Compounds with primary amino groups react with o-phthalaldehyde in solution to yield highly fluorescent products. This reaction is now in wide use for detecting amino acids and amines in liquid-chromatography effluents. We report here a method for using o-phthalaldehyde to detect primary amines in the effluent of a gas-liquid-chromatographic column. The effluent gas, nominally 20 mL/min, is delivered to a scrubber consisting of a small-bore, simulated capillary chromatographic column that is simultaneously supplied with 1 mL/min of the reagent solution. The liquid effluent of the scrubber, separated from the gas, is drawn through the flow cell of a fluorometer. Short-chain amines and ammonia were quantitatively scrubbed. The response of the fluorometer was directly proportional to the number of nanomoles injected into the column. Less than a nanomole of amine was detectable. Comparison of results with those from a hydrogen flame-ionization detector showed minimal additional peak broadening or compromise of resolution. These results demonstrate the feasibility of using highly specific, as well as sensitive, liquid-chromatographic detection methods for gas-liquid chromatography.
Abstract. A modified reversed passive hemagglutination test for the detection of hepatitis B surface antigen HBsAg is described. Sera and reagent cells coated with antibody to HBsAg (anti‐HBs) are loaded separately into the rotor of a miniature centrifugal fast analyzer. The rotor is centrifuged briefly to transfer the components into its cuvettes. After mixing, the suspensions are allowed to stand at room temperature for 30 min, following which the rotor is again centrifuged and the absorbance of each cuvette is monitored. Cells suspended in serum containing HBsAg leave the light path more rapidly than cells suspended in sera free of antigen. The magnitude of change in absorbance varies directly with the concentration of the antigen. In 45 sera tested by the conventional V‐plate technique, findings were as follows: 21 positive, 19 false positive and 5 negative. The automated procedure unequivocally differentiated the 21 positives; results for the false positive and negative specimens were identical and clearly distinguishable from the positive results. The automated procedure enhances specificity, offers equivalent sensitivity, and results that are quantitative and objective.
Frequent use of vancomycin in the neonatal intensive care unit results from a high prevalence of infections caused by methicillin-resistant staphylococci.1 Potential benefits to measuring vancomycin in most children and adults with normal renal function appear minimal when compared with cost of testing.2, 3 In contrast, in neonates prediction of vancomycin concentration according to the Bayesian method yields a relatively large standard error.4 The present study was designed to assess whether it is possible to reliably predict peak vancomycin serum concentrations from trough values in neonates. Methods.Patient population. The patients were hospitalized at Jacobi Medical Center, Weiler Hospital of Montefiore Medical Center and North Central Bronx Hospital, from December, 1996, to March, 1998. The study population included all neonates who received vancomycin for >3 days and had both peak and trough vancomycin serum concentrations documented in the chart. Design. This study was a retrospective review of the charts and laboratory computer database of patients receiving vancomycin. We collected information on gestational age, birth weight, postnatal age, postconceptional age (i.e. gestational age plus postnatal age), current weight, dose and interval of administration, peak and trough vancomycin concentrations and plasma creatinine concentration. The study was divided into 2 parts. The first included 100 paired vancomycin concentrations in 87 infants and was designed to compare peak and trough values and to yield a formula to predict peak vancomycin concentration from trough value and either volume of distribution or renal function. The second part was designed to evaluate the prediction formula developed from the first group. To limit inaccuracies we held several information sessions for nurses, nurse practitioners and physician assistants in the neonatal intensive care units to stress the importance of accurately documenting dosage and timing and of informing us of possible deviations from the protocol. Method of vancomycin administration and measurement. Vancomycin was administered according to standard guidelines.5 Recommendations include five different schedules of vancomycin (10 to 20 mg/kg every 6 to 24 h) based on gestational age and postconceptional age.5 Vancomycin was infused intravenously during 1 h by minipump equipped with a narrow tubing inserted close to the skin entry. The patients were studied at the fourth dose or after 72 h of treatment, whichever came first, to ensure steady state conditions. Trough samples were obtained just before the fourth dose and peak samples at 1 h after the end of the infusion, i.e. presumably after the end of the distribution phase,6 thereby allowing single compartment pharmacokinetics. The vancomycin concentrations were measured by Syva EMIT or Abbott TDX immunoassay procedure. This immunoassay yields results that are comparable to those obtained by high performance liquid chromatography, the