Penalized maximum likelihood inference under the mixture cure model in sparse data.

INTRODUCTION: When a study sample includes a large proportion of long-term survivors, mixture cure (MC) models that separately assess biomarker associations with long-term recurrence-free survival and time to disease recurrence are preferred to proportional-hazards models. However, in samples with few recurrences, standard maximum likelihood can be biased. OBJECTIVE AND METHODS: We extend Firth-type penalized likelihood (FT-PL) developed for bias reduction in the exponential family to the Weibull-logistic MC, using the Jeffreys invariant prior. Via simulation studies based on a motivating cohort study, we compare parameter estimates of the FT-PL method to those by ML, as well as type 1 error (T1E) and power obtained using likelihood ratio statistics. RESULTS: In samples with relatively few events, the Firth-type penalized likelihood estimates (FT-PLEs) have mean bias closer to zero and smaller mean squared error than maximum likelihood estimates (MLEs), and can be obtained in samples where the MLEs are infinite. Under similar T1E rates, FT-PL consistently exhibits higher statistical power than ML in samples with few events. In addition, we compare FT-PL estimation with two other penalization methods (a log-F prior method and a modified Firth-type method) based on the same simulations. DISCUSSION: Consistent with findings for logistic and Cox regressions, FT-PL under MC regression yields finite estimates under stringent conditions, and better bias-and-variance balance than the other two penalizations. The practicality and strength of FT-PL for MC analysis is illustrated in a cohort study of breast cancer prognosis with long-term follow-up for recurrence-free survival.

DNA methylation age calculators reveal association with diabetic neuropathy in type 1 diabetes.

BACKGROUND: Many CpGs become hyper or hypo-methylated with age. Multiple methods have been developed by Horvath et al. to estimate DNA methylation (DNAm) age including Pan-tissue, Skin & Blood, PhenoAge, and GrimAge. Pan-tissue and Skin & Blood try to estimate chronological age in the normal population whereas PhenoAge and GrimAge use surrogate markers associated with mortality to estimate biological age and its departure from chronological age. Here, we applied Horvath's four methods to calculate and compare DNAm age in 499 subjects with type 1 diabetes (T1D) from the Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) study using DNAm data measured by Illumina EPIC array in the whole blood. Association of the four DNAm ages with development of diabetic complications including cardiovascular diseases (CVD), nephropathy, retinopathy, and neuropathy, and their risk factors were investigated. RESULTS: Pan-tissue and GrimAge were higher whereas Skin & Blood and PhenoAge were lower than chronological age (p < 0.0001). DNAm age was not associated with the risk of CVD or retinopathy over 18-20 years after DNAm measurement. However, higher PhenoAge (ß = 0.023, p = 0.007) and GrimAge (ß = 0.029, p = 0.002) were associated with higher albumin excretion rate (AER), an indicator of diabetic renal disease, measured over time. GrimAge was also associated with development of both diabetic peripheral neuropathy (OR = 1.07, p = 9.24E-3) and cardiovascular autonomic neuropathy (OR = 1.06, p = 0.011). Both HbA1c (ß = 0.38, p = 0.026) and T1D duration (ß = 0.01, p = 0.043) were associated with higher PhenoAge. Employment (ß = - 1.99, p = 0.045) and leisure time (ß = - 0.81, p = 0.022) physical activity were associated with lower Pan-tissue and Skin & Blood, respectively. BMI (ß = 0.09, p = 0.048) and current smoking (ß = 7.13, p = 9.03E-50) were positively associated with Skin & Blood and GrimAge, respectively. Blood pressure, lipid levels, pulse rate, and alcohol consumption were not associated with DNAm age regardless of the method used. CONCLUSIONS: Various methods of measuring DNAm age are sub-optimal in detecting people at higher risk of developing diabetic complications although some work better than the others.

Gene-based and pathway-based testing for rare-variant association in affected sib pairs.

