Accumulating research: a systematic account of how cumulative meta-analyses would have provided knowledge, improved health, reduced harm and saved resources

BACKGROUND: "Cumulative meta-analysis" describes a statistical procedure to calculate, retrospectively, summary estimates from the results of similar trials every time the results of a further trial in the series had become available. In the early 1990 s, comparisons of cumulative meta-analyses of treatments for myocardial infarction with advice promulgated through medical textbooks showed that research had continued long after robust estimates of treatment effects had accumulated, and that medical textbooks had overlooked strong, existing evidence from trials. Cumulative meta-analyses have subsequently been used to assess what could have been known had new studies been informed by systematic reviews of relevant existing evidence and how waste might have been reduced.

METHODS AND FINDINGS: We used a systematic approach to identify and summarise the findings of cumulative meta-analyses of studies of the effects of clinical interventions, published from 1992 to 2012. Searches were done of PubMed, MEDLINE, EMBASE, the Cochrane Methodology Register and Science Citation Index. A total of 50 eligible reports were identified, including more than 1,500 cumulative meta-analyses. A variety of themes are illustrated with specific examples. The studies showed that initially positive results became null or negative in meta-analyses as more trials were done; that early null or negative results were over-turned; that stable results (beneficial, harmful and neutral) would have been seen had a meta-analysis been done before the new trial; and that additional trials had been much too small to resolve the remaining uncertainties.

CONCLUSIONS: This large, unique collection of cumulative meta-analyses highlights how a review of the existing evidence might have helped researchers, practitioners, patients and funders make more informed decisions and choices about new trials over decades of research. This would have led to earlier uptake of effective interventions in practice, less exposure of trial participants to less effective treatments, and reduced waste resulting from unjustified research.

Meta-analysis of prostate cancer gene expression data identifies a novel discriminatory signature enriched for glycosylating enzymes

BACKGROUND: Tumorigenesis is characterised by changes in transcriptional control. Extensive transcript expression data have been acquired over the last decade and used to classify prostate cancers. Prostate cancer is, however, a heterogeneous multifocal cancer and this poses challenges in identifying robust transcript biomarkers.

METHODS: In this study, we have undertaken a meta-analysis of publicly available transcriptomic data spanning datasets and technologies from the last decade and encompassing laser capture microdissected and macrodissected sample sets.

RESULTS: We identified a 33 gene signature that can discriminate between benign tissue controls and localised prostate cancers irrespective of detection platform or dissection status. These genes were significantly overexpressed in localised prostate cancer versus benign tissue in at least three datasets within the Oncomine Compendium of Expression Array Data. In addition, they were also overexpressed in a recent exon-array dataset as well a prostate cancer RNA-seq dataset generated as part of the The Cancer Genomics Atlas (TCGA) initiative. Biologically, glycosylation was the single enriched process associated with this 33 gene signature, encompassing four glycosylating enzymes. We went on to evaluate the performance of this signature against three individual markers of prostate cancer, v-ets avian erythroblastosis virus E26 oncogene homolog (ERG) expression, prostate specific antigen (PSA) expression and androgen receptor (AR) expression in an additional independent dataset. Our signature had greater discriminatory power than these markers both for localised cancer and metastatic disease relative to benign tissue, or in the case of metastasis, also localised prostate cancer.

CONCLUSION: In conclusion, robust transcript biomarkers are present within datasets assembled over many years and cohorts and our study provides both examples and a strategy for refining and comparing datasets to obtain additional markers as more data are generated.

Vaccinations and childhood type 1 diabetes mellitus: a meta-analysis of observational studies

Aims/hypothesis The aim of this study was to investigate the association between routine vaccinations and the risk of childhood type 1 diabetes mellitus by systematically reviewing the published literature and performing meta-analyses where possible.

Methods A comprehensive literature search was performed of MEDLINE and EMBASE to identify all studies that compared vaccination rates in children who subsequently developed type 1 diabetes mellitus and in control children. ORs and 95% CIs were obtained from published reports or derived from individual patient data and then combined using a random effects meta-analysis.

