Exp Biol Med 1984;176:485-9. A Computerized. Classification Technique for Screening for the Presence of Breath Biomarkers in Lung Cancer. H. J. O'NeIll,'5.
BS, Desai NC, Bhatnagar R, Garg SP. Study of the relationship between free amino acids and cataract in human lenses. Exp Eye Res 1984;38:177-9. 5. Chauhan
GW. Free amino acids in senile cataractous lenses: osmotic etiology. Invest Ophthalmol 1968;7:564-83. 7. Ron XII. Analysis of aqueous humor proteins and amino acids in senile cataract. Chung-Hua Yen Ko Tsa Chih (Peking) 1984;20:262-3 [in Chinese]. 8. Wen LY, Juang TY, Wuu JA, Chang GO. Serum ornithine levels in some ocular disorders. Biochem Med Metab Biol 1986;35:83-7. 9. Raviola E, Wiesel TN. An animal model of myopia. N Engl J Med 1985;312:1609-15.
6. Barber possible
10. Kolata
G. What
causes
nearsightedness?
[Review].
Science
14. Potts AM. Selective action of chemical agents on individual retinal layers. In: Graymore CN, ed. Biochemistry of the retina. New York: Academic Press, 1965:155-62. 15. Casper DS, Reif.Lehrer L. Effects of alpha-aminoadipate isomers on the morphology of the isolated chick embryo retina. Invest Ophthalmol Vie Sci 1983;24:1480-8. 16. McLerman H, Hall JG. The action of D-cr-aminoadipate on excitatory amino acid receptors of rat thalamic neurones. Brain Res 1978;149:541-5. 17. Sch#{248}nheyderF, Ehlers N, Hust B. Remarks on the aqueous humor/plasma ratios for amino acids and related compounds in patients with various chronic ocular disorders. Acta Ophthahnol 1975;53:627-34.
18. Ehlers
1985;229:1249-50.
retinal
11. Dickinson chromatography
19. Bunce
JC, Rosenblum H, Hamilton PB. Ion exchange of the free amino acids in the plasma of the newborn infant. Pediatrics 1965;36:2-13. 12. Shichi H. Biochemistry of vision. New York: Academic Press, 1983. 13. Watling KJ. Transmitter candidates in the retina [Review]. Trends Pharmacol Sci 1981;2:244-7.
CLIN. CHEM. 34/8, 1613-1618
A Computerized in Lung Cancer H. J. O’NeIll,’5
program
Classification
S. N. Gordon,1
installed
Technique
R. 0. GIbbons,3
N. H. OI1eIII,2
into a standard
‘Department Street, Chicago, 2Continental
reduces
the
of Chemistry,
personal-computer
amount
of
11T Research
IL 60616.
illinois
spread-
National
Bank,
data
Institute,
lOWest
231 South LaSalle,
35th Chica-
3Departments of Biometry and Psychiatry, University of illinois 912 South Wood Street, Chicago, IL 60680. ‘Department of Medicine, Rush-Presbyterian-St. Lukes Hospital and Medical Center, 1753 West Congress Parkway, Chicago, IL
at Chicago,
to whom
Received March
20. Anonymous. The nutritional origin of cataracts [Editorial]. Nutr Review 1984;42:377-9. 21. Bunce GE, Hess JL, Davis D. Cataract formation following limited amino acid intake during gestation and lactation. Proc Soc Exp Biol Med 1984;176:485-9.
correspondence
should be addressed. 22, 1988.
9, 1988; accepted April
for the Presence
of Breath Biomarkers
and J. P. Szldon4
go, IL 60697.
60612. 6Author
GE.
1979;37:337-43.
for Screening
required for statistical treatment. Such a sort routine can also be applied as easily to other complex GC/MS data files for the purpose of data reduction. greatly
humour and plasma amino acids in tapedo. Acta Ophthalmol 198 1;59:576-86. Nutrition and cataract [Review]. Nutr Review
Aqueous
(1988)
A simple computer-based screening technique has been developed for classifying human expired air components into 16 chemical classes, based on empirical formulas. The sort procedure was developed to simplify the screening of the composition of expired air samples by sorting all components into chemical classes and classifying components at the >75% and >90% occurrence levels. Both occurrence-rate components are then evaluated as diagnostic markers in a discnminant function model for their ability to detect lung cancer. Of the 386 components detected in the gas chromatography/mass spectrometry (GC/MS) data files, 45 components were present at the >75% occurrence level and 28 components at the >90% occurrence level. Thus, this preliminary sort routine, performed by using a simple macro sheet,
N.
degenerations.