Next generation sequencing technologies have made it possible to investigate the role of rare variants (RVs) in disease etiology. Because RVs associated with disease susceptibility tend to be enriched in families with affected individuals, study designs based on affected sib pairs (ASP) can be more powerful than case-control studies. We construct tests of RV-set association in ASPs for single genomic regions as well as for multiple regions. Single-region tests can efficiently detect a gene region harboring susceptibility variants, while multiple-region extensions are meant to capture signals dispersed across a biological pathway, potentially as a result of locus heterogeneity. Within ascertained ASPs, the test statistics contrast the frequencies of duplicate rare alleles (usually appearing on a shared haplotype) against frequencies of a single rare allele copy (appearing on a nonshared haplotype); we call these allelic parity tests. Incorporation of minor allele frequency estimates from reference populations can markedly improve test efficiency. Under various genetic penetrance models, application of the tests in simulated ASP data sets demonstrates good type I error properties as well as power gains over approaches that regress ASP rare allele counts on sharing state, especially in small samples. We discuss robustness of the allelic parity methods to the presence of genetic linkage, misspecification of reference population allele frequencies, sequencing error and de novo mutations, and population stratification. As proof of principle, we apply single- and multiple-region tests in a motivating study data set consisting of whole exome sequencing of sisters ascertained with early onset breast cancer.

Osteosarcoma and soft-tissue sarcomas with an immune infiltrate express PD-L1: relation to clinical outcome and Th1 pathway activation.

Immune checkpoint proteins, such as PD-L1 and PD-1, are important in several cancers; however, their role in osteosarcoma (OSA) and soft tissue sarcoma (STS) remains unclear. Our aims were to determine whether subsets of OSA/STS harbor tumor-infiltrating lymphocytes (TILs) and express PD-L1, and how PD-L1 expression is related to clinical outcome. Tissue sections of 25 cases each of untreated undifferentiated pleomorphic sarcoma (UPS), myxofibrosarcoma (MFS), liposarcoma (LPS) and 24 of leiomyosarcoma (LMS) were subjected to immunohistochemistry (IHC) for immune cells, PD-L1 and PD-1. RT-qPCR was utilized to quantify levels of PD-L1 mRNA from 33 UPS, 57 MFS and 79 OSA primary-untreated specimens. PD-L1 mRNA levels were tested for their correlation with overall survival in patients presenting without metastases. Transcriptome analysis evaluated biological pathway differences between high and low PD-L1 expressers. A subset of UPS and MFS contained TILs and expressed PD-L1 and PD-1; LMS and LPS did not. PD-L1 levels by IHC and RT-qPCR were positively correlated. PD-L1 over-expression was associated with better survival for UPS and OSA, but not MFS. The Th1 pathway was significantly activated in UPS with high levels of PD-L1 and improved survival. Some sarcoma subtypes harbor TILs and express PD-L1. Patients with UPS and OSA with high levels of PD-L1 had better overall survival than those with low expression levels. Important biological pathways distinguish PD-L1 high and low groups. The stratification of patients with OSA/STS with respect to potential immune therapies may be improved through investigation of the expression of immune cells and checkpoint proteins.

Fresh Cut Versus Stored Cut Paraffin-embedded Tissue: Effect on Immunohistochemical Staining for Common Breast Cancer Markers.

The proper handling of unstained paraffin slides for immunohistochemistry has been a matter of debate, with several studies demonstrating loss of antigenicity with prolonged storage at room temperature, 4°C and -20°C. The purpose of this study was to determine whether long-term storage of unstained slides at -80°C would impact the staining intensity and expression distribution of markers used to molecularly subtype breast cancer specimens [estrogen receptor (ER), human epidermal growth factor receptor 2 (HER2), cytokeratin 5 (CK5), epidermal growth factor receptor (EGFR), and Ki67]. The staining pattern of previously unstained breast tumor slides (n=39 to 64) stored at -80°C for a minimum of 9.93 years (avg., 12.8 y) was compared with the staining pattern of fresh cut slides from the same tumors. The Allred scoring method was used to score ER (0 to 2, negative; 3 to 8, positive), CK5 (≥4, positive), and EGFR (≥4, positive). ASCO/CAP guidelines were used to assess HER2 (0/1+, 2+, or 3+). Ki67 scores were determined based on the proportion of cells stained of any intensity, with 20% staining used as a cut-off. Agreement was assessed using concordance rates and chance-corrected agreement statistics. The chance-corrected agreements were as follows: 0.94 (38/39) for ER, 0.92 (53/55) for CK5, 0.87 (61/64) for EGFR, 0.86 (37/39) for HER2, and 0.67 (46/54) for Ki67. Long-term storage of cut unstained slides at -80°C does not significantly impact the scoring interpretation of ER, CK5, EGFR, and HER2.

Tumoral BRD4 expression in lymph node-negative breast cancer: association with T-bet+ tumor-infiltrating lymphocytes and disease-free survival.

