Dynamic pseudo-observations: a robust approach to dynamic prediction in competing risks.

In this article, we propose a new approach to the problem of dynamic prediction of survival data in the presence of competing risks as an extension of the landmark model for ordinary survival data. The key feature of our method is the introduction of dynamic pseudo-observations constructed from the prediction probabilities at different landmark prediction times. They specifically address the issue of estimating covariate effects directly on the cumulative incidence scale in competing risks. A flexible generalized linear model based on these dynamic pseudo-observations and a generalized estimation equations approach to estimate the baseline and covariate effects will result in the desired dynamic predictions and robust standard errors. Our approach has a number of attractive features. It focuses directly on the prediction probabilities of interest, avoiding in this way complex modeling of cause-specific hazards or subdistribution hazards. As a result, it is robust against departures from these omnibus models. From a computational point of view an advantage of our approach is that it can be fitted with existing statistical software and that a variety of link functions and regression models can be considered, once the dynamic pseudo-observations have been estimated. We illustrate our approach on a real data set of chronic myeloid leukemia patients after bone marrow transplantation.

Dynamic prediction by landmarking in competing risks.

We propose an extension of the landmark model for ordinary survival data as a new approach to the problem of dynamic prediction in competing risks with time-dependent covariates. We fix a set of landmark time points tLM within the follow-up interval. For each of these landmark time points tLM , we create a landmark data set by selecting individuals at risk at tLM ; we fix the value of the time-dependent covariate in each landmark data set at tLM . We assume Cox proportional hazard models for the cause-specific hazards and consider smoothing the (possibly) time-dependent effect of the covariate for the different landmark data sets. Fitting this model is possible within the standard statistical software. We illustrate the features of the landmark modelling on a real data set on bone marrow transplantation.

Robust age at onset linkage analysis in nuclear families.

OBJECTIVE: Standard methods for linkage analysis ignore the phenotype of the parents when they are not genotyped. However, this information can be useful for gene mapping. In this paper we propose methods for age at onset genetic linkage analysis in sibling pairs, taking into account parental age at onset. METHODS: Two new score statistics are derived, one from an additive gamma frailty model and one from a log-normal frailty model. The score statistics are classical non-parametric linkage (NPL) statistics weighted by a function of the age at onset of the four family members. The weight depends on information from registries (age-specific incidences) and family studies (sib-sib and father-mother correlation). RESULTS: In order to investigate how age at onset of sibs and their parents affect the information for linkage analysis the weight functions were studied for rare and common disease models, realistic models for breast cancer and human lifespan. We studied the performance of the weighted NPL methods by simulations. As illustration, the score statistics were applied to the GAW12 data. The results show that it is useful to include parental age at onset information in genetic linkage analysis.

Meta-analysis of pairs of survival curves under heterogeneity: a Poisson correlated gamma-frailty approach.

We address the problem of meta-analysis of pairs of survival curves under heterogeneity. Starting point for the meta-analysis is a set of studies, each comparing the same two treatments, containing information about multiple survival outcomes. Under heterogeneity, we model the number of events using an extension of the Poisson correlated gamma-frailty model with serial within-arm and positive between-arm correlations. The parameters of the models are estimated following a two-stage estimation procedure. In the first stage the underlying hazards and between-study variance are estimated using the marginals, while a second stage is used to estimate both within-arm and between-arm correlations. The methodology is illustrated with an observational study on breast cancer.

Weighted statistics for aggregation and linkage analysis of human longevity in selected families: the Leiden Longevity Study.

Typically long-lived sibling pairs have been collected for linkage analysis of human longevity and information on life span of first-degree relatives is available to assess familial aggregation of life span. We propose a new weighted statistic for aggregation analysis, which tests for a relationship between a family history of excessive survival of the sibships of the long-lived pairs and the survival of their parents and their offspring. For linkage analysis, we derive a new weighted score statistic from a simple gamma frailty model, which assigns more weight to excessive long-lived pairs. We apply the methods to data from the Leiden Longevity Study, which consists of sibling pairs of age 90 years or above and their first-degree relatives. The pairs have been genotyped for microsatellite markers in a candidate region. Association was present between survival within the sibships and survival of the offspring, but not with the parental generation. For linkage analysis, weighting increased the value of the test statistic, but the result was not statistically significant. About the methods we conclude that the statistic for aggregation provides insight into clustering of life span and the statistic for linkage provides a new tool to include demographic information into the analysis.

