eDoctor: machine learning and the future of medicine.

Machine learning (ML) is a burgeoning field of medicine with huge resources being applied to fuse computer science and statistics to medical problems. Proponents of ML extol its ability to deal with large, complex and disparate data, often found within medicine and feel that ML is the future for biomedical research, personalized medicine, computer-aided diagnosis to significantly advance global health care. However, the concepts of ML are unfamiliar to many medical professionals and there is untapped potential in the use of ML as a research tool. In this article, we provide an overview of the theory behind ML, explore the common ML algorithms used in medicine including their pitfalls and discuss the potential future of ML in medicine.

Assessment of body composition in dialysis patients by arm bioimpedance compared to MRI and 40K measurements.

This study used multi-frequency bioimpedance spectroscopy (BIS) of the arm and whole body to estimate muscle mass (MM) and subcutaneous adipose tissue (SAT) in 31 hemodialysis (HD) patients comparing these results with magnetic resonance imaging (MRI) and body potassium ((40)K) as gold standards. Total body and arm MM (MM(MRI)) and SAT (SAT(MRI)) were measured by MRI. All measurements were made before dialysis treatment. Regression models with the arm (aBIS) and whole body (wBIS) resistances were established. Correlations between gold standards and the BIS model were high for the arm SAT (r(2) = 0.93, standard error of estimate (SEE) = 3.6 kg), and whole body SAT (r(2) = 0.92, SEE = 3.5 kg), and for arm MM (r(2) = 0.84, SEE = 2.28 kg) and whole body MM (r(2) = 0.86, SEE = 2.28 kg). Total body MM and SAT can be accurately predicted by arm BIS models with advantages of convenience and portability, and it should be useful to assess nutritional status in HD patients.

Newer strategies for anemia prevention in hemodialysis.

Anemia prevention for hemodialysis relies primarily on supplemental erythropoietin (EPO) and intravenous iron (IV-iron). The doses of EPO utilized are somewhat higher than normal endogenous rates of EPO production in healthy subjects, and the amount of IV-iron used to boost red blood cell (RBC) production may be greater than the amounts used for erythropoiesis. EPO and IV-iron might be used more efficiently if two fundamental problems were solved in the management of dialysis patients: better vitamin C status, and avoidance of chronic inflammation. The low levels of plasma vitamin C commonly observed in dialysis patients restrict mobilization of stored iron from the reticuloendothelial system (RES), and inflammation has a very similar effect. The impact of low vitamin C levels and concurrent inflammation causes a large amount of iron to be stored, with relatively inefficient utilization for erythropoiesis. Vitamin C intake for dialysis patients is often restricted because of avoidance of vitamin C-rich foods, and because of concerns about oxalosis. Inflammation is a chronic feature of renal disease, which is compounded by infections from use of catheters. Research strategies to improve vitamin C status and to decrease inflammation would lead to better utilization of iron and EPO, and could have parallel benefits for the long-term health of patients on hemodialysis.

A kinetic model of calcium mass balance during dialysis therapy.

A kinetic model of Ca mass balance during dialysis has been developed. It is a single-compartment, variable-volume model to compute Ca mass balance during dialysis in its volume of distribution, the extracellular fluid. The model was used to analyze literature data which were suitable for the assessment of Ca mass balance over the course of dialysis. The modeled analyses predicted the serial plasma Ca concentrations very well. The mass balance analyses revealed a pool of rapidly diffusible Ca beyond the extracellular fluid distribution volume where Ca could be mobilized (M+(Ca)) or sequestered (M-(Ca)) very rapidly at rate equal but opposite in sign to dialyzer flux and thus effectively maintain near constant plasma Ca in the face of dialyzer Ca concentration gradients. This pool is likely the large pool of diffusible (miscible) Ca in connective tissue and on bone surfaces. Analysis of net Ca flux during dialysis with Cdi(Ca) = 2.50 mEq/l suggests that 80% of patients are in positive Ca balance during dialysis. Further studies are required to verify the model and to develop a model of interdialytic Ca mass balance.

Oxidative stress and inflammation in hemodialysis patients.

