A deep-learning model using enhanced chest CT images to predict PD-L1 expression in non-small-cell lung cancer patients

AIM

To develop a deep-learning model using contrast-enhanced chest computed tomography (CT) images to predict programmed death-ligand 1 (PD-L1) expression in patients with non-small-cell lung cancer (NSCLC).

MATERIALS AND METHODS

Preoperative enhanced chest CT images and immunohistochemistry results for PD-L1 expression (<1% and ≥1% were defined as negative and positive, respectively) were collected retrospectively from 125 NSCLC patients to train and validate a deep-learning radiomics model (DLRM) for the prediction of PD-L1 expression in tumours. The DLRM was developed by combining the deep-learning signature (DLS) obtained from a convolutional neural network and clinicopathological factors. The indexes of the area under the curve (AUC), integrated discrimination improvement (IDI), and decision curve analysis (DCA) were used to evaluate the efficiency of the DLRM.

RESULTS

DLS and tumour stage were identified as independent predictors of PD-L1 expression by the DLRM. The AUCs of the DLRM were 0.804 (95% confidence interval: 0.697-0.911) and 0.804 (95% confidence interval: 0.679-0.929) in the training and validation cohorts, respectively. IDI analysis showed the DLRM had better diagnostic accuracy than DLS (0.0028 [p<0.05]) in the validation cohort. Additionally, DCA revealed that the DLRM had more net benefit than the DLS for clinical utility.

CONCLUSION

The proposed DLRM using enhanced chest CT images could function as a non-invasive diagnostic tool to differentiate PD-L1 expression in NSCLC patients.

A real-world study in advanced non-small cell lung cancer with de novo brain metastasis

Brain metastases are the major cause of life-expectancy shortened for patients with lung cancer. The prognostic value of EGFR mutation subtypes and survival benefit of EGFR-tyrosine kinase inhibitors (TKIs) in advanced non-small cell lung cancer (NSCLC) patients with de novo brain metastasis is still not clear. Here, we present a real-world study nation-wide focusing on the prognostic value of genomic and therapeutic factors in overall survival (OS) of those patients. We enrolled a total of 233 patients diagnosed with advanced NSCLC and de novo BM from multi-medical centers across China. The enrolled patients were divided into 4 groups, including EGFR 19del, EGFR L858R, EGFR wild-type, and EGFR unknown groups. The median OS of patients with EGFR mutations and all patients were 29.0 and 25.0 months, respectively. There was significant difference in OS of patients among EGFR 19del (n=76), EGFR L858R (n=94), EGFR wild-type (n=46) and EGFR unknown (n=17) groups (30.5 vs 27.5 vs 16.0 vs 25.0, P=0.025). Patients treated by icotinib showed better OS than gefitinib and erlotinib (31.0 vs 25.5 vs 26.5, P=0.02). There was a difference in OS of patients received the whole-brain radiotherapy (WBRT), stereotactic radiosurgery (SRS), or WBRT+SRS (20.0 vs 31.0 vs 30.0 months, P<0.001), respectively. In multivariate analysis, patients treated with icotinib had superior iPFS benefit than gefitinib and erlotinib (HR=0.86[95%CI (0.74-1.0)], P=0.04). Besides, the histology of non-adenocarcinomas, the number of BM (>3), and extracranial metastases status could have an independent negative impact on the OS of all patients (P<0.001). EGFR mutant NSCLC patients with de novo BM had a better OS than patients with EGFR wild type. Patients treated with icotinib had longer iPFS than gefitinib and erlotinib but not in OS. Non-adenocarcinomas, number of BM (>3) and extracranial metastases were independent negative prognostic factors in iPFS and OS of all patients. Prospective clinical trials are warranted to explore more effective multimodality in this population.