AdditIonal
Keyphrases:
computers
data handling
smok-
ing
One of the most desirable and generally accepted approaches for the clinical chemist in performing a diagnostic procedure is the use of noninvasive techniques. Such procedures are readily acceptable to the subject and require minimal medical intervention and control. Toward this end, several laboratories, including our own, have been pursuing efforts toward the use of a gas chromatographic/mass spectrometric (GC/MS) technique for exploring the use of human expired air as a biological matrix for various clinical purposes. There are several reviews of such applications (1-3), while direct applications to renal disease (4), liver dysfunction (5, 6), lung cancer (7), acetone metabolism (8), and a possible shunt mechanism in the sterol pathway (9) have also been reported. In many of these applications, one is confronted with the monumental problem of seeking a biochemical marker(s) among three to four hundred components, all of which are present in sub-microgram per liter to microgram per liter concentrations. Because most of the components present in human expired air are irrelevant environmental pollutants, a successful diagnostic technique must be capable of either selectively identi1ring a unique species in this complex matrix or of monitoring its change in concentration without identification. The latter is particularly true when the change in concentration involves more than one component. In these cases, statistical or pattern recognition (7, 10-12) approaches can be more successfully applied. In the former case, specific component(s) can be identified, if such components can be isolated from the complex background and identified. CLINICAL CHEMISTRY,
Vol. 34, No. 8, 1988
1613
In our current investigation of biochemical markers for the detection of lung cancer, it is necessary to look at the entire spectrum of volatile organic compounds exhaled by each subject in order to establish differences between the various patient categories (lung cancers, cancers other than lung, lung pathologies other than cancer) and a normal control population. Our primary statistical approach for evaluating the CC/MS data from these four subject categories is to develop a suitable discriminant-function model (7, 13). In the development of such a model, the large number of variables (peaks) encountered makes the need to reduce the dimensionality
of the
data
a prerequisite
to establishing
relationships between the data sets. Therefore a preliminary screening (sorting) technique was developed, primarily to isolate those components of potential diagnostic value from the vast assortment of environmental pollutants (indoor and outdoor), cigarette smoke, and normal metabolic components that are typically found in human expired air. By applying a preliminary chemical class-sort procedure to the identified CC/MS peaks and establishing their relative occurrence rates in each of the various subject data sets, 80-90% of all components can be eliminated from consideration as potential biomarkers in the discriminant function. This approach also permits the classification of the potential biochemical markers into subgroups corresponding to >75% and >90% occurrence rates for the four subject populations. This information, which is obtained directly from the CC/MS data, can then be factored into the discriminantfunction model, in order to obtain a solution that is not simply due to fortuitous associations that are invariably found when examining large numbers of variables. The data presented herein were obtained from eight lungcancer patients who provided two duplicate sets (morning and afternoon) of expired air samples; i.e., the screening procedure is applied to 32 CC/MS data ifies. In practice, the technique involves downloading the individual CC/MS data ifies [after clean-up by a computerized spectrum-enhancement algorithm (14)] into the spreadsheet on a personal computer. The sorting procedure is then performed first, using a macro incorporated into the spreadsheet, to perform the preliminary sort of the individual components automatically on the basis of their empirical formula, then using the resident sort capability incorporated into the particular operating system in use.