Results In total, 23 studies investigating 16 vaccinations met the inclusion criteria. Eleven of these contributed to meta-analyses which included data from between 359 and 11,828 childhood diabetes cases. Overall, there was no evidence to suggest an association between any of the childhood vaccinations investigated and type 1 diabetes mellitus. The pooled ORs ranged from 0.58 (95% CI 0.24, 1.40) for the measles, mumps and rubella (MMR) vaccination in five studies up to 1.04 (95% CI 0.94, 1.14) for the haemophilus influenza B (HiB) vaccination in 11 studies. Significant heterogeneity was present in most of the pooled analyses, but was markedly reduced when analyses were restricted to study reports with high methodology quality scores. Neither this restriction by quality nor the original authors’ adjustments for potential confounding made a substantial difference to the pooled ORs.

Conclusions/interpretation This study provides no evidence of an association between routine vaccinations and childhood type 1 diabetes.

Does childhood bullying lead to the development of psychotic symptoms? A meta-analysis and review of prospective studies

Purpose: Researchers have demonstrated associations between trauma and psychosis. Childhood trauma, in particular, appears to be an important determinant. Recently, bullying has become considered a traumatic experience in its own right. This review aims to analyse research with prospective designs, which will enable conclusions about whether or not bullying causes psychosis.

Methods: A systematic review of the literature was carried out independently by two reviewers. Eligibility and quality assessment criteria were applied. A meta-analysis and narrative synthesis were then completed.

Results: Ten studies met inclusion criteria. Four used data from the same large database, and were combined as one. The majority provided confirmation that bullying appears to cause later development of psychosis. A meta-analysis yielded an unadjusted odds ratio (OR) of 2.148 [95% confidence interval (CI) 1.140–4.044].

Conclusions: The studies reviewed here suggest that bullying does predict the later development of psychotic symptoms. What is lacking from the literature is adequate investigation into other potential mediating factors. The current review highlights the significant role of bullying within this complex interaction. Potential mediating variables are explored, including a dose–response effect for the severity and frequency of victimization. Suggestions for targeting intervention are also suggested alongside clinical implications and recommendations for future research.

Supervised Aggregative Feature Extraction for Big Data Time Series Regression

In many applications, and especially those where batch processes are involved, a target scalar output of interest is often dependent on one or more time series of data. With the exponential growth in data logging in modern industries such time series are increasingly available for statistical modeling in soft sensing applications. In order to exploit time series data for predictive modelling, it is necessary to summarise the information they contain as a set of features to use as model regressors. Typically this is done in an unsupervised fashion using simple techniques such as computing statistical moments, principal components or wavelet decompositions, often leading to significant information loss and hence suboptimal predictive models. In this paper, a functional learning paradigm is exploited in a supervised fashion to derive continuous, smooth estimates of time series data (yielding aggregated local information), while simultaneously estimating a continuous shape function yielding optimal predictions. The proposed Supervised Aggregative Feature Extraction (SAFE) methodology can be extended to support nonlinear predictive models by embedding the functional learning framework in a Reproducing Kernel Hilbert Spaces setting. SAFE has a number of attractive features including closed form solution and the ability to explicitly incorporate first and second order derivative information. Using simulation studies and a practical semiconductor manufacturing case study we highlight the strengths of the new methodology with respect to standard unsupervised feature extraction approaches.

LD Score regression distinguishes confounding from polygenicity in genome-wide association studies

Both polygenicity (many small genetic effects) and confounding biases, such as cryptic relatedness and population stratification, can yield an inflated distribution of test statistics in genome-wide association studies (GWAS). However, current methods cannot distinguish between inflation from a true polygenic signal and bias. We have developed an approach, LD Score regression, that quantifies the contribution of each by examining the relationship between test statistics and linkage disequilibrium (LD). The LD Score regression intercept can be used to estimate a more powerful and accurate correction factor than genomic control. We find strong evidence that polygenicity accounts for the majority of the inflation in test statistics in many GWAS of large sample size.

