No abstract

It is important to identify the patients with high-risk progression to develop severe acute pancreatitis (SAP). The study was to assess whether neutrophil to lymphocyte ratio (NLR) and fluid sequestration (FS) could represent useful markers for predicting the severity. A total of 1639 patients who underwent clinical diagnosis of AP was performed. Various serologic and clinical parameters on admission were investigated. Chronologic change in NLR and FS were analyzed, and theirs utility for predicting severity of AP was evaluated by receiver operator characteristic (ROC) curve analysis. Correlation analysis was assessed by Spearman’s rank test. NLR and FS levels were both increased significantly in SAP and positively correlated with Ranson score and hospital stays. The ROC curve analyses showed the optimal cut-off values of NLR for admission with day0, day1, day2 were 9.64, 6.66 and 6.50, giving sensitivity of 77–82%. The optimal cut-off values of FS for admission with day1, day2, day3 were 1375 ml, 2345 ml and 3424 ml, giving sensitivity of 62–75%. Moreover, measurement of NLR and FS together exhibited a similar area under curve (AUC) and sensitivity for SAP prediction compared with the those of Ranson score. Increase of NLR and FS are correlated with severity and can be suggested as a predictive factor in an early stage of AP.

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT• Human pregnane X receptor (PXR/NR1I2) is a key regulator of cytochrome P450 3A4. • To date, there are 198 reported SNPs for the human PXR/NR1I2 gene.• Some of these SNPs are found to affect the inducing ability of PXR to CYP3A4. WHAT THIS STUDY ADDS• This study, for the first time, has investigated the effect of PXR haplotype on basal and St John's wort-induced CYP3A4 activity in humans.• H1/H1 of the PXR gene had weaker basal transcriptional activity but greater inducible transcriptional activity to CYP3A4 than H1/H2 and H2/H2. AIMSHuman pregnane X receptor (PXR/NR1I2) is the master regulator of CYP3A4, which metabolizes >50% of drugs on the market. This study investigated the relationship between the two most frequent haplotypes [H1 (TCAGGGGCCACC) and H2 (CCGAAAACTAAT)] of PXR and basal and St John's wort (SJW)-induced CYP3A4 activity. METHODSTen healthy subjects carrying H1 and H2 haplotypes (three subjects with H1/H1, four with H1/H2 and three with H2/H2) entered this study. The 10 subjects did not carry CYP3A4*4, *5 and *6. All subjects were administrated a 300-mg SJW tablet three times daily for 14 days, and CYP3A4 activity was measured using nifedipine (NIF) as a probe. The plasma concentrations of NIF and dehydronifedipine (DNIF) were determined by a validated liquid chromatography/mass spectrometry/mass spectrometry method. RESULTSAfter administration of SJW, the AUC0-• of NIF decreased significantly, and the AUC0-• of DNIF increased significantly (P < 0.05). For H1/H2, the AUC0-• of NIF decreased by 42.4%, and the AUC0-• of DNIF increased by 20.2%; for H2/H2, the AUC0-• of NIF decreased by 47.9%, and the AUC0-• of DNIF increased by 33.0%; for H1/H1, the AUC0-• of NIF decreased by 29.0%, yet the AUC0-• of DNIF increased by 106.7%. The increase of the AUC0-• of DNIF in H1/H1 was significantly different from the other two haplotype pairs (P < 0.05). Meanwhile, before administration of SJW, the ratio of AUC0-•(DNIF)/AUC0-•(NIF) was the lowest for H1/H1 (22.1%), compared with H1/H2 (58.7%) and H2/H2 (30.0%). CONCLUSIONSH1/H1 of the human PXR gene had weaker basal transcriptional activity but greater inducible transcriptional activity to CYP3A4 than H1/H2 and H2/H2.

In this paper, a channel state information (CSI) feedback method is proposed based on deep transfer learning (DTL). The proposed method addresses the problem of high training cost of downlink CSI feedback network in frequency division duplexing (FDD) massive multiple-input multiple-output (MIMO) systems. In particular, we obtain the models of different wireless channel environments at low training cost by finetuning the pre-trained model with a relatively small number of samples. In addition, the effects of different layers on training cost and model performance are discussed. Furthermore, a modelagnostic meta-learning (MAML)-based method is proposed to solve the problem associated with large number of samples of a wireless channel environment required to train a deep neural network (DNN) as a pre-trained model. Our results show that the performance of the DTL-based method is comparable with that of the DNN trained with a large number of samples, which demonstrates the effectiveness and superiority of the proposed method. At the same time, although there is a certain performance loss compared with the DTL-based method, the MAML-based method shows good performance in terms of the normalized mean square error (NMSE).