary cytokine expression in RFC and IFC. Elevated thrombospondin 1 and epidermal growth factor may play an etiological role in erosion.