Global microRNA profiles and signaling pathways in the development of cardiac hypertrophy

Hypertrophy is a major predictor of progressive heart disease and has an adverse prognosis. MicroRNAs (miRNAs) that accumulate during the course of cardiac hypertrophy may participate in the process. However, the nature of any interaction between a hypertrophy-specific signaling pathway and aberrant expression of miRNAs remains unclear. In this study, Spague Dawley male rats were treated with transverse aortic constriction (TAC) surgery to mimic pathological hypertrophy. Hearts were isolated from TAC and sham operated rats (n=5 for each group at 5, 10, 15, and 20 days after surgery) for miRNA microarray assay. The miRNAs dysexpressed during hypertrophy were further analyzed using a combination of bioinformatics algorithms in order to predict possible targets. Increased expression of the target genes identified in diverse signaling pathways was also analyzed. Two sets of miRNAs were identified, showing different expression patterns during hypertrophy. Bioinformatics analysis suggested the miRNAs may regulate multiple hypertrophy-specific signaling pathways by targeting the member genes and the interaction of miRNA and mRNA might form a network that leads to cardiac hypertrophy. In addition, the multifold changes in several miRNAs suggested that upregulation of rno-miR-331*, rno-miR-3596b, rno-miR-3557-5p and downregulation of rno-miR-10a, miR-221, miR-190, miR-451 could be seen as biomarkers of prognosis in clinical therapy of heart failure. This study described, for the first time, a potential mechanism of cardiac hypertrophy involving multiple signaling pathways that control up- and downregulation of miRNAs. It represents a first step in the systematic discovery of miRNA function in cardiovascular hypertrophy.

Glutamatergic activation of the amygdala differentially mimics the effects of audiogenic seizure kindling in two substrains of genetically epilepsy-prone rats

Comparisons of neuronal network mechanisms in closely related inherited seizure models are providing novel insights into epileptogenic pathophysiology. Genetically epilepsy-prone rats (GEPRs) exist in two substrains that inherit long-term susceptibility to behaviorally distinct audiogenic seizures (AGS). GEPR-3s exhibit generalized clonic AGS, while GEPR-9s exhibit generalized tonic AGS. After AGS kindling the tonic AGS of GEPR-9s is followed by generalized posttonic clonus (PTC), while the generalized clonic AGS is followed by facial and forelimb (F&F) clonus in GEPR-3s. PTC and F&F clonus are very rare in GEPRs before AGS kindling. The neuronal network subserving AGS in GEPR-9s lies exclusively in brainstem sites, but amygdala (AMG) and other sites are recruited into the network after AGS kindling. The present study attempted to mimic the effects of AGS kindling by bilaterally microinjecting subconvulsive doses of N-methyl-D-aspartate (NMDA) into the AMG of nonkindled GEPRs. NMDA (10 nmol/side) microinjected into AMG reversibly induced susceptibility to F&F clonus immediately following generalized clonic AGS in most nonkindled GEPR-3s. NMDA (7.5 nmol/side), microinjected into AMG temporarily induced susceptibility to generalized PTC immediately following tonic AGS in most nonkindled GEPR-9s. No seizures were induced in normal rats by these treatments, and no seizures were seen in GEPRs with these NMDA doses except those induced by acoustic stimuli. These findings support a critical role in AGS kindling for the AMG in the neuronal networks for both forms of AGS. However, the behavioral effect of the treatment was different in the two AGS substrains, suggesting interrelated but not identical pathophysiological mechanisms in these closely related epilepsy models.

Modulation of audiogenically kindled seizures by gamma-aminobutyric acid-related mechanisms in the amygdala

Repetitive induction of audiogenic seizures (AGSs) ("AGS kindling") results in expansion of the AGS neuronal network from the brainstem to forebrain structures. AGSs in kindled genetically epilepsy-prone rats (GEPR-9s) exhibit a significant increase in the duration of posttonic clonus (PTC). The amygdala (AMG) does not appear to be a required network component before AGS kindling, but this structure is implicated in the seizure network after AGS kindling. gamma-Aminobutyric acid (GABA) is a major neurotransmitter in AMG, and histamine receptor activation is also reported to stimulate GABA release. The present study examined the effect on audiogenically kindled seizures of focal microinjections into the AMG of GEPR-9s. AGS kindling involved induction of 14 AGSs in GEPR-9s. Bilateral microinjection of a GABA(A) agonist, muscimol (0.3 nmol/side), into the AMG significantly reduced the duration of PTC, starting 0.5 h after drug infusion, with recovery by 24 h. Microinjection of histamine (60 nmol/side) suppressed PTC at 0.5 h, with total blockade at 24 h, but the seizure pattern did not revert to that observed before kindling until 120 h. This long duration suggests that mechanisms in addition to modulation of GABA function may be involved in the effect of histamine. The wild running and tonic components of AGS were never affected by microinjection of these agents into the AMG. These findings confirm previous work suggesting that the AMG is not a required nucleus in the AGS neuronal network before kindling. However, the AMG becomes critical in expansion of the seizure network during AGS kindling, and audiogenically kindled seizures are negatively modulated by increased GABA function.