Materials
and Methods
Lung-Cancer
Patients
The eight were selected
lung-cancer patients screened in this study from patients of both sexes entering RushPresbyterian-St. Lukes Hospital and Medical Center for diagnosis and treatment. The patients were diagnosed as having lung cancer by bronchoscopic biopsy, fine needle aspiration, mediastinoscopy, or lymph node biopsy. Expired air samples were taken as soon as possible after diagnosis and prior to the commencement of any treatment modality (radiation, chemotherapy, or surgery). Each subject was given a complete lung-function examination, blood chemistry, chest x-ray, and completed a detailed 10-page medical questionnaire. The questionnaire serves to establish a medical history for each subject as well as to match each lungcancer patient with a control subject by a propensity scoring technique. This technique allows each patient to be matched with a control subject based on age, sex, race, smoking history, and socioeconomic background. 1614
CLINICAL CHEMISTRY,
Vol. 34, No. 8, 1988
Each subject contributed duplicate breath samples in the morning and afternoon of the same day. The resulting four CC/MS data files generated for each of the eight lung cancer patients were used in the evaluation described below. Sample
Collection
Breath
samples were collected in the morning and early by having the subject breathe through a sterilized Rudolph valve connected to a heated manifold, which directs the sample into a 40-L Teflon sample bag. The subject inhales purified air from a reservoir and, with a springloaded pinch clamp on the nose, continues to take deep breaths and inflates the 40-L bag several times. This washout period involves inflating and deflating the bag twice, and the sample obtained on the third inflation is used for analysis. The washout period is used to help purge the lung and anatomical deadspace of environmental pollutants, which would tend to confound the already-complex sample matrix. The sample-preconcentration procedure has been reported elsewhere (7). Materials consist of a 40-L Teflon bag containing the (water-saturated) sample enclosed in a Plexiglas box, Tenax CC glass collectors, a flow-regulating needle valve, an air pump, a flowrneter, and a precision wet-test meter. Two sequential 20-L samples are pumped from the Teflon bag into a glass cartridge (6 mm i.d. x 30 cm long) containing 1.2 g of 60/80 mesh Tenax CC (2,6-diphenyl-pphenylene oxide). The sample is then thermally transferred onto a second glass sorbent cartridge (14 mm i.d. x 10 cm long) containing 1.6 g of 60/80 mesh Tenax CC. The second transfer is necessary to decrease the collected water vapor to a concentration acceptable for CC/MS analysis. afternoon
Sample
Analysis
The volatile organic components collected in the Tenax CC collectors are thermally desorbed at 220 #{176}C with helium and purged into a sample trap cooled with liquid nitrogen. The cold trap is then rapidly heated to 250 #{176}C, and the sample is injected directly into the sample inlet port of a Varian Model 3700 gas chromatograph equipped with a high-resolution capillary column (CP Wax 57CB; Chrompack Inc., Bridgeview, NJ). During CC analysis, the capillary column was held at 0 #{176}C for 5 mm, then temperature programmed at 4 #{176}C/min to 220 #{176}C. The magnet sector mass spectrometer (Finnigan MAT 311A, including the SpectroSystem SS-200 data system) was scanned repetitively from mlz 20 to mlz 300 every 2.2 s, and the resulting output was stored on disk prior to further computerized workup. Data
Analysis
The CC/MS analysis has been reported in detail earlie (7). Briefly, using a minicomputer (PDP-11/34) we subj the raw data generated in the CC/MS run to an efficien spectrum-enhancement algorithm (14), which locates an extracts components from the raw data and produces a set o “clean spectra” that are free of background contributio and contamination. The peak areas from the ci program were used to calculate the concentration of eac component in the sample by comparing the peak area of th component with the peak area of an internal standard. Th internal standard used for these studies was perfluo toluene, which was added to each sample cartridge befor CC/MS analysis. The clean spectra generated above were then composi
