Heterogeneity in meta-analysis: a comprehensive overview.

In recent years, meta-analysis has evolved to a critically important field of Statistics, and has significant applications in Medicine and Health Sciences. In this work we briefly present existing methodologies to conduct meta-analysis along with any discussion and recent developments accompanying them. Undoubtedly, studies brought together in a systematic review will differ in one way or another. This yields a considerable amount of variability, any kind of which may be termed heterogeneity. To this end, reports of meta-analyses commonly present a statistical test of heterogeneity when attempting to establish whether the included studies are indeed similar in terms of the reported output or not. We intend to provide an overview of the topic, discuss the potential sources of heterogeneity commonly met in the literature and provide useful guidelines on how to address this issue and to detect heterogeneity. Moreover, we review the recent developments in the Bayesian approach along with the various graphical tools and statistical software that are currently available to the analyst. In addition, we discuss sensitivity analysis issues and other approaches of understanding the causes of heterogeneity. Finally, we explore heterogeneity in meta-analysis for time to event data in a nutshell, pointing out its unique characteristics.

Two-stage meta-analysis of survival data from individual participants using percentile ratios.

Methods for individual participant data meta-analysis of survival outcomes commonly focus on the hazard ratio as a measure of treatment effect. Recently, Siannis et al. (2010, Statistics in Medicine 29:3030-3045) proposed the use of percentile ratios as an alternative to hazard ratios. We describe a novel two-stage method for the meta-analysis of percentile ratios that avoids distributional assumptions at the study level.

Sensitivity analysis for multiple right censoring processes: investigating mortality in psoriatic arthritis.

In a mortality study in psoriatic arthritis (PsA), censored observations are generated from the fact that patients fail to attend their scheduled appointments at the clinic. As a result, more than one types of right-censored observations are available. In survival analysis, the treatment of censored observations remains a concern. The assumption of ignorable censoring, although in many cases justified, is an important assumption made often for convenience rather than any other reason. In this paper we discuss a semi-parametric model for the analysis of survival data, where sensitivity analysis on quantities of interest can be performed when small levels of association between the failure and the censoring processes are assumed. Extension of the model allows for the presence of more than one censoring processes, where one may be characterized as ignorable and the other not. This model will be used to analyze the PsA mortality data, where a sensitivity analysis on parameters can be done under the assumption of non-ignorable censoring. Sensitivity analysis will also be performed in the presence of two censoring processes, one of which will be classified as non-ignorable.

A semi-competing risks model for data with interval-censoring and informative observation: an application to the MRC cognitive function and ageing study.

Semi-competing risks data occur frequently in medical research when interest is in simultaneous modelling of two or more processes, one of which may censor the others. We consider the analysis of semi-competing risks data in the presence of interval-censoring and informative loss-to-followup. The work is motivated by a data set from the MRC UK Cognitive Function and Ageing Study, which we use to model two processes, cognitive impairment and death. Analysis is carried out using a multi-state model, which is an extension of that used by Siannis et al. (Statist. Med. 2007; 26:426­442) to model semi-competing risks data with exact transition times, to data which is interval-censored. Model parameters are estimated using maximum likelihood. The role of a sensitivity parameter k, which influences the nature of informative censoring, is explored.

Modelling competing risks data with missing cause of failure.

When competing risks data arise, information on the actual cause of failure for some subjects might be missing. Therefore, a cause-specific proportional hazards model together with multiple imputation (MI) methods have been used to analyze such data. Modelling the cumulative incidence function is also of interest, and thus we investigate the proportional subdistribution hazards model (Fine and Gray model) together with MI methods as a modelling approach for competing risks data with missing cause of failure. Possible strategies for analyzing such data include the complete case analysis as well as an analysis where the missing causes are classified as an additional failure type. These approaches, however, may produce misleading results in clinical settings. In the present work we investigate the bias of the parameter estimates when fitting the Fine and Gray model in the above modelling approaches. We also apply the MI method and evaluate its comparative performance under various missing data scenarios. Results from simulation experiments showed that there is substantial bias in the estimates when fitting the Fine and Gray model with naive techniques for missing data, under missing at random cause of failure. Compared to those techniques the MI-based method gave estimates with much smaller biases and coverage probabilities of 95 per cent confidence intervals closer to the nominal level. All three methods were also applied on real data modelling time to AIDS or non-AIDS cause of death in HIV-1 infected individuals.

Assessment of somatic k-RAS mutations as a mechanism associated with resistance to EGFR-targeted agents: a systematic review and meta-analysis of studies in advanced non-small-cell lung cancer and metastatic colorectal cancer.