A new serially correlated gamma-frailty process for longitudinal count data.

We describe a new multivariate gamma distribution and discuss its implication in a Poisson-correlated gamma-frailty model. This model is introduced to account for between-subjects correlation occurring in longitudinal count data. For likelihood-based inference involving distributions in which high-dimensional dependencies are present, it may be useful to approximate likelihoods based on the univariate or bivariate marginal distributions. The merit of composite likelihood is to reduce the computational complexity of the full likelihood. A 2-stage composite-likelihood procedure is developed for estimating the model parameters. The suggested method is applied to a meta-analysis study for survival curves.

Bivariate random effects meta-analysis of ROC curves.

Meta-analysis of receiver operating characteristic (ROC)-curve data is often done with fixed-effects models, which suffer many shortcomings. Some random-effects models have been proposed to execute a meta-analysis of ROC-curve data, but these models are not often used in practice. Straightforward modeling techniques for multivariate random-effects meta-analysis of ROC-curve data are needed. The 1st aim of this article is to present a practical method that addresses the drawbacks of the fixed-effects summary ROC (SROC) method of Littenberg and Moses. Sensitivities and specificities are analyzed simultaneously using a bivariate random-effects model. The 2nd aim is to show that other SROC curves can also be derived from the bivariate model through different characterizations of the estimated bivariate normal distribution. Thereby the authors show that the bivariate random-effects approach not only extends the SROC approach but also provides a unifying framework for other approaches. The authors bring the statistical meta-analysis of ROC-curve data back into a framework of relatively standard multivariate meta-analysis with random effects. The analyses were carried out using the software package SAS (Proc NLMIXED).

Relative efficiency of haplotype frequency estimation in sibships and nuclear families compared to unrelated individuals.

The problem of estimating haplotype frequencies from unphased single nucleotide polymorphism (SNP) genotype data in sibships with and without parents is considered. We focus on the Fisher information of the haplotype frequencies of the parents in order to correctly deal with the dependence of haplotypes within sibships. We compare these Fisher information matrices with those obtained for unrelated individuals and study the relative efficiency of sibships with and without parents compared to unrelated individuals in estimating haplotype frequencies. Crudely summarizing, the second sib contributes half the information of the first, except for rare haplotypes, when the second sib counts almost as one. We argue that the relative efficiencies can also be used to correct for dependence in the calculation of standard errors after initially ignoring the dependence in the estimation phase.

Quality of life in couples living with Huntington's disease: the role of patients' and partners' illness perceptions.

Research suggests that chronically ill patients and their partners perceive illness differently, and that these differences have a negative impact on patients' quality of life (QoL). This study assessed whether illness perceptions of patients with Huntington's disease (HD) differ from those of their partners, and examined whether spousal illness perceptions are important for the QoL of the couples (n = 51 couples). Partners reported that their HD-patient spouses suffered more symptoms and experienced less control than the patients themselves reported. Illness perceptions of patients and partners correlated significantly with patient QoL. Partners' beliefs in a long duration of the patients' illness and less belief in cure, were associated with patient vitality scores. Suggestions for future research emphasize the importance of qualitative research approaches in combination with cognitive-behavioural approaches.

Potential bias in Generalized Estimating Equations linkage methods under incomplete information.

The mean identity-by-descent (IBD) specification used in the Generalized Estimating Equations (GEE) methodology for linkage is only valid, strictly speaking, under the assumption of fully polymorphic markers. In practice, markers often provide only partial IBD information, which can potentially result in inconsistency of the locus location and gene effect estimates obtained by the GEE method. Using both simulations and theory, we identify some realistic conditions about marker information under which the validity of the GEE linkage methods may be arguable. Namely, researchers should not trust the GEE parameters' estimates and their associated confidence intervals in areas of the genome where IBD information is sparse or when this information changes abruptly. We show that properly standardized statistics based on IBD sharing provide a valid alternative.

Anxious-retarded depression: relation to two-year outcome of major depressive disorder.