by a second program (nM5EK) into a master ifie, after retention-time scaling, and directed to a historical library matching program (xwc), with use of the Environmental Protection Agency/National Institutes of Health library (33 900 spectra). The composited CC/MS data, representing 32 data ifies (four files/set) for each of the eight subjects, were then downloaded from the PDP-11/34 to a Zenith personal computer, where they were placed into a single spreadsheet listing all 32 data files. Each data ifie in the spreadsheet lists the retention index of each compound, as well as the empirical formula, compound name, and normalized total ion count (peak area). Empirical
Formula
Classification
Scheme
The following macro identifies one method for clasaiiying chemical compounds into specific categories based on their empirical formula. This is a very basic macro, written in Lotus 1-2-3 Command Language, that can be modified to suit specific needs. For example, cell references are used where range names could be substituted, and a working storage area is set up where the logic is tested. This could be done within the table. Before running the macro, the data table must be properly formatted. Formatting consists of establishing a column for each element and placing the proper number of atoms of each element in the respective column, according to the empirical formula. In Table 1 these columns are represented from A to I. A separate column J (represented by an R) is used to identi1r a ring number classification for each peak. In this column, 1 is used for a single ring aromatic, and 2 for a condensed ring aromatic. Similarly, single and condensed cycloalkane ring systems are represented by 3 and 4,
Table
2. Macro
ABCO
to Scan Empirical
Used
EFGHIJK
2
C
H 0
4 5 6 7 8
5 6 6 7 8
8
N S
IN
Cl F
Br I
8 6
9
3
4
3
6
1 1
11
4
8
1
12
7
6
1
13
1
2
2
1
0
0
RI
3 0
0
766
0
0
0
0
0
Coound
P DCC
isoprene
32
C6H8 C6146 C7H8 C8H8 C3H40 C3H60
cyclohexadlene benlene toluene
2 30 31
styrene propenal acetone
29
C4H80 C7H60
2-butanone
CH2CL2
dlchiorneethane
30 32 30
C2HCL3
trlchioroethane
32
681 1300 700
3
0
For.ula C5H8
713 814 1028 646 628
2
0
9
500 648
1
10
22
U
3 3 3
8 8
20
R
Formula
N
29 31
benzaldehyde
0
23
24 25
\G
(LET
A22,8CELLPOINTER
‘CONTENTS’
((RIGHT)
‘CONTENTS’ ‘CONTENTS’
)(RIGHT)
26 27
(LET B22,8CELLPOINTER (LET C22.BCELLPOINTER
28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
(LET D22.BCELLPOINTER ‘CONTENTS’ )(RIGHT) (LET E22,8CELLPOINTER ‘CONTENTS’ ((RIGHT) (LET F22,@CELIPOINTER ‘CoNTENTS’ ((RIGHT) (LET G22,ICELI.POINTER ‘CONTENTS’ ((RIGHT) (LET H22,@CELLPOINTER ‘CONTCNTS )(RIGHT) (LET I22.BCELLPOINTER ‘CONTENTS’ ((RIGHT) (LET J22,8CELLPOINTER ‘CONTENTS’ ((RIGHT) (LET K22,BCELLPOINTER ‘CONTENTS’ )(CALC) (IF A22-’-’)(QUIT) (IF A22 O)(DOWN)(LEFT 1O)(SRMCH G) (IF B22o(A22*2+2))(DOWN)(LEFT ID) (BRANCH (IF C22.oO)(DOWN)(LEFT 10)(8RANCH \G) (IF D2290%) components of interest. As can be observed in Table 5, several chemical class categories can be immediately eliminated from consideration as diagnostic markers by inspection, since they do not contain components at either the 75% or 90% occurrence levels. These class categories are identified as 2, 4, 7, 9, 12, 13, and 16 and represent a total of 150 components, or 39% of the total observed. It is also of interest to note in Table 5 that, of the 102 components (26% of the total) that comprised category 16 (unknown), not one occurred at either the >75% or >90% level, indicating that further clarification of their structures would appear unnecessary. Alternatively, by summing the total number of components comprising both the >75% and >90% occurrence levels, one reduces the total number of potentially important diagnostic components to 45 (12%) and 28 (7%), respectively, of the total number of components detected. It is assumed in this initial screening technique that any useful diagnostic marker will occur at least at the 75% occurrence level and possibly the 90% occurrence level, although for this small sample population this may not be an appropriate assumption for selection. Because the total composition of each of the two occur rence levels can be isolated from the remaining componen by the resident sort procedure, it is possible to print out listing of all the components comprising each of the tw categories. Proceeding with this printout, Tables 7 and identify all the components at the >75% and >90% occur rence levels from all the remaining background peaks. I will now be possible to further explore this reduced numbs of components in statistical studies and to test their rep ducibility when larger numbers of data sets become avail able.