BACKGROUND: Somatic mutations of the k-RAS oncogene have been assessed as a mechanism of de-novo resistance to epidermal growth factor receptor (EGFR) tyrosine-kinase inhibition in patients with non-small-cell lung cancer (NSCLC), and to anti-EGFR monoclonal antibodies in patients with metastatic colorectal cancer (mCRC). The aim of this systematic review and meta-analysis was to assess if k-RAS mutations represent a candidate predictive biomarker for anti-EGFR-targeted therapeutic strategies in mCRC and NSCLC. METHODS: We systematically identified articles pertaining to k-RAS mutational status in patients with NSCLC treated with tyrosine-kinase inhibitors (TKI), and patients with mCRC treated with any anti-EGFR-based regimens. Eligible studies had to report complete responses (CR) and partial responses (PR), stratified by k-RAS mutational status. Potential between-study heterogeneity was accommodated by use of random-effects models for bivariable meta-analysis of sensitivity and specificity (the primary endpoints). The positive and negative likelihood ratios (+LR and -LR, respectively) of k-RAS mutations for predicting an absence of response were considered as secondary endpoints and were calculated by use of pooled estimates for sensitivity and specificity. FINDINGS: Of 252 retrieved manuscripts, 17 were deemed eligible for the NSCLC meta-analysis (165 of 1008 patients with mutated k-RAS). The presence of k-RAS mutations was significantly associated with an absence of response to TKIs (sensitivity=0.21 [95% CI 0.16-0.28], specificity=0.94 [0.89-0.97]; +LR=3.52; -LR=0.84). Of 68 retrieved manuscripts reporting on anti-EGFR monoclonal-antibody-based treatment of mCRC, eight studies were deemed eligible for the final analysis (306 of 817 patients with mutated k-RAS). The presence of k-RAS mutations was significantly associated with an absence of response to anti-EGFR monoclonal-antibody-based treatments (sensitivity=0.47 [0.43-0.52]; specificity=0.93 [0.83-0.97]; +LR=6.82; -LR=0.57). INTERPRETATION: This analysis provides empirical evidence that k-RAS mutations are highly specific negative predictors of response (de-novo resistance) to single-agent EGFR TKIs in advanced NSCLC; and similarly to anti-EGFR monoclonal antibodies alone or in combination with chemotherapy in patients with mCRC. The low sensitivity and relatively high -LR of k-RAS mutations for determining non-responsiveness clearly shows that additional mechanisms of resistance to EGFR inhibitors exist.

Preservation of the mitral valve apparatus: evidence synthesis and critical reappraisal of surgical techniques.

Sub-valvular apparatus preservation after mitral valve replacement is not a new concept, yet to date there has been no quantification of its clinical effectiveness as a procedure and no consensus as to which surgical preservation technique should be adopted to achieve the best immediate and midterm clinical outcomes. This systematic review of current available literature aims to use an evidence synthesis and meta-analytic approach to compare outcomes following replacement of the mitral valve with (MVR-P) or without preservation (MVR-NP) of its apparatus. It considers all the relevant anatomical, experimental, echocardiographic, and clinical studies published in the literature and appraises all reported mitral valve sub-valvular apparatus preservation techniques. The results of this review strongly suggest that MVR-P is superior to MVR-NP with regards to the incidence of early postoperative low-cardiac output requiring inotropic support, and early or mid-term survival. They also suggest that the operative decision should be individualised based on patient's anatomy, pathology and ventricular function and therefore surgeons should be familiar with more than one surgical preservation technique. Finally, this paper highlights the need for further high quality research focusing particularly on the long-term assessment of quality of life and health utility following MVR-P.

Folate pathway enzyme gene polymorphisms and the efficacy and toxicity of methotrexate in psoriatic arthritis.

