Refining outcome prediction after traumatic brain injury with machine learning algorithms.

Outcome after traumatic brain injury (TBI) is typically assessed using the Glasgow outcome scale extended (GOSE) with levels from 1 (death) to 8 (upper good recovery). Outcome prediction has classically been dichotomized into either dead/alive or favorable/unfavorable outcome. Binary outcome prediction models limit the possibility of detecting subtle yet significant improvements. We set out to explore different machine learning methods with the purpose of mapping their predictions to the full 8 grade scale GOSE following TBI. The models were set up using the variables: age, GCS-motor score, pupillary reaction, and Marshall CT score. For model setup and internal validation, a total of 866 patients could be included. For external validation, a cohort of 369 patients were included from Leuven, Belgium, and a cohort of 573 patients from the US multi-center ProTECT III study. Our findings indicate that proportional odds logistic regression (POLR), random forest regression, and a neural network model achieved accuracy values of 0.3-0.35 when applied to internal data, compared to the random baseline which is 0.125 for eight categories. The models demonstrated satisfactory performance during external validation in the data from Leuven, however, their performance were not satisfactory when applied to the ProTECT III dataset.

Valveless pumping mechanics of the embryonic heart during cardiac looping: Pressure and flow through micro-PIV.

Cardiovascular development is influenced by the flow-induced stress environment originating from cardiac biomechanics. To characterize the stress environment, it is necessary to quantify flow and pressure. Here, we quantify the flow field in a developing zebrafish heart during the looping stage through micro-particle imaging velocimetry and by analyzing spatiotemporal plots. We further build upon previous methods to noninvasively quantify the pressure field at a low Reynolds number using flow field data for the first time, while also comparing the impact of viscosity models. Through this method, we show that the atrium builds up pressure to ~0.25mmHg relative to the ventricle during atrial systole and that atrial expansion creates a pressure difference of ~0.15mmHg across the atrium, resulting in efficient cardiac pumping. With these techniques, it is possible to noninvasively fully characterize hemodynamics during heart development.

Effect of zidovudine on human placental trophoblast and Hofbauer cell functions.

We have optimized a procedure to isolate placental trophoblasts and Hofbauer cells simultaneously in a quantity sufficient for short-term cultures and then used these placental cells to investigate the effects of zidovudine (ZDV) on trophoblast and Hofbauer cell functions. Of more than 10 term placentas tested, ZDV inhibits DNA synthesis of trophoblasts in a concentration-dependent manner with half the maximal inhibitory concentration (IC50) of 9.88 +/- 1.35 microM. Of the hormones evaluated, production of progesterone by trophoblasts is most sensitive to ZDV (IC50 = 3.65 +/- 0.29 microM). The inhibitory effect of ZDV on the secretion of placental lactogen and choriogonadotropin by the trophoblasts was detected only at a much higher concentration (> or = 60 microM). ZDV does not affect trophoblast or Hofbauer cell protein synthesis. Collectively, our results indicate that at clinically relevant concentrations (< or = 10 microM), ZDV significantly inhibits both the DNA synthesis of placental trophoblasts and their production of progesterone, while having a minimal effect on protein synthesis of both types of placental cells.

Localized release of perivascular heparin inhibits intimal proliferation after endothelial injury without systemic anticoagulation.

Segmental endothelial desquamation of the common carotid artery was produced in 30 rats using a balloon catheter technique which produces consistent proliferation of intimal smooth muscle cells from 5 to 20 days after injury. Immediately after endothelial injury, 15 animals were treated with periadventitial application of heparin contained in a continuous-release drug-delivery system using the polymer polyvinyl alcohol (PVA) and PVA alone applied in a similar fashion to 15 control rats. Animals were killed at 5, 10 and 20 days, respectively, after surgery by intracardiac perfusion-fixation, and vessels were prepared for light microscopy, scanning electron microscopy, and immunohistochemistry with antibodies directed against actin. At all time periods, there was a significant reduction in intimal cross-sectional area in heparin/PVA-treated vessels compared to control vessels. Scanning electron microscopy showed complete absence of endothelial cells from the luminal surface in both control and treated arteries at all time periods without evidence of significant platelet aggregation. Immunohistochemistry demonstrated the presence of immunoreactive actin in the proliferating myointimal cells. Femoral venous prothrombin time and partial thromboplastin time were unchanged in heparin/PVA-treated animals compared to controls at 1, 5, and 10 days. Continuous-release polymer drug delivery can be used to apply heparin selectively to the adventitial surface of vessels and effect changes in the vessel wall over periods of up to 3 weeks. By this means, smooth muscle proliferation and subsequent vessel narrowing after endothelial injury were inhibited without systemic anticoagulation. This technique may be applicable to both clinical and research applications related to the pathophysiology of arterial injury.

