Learning Optimal Bayesian Networks: A Shortest Path Perspective

In this paper, learning a Bayesian network structure that optimizes a scoring function for a given dataset is viewed as a shortest path problem in an implicit state-space search graph. This perspective highlights the importance of two research issues: the development of search strategies for solving the shortest path problem, and the design of heuristic functions for guiding the search. This paper introduces several techniques for addressing the issues. One is an A* search algorithm that learns an optimal Bayesian network structure by only searching the most promising part of the solution space. The others are mainly two heuristic functions. The first heuristic function represents a simple relaxation of the acyclicity constraint of a Bayesian network. Although admissible and consistent, the heuristic may introduce too much relaxation and result in a loose bound. The second heuristic function reduces the amount of relaxation by avoiding directed cycles within some groups of variables. Empirical results show that these methods constitute a promising approach to learning optimal Bayesian network structures.

Investigation of 20 non-HLA (human leucocyte antigen) psoriasis susceptibility loci in Chinese patients with psoriatic arthritis and psoriasis vulgaris

BACKGROUND

Recently, a number of non-HLA (human leucocyte antigen) psoriasis genetic susceptibility loci have been identified through genome-wide association studies, but data on their association with psoriatic arthritis (PsA) are lacking.

OBJECTIVES

To investigate recently identified psoriasis susceptibility loci in a cohort of Chinese patients with PsA, psoriasis vulgaris (PsV) and healthy controls.

METHODS

Twenty single-nucleotide polymorphisms (SNPs) from 20 loci were selected for genotyping in 379 patients with PsA, 595 patients with PsV and 1181 healthy controls using the MassARRAY platform (Sequenom, San Diego, CA, U.S.A.). Data handling, quality control and association were performed using PLINK software, v. 1.07. The Cochran-Armitage trend test was used to test the genotype-phenotype association.

RESULTS

PsA showed a significant association with markers at TNIP1 (rs17728338, P = 2.20 × 10(-8)), IL28RA (rs4649203, P = 5.04 × 10(-6)), IL12B (rs2082412, P = 3.82 × 10(-5)), ERAP1 (rs27524, P = 1.25 × 10(-3)), PTTG1 (rs2431697, P = 1.22 × 10(-3)) and GJB2 (rs3751385, P = 1.48 × 10(-3)) when compared with the control group. In PsV a significant association was found for IL28RA (rs4649203, P = 9.53 × 10(-7)), TNIP1 (rs17728338, P = 1.21 × 10(-4)) and ERAP1 (rs27524, P = 1.17 × 10(-3)). The allele frequencies were not statistically different between PsA and PsV except for SNPs at IL12B and ZNF816A with a nominal P-value of 0.04 and 0·01, respectively.

CONCLUSIONS

This study provides evidence for the involvement of ERAP1, IL28RA, GJB2 and PTTG1 loci in PsA susceptibility and confirmed the previously reported association with PsA and PsV. These results support the hypothesis that genetic aetiology of psoriasis is the same in both PsA and PsV and also support the higher genetic component of PsA than PsV.

Sex differences of high-risk carotid atherosclerotic plaque with less than 50% stenosis in asymptomatic patients: an in vivo 3T MRI study

BACKGROUND AND PURPOSE

Men have a greater prevalence of high-risk carotid plaque features associated with stroke compared with women who have ≥50% stenosis, but little is known about these features in less significantly stenotic carotid arteries. This study aims to evaluate sex differences in complicated carotid plaque features in asymptomatic patients with <50% stenosis.

MATERIALS AND METHODS

Ninety-six patients (50 men, 46 women) with <50% carotid stenosis on MRA who had been referred for analysis of contralateral >50% carotid stenosis were included. The associations between sex and plaque features as identified by 3T MR carotid plaque imaging were examined by using logistic and linear regression models controlling for demographic characteristics, MRA stenosis, and the presence of contralateral plaque features.

RESULTS

The presence of a thin/ruptured fibrous cap (16% versus 2%, adjusted odds ratio = 8.57, P = .047), IPH (24% versus 6%, adjusted odds ratio = 4.53, P = .027), and American Heart Association type VI plaque (26% versus 6%, adjusted odds ratio = 5.04, P = .017) was significantly higher in men than in women. These associations remained significant following adjustment for contralateral plaque features. Men demonstrated a larger percentage volume of LR/NC (median, 1.66% versus -0.21%; P < .01). Calcification was not significantly associated with sex.