BACKGROUND: We previously observed that T-bet+ tumor-infiltrating T lymphocytes (T-bet+ TILs) in primary breast tumors were associated with adverse clinicopathological features, yet favorable clinical outcome. We identified BRD4 (Bromodomain-Containing Protein 4), a member of the  Bromodomain and Extra Terminal domain (BET) family, as a gene that distinguished T-bet+/high and T-bet-/low tumors. In clinical studies, BET inhibitors have been shown to suppress inflammation in various cancers, suggesting a potential link between BRD4 and immune infiltration in cancer. Hence, we examined the BRD4 expression and clinicopathological features of breast cancer. METHODS: The cohort consisted of a prospectively ascertained consecutive series of women with axillary node-negative breast cancer with long follow-up. Gene expression microarray data were used to detect mRNAs differentially expressed between T-bet+/high (n = 6) and T-bet-/low (n = 41) tumors. Tissue microarrays (TMAs) constructed from tumors of 612 women were used to quantify expression of BRD4 by immunohistochemistry, which was analyzed for its association with T-bet+ TILs, Jagged1, clinicopathological features, and disease-free survival. RESULTS: Microarray analysis indicated that BRD4 mRNA expression was up to 44-fold higher in T-bet+/high tumors compared to T-bet-/low tumors (p = 5.38E-05). Immunohistochemical expression of BRD4 in cancer cells was also shown to be associated with T-bet+ TILs (p = 0.0415) as well as with Jagged1 mRNA and protein expression (p = 0.0171, 0.0010 respectively). BRD4 expression correlated with larger tumor size (p = 0.0049), pre-menopausal status (p = 0.0018), and high Ki-67 proliferative index (p = 0.0009). Women with high tumoral BRD4 expression in the absence of T-bet+ TILs exhibited a significantly poorer outcome (log rank test p = 0.0165) relative to other subgroups. CONCLUSIONS: The association of BRD4 expression with T-bet+ TILs, and T-bet+ TIL-dependent disease-free survival suggests a potential link between BRD4-mediated tumor development and tumor immune surveillance, possibly through BRD4's regulation of Jagged1 signaling pathways. Further understanding BRD4's role in different immune contexts may help to identify an appropriate subset of breast cancer patients who may benefit from BET inhibitors without the risk of diminishing the anti-tumoral immune activity.

New Locus for Skin Intrinsic Fluorescence in Type 1 Diabetes Also Associated With Blood and Skin Glycated Proteins.

Skin fluorescence (SF) noninvasively measures advanced glycation end products (AGEs) in the skin and is a risk indicator for diabetes complications. N-acetyltransferase 2 (NAT2) is the only known locus influencing SF. We aimed to identify additional genetic loci influencing SF in type 1 diabetes (T1D) through a meta-analysis of genome-wide association studies (N = 1,359) including Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) and Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR). A locus on chromosome 1, rs7533564 (P = 1.9 × 10(-9)), was associated with skin intrinsic fluorescence measured by SCOUT DS (excitation 375 nm, emission 435-655 nm), which remained significant after adjustment for time-weighted HbA1c (P = 1.7 × 10(-8)). rs7533564 was associated with mean HbA1c in meta-analysis (P = 0.0225), mean glycated albumin (P = 0.0029), and glyoxal hydroimidazolones (P = 0.049), an AGE measured in skin biopsy collagen, in DCCT. rs7533564 was not associated with diabetes complications in DCCT/EDIC or with SF in subjects without diabetes (nondiabetic [ND]) (N = 8,721). In conclusion, we identified a new locus associated with SF in T1D subjects that did not show similar effect in ND subjects, suggesting a diabetes-specific effect. This association needs to be investigated in type 2 diabetes.

Validation of Intratumoral T-bet+ Lymphoid Cells as Predictors of Disease-Free Survival in Breast Cancer.

