Analysis of melanoma onset: assessing familial aggregation by using estimating equations and fitting variance components via Bayesian random effects models.

We investigate whether relative contributions of genetic and shared environmental factors are associated with an increased risk in melanoma. Data from the Queensland Familial Melanoma Project comprising 15,907 subjects arising from 1912 families were analyzed to estimate the additive genetic, common and unique environmental contributions to variation in the age at onset of melanoma. Two complementary approaches for analyzing correlated time-to-onset family data were considered: the generalized estimating equations (GEE) method in which one can estimate relationship-specific dependence simultaneously with regression coefficients that describe the average population response to changing covariates; and a subject-specific Bayesian mixed model in which heterogeneity in regression parameters is explicitly modeled and the different components of variation may be estimated directly. The proportional hazards and Weibull models were utilized, as both produce natural frameworks for estimating relative risks while adjusting for simultaneous effects of other covariates. A simple Markov Chain Monte Carlo method for covariate imputation of missing data was used and the actual implementation of the Bayesian model was based on Gibbs sampling using the free ware package BUGS. In addition, we also used a Bayesian model to investigate the relative contribution of genetic and environmental effects on the expression of naevi and freckles, which are known risk factors for melanoma.     

Heterogeneity analysis of breast cancer families by using age at onset as a covariate.

An extension of the usual mixture model of heterogeneity (two family types, one with and one without linkage) is proposed by introducing age at onset as a covariate. The extended model defines age-dependent penetrances where the exact parametrization of age-at-onset distributions depends on the given genotype and family type (linked or unlinked). This extension was applied to breast cancer families. We postulated that the mean age at onset in individuals affected by the linked gene was lower than the mean age at onset in all other affected individuals. Linkage heterogeneity for breast cancer families was detected at a significance level of .003.     

Familial predisposition to cancer and age at onset of disease in randomly selected cancer patients

Incidence of malignancy among close relatives was used to evaluate the relationship of early age at diagnosis and familial cancer predisposition in a general population of cancer patients. The occurrence of cancer and other conditions in families of more than 1,350 randomly selected patients with a wide variety of malignancies was ascertained. Each patient was assigned to one of four study groups based on comparison of his age at diagnosis with the distribution of ages at diagnosis for his cancer site compiled by the Third National Cancer Survey. These groups consisted of patients whose ages at diagnosis were in: (1) the lowest decile, (2) the median decile, (3) above the median decile, and (4) between the lowest and median deciles. Person-years and calendar time at risk were calculated for first-degree relatives in each group. The numbers of cancers expected among these relatives were calculated using age- and time-specific incidence rates of a standard population. Statistical analysis of (1) the numbers of reported vs. expected cancers in relatives and (2) the numbers of families reporting cancer in parents or siblings of patients showed that a familial tendency to develop cancer exists in this randomly selected population of cancer patients, regardless of age at onset of malignancy in the proband. Conversely, early age at diagnosis of cancer may indicate genetic predisposition to malignancy only in exceptional cases.     

Logistic regression for clustered binary data in proband studies with application to familial aggregation of sleep disorders.

Estimation is considered for the class of conditional logistic regression models for clustered binary data proposed by Qu et al. (Communications in Statistics, Series A 16, 3447-3476, 1987) when clusters are sampled on the basis of the outcome for one or more cluster members. The problem is suggested by data from a study designed to investigate familial aggregation of sleep disorders. After appropriate consideration of the mode of ascertainment of "cases" and "controls," it is shown that the model is preserved under this form of sampling, and a method of estimation is presented. The inconsistency of two alternative methods is demonstrated, and an example is provided.     

Estimation of morbid risk and age at onset with missing information.