BACKGROUND: Anxious-retarded depression is a two-dimensionally defined subcategory of depression derived from DSM-IV melancholia. It is related to increased plasma vasopressin, correlative plasma vasopressin and cortisol levels, and a positive family history. We now explored its relation with outcome. METHODS: Seventy depressed patients were included to follow-up for two years. Outcome was defined by time until full-remission. Cox regression analyses were used to compare anxious-retarded and non-anxious-retarded patients, as well as melancholic and non-melancholic patients. RESULTS: Anxious-retarded depression had poor outcome. LIMITATIONS: The number of patients was relatively small. CONCLUSION: The poor outcome of anxious-retarded depression further supports its validity.

Long-term survival with non-proportional hazards: results from the Dutch Gastric Cancer Trial.

Randomized clinical trials with long-term survival data comparing two treatments often show Kaplan-Meier plots with crossing survival curves. Such behaviour implies a violation of the proportional hazards assumption for treatment. The Cox proportional hazards regression model with treatment as a fixed effect can therefore not be used to assess the influence of treatment of survival. In this paper we analyse long-term follow-up data from the Dutch Gastric Cancer Trial, a randomized study comparing limited (D1) lymph node dissection with extended (D2) lymph node dissection. We illustrate a number of ways of dealing with survival data that do not obey the proportional hazards assumption, each of which can be easily implemented in standard statistical packages.

Cancer risks in BRCA2 families: estimates for sites other than breast and ovary.

BACKGROUND: In BRCA2 mutation carriers, increased risks have been reported for several cancer sites besides breast and ovary. As most of the families included in earlier reports were selected on the basis of multiple breast/ovarian cancer cases, it is possible that risk estimates may differ in mutation carriers with a less striking family history. METHODS: In the Netherlands, 139 BRCA2 families with 66 different pathogenic mutations were included in a nationwide study. To avoid testing bias, we chose not to estimate risk in typed carriers, but rather in male and female family members with a 50% prior probability of being a carrier (n = 1811). The relative risk (RR) for each cancer site with the exception of breast and ovarian cancer was determined by comparing observed numbers with those expected, based on Dutch cancer incidence rates. RESULTS: We observed an excess risk for four cancer sites: pancreas (RR 5.9; 95% confidence interval (CI) 3.2 to 10.0), prostate (2.5; 1.6 to 3.8), bone (14.4; 2.9 to 42.1) and pharynx (7.3; 2.0 to 18.6). A small increase was observed for cancer of the digestive tract (1.5; 1.1 to 1.9). Histological verification was available for 46% of the tumours. Nearly all increased risks reached statistical significance for men only. Cancer risks tended to be higher for people before the age of 65 years. Moreover, families with mutations outside the previously defined ovarian cancer cluster region tended to have a higher cancer risk. CONCLUSIONS: We found that BRCA2 carriers are at increased risk for cancers of the prostate and pancreas, and possibly bone and pharynx. Larger databases with extended follow up are needed to provide insight into mutation specific risks of selected carriers in BRCA2 families.

Score statistic to test for genetic correlation for proband-family design.

In genetic epidemiological studies informative families are often oversampled to increase the power of a study. For a proband-family design, where relatives of probands are sampled, we derive the score statistic to test for clustering of binary and quantitative traits within families due to genetic factors. The derived score statistic is robust to ascertainment scheme. We considered correlation due to unspecified genetic effects and/or due to sharing alleles identical by descent (IBD) at observed marker locations in a candidate region. A simulation study was carried out to study the distribution of the statistic under the null hypothesis in small data-sets. To illustrate the score statistic, data from 33 families with type 2 diabetes mellitus (DM2) were analyzed. In addition to the binary outcome DM2 we also analyzed the quantitative outcome, body mass index (BMI). For both traits familial aggregation was highly significant. For DM2, also including IBD sharing at marker D3S3681 as a cause of correlation gave an even more significant result, which suggests the presence of a trait gene linked to this marker. We conclude that for the proband-family design the score statistic is a powerful and robust tool for detecting clustering of outcomes.

Reduced rank proportional hazards model for competing risks.