We previously observed T-bet(+) lymphocytes to be associated with a good prognosis in a cohort of women with familial breast cancer. To validate this finding, we evaluated lymphocyte T-bet expression in an independent unselected prospectively accrued series of women with lymph node-negative breast carcinoma. T-bet and clinicopathologic data were available for 614 women. Hormone receptors, HER2, Ki-67, CK5, EGFR, p53, and T-bet status were determined using IHC and/or biochemical methods. Tumors were assigned to luminal A, luminal B, HER2, and basal subtypes based on the expression of IHC markers. Multiple cutpoints were examined in a univariate penalized Cox model to stratify tumors into T-bet(+/high) and T-bet(-/low). Fisher exact test was used to analyze T-bet associations with clinicopathologic variables, IHC markers, and molecular subtype. Survival analyses were by the Cox proportional hazards model. All tests were two sided. A test with a P value < 0.05 was considered statistically significant. T-bet(+/high) tumor status was significantly associated with large tumor size, high grade, hormone receptor negativity, CK5, EGFR and p53 positivity, high Ki-67, and basal subtype. With a median follow-up of 96.5 months, T-bet(-/low) tumor status was associated with a reduced disease-free survival compared with T-bet(+/high) tumor status in multivariate analysis (P = 0.0027; relative risk = 5.62; 95% confidence intervals, 1.48-50.19). Despite being associated with adverse clinicopathologic characteristics, T-bet(+) tumor-infiltrating lymphoid cells are associated with a favorable outcome. This supports their role in Th1-mediated antitumor activity and may provide insight for the development of new therapeutic strategies.

Menacalc, a quantitative method of metastasis assessment, as a prognostic marker for axillary node-negative breast cancer.

BACKGROUND: Menacalc is an immunofluorescence-based, quantitative method in which expression of the non-invasive Mena protein isoform (Mena11a) is subtracted from total Mena protein expression. Previous work has found a significant positive association between Menacalc and risk of death from breast cancer. Our goal was to determine if Menacalc could be used as an independent prognostic marker for axillary node-negative (ANN) breast cancer. METHODS: Analysis of the association of Menacalc with overall survival (death from any cause) was performed for 403 ANN tumors using Kaplan Meier survival curves and the univariate Cox proportional hazards (PH) model with the log-rank or the likelihood ratio test. Cox PH models were used to estimate hazard ratios (HRs) for the association of Menacalc with risk of death after adjustment for HER2 status and clinicopathological tumor features. RESULTS: High Menacalc was associated with increased risk of death from any cause (P=0.0199, HR (CI)=2.18 (1.19, 4.00)). A similarly elevated risk of death was found in the subset of the Menacalc cohort which did not receive hormone or chemotherapy (n=142) (P=0.0052, HR (CI)=3.80 (1.58, 9.97)). There was a trend toward increased risk of death with relatively high Menacalc in the HER2, basal and luminal molecular subtypes. CONCLUSIONS: Menacalc may serve as an independent prognostic biomarker for the ANN breast cancer patient population.

Bivariate genetic association analysis of systolic and diastolic blood pressure by copula models.

We conduct genetic association analysis in the subset of unrelated individuals from the San Antonio Family Studies pedigrees, applying a two-stage approach to take account of the dependence between systolic and diastolic blood pressure (SBP and DBP). In the first stage, we adjust blood pressure for the effects of age, sex, smoking, and use of antihypertensive medication based on a novel modification of censored regression. In the second stage, we model the bivariate distribution of the adjusted SBP and DBP phenotypes by a copula function with interpretable SBP-DBP correlation parameters. This allows us to identify genetic variants associated with each of the adjusted blood pressures, as well as variants that explain the association between the two phenotypes. Within this framework, we define a pleiotropic variant as one that reduces the SBP-DBP correlation. Our results for whole genome sequence variants in the gene ULK4 on chromosome 3 suggest that inference obtained from a copula model can be more informative than findings from the SBP-specific and DBP-specific univariate models alone.

Distinguishing luminal breast cancer subtypes by Ki67, progesterone receptor or TP53 status provides prognostic information.

The objectives of this study were to determine the prognostic significance of subgrouping estrogen receptor (ER)-positive breast tumors into low- and high-risk luminal categories using Ki67 index, TP53, or progesterone receptor (PR) status. The study group comprised 540 patients with lymph node negative, invasive breast carcinoma. Luminal A subtype was defined as being ER positive, HER2 negative, and Ki67 low (<14% cells positive) and luminal B subtype as being ER positive, HER2 negative, and Ki67 high (≥ 14% cells positive). Luminal tumors were also subgrouped into risk categories based on the PR and TP53 status. Survival analysis was performed. Patients with luminal B tumors (n=173) had significantly worse disease-free survival compared to those with luminal A tumors (n=186) (log rank P-value=0.0164; univariate Cox regression relative risk 2.00; 95% CI, 1.12-3.58; P=0.0187). Luminal subtype remained an independent prognostic indicator on multivariate analysis including traditional prognostic factors (relative risk 2.12; 95% CI, 1.16-3.88; P=0.0151). Using TP53 status or PR negativity rather than Ki67 to classify ER-positive luminal tumors gave similar outcome results to those obtained using the proliferation index. However, it was a combination of the three markers, which proved the most powerful prognostically. Ki67 index, TP53 status, or PR negativity can be used to segregate ER-positive, HER2-negative tumors into prognostically meaningful subgroups with significantly different clinical outcomes. These biomarkers particularly in combination may potentially be used clinically to guide patient management.