Investigators of genetic illnesses are currently employing life-table techniques to estimate the lifetime risk of disease and the age-at-onset distribution. This methodology assumes that onset ages are known for affected individuals and that censoring ages are known for unaffected individuals. We extend these methods to incorporate affected individuals with unknown onset ages and unaffected persons with unknown censoring ages and illustrate how conventional life-table methods can produce seriously biased estimates, particularly of lifetime risk. The methodology is not restricted to genetic illnesses and can be applied to more complex illnesses with unknown etiology. We present an example for Huntington disease, which is generally assumed to be a Mendelian autosomal dominant disease, yielding estimates of lifetime risk of .503 +/- .70 and mean onset age of 47.7 +/- 3.1 years for offspring with a single affected parent. When conventional life-table techniques are employed, these estimates are .238 +/- .032 and 43.2 +/- 2.2.     

Analyzing the relationship between age at onset and risk to relatives.

Correlations in age at onset between relatives affect risk to relatives of a given age. Either an increase or a decrease in risk may be observed for a relative of a proband, according to whether there is a causal relationship between liability to disease and age at onset. Likelihood formulas are given for pairs of relatives under a number of different sampling schemes, and it is shown how data collected from relatives enable maximum-likelihood estimation of parameters of a linear model relating disease liability and age at onset. A genotype-environment extension of this model was fitted to data on age at onset for schizophrenia that were obtained from the National Academy of Sciences-National Research Council Twin Registry. Age at onset is correlated between twins, but this correlation appears to be associated with factors that are separate from those which affect liability to disease. However, even this relatively large sample of twins is too small to draw firm conclusions about any causal relationship between disease liability and onset.     

Theory of aggregation in solution. I. General equations and application to the stacking of bases,nucleosides, etc.

Earlier theoretical work on aggregation in imperfect gases and in solutions of protein molecules is reformulated here in such a way as to be applicable to a number of aggregation systems in solution. The theory is most easily developed for an “osmotic” solution. It is exact, in principle, and in particular does not treat the aggregates as an ideal mixture as is usually done. If osmotic virial coefficients are diffcult to measure in a particular case, it is then necessary to relate these coefficients to properties of the more conventional “constant pressure” solution. An important special case is used as an example in the last three sections: The stacking of bases, nucleosides, etc., is represented by a statistical mecchanical model in which the aggregates resemble vertical stacks of checkers.     

Boosting GWAS using biological networks: A study on susceptibility to familial breast cancer

Genome-wide association studies (GWAS) explore the genetic causes of complex diseases. However, classical approaches ignore the biological context of the genetic variants and genes under study. To address this shortcoming, one can use biological networks, which model functional relationships, to search for functionally related susceptibility loci. Many such network methods exist, each arising from different mathematical frameworks, pre-processing steps, and assumptions about the network properties of the susceptibility mechanism. Unsurprisingly, this results in disparate solutions. To explore how to exploit these heterogeneous approaches, we selected six network methods and applied them to GENESIS, a nationwide French study on familial breast cancer. First, we verified that network methods recovered more interpretable results than a standard GWAS. We addressed the heterogeneity of their solutions by studying their overlap, computing what we called the consensus. The key gene in this consensus solution was COPS5, a gene related to multiple cancer hallmarks. Another issue we observed was that network methods were unstable, selecting very different genes on different subsamples of GENESIS. Therefore, we proposed a stable consensus solution formed by the 68 genes most consistently selected across multiple subsamples. This solution was also enriched in genes known to be associated with breast cancer susceptibility (BLM, CASP8, CASP10, DNAJC1, FGFR2, MRPS30, and SLC4A7, P-value = 3 × 10⁻⁴). The most connected gene was CUL3, a regulator of several genes linked to cancer progression. Lastly, we evaluated the biases of each method and the impact of their parameters on the outcome. In general, network methods preferred highly connected genes, even after random rewirings that stripped the connections of any biological meaning. In conclusion, we present the advantages of network-guided GWAS, characterize their shortcomings, and provide strategies to address them. To compute the consensus networks, implementations of all six methods are available at https://github.com/hclimente/gwas-tools.     