OBJECTIVE: To determine the association between folate pathway gene polymorphisms and the effectiveness, toxicity, and drug survival of methotrexate (MTX) in psoriatic arthritis (PsA). METHODS: Data were obtained from a longitudinal cohort of PsA patients evaluated according to a standard protocol. Data on duration of drug therapy, dose, side effects, and reasons for discontinuation are systematically recorded. Patients treated with MTX after clinic admission who had > or = 3 swollen joints prior to initiating MTX therapy were selected for evaluation of effectiveness. Response to MTX treatment was assessed at 6 months. Data from all patients treated in the clinic with MTX were used in evaluation of toxicity and drug survival. The following single-nucleotide polymorphisms (SNP) were measured using the Sequenom platform: MTHFR 677C>T (rs1801133), MTHFR 1298A>C (rs1801131), DHFR -473T>C (rs1650697), DHFR 35289A>G (rs1232027), and RFC 80G>A (rs1051266). Fisher's exact test, logistic regression, and Cox proportional hazard analyses were used to determine association. RESULTS: Two hundred eighty-one patients were identified from the database. All patients were included in the analysis for side effects and drug survival, and 119 patients were included in the effectiveness analysis. The minor A allele of DHFR gene at +35289 was the only SNP demonstrating association with response to MTX therapy (OR 2.99, p = 0.02). Patients homozygous for the minor allele of MTHFR 677C/T (677TT) had more liver toxicity (Fisher exact test, p = 0.04). CONCLUSION: Polymorphisms of the DHFR gene may be associated with MTX efficacy. MTHFR 677TT may have a relationship with MTX-induced liver toxicity in PsA.

Somatic EGFR mutation and gene copy gain as predictive biomarkers for response to tyrosine kinase inhibitors in non-small cell lung cancer.

PURPOSE: The aim of this systematic review and meta-analysis was to characterize common EGFR molecular aberrations as potential predictive biomarkers for response to monotherapy with tyrosine kinase inhibitors (TKI) in non-small cell lung cancer (NSCLC). EXPERIMENTAL DESIGN: We systematically identified articles investigating EGFR status [somatic mutational and gene copy aberrations (copy number)] in patients with NSCLC treated with TKIs. Eligible studies had to report complete and partial response rates stratified by EGFR status. We used random effects models for bivariable meta-analysis of sensitivity and specificity; positive and negative likelihood ratios (+LR and -LR, respectively) were also calculated and were considered as secondary end points. RESULTS: Among 222 retrieved articles, 59 were considered eligible for the somatic EGFR mutation meta-analysis (1,020 mutations among 3,101 patients) and 21 were considered eligible for the EGFR gene copy number meta-analysis (542 gene gain among 1,539 patients). EGFR mutations were predictive of response to single-agent TKIs [sensitivity, 0.78; 95% confidence interval (95% CI), 0.74-0.82; specificity, 0.86; 95% CI, 0.82-0.89; +LR, 5.6; -LR, 0.25]. EGFR gene gain was also associated with response to TKIs, albeit with lower sensitivity and specificity. In subgroup analysis, the only recognized trend was for a higher predictive value in Whites compared with East Asians for both mutation and gene copy number. CONCLUSION: This analysis provides empirical evidence that EGFR mutations are sensitive and specific predictors of response to single-agent epidermal growth factor receptor TKIs in advanced NSCLC. The diagnostic performance of mutations seems better than that of EGFR gene gain.

Trastuzumab in the adjuvant treatment of early-stage breast cancer: a systematic review and meta-analysis of randomized controlled trials.

BACKGROUND: We performed a systematic review and meta-analysis to compare treatment outcomes for human epidermal growth factor receptor (HER)-2-positive breast cancer patients receiving adjuvant chemotherapy with or without trastuzumab. METHODS: We identified randomized clinical trials comparing adjuvant chemotherapy with or without trastuzumab in patients with resectable breast cancer. Fixed-effects meta-analysis was used to combine data. RESULTS: Five eligible trials were identified, reporting outcomes on 13,493 women. Fixed-effects analysis showed disease-free survival to be superior for trastuzumab-treated patients (risk ratio [RR], 0.62; 95% confidence interval [CI], 0.56-0.68). Superiority was also observed for patients receiving trastuzumab with respect to mortality (RR, 0.66; 95% CI, 0.57-0.77), locoregional recurrence (RR, 0.58; 95% CI, 0.43-0.77), and distant recurrence (RR, 0.60; 95% CI, 0.52-0.68). Patients receiving trastuzumab with chemotherapy had a higher risk for congestive heart failure (RR, 7.60; 95% CI, 4.07-14.18) and left ventricular ejection fraction decline (RR, 2.09; 95% CI, 1.84-2.37). A higher risk for central nervous system metastasis as the first recurrence event (RR, 1.60; 95% CI, 1.06-2.40) was also noted in patients receiving trastuzumab. CONCLUSIONS: The use of trastuzumab should be considered an integral part of the adjuvant therapy of HER-2-positive breast cancer patients.

Does preservation of the sub-valvular apparatus during mitral valve replacement affect long-term survival and quality of life? A Microsimulation Study.

