Utilizing machine learning to improve clinical trial design for acute respiratory distress syndrome.

Heterogeneous patient populations, complex pharmacology and low recruitment rates in the Intensive Care Unit (ICU) have led to the failure of many clinical trials. Recently, machine learning (ML) emerged as a new technology to process and identify big data relationships, enabling a new era in clinical trial design. In this study, we designed a ML model for predictively stratifying acute respiratory distress syndrome (ARDS) patients, ultimately reducing the required number of patients by increasing statistical power through cohort homogeneity. From the Philips eICU Research Institute (eRI) database, no less than 51,555 ARDS patients were extracted. We defined three subpopulations by outcome: (1) rapid death, (2) spontaneous recovery, and (3) long-stay patients. A retrospective univariate analysis identified highly predictive variables for each outcome. All 220 variables were used to determine the most accurate and generalizable model to predict long-stay patients. Multiclass gradient boosting was identified as the best-performing ML model. Whereas alterations in pH, bicarbonate or lactate proved to be strong predictors for rapid death in the univariate analysis, only the multivariate ML model was able to reliably differentiate the disease course of the long-stay outcome population (AUC of 0.77). We demonstrate the feasibility of prospective patient stratification using ML algorithms in the by far largest ARDS cohort reported to date. Our algorithm can identify patients with sufficiently long ARDS episodes to allow time for patients to respond to therapy, increasing statistical power. Further, early enrollment alerts may increase recruitment rate.

Plasmodium falciparum and Plasmodium yoelii: effect of the iron chelation prodrug dexrazoxane on in vitro cultures.

To determine if an iron-chelating prodrug that must undergo intracellular hydrolysis to bind iron has antimalarial activity, we examined the action of dexrazoxane on Plasmodium falciparum cultured in human erythrocytes and P. yoelii cultured in mouse hepatocytes. Dexrazoxane was recently approved to protect humans from doxorubucin-induced cardiotoxicity. Using the fluorescent marker calcein, we confirmed that the iron-chelating properties of dexrazoxane are directly related to its ability to undergo hydrolysis. As a single agent, dexrazoxane inhibited synchronized cultures of P. falciparum in human erythrocytes only at suprapharmacologic concentrations (> 200 microM). In combination with desferrioxamine B, dexrazoxane in pharmacologic concentrations (100-200 microM) moderately potentiated inhibition by approximately 20%. In contrast, pharmacologic concentrations of dexrazoxane (50-200 microM) as a single agent inhibited the progression of P. yoelli from sporozoites to schizonts in cultured mouse hepatocytes by 45 to 69% (P < 0.001). These results are consistent with the presence of a dexrazoxane-hydrolyzing enzyme in hepatocytes but not in erythrocytes or malaria parasites. Furthermore, these findings suggest that dexrazoxane must be hydrolyzed to an iron-chelating intermediate before it can inhibit the malaria parasite, and they raise the possibility that the iron chelator prodrug concept might be exploited to synthesize new antimalarial agents.

Protective effects of calcium channel blockers against free radical-impaired endothelial cell proliferation.

We have shown previously that the lipophilic calcium channel blockers exhibit membrane antioxidant activity. In the present study, when attached bovine aortic endothelial cells were exposed for 20 min to a low concentration of oxy-radicals generated from dihydroxyfumarate + Fe-ADP, no loss of glutathione or viability was detected; however, cell number, determined 48 hr later by the tetrazolium salt MTT assay, decreased to 45% of controls. Treatment of the cells for 1 hr with the calcium blockers (2-20 microM) prior to free radical exposure protected against the impaired cell growth in a concentration-dependent manner. The order of potency was nicardipine > or = nifedipine > or = verapamil > diltiazem, which appears to parallel their antioxidant potency. In addition, (+)-nicardipine, and its pharmacologically inactive isomer, (-)-nicardipine, were similarly effective. We conclude that it was the antioxidant activity of the calcium channel blockers that preserved the cell growth capacity against free-radical damage; such protective effects may contribute to their antiatherogenic effects.

Antioxidant effects of calcium channel blockers against free radical injury in endothelial cells. Correlation of protection with preservation of glutathione levels.

The effects of four calcium channel blockers (nicardipine, nifedipine, verapamil, and diltiazem) on free radical injury in cultured endothelial cells were studied and compared with those of butylated hydroxytoluene. When the cultured cells were exposed to a superoxide and hydroxyl radical generating system for up to 60 minutes, lipid peroxidation occurred, and cellular viability decreased by 60% at 30 minutes. Concomitantly, total cellular glutathione decreased by 40%, whereas total protein thiols changed minimally. Preincubation of the cells with each of the calcium blockers (5 and 20 microM) before free radical addition resulted in various degrees of significant protection against cell death, and losses of glutathione correlated significantly (r = 0.89, p less than 0.001). The order of efficacy was nicardipine greater than nifedipine greater than verapamil greater than diltiazem; butylated hydroxytoluene was about fourfold more potent than nicardipine. Because none of the agents affected the level of hydroxyl radicals generated in the aqueous phase, the data suggest that the protective mechanisms were mediated by their lipid antiperoxidative activities, which also prevented the glutathione decrease caused by inhibition of peroxide generation.