Expected estimating equations for missing data, measurement error, and misclassification, with application to longitudinal nonignorable missing data

Summary .   Missing data, measurement error, and misclassification are three important problems in many research fields, such as epidemiological studies. It is well known that missing data and measurement error in covariates may lead to biased estimation. Misclassification may be considered as a special type of measurement error, for categorical data. Nevertheless, we treat misclassification as a different problem from measurement error because statistical models for them are different. Indeed, in the literature, methods for these three problems were generally proposed separately given that statistical modeling for them are very different. The problem is more challenging in a longitudinal study with nonignorable missing data. In this article, we consider estimation in generalized linear models under these three incomplete data models. We propose a general approach based on expected estimating equations (EEEs) to solve these three incomplete data problems in a unified fashion. This EEE approach can be easily implemented and its asymptotic covariance can be obtained by sandwich estimation. Intensive simulation studies are performed under various incomplete data settings. The proposed method is applied to a longitudinal study of oral bone density in relation to body bone density.     

Case-control and case-only designs with genotype and family history data: estimating relative risk, residual familial aggregation, and cumulative risk

In case-control studies of inherited diseases, participating subjects (probands) are often interviewed to collect detailed data about disease history and age-at-onset information in their family members. Genotype data are typically collected from the probands, but not from their relatives. In this article, we introduce an approach that combines case-control analysis of data on the probands with kin-cohort analysis of disease history data on relatives. Assuming a marginally specified multivariate survival model for joint risk of disease among family members, we describe methods for estimating relative risk, cumulative risk, and residual familial aggregation. We also describe a variation of the methodology that can be used for kin-cohort analysis of the family history data from a sample of genotyped cases only. We perform simulation studies to assess performance of the proposed methodologies with correct and mis-specified models for familial aggregation. We illustrate the proposed methodologies by estimating the risk of breast cancer from BRCA1/2 mutations using data from the Washington Ashkenazi Study.     

Assessing agreement between multiple raters with missing rating information, applied to breast cancer tumour grading

Background

We consider the problem of assessing inter-rater agreement when there are missing data and a large number of raters. Previous studies have shown only ‘moderate’ agreement between pathologists in grading breast cancer tumour specimens. We analyse a large but incomplete data-set consisting of 24177 grades, on a discrete 1–3 scale, provided by 732 pathologists for 52 samples.

Methodology/Principal Findings

We review existing methods for analysing inter-rater agreement for multiple raters and demonstrate two further methods. Firstly, we examine a simple non-chance-corrected agreement score based on the observed proportion of agreements with the consensus for each sample, which makes no allowance for missing data. Secondly, treating grades as lying on a continuous scale representing tumour severity, we use a Bayesian latent trait method to model cumulative probabilities of assigning grade values as functions of the severity and clarity of the tumour and of rater-specific parameters representing boundaries between grades 1–2 and 2–3. We simulate from the fitted model to estimate, for each rater, the probability of agreement with the majority. Both methods suggest that there are differences between raters in terms of rating behaviour, most often caused by consistent over- or under-estimation of the grade boundaries, and also considerable variability in the distribution of grades assigned to many individual samples. The Bayesian model addresses the tendency of the agreement score to be biased upwards for raters who, by chance, see a relatively ‘easy’ set of samples.

Conclusions/Significance

Latent trait models can be adapted to provide novel information about the nature of inter-rater agreement when the number of raters is large and there are missing data. In this large study there is substantial variability between pathologists and uncertainty in the identity of the ‘true’ grade of many of the breast cancer tumours, a fact often ignored in clinical studies.     

Factors related to onset age of Huntington disease.