A rat femoral artery model for vasospasm.

A new animal model for vasospasm using rat femoral artery has been developed. Whole blood, washed erythrocytes, or leukocytes in platelet-rich plasma were selectively applied to the adventitial surface of the femoral artery for 7 days in 15 rats, after which the vessels were perfusion-fixed and examined by light and transmission electron microscopy and immunohistochemistry. As compared with matched control arteries, there was a prominent reduction in luminal cross-sectional area after 7 days in vessels exposed to whole blood or washed erythrocytes, but not in those exposed to leukocytes in platelet-rich plasma. In arteries with luminal narrowing, light and transmission electron microscopy demonstrated marked morphological changes throughout the vessel wall similar to those seen in cerebral vasospasm after subarachnoid hemorrhage. Immunohistochemistry disclosed a prominent loss of immunoreactive actin in smooth muscle cells of arteries exposed to whole blood or erythrocytes. To assess the time course of arterial narrowing in this model, whole blood was selectively applied to the adventitial surface of femoral arteries in 23 rats for periods from 2 to 20 days. As compared with control arteries, arterial narrowing was variably present at 2 days, progressively increased by 5 days, was maximal at 7 to 10 days, and returned to near control levels by 20 days. The presence and severity of ultrastructural changes in vessel wall corresponded to the degree of arterial narrowing over time. These results suggest that chronic narrowing in rat femoral artery exposed to periadventitial blood is analogous to that observed in cerebral arterial vasospasm after subarachnoid hemorrhage. This new model represents a simple and reliable means to investigate pathogenic mechanisms and potential therapies for vasospasm.

The significance of morphological changes in cerebral arteries after subarachnoid hemorrhage.

A porcine model was developed to allow quantitative assessment of morphological changes in cerebral arteries after subarachnoid hemorrhage and to determine the significance of structural changes in producing arterial narrowing. Whole blood was selectively applied to the middle cerebral artery (MCA) of seven pigs. After 10 days, vessels were perfusion-fixed and examined by light and transmission electron microscopy and immunohistochemistry. The MCA's exposed to whole blood for 10 days showed prominent luminal narrowing associated with profound ultrastructural changes affecting all layers of the vessel wall. Morphometric analysis, however, demonstrated that significant reductions in the luminal cross-sectional area (-55.8% +/- 12.5%, p less than 0.005) and increases in radial wall thickness (75.1% +/- 10.5%, p less than 0.005) were associated with only minimal increase in the cross-sectional area of the vessel wall (12.5% +/- 15%, p less than 0.025). By stereological analysis, the volume density of individual components of the arterial wall was unchanged in MCA's exposed to blood. Vessels exposed to blood showed a 44% reduction in smooth-muscle cell immunoreactive actin and increased collagen in the extracellular matrix of the vessel wall. These data suggest that structural changes in cerebral arteries after subarachnoid hemorrhage do not directly contribute to vessel narrowing through increases in wall mass. Nevertheless, such changes may reflect pathological mechanisms which act to augment prolonged vasoconstriction or inhibit the maintenance of normal vascular tone.

The role of hemoglobin in arterial narrowing after subarachnoid hemorrhage.