Discussion It is clear from the data presented herein and earlie reports from this laboratory (15,16) that human expired ai represents an extremely complex matrix in which to searc for diagnostic markers for various disease states. However the availability of the powerful data-enhancement alg rithms for processing GC/MS data, along with very simpl
Table
C
H
4 4 6 6 10 10 10 13
10 10 14 14 22 22 22 28 8 6
7. LIsting of Compounds In Expired-Air from Lung-Cancer Patients (Occurrence >75%) N
0
S Cl
F Br
I
P
0
433 450 486 470 786 856 558 1567 500 713 414 900 960 1001 1109 1019 1056 1037 986 949 1098 1028 1500 1555 648 628 615 681 1600 1300 1059 1413 1170 1154 765 1202 776 1148 1597 700 1200 766 476 800 483
8 10
10 12
10 8 10 11 3 3 3 4 6 7 8 8 8 9 5 5 5 6
6 2 2 2
12 12 12 12 12 12 14 8 8 10 4 1 6 1 6 1 8 1 6 1 6 1 16 I 8 1 18 1 18 1 8 2 4 2 8 2 14 2 5 2 4 I 1
1
1 2 2 3 3 4 3
0
RI
1
Fornola 1 1 1 1 1 1 1 1 3 5 5
5 5
7 8 10 10 10 10 10 10 10 10 10 10 11 Il 11 11 14 15 15 15 15 15 15
8. LIsting of Compounds from Lung-Cancer
Table
(Occurrence C
H
6 6 10 5 6 7 8 9
14 14 22 8 6 8 10 12 12 12 12 8 8 10
9
9 9 8 10 11
0
N
S Cl
F Br
I
P
U
RI
Fornola 1
10 10 10 10 10 10 10 11 11 15 15 15 15 15
3
1
646
3 4 6 7 8 9 1 5 5 1 6 2 2 2
1 1 1 1 1 1 2 2
628 681 1600 1300 1413 1154 765 776 700 1200 766 476 800
2 2 3 3 4
9
486 470 856 500 713 814 900 1001 1019 1056 1037 1028 1500 1555
Coound
C4HIO C4810 C61114 C6014 C10H22 C10H22 C10H22 C13H28 CSH8 C6H6 C7H8 C8H10 C81410 CR812 C9H12 CR812 C9H12 C9H12 C9812 C9812 C1OHI4 C8148 CIOH8 CI1HIO C3840 C3060 C3H60 C4H80 C6H60 CIH6O C8H160 C8H80 C88180 C9H180 C5H802 C5H402 C5H802 C6H1402 C7HSWS C82CL2 C6H4CI2 C2HCL3 C2HCL3 C2CI4 CC13F
C6814 C6814 C10022 C5NB C&46 C7HB C8*I10 C9$112 09652 091112 091112 CUH8 ClINIB CI1N1O C31140 C31160 C4H80 C6H60 C7N60 00180 CR14180 CR14802 CR6802 OH2CL.2 C684C12 C2HCL3 C244C13 CZCI4
Samples
Occur
butane 2-.ethylpropane heoane nethylpentune ethyl,,ethylheptane tri.ethylheptane alkane trlmethyldecane Isoprene benzene toluene ethylbenzene dlmethylbeezene methylethylbenzene trimethylbenzene trl.ethylbenzene ulkylbeozene alkylbenzene propylbenzene ethyl.ethylbenzene dl.ethylethylbeniene styrene naphthalene methylnaphthalene propenal acetone propanal 2-butanone phenol benzaldehyde octanal acetophenone methylethylpentanol nonanal ethyipropanoate Furancurboaaldehyde .ethyllsobutenoute 2-n-butooyethanol b4nzothlazole dlchloromethane dlChlorObenZene tnlchloroethane tnichlorofluoromethane tetrachioroethylene trlchlorofluoromethane
in Expired-Air Patients >90%) Cound hexane methylpentane trl.ethylheptane Isoprene benzene toluene ethylbenzene methylethylbeozene trimethylbenzene alkylbenzene alkylbeezene styrene naphthulene methyinapflthalene propenal acetone 2-butenone phenol benzuldehyde acetophenone nonanal ethyipropanoate methyllsobutenoate dlchioromethane dlchlorobenzene trlchloroethane trlchlorofluoromethane tetrachioroethylene
27 26 32 32 26 29 27 26 32 30 31 32
28 29 27 30 30 29 28 26 26 29 32 29 29 31 25 30 31 32 28 31 28 31 30 28 32 26 25 30 32 32 32 32 28
markers. The remaining 19 components ostensibly reprosent environmentally related pollutants that appear to bear little relevance to a biochemical marker. The lack of any nitrogenor sulfur-related components in Table 8 is also noteworthy. In surveying both Tables 7 and 8, only one component in Table 7 was identified as bearing such hetero atoms, namely benzothiazole. This component, widely seen in human expired air samples in both normal and various disease states, was present in 25 of 32 data files, a 78% occurrence level. It might be expected that during tumor proliferation, poor vascularization and subsequent tissue necrosis would generate an array ofboth sulfur- and nitrogenous-type products arising from catabolic processes. However, owing to the fact that the patients identified herein were associated with early stages ofdiagnosis, such degenerative processes might not yet have been in place. We emphasize that, although the application proposed herein was directed toward screening human expired air samples, the overall application of this technique would be equally useflul in screening any type of complex GCIMS spectral data. Its application to such areas as human breath and ambient air makes it particularly appealing, because one could systematically screen which classes of volatile organic compounds appear to be selectively removed from the air, and thus permit an approximation of the body burden based on simple database (chemical class) comparisons.