gold standard method, even in patients with poor renal function.7 The immunoassay has a 95% recovery and limited variability. Serum creatinine was measured by the kinetic Jaffe method; this method has good accuracy but limited precision for values <1 mg/dl.8 Pharmacokinetic analysis. We calculated the Vd, half-life (t1/2) and clearance (Cl) for each infant.9 Only data for which timing of medication administration and of blood sampling could be estimated within 15 min were kept for analysis. Statistical analysis. We assessed agreement between peak and trough values (categorized as low, therapeutic and high) using Cohen kappa. This test analyzes whether the distribution of paired low, therapeutic and high values is different from that expected by chance. Means were compared by Student's t test or by analysis of variance, as appropriate. We considered P < 0.05 as statistically significant. Values are presented as mean ± SD. We used hierarchical multiple regression analysis to yield a formula for predicting peaks from trough values, entering successively trough value and dose (milligrams/kg), postconceptional age and plasma creatinine concentration. Results.Population. We obtained 100 paired vancomycin concentrations in 72 patients (including 62 premature infants, i.e. those with a gestational age <37 weeks) in the first group and 33 paired values in 17 patients (including 15 premature infants) in the second group (Table 1).TABLE 1: Characteristics of the two groups Part 1. Mean peak and trough vancomycin concentrations were 29.0 ± 9.8 and 6.8 ± 5.2 μg/ml, respectively. The interval of administration of vancomycin at the time of sampling could be determined from the chart in 84 of 100 samples. The agreement between peak and trough values (categorized as low, therapeutic and high) was poor (Cohen kappa, 0.18; 95% confidence interval, 0.03 to 0.33). Three patients had a high peak vancomycin concentration but a low trough value, and another three demonstrated a low peak concentration but a high trough value. We developed a formula to predict peak values using multivariate analysis. The factors that entered significantly in this formula were the trough concentration (micrograms/ml), dose (milligrams/kg) and postconceptional age (weeks). Predicted peak = 19.02 + 1.07 dose + 1.05 trough − 0.47 postconceptional age. The correlation coefficient was only 0.71 (n = 79, P < 0.05), and the residual standard deviation was 6.3. This meant that with a predicted peak value of 30, 95% of the actual peak values ranged between 20 and 50 μg/ml, thus including values outside the therapeutic range. The prediction of peak values did not improve by using trough vancomycin concentrations and plasma creatinine measurements. Part 2. The peak vancomycin concentrations were weakly correlated (r = 0.38, n = 33, P < 0.05) with the values calculated using the formula obtained during the first part of the study. Discussion. Mean values and standard deviations of vancomycin pharmacokinetic variables obtained with a single compartment model were similar to those obtained previously with a model-independent method or a two-compartment model,10 in agreement with the assumption that the vancomycin distribution is almost completed within 1 h after the end of the infusion. Nevertheless some of the variability in this retrospective study may have resulted from lack of obtaining exact times of medication administration and of blood sampling. Lack of improvement of the prediction of peak concentrations from trough values with the use of serum creatinine concentration might be attributed to the poor reliability of creatinine measurements with the use of the kinetic Jaffe method for low values.8 We conclude that peak vancomycin concentrations cannot reliably be predicted from trough values in neonatal clinical practice. Acknowledgments. Preliminary data were presented at the Mead Johnson Nutritionals Greater NY Conference on Perinatal Research, Hauppauge, NY, November 7, 1997, and at the Annual Meeting of the Eastern Society of Pediatric Research, Atlantic City, NJ, March 1, 1998. Preliminary results were published in abstract form.11, 12 Onajovwe O. Fofah, M.D.* Arthur Karmen, M.D. Janet Piscitelli, M.D. Luc P. Brion, M.D. Departments of Pediatrics (OOF, LPB), Laboratory Medicine (AK) and Pathology (JP); Albert Einstein College of Medicine and Montefiore Medical Center; Bronx, NY
Abstract A method has been developed for performing rapid, quantitative analyses by liquid chromatography. Microgram quantities of lipids were separated on semimicro silicic acid columns by eluting them with a succession of solvents. Separations similar to those performed by thin-layer chromatography were accomplished in similar time periods. The analyses were quantified using a liquid-chromatography detector based on the difference in volatility between the compounds to be detected and the eluting solvents. The column effluent was deposited on a continuously moving metal chain. The solvent was evaporated at a controlled temperature. The residue was then carried into a heated tube in which it was volatilized or pyrolyzed in an atmosphere of nitrogen. The resulting vapors and pyrolysis products were aspirated into a hydrogen flame-ionization detector. Both high-boiling and nonvolatile lipids were detected quantitatively with fairly uniform sensitivity.