A porcine model for subarachnoid hemorrhage has been developed to allow the selective application of blood and its components to cerebral arteries. Whole blood was centrifuged to produce two fractions consisting of washed erythrocytes (red blood cells, RBC's) and white blood cells (WBC) plus platelet-rich plasma (PRP); the RBC fraction was subsequently separated into hemoglobin (Hb)-containing cytosol and erythrocyte membranes. Each fraction was selectively applied to the middle cerebral artery (MCA) of pigs for 10 days; after which, vessels were perfusion-fixed and examined by light and transmission electron microscopy and immunohistochemical studies. By morphometric analysis, a marked reduction in the MCA lumen cross-sectional area was observed after selective application of RBC's or Hb/cytosol but not of WBC/PRP or erythrocyte membranes. In both RBC- and Hb/cytosol-treated vessels, luminal narrowing was associated with a differential increase in vessel wall thickness of the ventral (subarachnoid) compared to the dorsal (brain) aspect of the artery, but no significant change in cross-sectional area of the vessel wall. After 10 days of exposure to RBC's or Hb/cytosol, there was a spectrum of ultrastructural changes in the vessel wall comparable to those seen after periadventitial application of whole blood. Selective application of commercially available Hb to MCA produced similar structural and morphometric changes. The degree of luminal narrowing after exposure to whole blood or RBC's was proportional to the volume of the erythrocyte mass adjacent to the vessel at sacrifice. These data suggest that arterial narrowing after SAH is mediated by mechanisms related to prolonged exposure of the vessel wall to hemoglobin or its catabolites from lysing subarachnoid erythrocytes.

Morphologic changes in cerebral arteries after subarachnoid hemorrhage.

The premise of this article is that morphologic changes observed in cerebral arteries after subarachnoid hemorrhage play an important role in the pathogenesis of associated ischemic deficits observed in this disorder. Secondly, the arteriopathic response of cerebral arteries to subarachnoid blood is similar in many respects to that observed in systemic vessels under various pathologic conditions, and common pathogenic mechanisms may exist. The data supporting these premises may be summarized as follows: 1. Morphologic changes in human and animal cerebral arteries after subarachnoid hemorrhage are temporally associated with angiographic and clinically significant vasospasm. 2. Profound morphologic changes in cerebral arteries after subarachnoid hemorrhage do not contribute to structural narrowing of the lumen through increases in vessel wall mass. Nevertheless, structural changes may act in concert with contractile mechanisms to alter normal physiologic responses and maintain a narrowed lumen. 3. The agent responsible for arterial narrowing and morphologic changes in cerebral arteries after subarachnoid hemorrhage is contained in the erythrocyte component of whole blood and is most likely hemoglobin. 4. The volume and duration of exposure of subarachnoid blood to the artery appears to be significant in the development of the angiopathic response. 5. Ultrastructural abnormalities in systemic vessels associated with hypertension, atherogenesis, and endothelial damage are similar in many respects to those seen after subarachnoid hemorrhage.

Rapid platelet accumulation leading to thrombotic occlusion.

Platelet thrombosis under arterial conditions remains a large clinical problem. Previous in vitro experiments have concentrated on early adherence without thrombotic occlusion. We have developed a controllable hemodynamic system that creates intravascular thrombosis to occlusion. Lightly heparinized (3.5 USP units/mL), whole, porcine blood is perfused through a 1.5 mm inner diameter, tubular, collagen-coated stenosis. The microscopic growth of thrombus is optically recorded using a high resolution CCD camera. Occlusive thrombus is examined using microcomputed tomography and histology. Thrombus consistently formed in the throat of the stenosis where wall shear rates were greatest. Rapid platelet accumulation (RPA) reached rates as high as 13.7 µm(3 )µm(-2) min(-1). Total occlusion of flow occurred after 17 ± 2.6 min (n = 6). The average thrombus volume accumulation of 7.8 ± 3.5 µm(3) µm(2) min(-1) occurred under very high wall shear rates exceeding 100,000 s(-1). Significant volumes of thrombus did not form until 7.6 ± 3.6 min after the onset of flow, a delay consistent with activation of adherent mural platelets. Platelets did not accumulate with large volume for normal wall shear rates <2000 s(-1). Very high wall shear rates stimulate the capture of millions of circulating platelets with exposure times <2 ms in an arterial stenosis.

Creatine phosphokinase activities in rainbow trout, Salmo gairdneri, associated with rapid decompression.

Serum creatine phosphokinase (CPK) activity levels were monitored in rainbow trout following rapid decompression. Serum activity was elevated 8.5-fold at 24 h and 15-fold at 48 h after decompression from 100-feet sea water (fsw) dives of 1 1/2 h duration. Serum enzyme activity was unchanged in the 66-fsw group. Stress results indicated a proportional relationship exists between serum CPK activity and muscular activity. CPK activities were found to be 1,023 in skeletal (white) muscle and 255 mumol . min-1 . g-1 in heart ventricle. Elevated CPK activity was not associated with tissue necrosis and is considered a consequence of increased anaerobic respiration stimulated by muscular activity and pressure-induced peripheral hypoxia.