CONCLUSIONS

There is a sex difference of higher risk carotid plaque features during the early stage of disease seen in patients recruited for MR imaging evaluation of contralateral moderate-to-severe stenosis. Given the potential of using LR/NC without or with IPH to monitor therapy, these results indicate the possible importance of sex-based management in patients with asymptomatic carotid atherosclerosis across all stages of carotid stenosis.

Case-control association analysis of polymorphisms in the δ-opioid receptor, OPRD1, with cocaine and opioid addicted populations

BACKGROUND

Addiction susceptibility and treatment responsiveness are greatly influenced by genetic factors. Sequence variation in genes involved in the mechanisms of drug action have the potential to influence addiction risk and treatment outcome. The opioid receptor system is involved in mediating the rewarding effects of cocaine and opioids. The μ-opioid receptor (MOR) has traditionally been considered the primary target for opioid addiction. The MOR, however, interacts with and is regulated by many known MOR interacting proteins (MORIPs), including the δ-opioid receptor (DOR).

METHODS

The present study evaluated the contribution of OPRD1, the gene encoding the DOR, to the risk of addiction to opioids and cocaine. The association of OPRD1 polymorphisms with both opioid addiction (OA) and cocaine addiction (CA) was analyzed in African American (OA n=336, CA n=503) and European American (OA n=1007, CA n=336) populations.

RESULTS

The primary finding of this study is an association of rs678849 with cocaine addiction in African Americans (allelic p=0.0086). For replication purposes, this SNP was analyzed in a larger independent population of cocaine addicted African Americans and controls and the association was confirmed (allelic p=4.53 × 10(-5); n=993). By performing a meta-analysis on the expanded populations, the statistical evidence for an association was substantially increased (allelic p=8.5 × 10(-7)) (p-values non-FDR corrected).

CONCLUSION

The present study suggests that polymorphisms in OPRD1 are relevant for cocaine addiction in the African American population and provides additional support for a broad role for OPRD1 variants in drug dependence.

High-risk human papillomavirus reduces the expression of microRNA-218 in women with cervical intraepithelial neoplasia

This study was designed to investigate whether there is a correlation between the down-regulation of microRNA-218 (miR-218) and the presence of human papillomavirus (HPV) infection in the pathogenesis of cervical cancer. The participants comprised 78 women with cervical intraepithelial neoplasia (CIN); 22 (28.2%) had CIN 1, 27 (34.6%) had CIN 2 and 29 (37.2%) had CIN 3. MiR-218 expression was determined by reverse transcriptase polymerase chain reaction and HPV genotypes in tissue specimens were identified with a microarray test kit. The findings showed that miR-218 levels in patients with high-risk HPV infection were lower than in those infected with low-risk or intermediate-risk HPV, or in those who were HPV-free. MiR-218 levels in patients with high-risk CIN were lower than in those with low-risk CIN. We concluded that infection with high-risk HPV lowered the expression of miR-218 and that down-regulation of miR-218 was involved in the pathogenesis of cervical cancer.

Evaluation of patient self-management outcomes in health care: a systematic review

BACKGROUND

The importance of self-management and its intervention for improving the ability and skill of self-management has been discussed in literatures. It is, however, unclear how to choose the fitted, objective and accurate evaluation system when assessing the outcome.

AIM

The aim of this article was to establish a general evaluation system for skill and ability of self-management in chronic diseases through systematic review on different evaluation indicators and scales.

METHOD

A systematic search of six electronic databases was conducted. Two authors independently reviewed each qualified study for relevance and significance. Subsequently, main evaluation indicators and scales were identified and categorized into themes and sub-themes.

RESULTS

Nineteen articles were identified in this review. Among them, six main evaluation indicators of self-management, including frequently used scales, were extracted and tabulated. Self-efficacy, health behaviour/attitude, health status, health service utilization, quality of life and psychological indicators were the main indicators in evaluating self-management outcome, and they could be used alone or in combination flexibly according to the different goals of programmes.

CONCLUSION

Accurate evaluation of skill and ability of self-management is crucial not only in baseline data collection but also in proving the effectiveness of intervention. The outcomes of this study provide future researchers or caregivers with a better understanding and a series of good choices in self-management outcome evaluation.