Dialysis is associated with an increased generation of oxidants, which play an important part in the development of endothelial dysfunction and atherogenesis. Markers of oxidative stress include F2-isoprostanes and ethane. Measurements in dialysis patients before dialysis showed higher levels of esterified plasma F2-isoprostanes (1.62 +/- 0.73 ng/mL) than in control subjects (0.27 +/- 0.10 ng/mL) (P < 0.001). Furthermore, levels also correlated with high plasma C-reactive protein (CRP) levels (r =.48, P = 0.015). Breath ethane levels for dialysis patients (N = 19) were 6.32 +/- 3.16 pmol/kg-min, in contrast to 3.08 +/- 1.50 pmol/kg-min in control subjects (N = 11, P < 0.005). Analysis to investigate the relationship between CRP levels and outcome indicated that there was a significant difference in mortality rate over a 3-year period between patients with low and high CRP values (P < 0.001). Patients with high CRP (> 16.8 mg/L) levels were more than twice as likely to die as patients with low CRP levels (relative risk [RR] = 2.16; 95% confidence interval [CI], 1.50-3.09). CRP values were a significant predictor of mortality even after controlling for diabetes, albumin, ferritin, and age at commencement of dialysis. The RR for CRP after adjustment was 1.58 (95% CI, 1.06-2.34, P = 0.024). There were no significant interactions between CRP and other predictors of mortality, indicating that high CRP levels have an additive effect on the mortality risk. These findings show that hemodialysis patients are exposed to both oxidative stress and inflammation.

The evolving role of carotenoids in human biochemistry.

The growth of our knowledge of carotenoid biochemistry has opened new and divergent paths for research. The earliest role established for beta-carotene in animals was as a vitamin A precursor, a role it shares with several other pro-vitamin A carotenoids. Additional studies have continued to refine our understanding of this function. Because carotenoids are excellent scavengers of singlet oxygen and respectable scavengers for other reactive oxygen species, substantial work was done concerning their potential role as antioxidants. In an unexpected twist, the ability of radicals in cigarette smoke to degrade carotenoids might be responsible for the finding that high-dose dietary beta-carotene increased the incidence of lung cancer in smokers. A new role for the polar carotenoids lutein and zeaxanthin was identified, when those carotenoids were found to constitute the macular pigment (the yellow spot at the center of the human retina). Many different carotenoids can be metabolized to products with retinoid activity, which might affect gene expression and cell differentiation. The formation of retinoids from diverse carotenoids might account for a portion of their activities as anticancer agents. Studies of lycopene in prostate cancer prevention have been very promising, and clinical studies of lycopene are underway. Carotenoids have emerged as the best single tissue marker for a diet rich in fruits and vegetables, and measurements of plasma and tissue carotenoids have an important role in defining the optimal diets for humans.

Dietary effects on cardiovascular disease risk factors: beyond saturated fatty acids and cholesterol.

Hypercholesterolemia represents a significant risk for cardiovascular disease (CVD). While diet intervention remains the initial choice for the prevention and treatment of CVD, the nature of the dietary modification remains controversial. For example, reducing calories from total fat, without decreasing saturated fat intake results in insignificant changes in low density lipoprotein cholesterol (LDL-C). Similarly, diet interventions that focus solely on lowering dietary cholesterol and saturated fat intake not only decrease LDL-C, but also high density lipoprotein cholesterol (HDL-C) and therefore may not improve the lipoprotein profile. This brief review summarizes dietary interventions that lower LDL-C without affecting HDL-C levels. These interventions include soy protein, soluble fiber, soy lecithin and plant sterols. This review also includes some of the reported dietary interventions, such as polyphenols, isoflavones, folic acid and vitamins B6 and B12, which reduce the risk of CVD without changes in lipoprotein cholesterol.

Elevated plasma F2-isoprostanes in patients on long-term hemodialysis.