We acknowledge the valuable assistance ofM. Kathy McCloskey in sample collection, Ma. Louise Brousek for performing the GC/MS semple analysis, and Mrs. Melissa Love for data analysis. The study was supported by Grant No. CA37056 from the National Cancer Institute, NIH, Bethesda, MD.
Samples
Occur 32 32 29 32 30 31 32 29 30 30 29 29
32 29 29 31 30 31 32 31 31 30 32 30 32 32 32 32
personal-computer software macro sort techniques, makes the initial screening of these complex profiles relatively straightforward. Nevertheless, despite the relative simplicity of this data-reduction technique, the clinical association of such peaks with lung cancer is only advanced by a rigorous statistical evaluation, where much larger patient populations are available. Such screening is currently tinderway. The approach presented herein is only intended to identify a useful screening technique that can support the statistical evaluations. Of particular interest, arising from the above data-reduction programs, is the fact that only nine components (Classes 10 and 11) of the 28 components comprising the >90% occurrence level listing (Table 8) and representing only 2% of the total 386 components detected are potential diagnostic
References 1. Stewart RD. The use of breath analysis in clinical toxicology. Chap. 5 in: Hayes Jr WJ, ed. Essays in toxicology. Vol. 5. New York: Academic Press, 1974:121-47. 2. Pierce SK, Gearhart HL, Payne-Bose D. The analysis of human breath and urine for organic components with chromatographic and mass spectrometric techniques: a review. Talanta 1977;24:473-81. 3. Manolis A. The diagnostic potential ofbreath analysis [Review]. Cliii Chem 198329:5-15. 4. Simenhof ML, Burke JF, Saukkonen JJ, Ordinario AT, Doty It Biochemical profile of uremic breath. N Engl J Med 1977297:132-5. 5. Kavin H, Krotoszynski BK, Gordon SM, O’Neill RI, Szidon JP. A non-invasive
diagnostic
method
in liver disease:
analysis
of
expired air [Abstract]. Hepatology 1983;3:871. 6. Kavin H, Krotoszynaki BK, Gordon SM, O’Neill HJ. Further evaluation of human expired air: gas chromatography-mass spectrometry profiles in liver disease [Abstract]. Gastroenterology 1984;86:1131. 7. Gordon SM, Szidon JP, Krotoszynski BK, Gibbons RD. O’Neill NJ. Volatile organic compounds in exhaled air from patients with lung cancer. Clin Chem 1985;31:1278-82. 8. Reichard Jr GA, Haff AC, Skutches CL, Paul F, Holroyde CP, Owen OE. Plasma acetone metabolism in the fasting human. J Cliii Invest 1979;63:619-26. 9. O’Neill HJ, Gordon SM, KrOtOsZynBki BK, Kavin H, Szidon JP. Identification of isoprenoid-type components in human expired air: a possible shunt pathway in sterol metabolism. Biomed Chromatogr 1987;2:66-70. 10. Ton JT, Gonzalez RC. Pattern-recognition principles. Reading, MA: Addison-Wesley, 1974. 11. Jurs PC, Isenhour TL. Chemical applications of pattern recognition. New York: Wiley-Interscience, 1975. 12. Smith AB, Beicher AM, Epple G, Jurs PC, Lavine B. Computerized pattern recognition: a new technique for the analysis of CLINICAL CHEMISTRY,
Vol. 34, No. 8, 1988
1617
chemical communication. Science 1985;228:175-7. 13. Discon WJ, Brown MB, eds. BMDP biomedical computer programs P-series 1979. Berkeley, CA: University of California Press, 1979. 14. Dromey RG, Stefik MJ, Rindfieisch TC, Dufileld AM. Extraction of mass spectra free of background and neighboring component contributions from gas chromatography/mass spectrometry
CLIN. CHEM. 34/8, 1618-1621
data. Anal Chem 1976;48:1368-75. 15. Krotoszynski BK, Bruneau BM, O’Neill HJ. Measurement of chemical inhalation exposure in urban pollution in the presence of endogenous effluents. J Anal Toxicol 1979;3:225-34. 16. Krotoszynski BK, O’Neill HJ. Involuntary bioaccumulation of environmental pollutants in non-smoking heterogeneous human population. J Environ Sci Health A17 1982;6:855-83.