Meta-analysis of Positive and Negative Symptoms Reveals Schizophrenia Modifier Genes

BACKGROUND: Evidence suggests that genetic factors may influence both schizophrenia (Scz) and its clinical presentation. In recent years, genome-wide association studies (GWAS) have demonstrated considerable success in identifying risk loci. Detection of "modifier loci" has the potential to further elucidate underlying disease processes.

METHODS: We performed GWAS of empirically derived positive and negative symptom scales in Irish cases from multiply affected pedigrees and a larger, independent case-control sample, subsequently combining these into a large Irish meta-analysis. In addition to single-SNP associations, we considered gene-based and pathway analyses to better capture convergent genetic effects, and to facilitate biological interpretation of these findings. Replication and testing of aggregate genetic effects was conducted using an independent European-American sample.

RESULTS: Though no single marker met the genome-wide significance threshold, genes and ontologies/pathways were significantly associated with negative and positive symptoms; notably, NKAIN2 and NRG1, respectively. We observed limited overlap in ontologies/pathways associated with different symptom profiles, with immune-related categories over-represented for negative symptoms, and addiction-related categories for positive symptoms. Replication analyses suggested that genes associated with clinical presentation are generalizable to non-Irish samples.

CONCLUSIONS: These findings strongly support the hypothesis that modifier loci contribute to the etiology of distinct Scz symptom profiles. The finding that previously implicated "risk loci" actually influence particular symptom dimensions has the potential to better delineate the roles of these genes in Scz etiology. Furthermore, the over-representation of distinct gene ontologies/pathways across symptom profiles suggests that the clinical heterogeneity of Scz is due in part to complex and diverse genetic factors.

Long working hours and risk of coronary heart disease and stroke: a systematic review and meta-analysis of published and unpublished data for 603 838 individuals

Background: Long working hours might increase the risk of cardiovascular disease, but prospective evidence is scarce, imprecise, and mostly limited to coronary heart disease. We aimed to assess long working hours as a risk factor for incident coronary heart disease and stroke. 

Methods We identified published studies through a systematic review of PubMed and Embase from inception to Aug 20, 2014. We obtained unpublished data for 20 cohort studies from the Individual-Participant-Data Meta-analysis in Working Populations (IPD-Work) Consortium and open-access data archives. We used cumulative random-effects meta-analysis to combine effect estimates from published and unpublished data. 

Findings We included 25 studies from 24 cohorts in Europe, the USA, and Australia. The meta-analysis of coronary heart disease comprised data for 603 838 men and women who were free from coronary heart disease at baseline; the meta-analysis of stroke comprised data for 528 908 men and women who were free from stroke at baseline. Follow-up for coronary heart disease was 5·1 million person-years (mean 8·5 years), in which 4768 events were recorded, and for stroke was 3·8 million person-years (mean 7·2 years), in which 1722 events were recorded. In cumulative meta-analysis adjusted for age, sex, and socioeconomic status, compared with standard hours (35-40 h per week), working long hours (≥55 h per week) was associated with an increase in risk of incident coronary heart disease (relative risk [RR] 1·13, 95% CI 1·02-1·26; p=0·02) and incident stroke (1·33, 1·11-1·61; p=0·002). The excess risk of stroke remained unchanged in analyses that addressed reverse causation, multivariable adjustments for other risk factors, and different methods of stroke ascertainment (range of RR estimates 1·30-1·42). We recorded a dose-response association for stroke, with RR estimates of 1·10 (95% CI 0·94-1·28; p=0·24) for 41-48 working hours, 1·27 (1·03-1·56; p=0·03) for 49-54 working hours, and 1·33 (1·11-1·61; p=0·002) for 55 working hours or more per week compared with standard working hours (p_trend<0·0001).

Interpretation Employees who work long hours have a higher risk of stroke than those working standard hours; the association with coronary heart disease is weaker. These findings suggest that more attention should be paid to the management of vascular risk factors in individuals who work long hours. 