BACKGROUND: End-stage renal disease (ESRD) patients on long-term hemodialysis (HD) may be under increased oxidative stress, caused by either HD or renal failure. Plasma F2-isoprostanes have been established as an important indicator of in vivo oxidative stress. METHODS: Plasma esterified F2-isoprostanes were measured in 25 HD patients and 23 controls with normal renal function, employing gas chromatography-mass spectrometry with negative chemical ionization (GC-MS-NCI). C-reactive protein (CRP) was determined concurrently in patients and controls by enzyme-linked immunosorbent assay (ELISA). alpha-Tocopherol, retinol, albumin and creatinine were also determined. RESULTS: The average total esterified F2-isoprostanes in the ESRD patients was 1.62 +/- 0.73 vs. 0.27 +/- 0.10 ng/mL in controls (P < 0.001), with no overlap between patients and controls. Plasma F2-isoprostanes in diabetic ESRD patients were similar to F2-isoprostanes in patients with other causes for renal failure. In a subset of 10 of these ESRD patients evaluated eight months after the initial measurement, plasma-esterified F2-isoprostanes were not altered by an individual dialysis session. Average plasma CRP values were also higher in HD patients (P < 0.02), but some patients had CRP values that were similar to controls. In the HD patients, total plasma F2-isoprostanes and plasma CRP were correlated (r = 0.48, P = 0.015). Plasma alpha-tocopherol did not differ between patients and controls, but plasma retinol was higher in patients (3.15 +/- 1.71 micromol/L) than in controls (1.97 +/- 0.51 micromol/L, P < 0.05). CONCLUSIONS: These results are consistent with the hypothesis that oxidative stress in ESRD patients contributes to increased values of esterified plasma F2-isoprostanes, with concurrent increases in plasma CRP levels in some patients. Impaired clearance of esterified F2-isoprostanes may contribute to the elevated levels in renal failure. Plasma esterified F2-isoprostanes may be a useful indicator to evaluate effectiveness of interventions to decrease oxidative stress and associated inflammation.

Relative reactivity of lysine and other peptide-bound amino acids to oxidation by hypochlorite.

Antibacterial and inflammatory responses of neutrophils and macrophages produce hypochlorite as a major oxidant. Numerous side chains of amino acids found in extracellular proteins can be modified by hypochlorite, including His, Arg, Tyr, Lys, Trp, and Met. We studied the relative reactivity of each of these amino acid residues in short N-blocked peptides, where other residues in the peptide were highly resistant to hypochlorite attack. Hypochlorite treatment led to modified peptides in each case, which were detected by changes in retention on reversed-phase HPLC. A distinct single product, consuming two equivalents of hypochlorite per equivalent of peptide, was obtained from the Lys-containing peptides. UV spectroscopy, nuclear magnetic resonance (NMR), and electrospray/mass spectroscopy identified this product as the dichloramine at the epsilon-amino group of the Lys side chain. The dichloramine at Lys did not decompose to form a detectable amount of carbonyl reactive with dinitrophenylhydrazine. The dichloramine at Lys did however quantitatively revert back to Lys during HCl digestion of the tetrapeptide for amino acid analysis, with simultaneous modification of the adjacent Phe residue. The formation of the dichloramine at Lys was not blocked by peptides or acetylated amino acids that contained Tyr, His, or Arg. In contrast, the presence of equimolar Met-containing peptide, or N-Acetyl-Trp, both inhibited the formation of the dichloramine at Lys. Thus, Met and Trp side chains of proteins might be able to protect Lys from chloramine formation under some circumstances, but this interpretation must consider that Met and Trp are typically found in relatively inaccessible hydrophobic sites, whereas lysine is typically exposed on the protein surface. The hierarchy of amino acid reactivities examined here will aid in the prediction of residues in biological samples most likely to be modified by hypochlorite.

Streamlined F2-isoprostane analysis in plasma and urine with high-performance liquid chromatography and gas chromatography/mass spectroscopy.

F2-Isoprostanes in plasma and urine are generally determined by labor-intensive methods requiring sample purification by solid-phase extraction and thin-layer chromatography (TLC). A streamlined and more sensitive method for the measurement of esterified plasma F2-isoprostanes was developed by replacing these steps with high-performance liquid chromatography (HPLC) using an amino column with a hexane/2-propanol gradient. Pentafluorobenzyl esters of F2-isoprostanes were prepared and purified by HPLC, silylated, and then analyzed by gas chromatography (GC) with negative chemical ionization mass spectroscopy (NCI-MS). This method permits analysis with lower plasma volumes (100 microL) and greater sensitivity (to 10 pg; allowing detection to 50 pg/mL) than provided by other methods. Urinary F2-isoprostanes can also be efficiently quantified by this method, with 8-iso-PGF2alpha being identified as a major isomer. With this procedure, esterified plasma F2-isoprostanes were found to be 8.3-fold higher in an end-stage renal failure patient on hemodialysis and urinary 8-iso-PGF2alpha was 7.1-fold higher in a cigarette smoker than respective control subjects. This method, particularly the substitution of the TLC step common to other methods with HPLC, results in a more sensitive and reproducible assay.