(1988)
Age- and Sex-Specific Pediatric Reference Intervals: Study Design and Methods Measurement of Serum Proteins with the Behnng LN Nephelometer GIlilan Lockltch,1
2
Anne
C. Halstead,1
2
Gayle
Qulgiey,1
and Carol
We analyzed blood from 450 healthy children and adolescents, ages one to 19 y, as well as term and preterm infants, to define age- and sex-specific reference intervals for numerous blood constituents. Reference intervals were derived by using nonparametric methods to determine the 0.025 and 0.975 fractiles. Ten serum proteins were measured with the Behnng LN Nephelometer. Girls over 10 years of age had higher concentrations of ceruloplasmin and alpha1 -antitrypsin than other children had. There was no sex-related difference in reference intervals for the other proteins tested. Reference intervals are presented for immunoglobulins G, A, and M, complement fractions C3c and C4, ceruloplasmin, transferrin, alpha1 -antitrypsin, retinol-binding protein, and prealbumin (transthyretin).
Additional
Keyphrases:
ceruloplasmin transferrin protein . franstf7yrefifl
immunoglobulins
antitrypsin
complement
retinol-binding
healthy children from three local schools and outpatient dental clinic to determine normal reference intervals that would be valid for the diverse ethnic groups represented in our patient population. We also studied healthy term and preterm infants from neonatal nurseries. We studied
our
Materials Study
and Methods
Population
The study population comprised three different groups of healthy normal children studied over the same period. Approximately half of the children were white and the other half were of Oriental, East Indian, American Indian, or other ethnic origins. The first group consisted of 350 healthy, fasting, ambulatory school children and adolescents, ages 5 to 19 y, from an elementary school and two secondary schools in the Vancouver area. Children were excluded from the study for chronic or acute illness, current medication use, or not fasting. Information was collected regarding recent exercise, cigarette smoking, and alcohol ingestion, but children were not excluded from the study for these reasons. tDepartment of Pathology, B.C. Children’s Hospital, 4480 Oak St., Vancouver, BC, V6H 3V4, Canada. 2University of British Columbia, Vancouver, BC, Canada. Received January 19, 1988; accepted April 21, 1988.
1618
CLINICAL
CHEMISTRY,
Vol. 34, No. 8, 1988
Illustrated by
MacCailum1 Blood was sampled between 0800 and 1030 h, by veni puncture, into heparmn-, citrate-, or EDTA-containing Vacu tamer Tubes (Becton-Dickinson, Mississauga, Ontario, L5,J 2M8, Canada) and into a siliconized Vacutainer Tube con taming no anticoagulant and suitable for trace elemeni determination. The procedure was explained to each chilc before venipuncture to minimize fear. Similar samples were also obtained from the seconc group: 100 children and adolescents, ages 1 to 19 y, wh attended an outpatient dental clinic at the hospital. Sample in this group were collected between 0800 and 1400 h. Data on neonates were obtained from infants in tIn nurseries of the Grace Hospital and from preterm infants ir the special-care nursery of the Children’s Hospital. Samplei were obtained from almost 300 babies, but only a few test were performed on an individual sample, owing to volum constraints. The term infants were babies weighing >2.5 k who required no special nursing care and were in most case breastfed. They were either well babies having blood collect ed for the Provincial Neonatal Screening Program or babie with mild physiological jaundice not requiring photother apy. Samples left over when ordered tests for preterix infants (