Killer-cell immunoglobulin-like receptor gene polymorphisms and susceptibility to psoriatic arthritis.

OBJECTIVES: We conducted a case-control study to determine the association between KIR2D and KIR3D gene polymorphisms and their interaction with HLA alleles in PsA. METHODS: A total of 678 subjects with PsA and 688 healthy controls were studied. Differences between cases and controls in the frequency of individual KIR polymorphisms were tested for significance by an asymptotic χ(2) test and Fisher's exact test. Trends for increasing susceptibility to PsA from combined genotypes (HLA-KIR and HLA) were evaluated by the Cochran-Armitage trend test. Multigene logistic regression analysis was conducted to identify independent associations and interactions. RESULTS: In univariate analyses, KIR2DL2 and KIR2DS2 polymorphisms were significantly associated with PsA. Only KIR2DS2 was associated with PsA compared with healthy controls in multivariate analysis [odds ratio (OR) 1.25, 95% CI 1.01, 1.54, P = 0.044]. The presence of HLA-C group 2 alleles was associated with a higher risk of PsA (trend test P = 0.006). The risk of PsA is higher when KIR2DS2 is present with the HLA-C ligands (C group 1) for the corresponding inhibitory KIRs, and is highest when KIR2DS2 is present in the absence of HLA-C ligands for homologous inhibitor KIRs, compared with the state when KIR2DS2 is absent (trend test P = 0.027). The presence of HLA-C alleles that have high cell surface expression was also associated with a higher risk of PsA (trend test P < 0.001). HLA-B Bw4 and HLA-B Bw4 80ile allele groups were associated with a higher PsA risk (trend test P < 0.0001 for both analyses). CONCLUSION: This study confirms the association of the KIR2DS gene, especially KIR2DS2, with PsA.

Human leukocyte antigen alleles and susceptibility to psoriatic arthritis.

OBJECTIVE: Our purpose was to determine associations between HLA alleles and psoriatic arthritis (PsA). METHODS: 678 PsA cases and 688 healthy controls were analyzed in a case-control design. The difference in the proportion of cases and controls with at least 1 copy of HLA alleles were tested for significance using χ(2) test and Fisher's exact test. Association analyses of haplotypes inferred by the Expectation-Maximization algorithm were performed. In the family-based association study, data from 283 families were analyzed. RESULTS: Univariate analysis revealed that cases were more likely to be carriers of HLA-C*01, -C*02, -C*06, -C*12, -B*27, -B*38 and -B*57, whereas controls were more likely to be carriers of HLA-C*03, -C*07, -B*07, -B*51, -DRB1*15 and -DQB1*0602. In haplotype analyses, PsA cases were more likely to be carriers of the HLA haplotypes -C*01/-B*27, -C*02/-B*27, -C*12/-B*38, and -C*06/-B*57, while controls were more likely to be carriers of the haplotypes -C*07/-B*07 and -C*15/-B*51. In the family-based association analysis, the HLA alleles -A*02, -B*27 and -DRB1*07 were preferentially transmitted to cases, whereas the alleles -A*03, -A*28, -B*51, -DRB1*11 and -DQB1*0301 were under transmitted. CONCLUSION: This large case-control and family based association study shows that HLA-C*12/B*38, HLA-B*27 and HLA-C*06/B*57 are haplotypes (alleles) robustly associated with PsA. However, since patients with PsA also have psoriasis it is difficult to determine whether the primary association is with arthritis or psoriasis.

Re-ranking sequencing variants in the post-GWAS era for accurate causal variant identification.