Dyslipidemia and Diabetic Macular Edema: A Systematic Review and Meta-Analysis

Topic: A systematic review and meta-analysis of dyslipidemia and diabetic macular edema (DME). 

Clinical Relevance: Diabetic macular edema causes impairment of vision in patients with diabetes, and dyslipidemia has been reported as a risk factor for its development. A systematic review with a meta-analysis was undertaken to examine the evidence of an association between dyslipidemia and DME. 

Methods: We defined eligibility criteria as randomized controlled trials (RCTs) and cohort, case-control, and cross-sectional studies reporting on the relationship between blood lipid levels and DME. We performed a literature search in MEDLINE, PubMed, and Embase from inception to September 2014. We used the NewcastleeOttawa scale to assess the quality of case-control, cross-sectional, and cohort studies, and the Cochrane risk of bias tool for RCTs. 

Results: The search strategy identified 4959 publications. After screening, we selected 21 articles for review (5 cross-sectional, 5 cohort, 7 case-control, and 4 RCTs). Meta-analysis of case-control studies revealed that mean levels of total serum cholesterol (TC), low-density lipoproteins (LDLs), and serum triglycerides (TGs) were significantly higher in patients with DME compared with those without DME (TC: 30.08; 95% confidence interval [CI], 21.14e39.02; P < 0.001; LDL: 18.62; 95% CI, 5.80e31.43; P < 0.05; TG: 24.82; 95% CI, 9.21e40.42; P < 0.05). Meta-analysis of RCTs did not show significant risk in worsening of hard exudates and severity of DME in the lipid-lowering group compared with placebo (hard exudates: relative risk, 1.00; 95% CI, 0.47e2.11; P ¼ 1.00; DME: relative risk, 1.18; 95% CI, 0.75e1.86; P ¼ 0.48). 

Conclusions: Despite evidence from the cohort studies and meta-analysis of the case-control studies suggesting a strong relationship between lipid levels and DME, this was not confirmed by the meta-analysis that included only prospective RCTs. Therefore, given the significant public health relevance of the topic, the relationship between lipid levels and DME deserves further investigation.

Natural Killer Cell-like signature observed in locally advanced rectal cancer after neoadjuvant chemoradiotherapy in a tumour regression grade dependent manner

Background: Around 10-15% of patients with locally advanced rectal cancer (LARC) undergo a pathologically complete response (TRG4) to neoadjuvant chemoradiotherapy; the rest of patients exhibit a spectrum of tumour regression (TRG1-3). Understanding therapy-related genomic alterations may help us to identify underlying biology or novel targets associated with response that could increase the efficacy of therapy in patients that do not benefit from the current standard of care.
Methods: 48 FFPE rectal cancer biopsies and matched resections were analysed using the WG-DASL HumanHT-12_v4 Beadchip array on the illumina iScan. Bioinformatic analysis was conducted in Partek genomics suite and R studio. Limma and glmnet packages were used to identify genes differentially expressed between tumour regression grades. Validation of microarray results will be carried out using IHC, RNAscope and RT-PCR.
Results: Immune response genes were observed from supervised analysis of the biopsies which may have predictive value. Differential gene expression from the resections as well as pre and post therapy analysis revealed induction of genes in a tumour regression dependent manner. Pathway mapping and Gene Ontology analysis of these genes suggested antigen processing and natural killer mediated cytotoxicity respectively. The natural killer-like gene signature was switched off in non-responders and on in the responders. IHC has confirmed the presence of Natural killer cells through CD56+ staining.
Conclusion: Identification of NK cell genes and CD56+ cells in patients responding to neoadjuvant chemoradiotherapy warrants further investigation into their association with tumour regression grade in LARC. NK cells are known to lyse malignant cells and determining whether their presence is a cause or consequence of response is crucial. Interrogation of the cytokines upregulated in our NK-like signature will help guide future in vitro models.