Competing events concerning individual subjects are of interest in many medical studies. For example, leukemia-free patients surviving a bone marrow transplant are at risk of developing acute or chronic graft-versus-host disease, or they might develop infections. In this situation, competing risks models provide a natural framework to describe the disease. When incorporating covariates influencing the transition intensities, an obvious approach is to use Cox's proportional hazards model for each of the transitions separately. A practical problem then is how to deal with the abundance of regression parameters. Our objective is to describe the competing risks model in fewer parameters, both in order to avoid imprecise estimation in transitions with rare events and in order to facilitate interpretation of these estimates. Suppose that the regression parameters are gathered into a p x K matrix B, with p and K as the number of covariates and transitions, respectively. We propose the use of reduced rank models, where B is required to be of lower rank R, smaller than both p and K. One way to achieve this is to write B = AGamma(intercal) with A and Gamma matrices of dimensions p x R and K x R, respectively. We shall outline an algorithm to obtain estimates and their standard errors in a reduced rank proportional hazards model for competing risks and illustrate the approach on a competing risks model applied to 8966 leukemia patients from the European Group for Blood and Marrow Transplantation.

Studying the biological and technical sources of variation in telomere length of individual chromosomes.

BACKGROUND: Consistent average length differences between species and chromosome arm differences within species indicate that telomere length is genetically determined. This seems to contradict an observed large variation in lengths of the same human telomere between metaphases of the same individual. We examined the extent to which the variation in the telomeres of the human X and Y chromosomes is heritable, induced, or technical in origin. METHODS: Metaphase chromosomes were stained by fluorescence in situ hybridization with a telomere repeat-specific probe, and fluorescence intensities of the X and Y chromosomes were measured. If telomere length variation is predominantly genetically determined and a 50% probability of meiotic recombination between the pseudo-autosomal regions of Yp and Xp in the father is taken into account, one expects an equal chance that the Yp telomere of a son is derived from his father's Xp or Yp telomere. This implies that the Yp/Yq telomere ratios in fathers and sons will be identical in the absence of paternal meiotic recombination and different when recombination occurs. RESULTS: Among five father-son pairs, four showed similar Yp/Yq ratios (P > 0.05), whereas one pair exhibited a large difference in the Yp/Yq ratio that was attributable to a significantly longer Xp than Yp telomere in the father and a presumptive meiotic exchange between X and Y during paternal meiosis. Further, the Xq telomere exhibited a generally shorter telomere length than the others. CONCLUSIONS: The high variation in telomere length appeared to be intracellular (between sister chromatids) and, hence, technical in nature. We found no measurable induced variation in the cells studied, implying that, if induced variation exists, it is small compared with the technical variation.

Inhaled corticosteroids and growth of airway function in asthmatic children.

Airway inflammation and remodelling play an important role in the pathophysiology of asthma. Remodelling may affect childhood lung function, and this process may be reversed by anti-inflammatory treatment. The current study assessed longitudinally whether asthma affects growth of airway function relative to airspaces, and if so whether this is redressed by inhaled corticosteroids (ICS). Every 4 months for up to 3 yrs, lung function was assessed in 54 asthmatic children (initial age 7-16 yrs), who inhaled 0.2 mg salbutamol t.i.d. and 0.2 mg budesonide t.i.d. (beta2-agonist (BA)+ICS), or placebo (PL) t.i.d. (BA+PL) in a randomised, double-blind design. Measurements were carried out before and after maximal bronchodilation. Airway growth was assessed from the change of forced expiratory volume in one second and of maximal expiratory flows (at 60% and 40% of total lung capacity (TLC) remaining in the lung) relative to TLC, as measures of more central, intermediate and more peripheral airways. Growth patterns were compared with the longitudinal findings in 376 healthy children. Airway patency after maximal bronchodilation in patients on BA+PL remained reduced compared to healthy subjects, whereas in patients on BA+ICS a marked improvement was observed to subnormal. No differences between patients and controls could be demonstrated for growth patterns of central and intermediate airway function. Compliance with BA+ICS was 75% of the prescribed dose, resulting in significant, sustained improvement of symptoms and postbronchodilator calibre of central and intermediate airways to subnormal within 2 months, but postbronchodilator small airway patency remained reduced, though improved compared to patients on BA+PL. Anti-inflammatory treatment of asthmatic children is associated with normal functional development of central and intermediate airways. The persistently reduced postbronchodilator patency of peripheral airways may reflect remodelling, or insufficient anti-inflammatory treatment.