Next generation sequencing has dramatically increased our ability to localize disease-causing variants by providing base-pair level information at costs increasingly feasible for the large sample sizes required to detect complex-trait associations. Yet, identification of causal variants within an established region of association remains a challenge. Counter-intuitively, certain factors that increase power to detect an associated region can decrease power to localize the causal variant. First, combining GWAS with imputation or low coverage sequencing to achieve the large sample sizes required for high power can have the unintended effect of producing differential genotyping error among SNPs. This tends to bias the relative evidence for association toward better genotyped SNPs. Second, re-use of GWAS data for fine-mapping exploits previous findings to ensure genome-wide significance in GWAS-associated regions. However, using GWAS findings to inform fine-mapping analysis can bias evidence away from the causal SNP toward the tag SNP and SNPs in high LD with the tag. Together these factors can reduce power to localize the causal SNP by more than half. Other strategies commonly employed to increase power to detect association, namely increasing sample size and using higher density genotyping arrays, can, in certain common scenarios, actually exacerbate these effects and further decrease power to localize causal variants. We develop a re-ranking procedure that accounts for these adverse effects and substantially improves the accuracy of causal SNP identification, often doubling the probability that the causal SNP is top-ranked. Application to the NCI BPC3 aggressive prostate cancer GWAS with imputation meta-analysis identified a new top SNP at 2 of 3 associated loci and several additional possible causal SNPs at these loci that may have otherwise been overlooked. This method is simple to implement using R scripts provided on the author's website.

Insights from mixture cure modeling of molecular markers for prognosis in breast cancer.

With the ultimate aim of improving clinical management of breast cancer, investigators have sought to identify molecular genetic markers that stratify newly diagnosed patients into subtypes differing in short- or long-term prognosis. Conventional survival models can fail to describe adequately the relationship between subtype and disease recurrence, particularly when there is a substantial proportion of long-term disease-free survivors. The observed patterns of disease-free survival in an undifferentiated patient cohort may be explained by an underlying mixture of two subgroups: patients who will remain free of disease in the long term (ie, cured), and those who will experience disease recurrence within their lifetime (ie, susceptible.) In this article, we review the concepts and methods of the mixture cure models and apply them in the analysis of molecular genetic prognostic factors for disease-free survival and time to disease recurrence in a cohort of patients with axillary lymph node-negative breast cancer.

Elevated expression of podocalyxin is associated with lymphatic invasion, basal-like phenotype, and clinical outcome in axillary lymph node-negative breast cancer.

Lymphatic invasion (LVI) is associated with disease recurrence in axillary node-negative (ANN) breast cancer. Using gene expression profiling of 105 ANN tumors, we found that podocalyxin (PODXL) was more highly expressed in tumors with LVI (LVI+) than in those without LVI (LVI-). Differences in PODXL expression were validated using real-time polymerase chain reaction as well as by immunohistochemistry in an independent set of 652 tumors on tissue microarrays. Disease-free survival (DFS) analyses were conducted for association of high PODXL protein expression with risk of distant recurrence overall and within breast cancer subtypes using both Cox and cure-rate models. High PODXL expression was associated with poor prognosis features including large tumor size, high histological grade, estrogen and progesterone receptor negativity, and with clinical alterations characteristic of the basal-like breast cancer phenotype. Surprisingly, despite having other poor prognosis characteristics, women with high PODXL expressing tumors had better long-term DFS in multivariate analysis with traditional clinicopathologic factors including LVI and HER2 status (P = 0.001). PODXL has the potential to be a useful biomarker for identifying good prognosis patients in characteristically poor prognosis breast cancer groups and may impact treatment of women with this disease.

Gene network modular-based classification of microarray samples.

BACKGROUND: Molecular predictor is a new tool for disease diagnosis, which uses gene expression to classify diagnostic category of a patient. The statistical challenge for constructing such a predictor is that there are thousands of genes to predict for the disease categories, but only a small number of samples are available. RESULTS: We proposed a gene network modular-based linear discriminant analysis approach by integrating 'essential' correlation structure among genes into the predictor in order that the modules or cluster structures of genes, which are related to the diagnostic classes we look for, can have potential biological interpretation. We evaluated performance of the new method with other established classification methods using three real data sets. CONCLUSIONS: Our results show that the new approach has the advantage of computational simplicity and efficiency with relatively lower classification error rates than the compared methods in many cases. The modular-based linear discriminant analysis approach induced in the study has the potential to increase the power of discriminant analysis for which sample sizes are small and there are large number of genes in the microarray studies.