A HuGE Review and Meta-Analyses of Genetic Associations in New Onset Diabetes after Kidney Transplantation

PURPOSE: New onset diabetes after transplantation (NODAT) is a serious complication following solid organ transplantation. There is a genetic contribution to NODAT and we have conducted comprehensive meta-analysis of available genetic data in kidney transplant populations.

METHODS: Relevant articles investigating the association between genetic markers and NODAT were identified by searching PubMed, Web of Science and Google Scholar. SNPs described in a minimum of three studies were included for analysis using a random effects model. The association between identified variants and NODAT was calculated at the per-study level to generate overall significance values and effect sizes.

RESULTS: Searching the literature returned 4,147 citations. Within the 36 eligible articles identified, 18 genetic variants from 12 genes were considered for analysis. Of these, three were significantly associated with NODAT by meta-analysis at the 5% level of significance; CDKAL1 rs10946398 p = 0.006 OR = 1.43, 95% CI = 1.11-1.85 (n = 696 individuals), KCNQ1 rs2237892 p = 0.007 OR = 1.43, 95% CI = 1.10-1.86 (n = 1,270 individuals), and TCF7L2 rs7903146 p = 0.01 OR = 1.41, 95% CI = 1.07-1.85 (n = 2,967 individuals).

CONCLUSION: Evaluating cumulative evidence for SNPs associated with NODAT in kidney transplant recipients has revealed three SNPs associated with NODAT. An adequately powered, dense genome-wide association study will provide more information using a carefully defined NODAT phenotype.

HDAC1 and HDAC2 independently predict mortality in hepatocellular carcinoma by a competing risk regression model in a Southeast Asian population

Histone deacetylases (HDACs) are enzymes involved in transcriptional repression. We aimed to examine the significance of HDAC1 and HDAC2 gene expression in the prediction of recurrence and survival in 156 patients with hepatocellular carcinoma (HCC) among a South East Asian population who underwent curative surgical resection in Singapore. We found that HDAC1 and HDAC2 were upregulated in the majority of HCC tissues. The presence of HDAC1 in tumor tissues was correlated with poor tumor differentiation. Notably, HDAC1 expression in adjacent non-tumor hepatic tissues was correlated with the presence of satellite nodules and multiple lesions, suggesting that HDAC1 upregulation within the field of HCC may contribute to tumor spread. Using competing risk regression analysis, we found that increased cancer-specific mortality was significantly associated with HDAC2 expression. Mortality was also increased with high HDAC1 expression. In the liver cancer cell lines, HEP3B, HEPG2, PLC5, and a colorectal cancer cell line, HCT116, the combined knockdown of HDAC1 and HDAC2 increased cell death and reduced cell proliferation as well as colony formation. In contrast, knockdown of either HDAC1 or HDAC2 alone had minimal effects on cell death and proliferation. Taken together, our study suggests that both HDAC1 and HDAC2 exert pro-survival effects in HCC cells, and the combination of isoform-specific HDAC inhibitors against both HDACs may be effective in targeting HCC to reduce mortality.

Illness perceptions and coping in physical health conditions: A meta-analysis

Objective: There is a considerable body of research linking elements of Leventhal’s Common Sense Model (CSM) to emotional well-being/distress outcomes among people with physical illness. The present study aims to consolidate this literature and examine the evidence for the role of coping strategies within this literature.
Methods: A systematic review was conducted where the outcomes of interest were: depression, anxiety and quality of life. A total of 1050 articles were identified and 31 articles were considered eligible to be included in the review.
Results: Across a range of illnesses, perceptions of consequences of the illness and emotional representations were consistently the illness perceptions with the strongest relationship with the outcomes. Coping variables tend to be stronger predictors of outcomes than the illness perception variables. The evidence for the mediating effect of coping was inconsistent.
Conclusions: Illness perceptions and coping have an important role to play in the explanation of distress outcomes across a range of physical health conditions. However, some clarity about the theoretical position of coping in relation to illness perceptions, and further longitudinal work is needed if we are to apply this information to the design of interventions for the improvement of psychological health among people with physical health conditions.