Modulation of audiogenic seizures by histamine and adenosine receptors in the inferior colliculus

Susceptibility to behaviorally similar audiogenic seizures (AGS) occurs genetically and is inducible during ethanol withdrawal (ETX). Comparisons between AGS mechanisms of genetically epilepsy-prone rats (GEPR-9s) and ethanol-withdrawn rats (ETX-Rs) are yielding information about general pathophysiological mechanisms of epileptogenesis. The inferior colliculus (IC) is the AGS initiation site. Excitatory amino acid (EAA) abnormalities in the IC are implicated in AGS, and histamine and adenosine receptor activation each reduce EAA release and inhibit several seizure types. Previous studies indicate that focal infusion of an adenosine receptor agonist into the IC blocked AGS in GEPR-9s, but the effects of adenosine receptor activation in the IC on AGS in ETX-Rs are unknown. The effects of histamine receptor activation on either form of AGS are also unexamined. The present study evaluated effects of histamine or a nonselective adenosine A(1) agonist, 2-chloroadenosine, on AGS by focal microinjection into the IC. Ethanol dependence and AGS susceptibility were induced in normal rats by intragastric ethanol. Histamine (40 or 60 nmol/side) significantly reduced AGS in GEPR-9s, but histamine in doses up to 120 nmol/side did not affect AGS in ETX-Rs. 2-Chloroadenosine (5 or 10 nmol/side) did not affect AGS in ETX-Rs, despite the effectiveness of lower doses of this agent in GEPR-9s reported previously. Thus, histamine and adenosine receptors in the IC modulate AGS of GEPR-9s, but do not modulate ETX-induced AGS. The reasons for this difference may involve the chronicity of AGS susceptibility in GEPR-9s, which may lead to more extensive neuromodulation as compensatory mechanisms to limit the seizures compared to the acute AGS of ETX-Rs.

The induction effect of rifampicin on activity of mephenytoin 4'-hydroxylase related to M1 mutation of CYP2C19 and gene dose

AIMS

To determine the induction effect of rifampicin on the activity of 4'-hydroxylase in poor metabolizers (PMs) with m1 mutation of S-mephenytoin 4'-hydroxylation and the relationship of the effect with gene dose.

METHODS

Seven extensive metabolizers (EMs) of S-mephenytoin 4'-hydroxylation and five PMs with m1 mutation were chosen to take rifampicin 300 mg day(-1) orally for 22 days. Prior to and after rifampicin treatment, each subject was given racemic mephenytoin 100 mg. The 4'-hydroxymephenytoin (4'-OH-MP) excreted in the 0-24 h urine and mephenytoin S/R ratio in the 0-8 h urine were determined by h.p.l.c. and GC, respectively.

RESULTS

In all EMs, the excretion of 4'-OH-MP in the 0-24 h urine was increased by 146.4 +/- 17.9%, 0-8 h urinary mephenytoin S/R ratio was decreased by 77.3 +/- 8.8%, the percentage increase in the 0-24 h excretion of 4'-OH-MP in those CYP2C19 homozygous (wt/wt) was greater than that in those heterozygous (wt/m1 and wt/m2) (203.9 +/- 42.5% vs 69.6 +/- 4.1%). 0-8 h urinary mephenytoin S/R ratio of those PMs with m1 mutation was decreased by 9.6%, the amount of 4'-OH-MP excreted in the 0-24 h urine was increased by 80.1 +/- 48.0%.

CONCLUSIONS

The activity of 4'-hydroxylase of PMs with m1 mutation of S-mephenytoin 4'-hydroxylation can be induced by rifampicin and the inducing effect of rifampicin on 4'-hydroxylase is gene dependent.