Generation of mammalian cell lines with gene knock-down for human MCHR2

Appetite regulating neuropeptide melanin-concentrating hormone (MCH) has been implicated in obesity. It functions through its two receptors MCHR1 and MCHR2. While MCH and MCHR1 have been studied more extensively, the function of MCHR2 remains largely unknown, due to the lack of suitable in vitro and in vivo models. To create an in vitro system of genetic knock-down of MCHR2 in mammalian cells, we constructed four small hairpin RNAs (shRNAs) against human MCHR2 in eukaryotic expression vector, and transfected the plasmids into CHO cells that stably express human MCHR2. Using the empty vector or a negative shRNA control plasmid, we show that MCHR2-shRNAs suppressed 45.8% - 66.4% of MCHR2 expression at both mRNA and protein levels. As the result, in cells carrying the MCHR2-shRNAs, binding of MCHR2 to MCH was decreased by 39.4% - 78.7% accompanied by a similar decrease in affinity of the receptor to ligand by 40.9% - 81.9%. These cells still respond to MCH treatment, but intracellular Ca2+ release as the downstream signaling event was also decreased by 114.8% - 822.4%. Together, this study generated a set of shRNAs and cell lines as valuable reagents for further study on MCHR2 functions. These results will ultimately help to advance our knowledge about appetite regulating neuropeptide receptors.

MR carotid plaque imaging and contrast-enhanced MR angiography identifies lesions associated with recent ipsilateral thromboembolic symptoms: an in vivo study at 3T

BACKGROUND AND PURPOSE

Recent research has suggested the importance of plaque composition to identify patients at risk for stroke. This study aims to identify specific plaque features on 3T carotid MR imaging and CE-MRA associated with recent carotid thromboembolic symptoms in patients with mild/moderate versus severe stenosis.

MATERIALS AND METHODS

Ninety-seven consecutive patients (symptomatic, 13; asymptomatic, 84) with 50%-99% stenosis by sonography or CT angiography underwent carotid plaque imaging combined with MRA at 3T. The symptomatic carotid artery or the most stenotic asymptomatic carotid artery was chosen as the index vessel to be analyzed. Plaque features were compared by symptomatic status in patients with mild/moderate (30%-70%) versus severe (70%-99%) stenosis on MRA.

RESULTS

Ninety (92.8%) patients had sufficient image quality for interpretation. In 50 patients with mild/moderate stenosis, there were significant associations between the presence of the following plaque characteristics and symptoms: thin/ruptured fibrous cap (100% versus 36%, P = .006) and lipid-rich necrotic core (100% versus 39%, P = .022), with marginal association with hemorrhage (86% versus 33%, P = .055). In 40 patients with severe stenosis, only the angiographic presence of ulceration (86% versus 36%, P = .039) was associated with symptoms.

CONCLUSIONS

Several plaque components identified on 3T MR imaging are correlated with recent ipsilateral carotid thromboembolic symptoms. These preliminary results also suggest that associations between plaque characteristics and symptom history may vary by degree of stenosis. If confirmed in larger studies, carotid MR imaging may distinguish stable from unstable lesions, particularly in individuals with mild/moderate stenosis in whom the role of surgical intervention is currently unclear.

A noninvasive imaging approach to assess plaque severity: the carotid atherosclerosis score

BACKGROUND AND PURPOSE

The presence of IPH and/or FCR in the carotid atherosclerotic plaque indicates a high-risk lesion. The aim of this multicenter cross-sectional study was to establish the characteristics of lesions that may precede IPH and/or FCR. We further sought to construct a CAS that stratifies carotid disease severity.

MATERIALS AND METHODS

Three hundred forty-four individuals from 4 imaging centers with 16%-99% carotid stenosis by duplex sonography underwent carotid MR imaging. In approximately 60% of the study sample (training group), multivariate analysis was used to determine factors associated with IPH and FCR. Statistically significant parameters identified during multivariate analysis were used to construct CAS. CAS was then applied to the remaining arteries (40%, test group), and the accuracy of classification for determining the presence versus absence of IPH or, separately, FCR was determined by ROC analysis and calculation of the AUC.

RESULTS

The maximum proportion of the arterial wall occupied by the LRNC was the strongest predictor of IPH (P < .001) and FCR (P < .001) during multivariate analysis of the training group. The subsequently derived CAS applied to the test group was an accurate classifier of IPH (AUC = 0.91) and FCR (AUC = 0.93). Compared with MRA stenosis, CAS was a stronger classifier of both IPH and FCR.

CONCLUSIONS

LRNC quantification may be an effective complementary strategy to stenosis for classifying carotid atherosclerotic disease severity. CAS forms the foundation for a simple imaging-based risk-stratification system in the carotid artery to classify severity of atherosclerotic disease.