BACKGROUND: Techniques to preserve the sub-valvular apparatus in order to reduce morbidity and mortality following mitral valve replacement have been frequently reported. However, it is uncertain what impact sub-valvular apparatus preservation techniques have on long-term outcomes following mitral valve replacement. This study investigated the effect of sub-valvular apparatus preservation on long-term survival and quality of life following mitral valve replacement. METHODS: A microsimulation model was used to compare long-term survival and quality-adjusted life years following mitral valve replacement after conventional valve replacement and sub-valvular apparatus preservation. Probabilistic sensitivity analysis and alternative analysis were performed to investigate uncertainty associated with the results. RESULTS: Our Analysis suggests that patients survive longer if the sub-valvular apparatus are preserved (65.7% SD 1.5%, compared with 58.1% SD 1.6% at 10 years). The quality adjusted life years gained over a 10 year period where also greater after sub-valvular apparatus preservation. (6.54 QALY SD 0.07 QALY, compared with 5.61 QALY, SD 0.07 QALY). The superiority of preservation techniques was insensitive to patient age, parameter or model uncertainty. CONCLUSION: This study suggests that long-term outcomes may be improved when the sub-valvular apparatus are preserved. Given the lack of empirical data further research is needed to investigate health-related quality of life after mitral valve replacement, and to establish whether outcomes differ between preservation techniques.

Sensitivity analysis for informative censoring in parametric survival models.

Most statistical methods for censored survival data assume there is no dependence between the lifetime and censoring mechanisms, an assumption which is often doubtful in practice. In this paper we study a parametric model which allows for dependence in terms of a parameter delta and a bias function B(t, theta). We propose a sensitivity analysis on the estimate of the parameter of interest for small values of delta. This parameter measures the dependence between the lifetime and the censoring mechanisms. Its size can be interpreted in terms of a correlation coefficient between the two mechanisms. A medical example suggests that even a small degree of dependence between the failure and censoring processes can have a noticeable effect on the analysis.

HLA markers for susceptibility and expression in scleroderma.

OBJECTIVE: Reported associations between HLA alleles and both susceptibility to and features of scleroderma have been conflicting. Our objective was (1) to determine the role of HLA alleles in the susceptibility to scleroderma; and (2) to determine the role of HLA alleles in various aspects of disease expression. METHODS: Consecutive patients were followed in the scleroderma clinic between 1996 and 1998. Clinical data were obtained through chart review. Healthy volunteers as well as cadaveric donors served as controls. Molecular HLA typing was performed (polymerase chain reaction/sequence-specific oligonucleotides). Statistical analysis included Fisher's exact test and multivariate analyses, using logistic and linear regression models. RESULTS: Ninety-five Caucasian patients (75 women, 20 men, age 43.9 yrs, disease duration 11.9 yrs) with scleroderma and 416 controls were studied. HLA-DRB1*01 and HLA-DRB1*11 were associated with susceptibility to scleroderma, whereas HLA-DRB1*07 was protective. HLA-A*30 and HLA-A*32 were also associated with susceptibility to scleroderma, while HLA-B*57 and HLA-Cw*14 were protective. HLA-B*62 and HLA-DRB1*07 had a significant correlation with the presence of diffuse skin involvement in both univariate and multivariate analyses. HLA-DRB1*11 was associated with high skin score values, while lower values were related to the presence of HLA-Cw*14 and HLA-DQB1*06. Both alleles retained significance in a linear regression model. High skin score values were related to the absence of anticentromere antibodies. Pulmonary fibrosis was associated with HLA-B*62 and HLA-Cw*0602, whereas pulmonary hypertension was associated with HLA-B*13 and HLA-B*65. CONCLUSION: HLA alleles play a role in susceptibility to scleroderma and its disease expression.

Applications of a parametric model for informative censoring.

In this article, we explore the use of a parametric model (for analyzing survival data) which is defined to allow sensitivity analysis for the presence of informative censoring. The dependence between the failure and the censoring processes is expressed through a parameter delta and a general bias function B(t, theta). We calculate the expectation of the potential bias due to informative censoring, which is an overall measure of how misleading our results might be if censoring is actually nonignorable. Bounds are also calculated for quantities of interest, e.g., parameter of the distribution of the failure process, which do not depend on the choice of the bias function for fixed delta. An application that relates to systematic lupus erythematosus data illustrates how additional information can result in reducing the uncertainty on estimates of the location parameter. Sensitivity analysis on a relative risk parameter is also explored.