One prominent feature of Huntington disease (HD) is the variable age at which the characteristic neurological or psychiatric symptoms appear. Ages of manifestation varying from 4 to 65 years are found in a sample of 95 HD pedigrees compiled since 1968 from the Southeastern United States. Significant parent-child correlations of age of onset indicate consistency of onset age within nuclear families. However, an average intrafamily range of 9 years and an average intrapedigree range of 12 years reveal substantial variability of onset age within these groups. Of the nine cases of juvenile-onset HD identified in this sample, seven were of paternal descent. The preponderance of juvenile patients inheriting the HD gene from a father confirms similar findings from other studies. In addition, a trend toward earlier onset in all offspring of paternal transmission suggests that the juvenile-onset phenomenon is only the tail of a shift in the curve of onset ages for this group. A trend toward earlier onset in successive generations was noted. This "anticipation" may reflect the finding that persons of early onset in prior generations are selectively nonreproductive as a result of manifestation of the disorder. By identifying familial factors influencing onset age of HD, it may be possible to more effectively evaluate environmental factors that influence the onset of the disorder.     

Neonatal necrotizing enterocolitis: the relation of age at the time of onset to prognosis.

A study was undertaken to evaluate the predisposing factors, age at the time of onset and prognosis of neonatal necrotizing enterocolitis in 62 patients treated in a neonatal intensive care unit during a 5-year period (1974-78). Because of a peak frequency during the first week of life, the cases were divided into those of early-onset illness (appearing within the first week of life) and those of late-onset illness (appearing after this week). The main differences between the two groups were in the age at the time of the first enteric feeding (1.6 +/- 0.9 d v. 40 +/- 2.4 d [mean +/- one standard deviation]; P less than 0.001) and the interval between this feeding and the onset of symptoms (3.0 +/- 1.5 d v. 10.6 +/- 6.0 d; P less than 0.01). Furthermore, the early-onset illness was more severe, more often necessitating surgical intervention and carrying a higher mortality, than the late-onset illness. Thus, this study demonstrates that there are two forms of neonatal necrotizing enterocolitis, differing in time of appearance after birth as well as in severity and prognosis.     

Molecular mechanisms of platelet activation and aggregation induced by breast cancer cells

Tumor cell-induced platelet aggregation represents a critical process both for successful metastatic spread of the tumor and for the development of thrombotic complications in cancer patients. To get further insights into this process, we investigated and compared the molecular mechanisms of platelet aggregation induced by two different breast cancer cell lines (MDA-MB-231 and MCF7) and a colorectal cancer cell line (Caco-2). All the three types of cancer cells were able to induce comparable platelet aggregation, which, however, was observed exclusively in the presence of CaCl₂ and autologous plasma. Aggregation was supported both by fibrinogen binding to integrin αIIbβ3 as well as by fibrin formation, and was completely prevented by the serine protease inhibitor PPACK. Platelet aggregation was preceded by generation of low amounts of thrombin, possibly through tumor cells-expressed tissue factor, and was supported by platelet activation, as revealed by stimulation of phospholipase C, intracellular Ca²⁺ increase and activation of Rap1b GTPase. Pharmacological inhibition of phospholipase C, but not of phosphatidylinositol 3-kinase or Src family kinases prevented tumor cell-induced platelet aggregation. Tumor cells also induced dense granule secretion, and the stimulation of the P2Y12 receptor by released ADP was found to be necessary for complete platelet aggregation. By contrast, prevention of thromboxane A₂ synthesis by aspirin did not alter the ability of all the cancer cell lines analyzed to induce platelet aggregation. These results indicate that tumor cell-induced platelet aggregation is not related to the type of the cancer cells or to their metastatic potential, and is triggered by platelet activation and secretion driven by the generation of small amount of thrombin from plasma and supported by the positive feedback signaling through secreted ADP.     

Absence of linkage to the ataxia telangiectasia locus in familial breast cancer

Heterozygotes for ataxia-telangiectasia (AT) are known to have an increased risk of breast cancer. The gene (or genes) responsible for almost all cases of AT has been localised to chromosome 11q by genetic linkage analysis. To examine the possibility that AT heterozygosity may account for a substantial proportion of familial breast cancer, we have typed five markers on chromosome 11q in 16 breast cancer families. We have found no evidence for linkage between breast cancer and chromosome 11q markers and conclude that the contribution of AT to familial breast cancer is likely to be minimal.