Local anticoagulation without systemic effect using a polymer heparin delivery system.

Polymeric drug delivery systems that allow the application of substances to a localized region for specified periods of time have been developed. A model for intravascular thrombosis in the rat common carotid artery was established using a combination of balloon catheter endothelial injury with 1-hour occlusion of the vessel. After endothelial injury in 11 Sprague-Dawley rats, the adventitial surface of the carotid artery was exposed to the polymer polyvinyl alcohol (PVA) containing heparin and was compared with exposure to PVA alone in the contralateral (control) vessel. Subsequent determinations of the coagulation parameters systemic prothrombin and partial thromboplastin times showed no systemic effect of heparin. All 11 control vessels demonstrated complete or partial thrombosis, whereas only one of 11 heparin/PVA-treated vessels showed a small thrombus. Morphometric analysis of the cross-sectional thrombus: lumen ratio in 10 rats showed a significant reduction (p less than 0.005) in thrombus size for treated vessels (4.1 +/- 9.6%) compared with control vessels (60.2 +/- 25.8%). Scanning electron microscopy verified the absence of thrombus in the treated vessels despite complete endothelial desquamation. In a second group of eight rats, endothelial injury without occlusion did not cause thrombosis in treated or control arteries. The coagulation parameters for this group of eight unoccluded rats were similarly unaffected by heparin/PVA treatment. Our observations suggest that a localized antithrombotic effect of heparin can be achieved without systemic anticoagulation using a polymeric drug delivery system. This technique may be applied to a variety of surgical and nonsurgical clinical conditions.

Embryotoxic effects of salicylates: role of biotransformation.

The three major metabolites of salicylate, o- hydroxyhippurate ( salicylglycine , salicyluric acid), 2,5-dihydroxybenzoate (gentisic acid), and 2,3-dihydroxybenzoate, were examined for their capacities to elicit dysmorphogenesis, embryolethality , and growth retarding effects in an embryo culture system. The effects were compared with those produced by the parent salicylate. At the highest concentrations tested (1.9 mM), none of the three metabolites produced significant increases in the number of malformed embryos or in embryolethality . At the same concentration, all three agents reduced crown-rump lengths and somite numbers slightly but significantly (p less than 0.01), and the dihydroxy metabolites also reduced the embryonic protein content (p less than 0.01). In contrast, the parent salicylate produced large increases in embryolethality ( embryolethality in controls was 6% or less) and malformed embryos at equivalent or lower concentrations. Preincubation of the parent salicylate with various biotransforming systems did not affect embryotoxicity significantly. The most rapid biotransformation of salicylate in vitro was achieved with mitochondrial preparations of monkey kidney as the enzyme source but quantities metabolized were not sufficient to prevent malformations in the culture system. Increased serum protein concentration in the culture medium, however, markedly reduced the capacity of added salicylate to cause malformations. An examination of the kinetics of the dysmorphogenic effects of parent salicylate indicated that 5 hr of exposure elicited nonsignificant increases in numbers of malformations. A significant malformation rate was produced by 9 hr of exposure. In contrast, effects on embryonic growth parameters and embryolethality were greatest after a 24-hr exposure period. The results strongly suggest that the parent salicylate, rather than generated metabolites, was primarily or solely responsible for the malformations observed and that the duration of exposure of embryos to unmetabolized salicylate may be the critical factor for determining teratogenic outcome.

Generation of reactive dysmorphogenic intermediates by rat embryos in culture: effects of cytochrome P-450 inducers.