Methods for measuring ethane and pentane in expired air from rats and humans.

Numerous studies in animals and humans provide evidence that ethane and pentane in expired air are useful markers of in vivo lipid peroxidation. The measurement of breath hydrocarbons, being noninvasive, is well suited for routine use in research and clinical settings. However, the lack of standardized methods for collecting, processing, and analyzing expired air has resulted in the use of a wide variety of different methods that have yielded highly disparate results among investigators. This review outlines the methods that we have developed and validated for measuring ethane and pentane in expired air from rats and humans. We describe the advantages of these methods, their performance, as well as potential errors that can be introduced during sample collection, concentration, and analysis. A main source of error involves contamination with ambient-air ethane and pentane, the concentrations of which are usually much greater and more variable than those in expired air. Thus, it appears that the effective removal of ambient-air hydrocarbons from the subject's lungs before collection is an important step in standardizing the collection procedure. Also discussed is whether ethane or pentane is a better marker of in vivo lipid peroxidation.

Antioxidant capacity of oat (Avena sativa L.) extracts. 1. Inhibition of low-density lipoprotein oxidation and oxygen radical absorbance capacity.

Milled oat groat pearlings, trichomes, flour, and bran were extracted with methanol and the fractions tested in vitro for antioxidant capacity against low-density lipoprotein (LDL) oxidation and R-phycoerythrin protein oxidation in the oxygen radical absorbance capacity (ORAC) assay. The oxidative reactions were generated by 2,2'-azobis(2-amidinopropane) HCl (AAPH) or Cu(2+) in the LDL assay and by AAPH or Cu(2+) + H(2)O(2) in the ORAC assay and calibrated against a Trolox standard to calculate Trolox equivalents (1 Trolox equivalent = 1 TE = activity of 1 micromol of Trolox). The antioxidant capacity of the oat fractions was generally consistent with a potency rank of pearlings (2.89-8.58 TE/g) > flour (1.00-3.54 TE/g) > trichome (1.74 TE/g) = bran (1.02-1.62 TE/g) in both LDL and ORAC assays regardless of the free radical generator employed. A portion of the oat antioxidant constituents may be heat labile as the greatest activity was found among non-steam-treated pearlings. The contribution of oat tocols from the fractions accounted for <5% of the measured antioxidant capacity. AAPH-initiated oxidation of LDL was inhibited by the oat fractions in a dose-dependent manner, although complete suppression was not achieved with the highest doses tested. In contrast, Cu(2+)-initiated oxidation of LDL stimulated peroxide formation with low oat concentrations but completely inhibited oxidation with higher doses. Thus, oats possess antioxidant capacity most of which is likely derived from polar phenolic compounds in the aleurone.

Purification of fatty acid methyl esters by high-performance liquid chromatography.

The determination by gas chromatography (GC) of fatty acid methyl esters (FAMEs) prepared from complex biological samples is subject to interference from cholesterol. During sample injection on the GC system of FAMEs prepared from tissues that contain cholesterol, we observed a major contaminant that co-eluted with docosahexaenoic acid (DHA, 22:6n-3). To address this problem, FAMEs were purified on an amino-phase high-performance liquid chromatography (HPLC) column using a hexane-isopropanol gradient. The HPLC retention times for both the FAME fraction and cholesterol were stable and reproducible when the amino column was used for sample purification. The purified extracts were analyzed by GC without artifacts or impurity peaks after 50 analytical runs. The method described here will be useful for measurement of 22:6n-3 and other fatty acids important for studies of nutrition or pathology.

Effect of dietary supplementation with black currant seed oil on the immune response of healthy elderly subjects.