Strategies for genetic association analyses combining unrelated case-control individuals and family trios.

In genetic association studies, analyses integrating data or estimates from unrelated case-control individuals and case trios (case offspring and their parents) can increase statistical power to identify disease susceptibility loci. Data on control trios may also be available, but how and when their use is advantageous is less familiar and is described here. In addition, the authors examine assumptions and properties of hybrid analyses combining association estimates from unrelated case-control individuals together with case and control family trios, focusing on low-prevalence disease. One such assumption is absence of population stratification bias (PSB), a potential source of confounding in case-control analyses. For detection of PSB, the authors discuss 4 possible tests that assess equality between individual-level and family-based estimates. Furthermore, a weighted framework is presented, in which estimates from analyses combining unrelated individuals and families (most powerful but subject to PSB) and family-based analyses (robust to PSB) are weighted according to the observed PSB test P value. In contrast to existing hybrid designs that combine individuals and families only if no significant PSB is detected, the weighted framework does not require specification of an arbitrary PSB testing level to establish significance. The statistical methods are evaluated using simulations and applied to a candidate gene study of childhood leukemia (Quebec Childhood Leukemia Study, 1980-2000).

Differential human leucocyte allele association between psoriasis and psoriatic arthritis: a family-based association study.

OBJECTIVE: A recent population-based study identified several HLA alleles as conferring a risk for psoriatic arthritis (PsA) among patients with psoriasis. The authors aimed to confirm these results using a family-based association study. METHODS: PsA probands, psoriasis probands and their first-degree family members were included. All participants were evaluated for the presence of psoriasis and inflammatory arthritis. HLA-B and -C genotyping was performed. The family-based association test was used to test for differences between PsA and psoriasis patients in transmission of candidate alleles from parents to offspring. RESULTS: A total of 178 PsA and 30 psoriasis probands and 561 first degree family members were analysed. The following HLA alleles were over-transmitted to PsA compared with psoriasis: HLA-C*12 (p=0.005), HLA-B*38 (p=0.04), HLA-B*39 (p=0.03), HLA-B*27 (p=0.002). CONCLUSIONS: HLA-B*27, HLA-B*38, HLA-B*39 and HLA-C*12 alleles are potential PsA-specific genetic markers among patients with psoriasis.

Human leucocyte antigen risk alleles for psoriatic arthritis among patients with psoriasis.

UNLABELLED: Aim Genes that differentiate patients with psoriatic arthritis (PsA) from those with cutaneous psoriasis (PsC) may serve as markers for the development of PsA in patients with psoriasis. The authors aimed to identify human leucocyte antigen (HLA) alleles that are associated with the development of PsA in patients with psoriasis. METHODS: 712 adult patients with PsA, 335 adult patients with PsC and 713 healthy controls were genotyped for HLA-A, HLA-B, HLA-C, HLA-DR and HLA-DQ alleles. Differences in allelic distributions for each of the HLA loci were compared using a likelihood ratio test. Logistic regression analysis of multiple loci was performed to account for linkage disequilibrium. Haplotype information was inferred using the expectation-maximisation algorithm (given HLA-C and HLA-B genotypes) and analysed similarly. RESULTS: The following HLA alleles were found to be significantly associated with patients with PsA compared to patients with PsC in multivariate regression analysis: B*08 (OR 1.61, p=0.009), B*27 (OR 5.17, p<0.0001), B*38 (OR 1.65, p=0.026) and C*06 (OR 0.58, p=0.0002). HLA-B*27, HLA-B*38 and HLA-C*06 frequencies were also significantly higher in patients with PsA than in healthy controls (B*27: OR 3.05, p<0.0001; B*38: OR 5.9, p<0.0001; HLA-C*06: OR 1.71, p<0.0001). The following haplotypes were independently associated with PsA compared to PsC: HLA-B*18-C*07 (OR 10.1, p=0.004), HLA-B*27-C*01 (OR 41.1, p<0.0001), HLA-B*27-C*02 (OR 19.9, p<0.0001), HLA-B*38-C*12 (OR 2.9, p=0.01), HLA-B*08-C*07 (OR 2.6, p=0.004) and HLA-B*57-C*06 (OR 0.5, p=0.03). CONCLUSIONS: Certain HLA-B and HLA-C alleles confer susceptibility to PsA among patients with psoriasis and may be used to identify patients with PsC who may develop PsA.