We have investigated the capacity of cultured whole rat embryos to convert 2-acetylaminofluorene (AAF) to reactive metabolites capable of eliciting dysmorphogenic effects in the same embryos. Cultured embryos (Sprague-Dawley) were exposed to AAF for periods of 2 or 24 hr, after which metabolites were isolated from the culture medium and identified with HPLC. Embryotoxic effects were evaluated in the same embryos. Day 10 embryos preexposed in utero to pregnenolone-16 alpha-carbonitrile (PCN) exhibited marked increases in capacity to convert AAF to a variety of hydroxylated metabolites. 3-Methylcholanthrene (3MC) was also a very effective inducer in utero but Aroclor 1254 (PCB), and isosafrole (ISF) evoked only minimal induction while phenobarbital (PB) was not demonstrably effective. Exogenously added hepatic postmitochondrial supernatant (S9) fractions from adult male rats pretreated with PCB, 3MC, or ISF exhibited induced monooxygenase activities as well as increased capacity to convert AAF to dysmorphogenic intermediates in the culture system. PB and PCN displayed much lesser effects. PCN was a very effective inducer of hepatic monooxygenases of pregnant rats but, when this tissue was utilized as an enzyme source, no significant increase in malformations was observed. Embryos with relatively high monooxygenase activities also displayed a high incidence of embryonic abnormalities when cocultured with AAF. Malformation incidence was strongly correlated with hydroxy metabolite generation, suggesting that induction in utero of P-450-dependent, embryonic monooxygenases resulted in the production of embryotoxic metabolites by the embryos own enzymes. The data also indicated that endogenous bioactivation (within the conceptus) was considerably more effective than bioactivation effected by an exogenous (hepatic) enzyme source.

Thrombospondin deposition in rat carotid artery injury.

The balloon catheter injury model was used to determine the relative contributions of vascular smooth muscle cells (SMC) and platelets to thrombospondin (TSP) antigen deposition in the artery wall. Rat carotid arteries were denuded of endothelium, exposing the thrombogenic subendothelial extracellular matrix (ECM) to the circulation. Rats were killed after 1 hour, or 5, 10, or 20 days. Thrombospondin antigen deposition in the injured arteries was assessed using a specific polyclonal antiserum raised in rabbit against rat platelet TSP and a sensitive silver-enhanced immunogold staining method. Faint immunostaining for TSP antigen was detected, associated mostly with cells, in the media of the carotid artery of the nonoperated controls. One hour after balloon catheter injury, however, prominent cell-associated immunostaining was evident in the media; extracellular matrix staining was negligible. At this time, large foci of immunostaining were present on the lumenal surface of the vessel. Intimal proliferation was evident on most stained sections of tissue taken 5 days after balloon injury. Thrombospondin antigen immunostaining was markedly increased compared to nonoperated controls in all sections, regardless of the degree of intimal thickening. Thrombospondin immunostaining remained associated with cells in the neointima and media; extracellular matrix staining remained negligible. Ten days after endothelial injury, immunostaining for TSP antigen was detected in all layers of the artery, but was greater in the neointima and media. Reaction product was still associated only with cells. Thrombospondin antigen levels, as detected by this procedure, remained high in the injured tissue through 10 days of observation but appeared less prominent 20 days after injury. At this time extracellular matrix staining was obvious and cell-associated staining was reduced. These data support the hypotheses that thrombospondin (TSP) expression by vascular smooth muscle cells is an early response to injury and that the primary source of TSP antigen in injured artery is the vascular smooth muscle cells (SMC). These results support data derived from in vitro studies of TSP secretion.

Release of neuronotrophic factor from rabbit corneal epithelium during wound healing and nerve regeneration.

Epithelial neuronotropic factor (ENF) is secreted by cultured epithelial cells of rabbit cornea and conjunctiva, and is active in promoting survival and inducing neurite outgrowth of cultured trigeminal neurons. This study evaluated the relation of ENF to corneal nerve regeneration utilizing a model of heptanol-induced epithelial wounding. The organ culture technique was used to collect ENF from the intact corneal epithelium, and a neuronal bioassay was utilized to quantify ENF. The results revealed no change in ENF secretion either during initial wound closure or after 1 week, when the epithelium had regenerated. However, ENF secretion was elevated 2.4 times in 2 weeks after wounding. Morphometric analysis of corneal nerves stained by gold chloride impregnation showed that the first sign of regeneration of intraepithelial nerves was observed after 2 weeks, and the normal pattern of epithelial neural density was re-established after 3 weeks. However, the neural density was still subnormal (35-47% less than the control) in the wounded epithelium up to 4 weeks after wounding. Thus it appears that a surge in ENF secretion occurred after epithelial regeneration but before nerve regeneration. The results suggest that ENF may mediate corneal nerve regeneration.