Sonographic assessment of non-malignant ovarian cysts: does sonohistology exist?

BACKGROUND: Transvaginal ultrasound (TVU) is feasible and accurate in the differentiation between non-malignant and malignant ovarian abnormalities. However, despite the clinical relevance, the accuracy of TVU in the differentiation between the many different non-malignant cysts is unknown. METHODS: Between 1992 and 2002, all women who had surgery at our centre because of a non-malignant ovarian cyst were included prospectively in this study. The sonographic characteristics as well as the expected histological diagnosis (the 'sonohistological diagnosis') were evaluated pre-operatively. This diagnosis was compared with the histopathological diagnosis, and diagnostic parameters [with 95% confidence interval (CI)] of the sonohistological diagnosis were calculated. Logistic models, with the sonographic characteristics as variables, were constructed for each histopathological diagnosis. RESULTS: A total of 406 women were included consecutively. The overall diagnostic accuracy of the sonohistological diagnosis was 60% (95% CI 0.56-0.65). Only in cases of simple ovarian cysts did the diagnostic accuracy of the respective logistic model exceed that of the sonohistological diagnosis (0.88 versus 0.81, P < 0.01). The diagnostic accuracy of the sonohistological diagnosis for endometriotic and dermoid ovarian cysts was significantly better compared with the respective logistic model (0.84 versus 0.71, P < 0.01 and 0.87 versus 0.82, P = 0.03, respectively). CONCLUSION: In approximately half of the non-malignant ovarian cysts, TVU is capable of distinguishing between the different histopathological diagnoses of non-malignant ovarian masses. Only in the diagnosis of simple ovarian cysts might use of the logistic models be helpful.

Risk estimation for healthy women from breast cancer families: new insights and new strategies.

Risk estimation in breast cancer families is often estimated by use of the Claus tables. We analyzed the family histories of 196 counselees; compared the Claus tables with the Claus, the BRCA1/2, the BRCA1/2/ models; and performed linear regression analysis to extend the Claus tables with characteristics of hereditary breast cancer. Finally, we compared the Claus extended method with the Claus, the BRCA1/2, and the BRCA1/2/u models. We found 47% agreement for Claus table versus Claus model; 39% agreement for Claus table versus BRCA1/2 model; 48% agreement for Claus table versus BRCA1/2/u model; 37% agreement for Claus extended method versus Claus model; 44% agreement for Claus extended model versus BRCA1/2 model; and 66% agreement for Claus extended method versus BRCA1/2/u model. The regression formula (Claus extended method) for the lifetime risk for breast cancer was 0.08 + 0.40 (*) Claus Table + 0.07 (*) ovarian cancer + 0.08 (*) bilateral breast cancer + 0.07 (*) multiple cases. This new method for risk estimation, which is an extension of the Claus tables, incorporates information on the presence of ovarian cancer, bilateral breast cancer, and whether there are more than two affected relatives with breast cancer. This extension might offer a good alternative for breast cancer risk estimation in clinical practice.

Health-related quality of life in patients with systemic lupus erythematosus: development and validation of a lupus specific symptom checklist.

Reliable and sensitive measures are needed to evaluate the quality of life (QoL) in patients with systemic lupus erythematosus (SLE). No lupus specific questionnaires are available. This study describes the development and validation of a disease-specific questionnaire for lupus patients, which assesses the presence and burden of 38 disease- and treatment-related symptoms: the SLE Symptom Checklist (SSC). Reliability and reproducibility were tested in respectively 87 and 28 stable SLE patients. The internal consistency (Cronbach's alpha coefficients 0.89) and test-retest reliability (Pearson product-moment correlation coefficient between 0.67 and 0.87) were satisfactory. Concurrent validity was supported by significant, but moderate correlations with other measures of subjective well-being and functional status. Responsiveness was measured in 17 patients with lupus nephritis treated with cyclophosphamide, at start of therapy and 1 year thereafter. A significant change in number of symptoms and total distress level was found. It is concluded that the SSC has satisfactory psychometric properties and appears suitable for both clinical and research purposes.