BACKGROUND: We have shown that the age-associated increase in prostaglandin E(2) production contributes to the decline in T cell-mediated function with age. Black currant seed oil (BCSO), rich in both gamma-linolenic (18:3n-6) and alpha-linolenic (18:3n-3) acids, has been shown to modulate membrane lipid composition and eicosanoid production. OBJECTIVE: Our objectives were to 1) test whether dietary supplementation with BCSO can improve the immune response of healthy elderly subjects, and 2) determine whether the altered immune response is mediated by a change in the factors closely associated with T cell activation. DESIGN: A randomized, double-blind, placebo-controlled (soybean oil) study was conducted to examine the effect of 2 mo of BCSO supplementation on the immune response of 40 healthy subjects aged >/=65 y. In vivo immune function was determined by delayed-type hypersensitivity skin response. Peripheral blood mononuclear cells (PBMCs) were tested for in vitro immune response. RESULTS: In subjects supplemented with BCSO, the total diameter of induration at 24 h and individual responses to tetanus toxoid and Trichophyton mentagrophytes were significantly higher than their baseline values. The change in response to tetanus toxoid was significantly different from that of the placebo group. The BCSO group showed a significant increase in proliferative response of PBMCs to the T cell mitogen phytohemagglutinin that was not significantly different from that observed in the placebo group. BCSO had no effect on concanavalin A-induced mitogenic response, interleukin 2 and -1beta production, and PBMC membrane fluidity. Prostaglandin E(2) production was significantly reduced in the BCSO-supplemented group, and this change was significantly different from that of the placebo group. CONCLUSION: BCSO has a moderate immune-enhancing effect attributable to its ability to reduce prostaglandin E(2) production.

Lutein and zeaxanthin concentrations in plasma after dietary supplementation with egg yolk.

BACKGROUND: The food matrix in which carotenoids are found affects their bioavailability. Lutein and zeaxanthin are abundant in egg yolks and accumulate in the macular region of the retina, where they may affect visual function. OBJECTIVE: We sought to determine whether plasma lutein and zeaxanthin concentrations are elevated after dietary supplementation with egg yolk. DESIGN: Eleven moderately hypercholesterolemic men and women consumed 2 separate baseline diets, which contained 29-33% of energy as total fat, with 20% of energy as either beef tallow or corn oil. These diets were supplemented with cooked chicken egg yolks (1.3 egg yolks/d for an intake of 10.4 MJ). Each subject consumed all 4 diets. Each diet was consumed for 4.5 wk, with a washout period of >/=2 wk between diet phases. At the end of each diet phase, fasting morning plasma samples were collected and stored for carotenoid analysis by HPLC. Commercial chicken egg yolks were analyzed for carotenoids and cholesterol. RESULTS: Egg yolk supplementation of the beef tallow diet increased plasma lutein by 28% (P < 0.05) and zeaxanthin by 142% (P < 0.001); supplementation of the corn oil diet increased plasma lutein by 50% (P < 0.05) and zeaxanthin by 114% (P < 0.001). Changes in plasma lycopene and beta-carotene were variable, with no consistent trend. Egg yolk supplementation increased plasma LDL-cholesterol concentrations by 8-11% (P < 0.05). CONCLUSIONS: Egg yolk is a highly bioavailable source of lutein and zeaxanthin. The benefit of introducing these carotenoids into the diet with egg yolk is counterbalanced by potential LDL-cholesterol elevation from the added dietary cholesterol.

High-performance liquid chromatography analysis of cholesterol linoleate hydroperoxide in oxidized low density lipoproteins: calibration by conjugated diene internal standard.

The cholesterol linoleate hydroperoxides formed in LDL after oxidant stress are measured by HPLC, with UV detection at 234 nm. Calibration is performed with a conjugated diene internal standard. This internal standard is synthesized by the transesterification of the methyl ester of conjugated diene linoleic acid with a long-chain alcohol, such as arachidyl alcohol (C20). Different long-chain alcohols can be used during the transesterification, to achieve internal standards with variable HPLC retention times. The method allows measurement of cholesterol linoleate hydroperoxide in LDL very early during attack with Cu2+ or other initiator, so that the kinetics of antioxidant loss and hydroperoxide formation can be concurrently monitored.

Modelling cortical cataractogenesis XX. In vitro effect of alpha-lipoic acid on glutathione concentrations in lens in model diabetic cataractogenesis.

In previous studies stereospecific protection against lens opacity was consistent with specific reduction of R-alpha-lipoic acid(R-alpha-LA) in mitochondria of the vulnerable cells at the lens equator where the first globular degeneration is seen in glucose cataract. In this study two further possible explanations of this effect were investigated: (1) increased glucose uptake by the lens, leading to increased glycolysis and release of lactate into the incubation medium and/or (2) maintenance of glutathione levels by the R-alpha-LA. The data did not support 1, but was consistent with 2, after 24 hr incubation. The concentrations of glutathione in normal lenses or lenses incubated with R- or racemic alpha-LA were not significantly different, but the concentration of glutathione in lenses incubated with S-alpha-LA was significantly lower than the R-alpha-LA-incubated lenses.

Protection by vitamin C of loss of vitamin E in cultured rat hepatocytes.

Results from in vivo studies of the capacity of vitamin C to spare and/or recycle vitamin E are equivocal. While some in vitro and membrane models reveal an interaction between vitamins C and E, the characterization of this relationship in biologically relevant systems is lacking. Thus, we investigated this relationship using hepatocytes isolated from 3- to 6-month-old male Sprague-Dawley rats. Cells were incubated for 18-20 h in medium supplemented with 0.1-4 mM ascorbic acid. The loss of alpha-tocopherol and the formation of its primary oxidized metabolite, alpha-tocopherolquinone, was determined by HPLC. Levels of alpha-tocopherol in hepatocytes incubated without ascorbic acid declined from 390 to 35 pmol/mg protein; hepatocyte ascorbic acid levels declined from 9 to 0.5 nmol/mg protein. alpha-Tocopherolquinone was undetectable in freshly isolated hepatocytes but following incubation in ascorbate-free medium reached 10 pmol/mg protein. The formation of alpha-tocopherolquinone was not detected in hepatocytes incubated with ascorbic acid. Dehydroascorbic acid (DHA) levels represented 10-20% of the total ascorbate content in freshly isolated hepatocytes but after 3 h incubation the proportion of DHA increased to 50%; after 18-20 h incubation DHA was undetectable. Hepatocytes incubated with 1.0, 2.0, 2.5, or 4.0 mM ascorbic acid lost significantly less alpha-tocopherol (62, 69, 67, and 56%, respectively) than unsupplemented controls (90%). Twelve percent of the alpha-tocopherol lost from hepatocytes during incubation was detected in the medium of cells incubated with ascorbic acid, but vitamin E was undetectable in the medium of cells incubated without ascorbic acid. These results demonstrate an interaction between vitamins C and E in cell culture and are not inconsistent with a potential recycling of oxidized alpha-tocopherol by ascorbic acid.

Plasma kinetics of an oral dose of [2H4]retinyl acetate in human subjects with estimated low or high total body stores of vitamin A.

The deuterated retinol dilution technique is an indirect method for quantitatively estimating total body stores of vitamin A by using the postequilibration plasma isotopic ratio [2H4]retinol:retinol and the prediction model described by Furr et al (Am J Clin Nutr 1989;49:713-6). Limited data are available on the time required for an oral dose of labeled vitamin A to mix with vitamin A body stores in human subjects. This article describes the plasma retinol kinetics of an oral dose of [2H4] retinyl acetate in 4 healthy adults (2 men and 2 women) and 1 healthy female child in the United States and in 4 Bangladeshi women. After an oral dose of [2H4]retinyl acetate was administered, plasma samples were collected at 6, 12, and 24 h postdose during the first day and at 15 time points during the subsequent 90-d period for measurement of plasma [2H4]retinol:retinol. The mean respective plasma isotopic ratios on day 20 for US and Bangladeshi subjects (0.02 +/- 0.02 and 0.17 +/- 0.12, P = 0.03) and estimated total body vitamin A reserves (1.03 +/- 0.45 and 0.10 +/- 0.11 mmol, P = 0.003) were significantly different. The fraction of dose in plasma was plotted against time, and biexponential equations were fit to the kinetic data by using the time points from 24 h through day 90. The mean equilibration time (time required for the fraction of dose in plasma to reach a plateau) for all subjects was 16.6 +/- 3.8 d (11-23 d). There was no difference in estimated equilibration time between the group of US and Bangladeshi adult subjects (17.5 +/- 4.4 and 16.3 +/- 3.9 d, respectively, P = 0.69). Thus, the size of hepatic vitamin A reserves does not appear to affect equilibration time within the range of values observed.